MBL U.S. Department of Commerce Volume 100 Number 1 January 2002 Fishery Bulletin U.S. Department of Commerce Donald L Evans Secretary National Oceanic and Atmospheric Administration Scott B. Gudes Acting Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Acting Assistant Administrator for Fisheries Scientific Editor Dr. John V. Merriner Editorial Assistant Sarah Shoffler Center for Coastal Fisheries and Habitat Researcln, 101 Pivers Island Road Beaufort, NC 28516 NOS ^ATES O^ 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 98 1 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 20402-937.3. 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. 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. Andrew E. Dizon 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 National Marine Fisheries Service 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 Fishenes 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 100 Number 1 January 2002 Fishery Bulletin Contents JAN 3 1 mi The conclusions and opinions expressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Manne 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 matenal mentioned in this puh- 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 propnetary 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- Articles 1-10 Blick, D. James, and Peter T. Hagen The use of agreement measures and latent class models to assess the reliability of classifying thermally marked otoliths 11-25 Carmona-Suarez, Carlos A., and Jesus E. Conde Local distribution and abundance of swimming crabs (Calllnectes spp. and Arenaeus cribrarius) on a tropical arid beach 26-34 Crabtree, Roy E., Peter B. Hood, and Derke Snodgrass Age, growth, and reproduction of permit (Trachinotus falcatus) in Florida waters 35-41 Denson, Michael R., Wallace E. Jenkins, Arnold G. Woodward, and Theodore I. J. Smith Tag-reporting levels for red drum (Saaenops ocellatus) caught by anglers in South Carolina and Georgia estuaries 42-50 Faunce, Craig H., Heather M. Patterson, and Jerome J. Lorenz Age, growth, and mortality of the Mayan cichlid (Cichlasoma urophthalmus) from the southeastern Everglades 51 -62 Hastings, Kelly K., and William J. Sydeman Population status, seasonal vanation in abundance, and long-term population trends of Steller sea lions (Eumetopias jubatus) at the South Farallon Islands, California 63-73 McBride, Richard S., Michael P. Fahay, and Kenneth W. Able Larval and settlement periods of the northern searobin (Prionotus carollnus) and the striped searobin (P. evolans) 74-80 Pennington, Michael, Liza-Mare Burmeister, and Vidar Hjellvik Assessing the precision of frequency distributions estimated from trawl-survey samples Fishery Bulletin 100(1) 81-89 Potts, Jennifer C, and Charles S. Manooch III Estimated ages of red porgy (Pagrus pagrus) from fishery-dependent and fishery-independent data and a comparison of growth parameters 90-105 Romanov, Evgeny V. Bycatch in the tuna purse-seine fisheries of the western Indian Ocean 106-116 Sainte-Marie, Bernard, and Denis Chabot Ontogenetic shifts in natural diet during benthic stages of American lobster (Homarus americanus), off the Magdalen Islands 117-127 Zug, George R., George H. Balazs, Jerry A. Wetherall, Denise M. Parker, and Shawn K. K. Murakawa Age and growth of Hawaiian seaturtles (Chelonia mydas): an analysis based on skeletochronology Notes 128-133 DiNardo, Gerard T., Edward E. DeMartini, and Wayne R. Haight Estimates of lobster-handling mortality associated with the Northwestern Hawaiian Islands lobster-trap fishery 134-142 Graves, John E., Brian E. Luckhurst, and Eric D. Prince An evaluation of pop-up satellite tags for estimating postrelease survival of blue marlin (Makaira nigricans) from a recreational fishery 143-148 Hazin, Fabio H. V., Paulo G. Oliveira, and Matt K. Broadhurst Reproduction of blacknose shark (Carcharliinus acronotus) in coastal waters off northeastern Brazil 149-152 Porch, Clay E., Charles A. Wilson, and David L. Nieland A new growth model for red drum (Sciaenops ocellatus) that accommodates seasonal and ontogenic changes in growth rates 153 Subscription form Abstract-Otolith thermal marking is an I'llii'it'nt method for mass mark- ing hatehcry-rcared salmon and can be used to estimate the proportion of hatchery fish captured in a mixed-stock fishery. Accuracy of the thermal pattern classification depends on the promi- nence of the pattern, the methods used to prepare and view the patterns, and the training and experience of the per- sonnel who determine the presence or absence of a particular pattern. Esti- mating accuracy rates is problematic when no secondary marking is avail- able and no error-free standards exist. Agreement measures, such as kappa I K). provide a relative measure of the reliability of the determinations when independent readings by two readers are available, but the magnitude of k can be influenced by the proportion of marked fish. If a third reader is used or if two or more groups of paired read- ings are examined, latent class models can provide estimates of the error rates of each reader. Applications of K and latent class models are illustrated by a program providing contribution esti- mates of hatchery-reared chum and sockeye salmon in Southeast Alaska. The use of agreement measures and latent class models to assess the reliability of classifying thermally marked otoliths* D. James Blick Peter T. Hagen Alaska Department of Fish and Game Division of Commercial Fisheries 10107 Bentwood Place Juneau, Alaska 99802-5526 E mail address ((or P T Hagen, contact author) peter hagenmifishgame state ak us Manuscript accepted 16 April 2001. Fish. Bull. 100:1-10(2002). The ability to induce patterns in salmon otoliths by manipulating water temper- atures has proved to be an efficient means for marking large numbers of salmon (Volket al., 1990). Wlien salmon embryos or alevins are exposed to a rapid drop in temperature, otolith growth is temporarily disrupted, and this results in a discontinuity in the otolith "s microstructure. When viewed under transmitted light microscopy, this discontinuity appears as a dark ring. By controlling the number of tem- perature drops and the timing between drops, a coded pattern of dark rings can be recorded on the otolith and this pattern can be recovered from otoliths of older fish by removing the overlay- ing material and exposing the otolith core. For hatcheries that release a large number of fish, this type of marking method has shown to be particularly cost effective for marking 100% of the releases (Munk et al. 1993). Several fisheries management pro- grams in Alaska use thermal marking to estimate hatchery contributions to commercial fisheries (Hagen et al., 1995). Typically, several hundred salm- on otoliths are systematically collected during each two- or three-day com- mercial opening during the fishing sea- son. The otoliths and sampling data are shipped to a processing laboratory where a subsample of otoliths (generally 50 to 100) are processed immediately to meet in-season management needs; a portion of the remaining otoliths are processed later to provide an overall es- timate of hatchery contribution to the fisheries. The process by which a reader de- termines the presence or absence of a thermal mark in an otolith can be char- acterized as one of pattern recognition and image matching. Prior to examin- ing otoliths of unknown origin, the read- ers gain familiarity with the patterns likely to be encountered by carefully examining fry otoliths that were ob- tained after thermal marking but prior to their release into the wild. Because there can be wide variation in the ap- pearance of the thermal marks within a mark group (due in part to differenc- es in developmental stages at marking), a single mark group may be represent- ed by a variety of patterns. As a result, secondary characteristics and measure- ments of the patterns are sometimes necessary to identify an otolith to a mark group. The examination is also used to confirm that all the hatchery fish have been successfully marked. The process of making a determina- tion on otoliths from returning adult salmon can become problematic be- cause wild salmon may also contain otolith patterns that can mimic the fea- tures imposed through thermal mark- ing. Referred to as "noisy patterns," their presence can increase the rate of false positives. Conversely, if the hatch- ery employs poor temperature control or unintended disruptions occur around the period of marking, it may be diffi- cult to identify the otolith as that of a * Contribution PP-184 of the Alaska De- partment of Fish and Game, Commercial Fisheries Division, Juneau, Alaska 99802- 5526. Fishery Bulletin 100(1) hatchery fish, and this would increase the rate of false negatives. Differences between readers in skill and train- ing level, and how they process otoliths, can add to the un- certainty in estimating the accuracy of the readings and the rates of false positives and negatives. Otolith marking generally takes place without any sec- ondary marking, such as fin-clipping or coded-wire-tag- ging; therefore the accuracy of a reading cannot directly be determined through conventional methods that make use of a "gold standard" (known origin sample) or other error-free classification methods. To ensure that the in- formation provided to the Alaskan fisheries managers is accurate, each otolith is independently examined by two readers, and a third reading is used to resolve differenc- es between the first two readings. The resolved readings are used to estimate the contribution of hatchery fish, and the presumption of accuracy is based on the premise that, through multiple readings, all marked fish are ei- ther correctly identified or that errors, if present, are in- consequential. Developing the analytical tools to deter- mine the veracity of that assumption is the objective of this investigation, and by establishing such tools, quality control standards for recovering thermal marks can be developed. In developing the tools to measure the quality of otolith readings, three questions are addressed: 1 How to assess the reliability of otolith readings when no standards are available. 2 How to estimate the proportion of hatchery marks when there is disagreement between two or more readers. 3 How the precision of the estimate of the proportion is influenced by classification error We discuss two approaches: 1 ) indices of agreement typi- cally used in reliability studies, and 2) latent class models where classification errors are estimated for each reader even though the true error rate is considered unknown. The data requirements and their attendant assumptions are presented for each approach. The methods are illus- trated by examining among-reader comparisons of chum salmon (Oncorhynchus keta) and sockeye (Oncorhynchus nerka) salmon otoliths collected from programs that moni- tor inseason contributions of hatchery fish in several com- mercial fisheries in Southeast Alaska (Hagen et al., 1995). The results are used to provide recommendations for mon- itoring the quality of otolith readings for thermal marking programs. Table 1 Notation used to show the cross-classification of a sample of fi otoliths by two readers to either hatchery (H) or wild stock (W) assignment. Row and column sums are indicated by the subscript "." Reader 1 H Reader 2 "h- H W "hh "hh W "WH "ww «w "•H "w /; 2 is infallible (or is considered a "gold standard"), unbiased estimates of the accuracy and error rates of reader 1 and the proportion of hatchery stocks (p) are given by '^HlH ~ "hh/"h- '^WjH ~ "\VH ^ " H - 1 ■'''hIH '^wjw ~ "vv\\7" w- '^Hiw ~ "hw I " w = 1~ ■'^w|w P = "h/". (where, for example, ;r\v|n refers to the probability that reader 1 classifies an otolith as W when its true state is H). These estimates reflect the fact that reader 2 is infallible; the accuracy rates CThih' '^wiw' ^"d the error rates CTwifi- Tc■^^,^^) are conditional on the numbers of hatchery or wild stock otoliths as determined by reader 2. No standard available If a standard is not available, an unbiased estimate of p can be obtained if the accuracy rates for reader 1 are known. The estimate is p* = ("n/«+^wr I'/f'^HI H|H ' W|W 1), where n■^^ is the number of otoliths classified as hatchery otoliths. If the accuracy rates are estimated, thenp* will no longer be unbiased, but will be much less biased than the estimator n■^^ln and will in general have a much smaller mean-squared error (Rogan and Gladen, 1978). For a Bayesian approach to this problem, see Viana et al. ( 1993 ) and Joseph et al. ( 1995). Methods Standard available A sample of /i otoliths, which are examined by two readers, can be cross-classified as hatchery (H) or wild stock (W) as in Table 1. Suppose we wish to estimate the accuracy rate (probability of making a correct classification) or con- versely, the error rate ( probability of making a wrong clas- sification). If we know nothing about reader 1, but reader Agreement measures When accuracy rates are unavail- able, statistics that measure "agreement" between readers are often calculated (e.g. Fleiss, 1981). One such index is simply the proportion of observed agreement (P„), defined as :(/(. ■ )ln. Another index, called kappa (k), corrects P„ for the degree of agi'eement that is expected by chance alone. It is defined as Blick and Hagen Use of agreement measures and latent class models to assess the reliability of classifying ttiermally marked otoliths 3 K = iP„-P,.)/(l-P^.), where P,, = expected agreement = ('!h"h + "w"w'^"^- ^^^ divisor, 1 - P., constrains k to be less than or equal to one, and if all agreement is due to chance {P^=PJ, then k: equals zero. Note that with k; independence between readers is assumed in order to calculate expected agreement. An example of how agreement indices can be used to monitor readings is shown in Figure 1, which displays k and its standard error for 2874 chum otoliths readings di- vided into 27 groups based on different reader pairs and capture locations. Included are P,'s for four of the groups. The results indicate that v levels were similar between the different groups, suggesting overall consistency in read- ings, although some of the groups had lower values, which in practice would invite further investigation. The Pg's in Figure 1 have a different rank order than the ic values. This apparent discrepancy highlights a potential problem in interpretation when using agreement indices to draw conclusions. To help illustrate this point, consider the following examples (Table 2). Table 2A is generated as the expected counts, given ;rj,|j^ = 0.9 and %|w = 1-0 for both readers, and p = 0.1. In this case, P, = 0.98 and k: = 0.89. On the other hand. Table 2B is generated under the same assumptions except that rt^n = 0.5. In this case P„ drops only slightly to 0.95, whereas v drops to 0.47. Be- cause the hatchery stock is rare, the inability of the read- ers to detect the mark is not well reflected by P„ whereas k reflects it better by correcting for the high level of chance agreement. Now let K, HIH 0.9 and /Twiw = 0.9 for both readers, and 0.64. On P= 0.5 (Table 2C). In this case, P, = 0.82 and k the other hand. Table 2D is generated under the same as- sumptions except that P= 0.05. In this case, P, remains unchanged at 0.82, but \' drops to 0.25. In none of the above examples is the index "wrong." Rather, as is the case with most indices, interpretation is affected by the values of the underlying parameters. In the latter example (Table 2, C-D), even though P, is the same for C and D, the scale it is being compared with has changed, thus changing the value of k. This increases the difficulty of comparing k across populations with differ- ent underlying proportions. Note also that Table 2D could have been derived from %|h = 0.5 and ttwiw = 0.944 for both readers, andp = 0.19. Thus, without additional infor- mation, it is impossible to draw reliable conclusions about reader accuracies or the proportion of hatchery marks. Although agreement measures can be ambiguously in- terpreted, in practice they can still sei've a useful moni- toring role during routine comparisons when the circum- stances of the readings are fairly well characterized. The interpretive difficulties with indices such k and P, become apparent when trying to translate agi"eement measures into statements about the accuracy of different readers and about the influence of reading error on the contribu- tion estimates. Latent class models An alternative approach is to try to estimate tTj^, j^ and tt^viw f""" each reader, along with p. Although at first thought this may seem impossible, it can 1 0 ra 0.6 04 02 T -^ 8si 920 tl J_ J_ 10 20 Group number 30 Figure 1 The values of k{±1 SE) from 27 gi'oups of paired read- ings of chum salmon otoliths (total=2874). The groups are based on pairs of different readers examining oto- liths collected at different times and locations. The pro- portion of agreement (P,) is shown next to group 4, 7, 9, and 12 for comparison with the value of k. be shown that either by setting a few constraints or by col- lecting additional information, estimation is indeed pos- sible. This problem falls into the category of latent class modeling (e.g. Everitt, 1984; Bartholomew, 1987; McCutch- eon, 1987; Clogg, 1995). Latent class models (LCMs) belong to a family of latent variable models that hypothesize the existence of unobservable "latent" variables, about which information can be obtained only though measurements on observable "manifest" variables. LCMs specifically restrict the latent and manifest variables to be categorical. In the present situation, the latent variable is the true class (H or W) to which the otolith belongs, whereas the mani- fest variables are the readers' classifications. Such models have been used for assessing reliability of diagnostic tests in the medical field over the last 20 years (see Walter and Irwig, 1988; Formann, 1996, for reviews). Returning to the problem with two readers, neither of which is a standard, there are five essential parameters to estimate: s-i)|H,^H|H'^w|w.'fw|'w ' andp, with only 3 df (four pieces of data, /i^H' "hw "wH' "ww- minus one because the sample size, n, is fixed). Thus, the model is overparameter- ized, and either constraints on the parameters or more da- ta are needed. Possible constraints include 1) considering that two of the parameters are known (e.g. /r^vjw = Tw|w = ^• i.e. both readers always call a wild stock correctly, there are no "false positives"), or 2) considering that two sets of parameters are equal (e.g. t1|'|'h , 7r|f|H , ;r\v|'\v ='fwi'w' i-^- the accuracy rates are the same for both readers). Although there may be times when such constraints are realistic, in general they will not be; therefore more infor- Fishery Bulletin 100(1) mation will be necessary. One way to generate more in- formation is to have a third independent reader (Walter, 1984). With three readers, there are seven essential pa- rameters; 7i'ii;^"-'''\r^'^\i,''-"" and p. There is also 2^ - 1 = 7 df, so that all the parameters are estimable. Estimation is most commonly done by the method of maximum likeli- hood. If readings are assumed to be independent among read- ers and among otoliths, the likelihood function is i = H,\V ) = H,\V*:H,VV This likelihood function must be maximized numerically and methods for this computation will be discussed later If more than three readers are used, there are extra de- gi-ees of freedom that can be used to assess goodness-of-fit. For example, with four readers there will be nine param- eters with 15 df leaving 6 df for goodness-of-fit. Pearson chi- square or likelihood ratio G'-^ tests would both be applicable. Another way to generate additional information was proposed by Hui and Walter ( 1980). Suppose there are two or more strata with different hatchery proportions in each strata. For example, catch could be stratified temporally or spatially. If it is assumed that ;r||||| and /Ty^iw remain constant over strata, then a solution for just two readers may be obtained. For example, if there are two readers and two strata, then there are six parameters; 'rH|H"'>'i'w|w' > Pj, and p.,, with 2(2'^ - 1) = 6 df Increasing the number of strata increases the degrees of freedom; e.g. three strata for two readers gives 3(2^ - 1) = 9 df for 7 parameters. The likelihood function for two readers and S strata is fin ni^^'^^iH'^^^+'i-^. (1) 12) 1" 'Iw'TjIWl g=l (=H,W_/ = H.W Table 2 Examples from cross-classification data generated as expected counts from a sample of 1000 otoliths based on different accuracy rates for identifying hatchery fish < tt,., | ,^ I and wild fish (/Twiw' under different mark proportions tp). The examples used illustrate differences between obsei"ved agreement IP, i and chance-corrected agi-eement U') under different underlying conditions. A H Reader 2 90 ^Hl H ~ 0.9 P„ = 0.98 ' H W Reader 1 81 9 W 9 901 910 % \V ~ 1.0 K- 0.89 Total 90 910 1000 /' = 0.1 B H Reader 2 50 ^11 n = 0.5 P, = 0.95 H W Reader 1 2.5 25 W 25 925 950 % w = 1.0 K- = 0.47 Total 50 950 1000 P = 0.1 C Reader 1 H Reader 2 500 '^ll|H = 0.9 P„ = 0.82 H W 410 90 W 90 410 500 %■ w - 0.9 V = 0.64 Total 500 500 1000 P = 0.5 D Reader 1 H Reader 2 140 'fH H = 0.9 P, = 0.82 H W 50 90 W 90 770 860 Tu |\V = -0.9 K = 0.25 Total 140 860 1000 P = 0.05 A third way to supply additional information is to take a Bayesian approach (see "Discussion" section). By speci- fying prior distributions of the model parameters, unique estimates can be obtained (Joseph et al., 1995). A critical assumption in the above models is that read- ings are independent. Specifically, the reading of each oto- lith by a given reader is independent of any other reading by the same reader, and each reading by various readers on a given otolith is independent given the true state of the otolith. In principle, the latter assumption may be dif- ficult to meet especially if all readers examine the same otolith. The fact that the otolith is not prepared indepen- dently by each reader could induce a dependence among the readers. Also, variability in the readability of the mark due to the marking process can induce a dependence. Such dependence can bias the estimators of n and p (Vacek, 1985). Note that this latter assumption of independence is also required for v. One remedy for the problem of dependence due to prepa- ration is to require independent preparations. This however, requires additional otoliths and with only two otoliths per fi.sh, this would limit the number of readers to two. But in practice, this may not be a large concern. Typically, the second reader has the option to provide additional process- ing effort to the first otolith or, if needed, to process the second otolith. In almost all cases additional preparation is not done and readers feel they are able to extract suf- ficient information about the presence or absence of a mark from each other's preparations. In addition, reader accura- cy rates obtained by LCM do not appear to vary systemati- cally with the reading order, which also suggests that prep- aration-induced dependency is not a significant factor Dependency associated with variability in the appear- ance of the mark may be harder to address. A general so- lution is to model the dependence with additional param- eters (e.g. Vacek, 1985; Qu et al., 1996; Yang and Becker, 1997; Qu and Hagdu; 1998; Albert et al., 2001). Modeling dependence requires either more readers or more strata. These modeling approaches are complicated and are cur- rently evolving (see Albert et al., 2001). Alternatively, ad- Blick and Hagen: Use of agreement measures and latent class models to assess the reliability of classifying tfiermally marked otolitfis ditional latent classes may be added (Christenson ct al., 1992; Forniann, 1994), e.g. a third class of otoliths from ambiguous sources. In the previous discussion concerning three or more readers, we implied that readers were different individu- als. This need not be so; what is required are three or more independent readings. If it were possible for the same in- dividual to read the same otolith more than once, indepen- dently, then the number of different readers could be re- duced. If independence could not be met, the dependence could be modeled, as discussed above. Another critical assumption, but one that should be met most of the time, is that the individual accuracy rates are known to be either greater than or less than the error rates (e.g. %|h > ^wm ^^'^ %-|W ^ %|W' which im- plies that ^Tj^iH and JT^,-^ are either greater than or less than 0.5) because of an inherent symmetry in the problem that results in the same likelihood function being gener- ated when the error rates are switched with the accuracy rates. Computation Formulas for estimating \'and its standard error are straightforward (Fleiss, 1981). Estimates can also be obtained from several software packages including PROC FREQ in SAS (SAS Institute, 1989). Maximizing either of the likelihood functions for the LCMs requires a numerical procedure. The most straight- forward is to use an optimization routine such as "Solver" in Excel (Microsoft Corporation, 1993) or "nlminb" in S- PLUS (Statistical Sciences, 1995). Alternatively, the EM algorithm (Dempster et al., 1977; Dawid and Skene, 1979; McLachlan and Krishnan, 1997) can be easily used. The simplicity of the EM algorithm follows from the recogni- tion that the LCM is an example of a finite mixture prob- lem, specifically, in this case, a mixture of multivariate Bernoulli distributions with mixing parameter p (Everitt, 1984). Use of the EM algorithm for such mixture prob- lems in fisheries is well documented, e.g. for stock compo- sition estimates (Millar, 1987; Pella et al., 1996) and for age-length keys (Kimura and Chikuni, 1987). A more ef- ficient alternative to the EM algorithm is to use iteratively reweighted least squares (Agresti, 1990). This method is relatively easy to implement in software such as PROC NLIN in SAS (SAS Institute, 1989). Perhaps the most di- rect and efficient way would be to use LCM software. We are not aware of any routines for LCMs in any major statistical package at present, but several independent LCM packages exist (for a review, see Clogg, 1995; and for an Internet listing see http://oui-world.compuserve.com/ homepages/jsuebersax/index.htm). As with many maximum likelihood problems, where nu- merical methods must be used, complications can arise. Constraints may at times be needed to ensure that pa- rameter estimates fall in acceptable intervals (e.g. [0,1] for p and [0.5,1] for the ;r's). Also the likelihood function may have local maxima, which means that several runs with varying starting values may be necessary to identify the global maximum. Finally, estimates of standard er- rors may entail additional computing. PROC NLIN in SAS provides asymptotic (i.e. large-sample) standard errors. Jackknife and bootstrap estimates are relatively easy to program, the jackknife being much less computationally intensive. Finally, the Bayesian programs discussed in Joseph et al. (1995) can be found at http://www.epi.mcgill.ca/Josepli/ software. html. Examples The first example analyzes the results of three readers examining 570 chum otoliths. The samples were taken from a common location, and the readers were familiar with the patterns. Each reading was made without knowl- edge of prior readings. The data, along with pairwise k estimates and the LCM parameter estimates (using PROC NLIN in SAS; see appendix for code) are presented in Table 3. These results indicate that the third reader is signifi- cantly (a=0.05) less able to correctly identify a hatchery mark when it is present and that there are no significant differences among readers in their ability to detect a wild mark when it is present. These conclusions are readily ap- parent from the table of results, and although the pairwise K"'s are consistent with these results, they are more dif- ficult to interpret. With the variance due to sampling es- timated to be (0.7379X1 - 0.7379)/(570 - 1) = 0.0003399, misclassification error contributes only 0.36% to the total variance. The second example consists of two readers with four spatial strata. Samples were obtained from sockeye salm- on caught in four neighboring Alaskan gillnet fisheries in central Southeast Alaska. The data and the LCM esti- mates are shown in Table 4. These estimates indicate that the readers are not statistically different in their ability to detect hatchery marks, whereas the second reader is bet- ter able to distinguish wild marks. With eight parameters and 12 df there are 4 df available for a goodness-of-fit test. Pearson's chi-square yields 4.83, which with 4 df, has a p-value of 0.306, thus indicating an acceptable model fit. Misclassification error contributes from about 8% to 14% to the total variance in the estimates of the proportion of hatchery stock. Design considerations Design of an otolith reading program is complicated by misclassification error. An important consideration is the precision of the estimates, in particular the precision of the estimate ofp. Table 5 shows the asymptotic standard error of p for various combinations ofp, /r^iH' ^^^ ^wiv! f'"' '-^e three-reader model with unknown accuracies, and the one-, two-, and three-reader models with accuracies assumed known. Although this table is derived for a sample of 1000 otoliths, the ratio of any two standard errors within the table would be the same for any sample size (assuming the sample size is large enough to approximate the asymptotic conditions). It is evident that misclassification inflates the standard error over the usual binomial case (right-most column). The table also makes clear the increase in the uncertainty of estimating p when the accuracies also have Fishery Bulletin 100(1) Table 3 Cross-classification data and results for 570 chum otoliths examined bv three readers showing the parameter estimates and stan- dard errors from the latent class model, followed by a comparison of the differences among reader pan's by jsing kapp 3 and the latent class model (LCM) accuracy rates. The data show that the high agreement among read ers as to hatcher V and wild ( lassifica- tion (e.g. HHH= 406 and WWW= = 135) is reflected in the overall high accuracy rates estimated from the LCM However the model also shows that reader 3 has a significantly lower accuracy rate in detecting hatchery marks (;rij5'|H=0.969) than the other readers. Reading Count LCM Parameter Estimate SE HHH 406 'Thih 0.998 0.002 HHW 13 'f'&IH 0.998 0.002 HWH 1 '^'^IH 0.969 0.008 WHH 1 f'^'jW 0.958 0.017 HWW 6 t'w/|w 0.986 0.010 WHW 2 *rl3t 0.957 0.017 WWH 6 P 0.738 0.018 WWW 135 Reader pairs K SE Difference in tTj^ih SE Difference in ^-^v SE land 2 0.954 0.014 0.000 0.004 -0.028 0.020 lands 0.882 0.022 0.029 0.009 0.000 0.024 2 and 3 0.901 0.021 0.029 0.009 0.028 0.020 Table 4 Cross-classification data for 2340 sockeye otoliths e.xamined bv two readers and stratified by four fishing districts showing the estimates of the latent class parameters and their standard errors. Between- reader comparison is based on whether the difference in accuracy estimates are significantly different th an zero. The result s indicate that the readers were not statistical! V different in detecting hatchery marks ' "^H 1 H ' ^"^'- were statistically different in detecting wild marks (;rw|w'LCM = latent class model. Fishing districts 108-30 108-50 106-41 106-30 HH 152 127 85 20 HW 11 9 21 5 WH 2 6 5 1 WW 271 382 832 411 n 436 524 943 437 LCM parameter Estimate SE Reader difference SE '^hih'" rr <2> "HjH 0.980 0.964 0.013 0.021 0.017 0.025 IT 11' "w 1 W TT 12' ''W|W 0.984 0.997 0.005 0.003 -0.013 0.006 P108-30 0.366 0.024 Pi 08-50 0.257 0.020 P1O6--H 0.096 0.010 P1O6-3O 0.047 0.011 to be estimated in the three-reader case. For example, if = 0.8 for all three readers, one would have to '^HlH ''wlw have almost twice (0.035/.019=1.84) the sample size to esti- mate ap of about 0.5. Once accuracy estimates for the read- ers are obtained, dropping one or even two readers may be appropriate, although the assumption must be made that the accuracy rates will be constant for the remainder of the program. Maintaining two readers will allow for that Blick and Hagen: Use of agreement measures and latent class models to assess the reliability of classifying thermally marked otoliths Table 5 AsyniptolK' Uand ard errors Ibr the cs timalcd ])! opor'tion of marked fish p, for various combinat ons of accuracy rates in identify- | iiifj; halclu'rv fish. ;r|,|„,and wild fisli, %|W'"«1 mark proportion p, for a sample of 1000 otoliths Val ues are reported foi the cases u liero accur icy r ites, K. are the same and assumed known or one, two. or three readers and for the case w lere ;r's are estimated lor three readers. Table illustrates how misclassification will increase standard errors in the estimate of hatchery proportion. '''n 1 11 0.8 0.9 1.0 '^W 1 w 0.8 0.9 1.0 0.8 0.9 1.0 0.8 0.9 1.0 ,'i readers P 0.1 0.032 0.016 0.011 0.023 0.013 0.010 0.018 0.011 0.009 1 rfs estimated) 0.3 0.034 0.021 0.017 0.024 0.017 0.015 0.020 0.015 0.014 0.5 0.035 0.023 0.019 0.023 0.018 0.016 0.019 0.016 0.016 0.7 0.034 0.024 0.020 0.021 0.017 0.015 0.017 0.015 0.014 0.9 0.032 0.023 0.018 0.016 0.013 0.011 0.011 0.010 0.009 3 readers 0.1 0.013 0.011 0.010 0.011 0.010 0.009 0.010 0.010 0.009 iffs known 1 0.3 0.018 0.016 0.015 0.017 0.015 0.015 0.015 0.015 0.014 0.5 0.019 0.018 0.016 0.018 0.017 0.016 0.016 0.016 0.016 0.7 0.018 0.017 0.015 0.016 0.015 0.015 0.015 0.015 0.014 0.9 0.013 0.011 0.010 0.011 0.010 0.010 0.010 0.009 0.009 2 readers 0.1 0.015 0.013 0.010 0.013 0.011 0.010 0.011 0.010 0.009 ( ;r's known ) 0.3 0.020 0.018 0.015 0.018 0.016 0.015 0.015 0.015 0.014 0.5 0.022 0.019 0.016 0.019 0.018 0.016 0.016 0.016 0.016 0.7 0.020 0.018 0.015 0.018 0.016 0.015 0.015 0.015 0.014 09 0.015 0.013 0.011 0.013 0.011 0.010 0.010 0.010 0.009 1 reader 0.1 0.023 0.017 0.011 0.020 0.015 0.010 0.018 0.014 0.009 (.It's known) 0.3 0.026 0.021 0.017 0.022 0.019 0.016 0.020 0.017 0.014 0.5 0.026 0.022 0019 0.022 0.020 0.017 0.019 0.017 0.016 0.7 0.026 0.022 0.020 0.021 0.019 0.017 0.017 0.016 0.014 0.9 0.023 0.020 0.018 0.017 0.015 0.014 0.011 0.010 0.009 assumption to be checked because there will now be extra degrees of freedom to assess goodness-of-fit (there are 3 df. but only one parameter. p, needs to be estimated). Esti- mates of p can still be obtained with one reader, but there can be no check of the assumptions. Also, there can be a significant increase in uncertainty in the estimate in using only one reader. Discussion There are numerous classification problems in fisheries that require the judgment of trained individuals. In many of those situations no "gold standard" is available to test those judgments, and it becomes necessary to apply other methods to determine the veracity of the classifications. Reading thermally marked otoliths is a particularly good example of this problem because thousands of classifica- tion decisions are needed each year to provide estimates of hatchery contributions. The common approach for assessing the quality of the readings, in the absence of having samples of known origin, has been to collect independent and multiple readings on the samples, and to presume that agreement between read- ings can serve as a proxy for reading accuracy. Agreement indices such as k" are very easy to compute, and they have utility in that they can serve as flags to indicate reading problems. However, as was shown here, they also suffer dif- ficulties in interpretation. Also, the indices in themselves do not provide inferences about the relative skill of differ- ent readers in pulling out a particular set of patterns. Latent class models provide an approach with readily interpretable quantities for a modest computational cost. Classification accuracies or errors are direct, meaningful parameters unlike an index of agreement. In addition, es- timates of p are available. These models can be readily ex- tended to the case of more than two outcomes, e.g. multiple hatchery marks. These models could also be useful in oth- er applications, such as in aging fish or in the identifica- tion of any character for which there is no "gold standard" (e.g. field identification of species or sex). A somewhat sim- ilar analysis has been proposed for aging (Richards et al., 1992), although the link to LCMs was not discussed. LCMs can handle fairly complicated situations, including ordered classes (Croon. 1990), continuous manifest vari- ables, and parameter constraints (see Clogg, 1995, and Krzanowski and Marriott, 1995. for reviews). We have not discussed the Bayesian approach to these problems in great detail, but we believe it has much to offer in that it can incorporate prior information, either Fishery Bulletin 100(1) in the form of expert opinion (e.g. Demissie et al., 1998) or in the form of results of earher analyses (e.g. Viana et al., 1993). Rather than assuming that estimated accu- racies are "known," one can incorporate the uncertainty in the estimates into the prior distributions. In addition, the Bayesian approach does not rely on asymptotic results that may behave poorly with small samples. We have also not assessed the possible bias due to the lack of indepen- dence in the readings. When suitable software becomes available, this assumption should be checked. In our examples above, misclassification error contribut- ed relatively little to the overall uncertainty. In these ap- plications, where estimates of hatchery contribution were used to make management decisions, the accuracy of read- ings were within an acceptable range. However, the criteria used to establish quality control standards in any program need to be developed in the context of how the information is to be used along with other sources of uncertainty. In conclusion, we believe that the use of agreement mea- sures in combination with latent class models can con- tribute significant information about both the proportions of interest and the quality control aspects of an otolith- marking program. Furthermore these approaches could have application to similar areas in fisheries which re- quire judgments that are not free of error. Acknowledgments We thank Bob Wilbur for editorial comments and three anonymous reviewers for valuable suggestions. Literature cited Agresti, A. 1990. Categorical data analysis. John Wiley, New York, NY, 576 p. Albert. P. S., L. M. McShane, and J. H. Shih. 2001. Latent class modeling approached for assessing diag- nostic error without a gold standard: with applications to p53 immunohistochemical assays in bladder tumors. Bio- metrics 57:610-619. Bartholomew, D. J. 1987. Latent variable models and factor analysis. Oxford Univ. Press, New York. NY'. 427 p. Christen.sen A. H., T. Gjorup, J. Hilden, C. Fenger. B. Henriksen, M. Vyberg, K. Ostergaard, and B. F. Hansen. 1992. Observer homogeneity in the histologic diagnosis of Helicobacter pylori: latent class analysis, kappa coefficient, and repeat frequencv Scand. J. Gastroenterol. 27:933-939. Clogg, C. C. 1995. Latent class models. Chapter 6 ;;; Handbook of sta- tistical modeling for the social and behavioral sciences (G. Arminger, C. C. Clogg, and M. E. Sobel, eds.), p. 311-359. Plenum Press, New York, NY. Croon, M. 1990. Latent class analysis with ordered classes. Brit. J. Math. Stat. Psych. 43:171-192. Dawid, A. P., and A. M. Skene. 1979. Maximum likelihood estimation of observer error- rates using the EM algorithm. Appl. Statist. 28:20-28. Demissie, K., N. White, L. Joseph, and P. Ernst. 1998. Bayesian estimation of asthma prevalence, and com- parison of exercise and questionnaire diagnostics in the absence of a gold standard. Ann. Epidemiol. 8:201-208. Dempster, A.P., N.M. Laird, and D.B. Rubin. 1977. Maximum likelihood from incomplete data via the EM algorithm (with discussion). J. Royal Stat. Soc. B 39: 1-38. Everitt, B. S. 1984. An introduction to latent variable models. Chapman and Hall. London. 107 p. Fleiss, J. L. 1981. Statistical methods for rates and proportions, 2"'' ed. John Wiley, New York, NY, 352 p Formann, A. K. 1994. Measurement errors in caries diagnosis: some further latent class models. Biometrics 50:865-871. 1996. Latent class analysis in medical research. Stat. Meth. Med. Res. 5:179-211. Hagen, P., K. Munk, B. Van Alen, and B. White. 1995. Thermal mark technology for inseason fisheries man- agement: a case study Alaska Fishery Res. Bull. 2:14.3- 158. Hui.S.L, and S.D.Walter 1980. Estimating the error rates of diagnostic tests. Bio- metrics 36:167-171. Joseph, L., T. Gyorkos, and L. Coupal. 1995. Bayesian estimation of disease prevalence and the parameters of diagnostic tests in the absence of a gold standard. Am. J. Epidemiol. 141:263-72. Kimura, D. K.. and S. Chikuni. 1987. Mixtures of empirical distributions: an iterative appli- cation of the age-length key. Biometrics 43:23-35. Ki-zanowski, W. J., and F. H. C. Marriott. 1995. Multivariate analysis, part 2: classification, cova- riance structures and repeated measurements. Arnold, London, 280 p. McCutcheon, A. L. 1987. Latent class analysis. Sage, Beverly Hills, CA, 96 p. McLachlan, G. J., and T. Ki'ishnan. 1997. The EM algorithm and extensions. John Wiley, New York, NY, 304 p. Microsoft Corporation. 1993. Microsoft Excel user's guide. Microsoft Corporation, Redmond, WA. Millar, R. B. 1987. Maximum likelihood estimation of mixed stock fish- ery composition. Can. J. Fish. Aquat. Sci. 44:583-590. Munk, K. M., W W. Smoker, D. R. Beard, and R. W. Mattson. 1993. A hatchery water-heating system and its application to 100'7f thermal marking of incubating salmon. Progi'es- sive Fish-Culturist 55:284-288. Fella, J., M. Masuda, and S. Nelson. 1996. Search algorithms for computing stock composition of a mixture from traits of individuals by maximum like- lihood. U.S. Dep. Commerce, NOAA Tech. Memo. NMFS- AFSC-61. Qu, Y., M. Tan, and M.H. Kutner 1996. Random effects models in latent class analysis for evaluating accuracy of diagnostic tests. Biometrics 52: 797-810. Qu.Y.andA. Hagdu. 1998. A model for evaluating sensitivity and specificity for correlated diagnostic tests in efficacy studies with an imperfect reference test. J. Am. Stat. Assoc. 93:920-928. Blick and Hagen: Use of agreement measLires and latent class models to assess the reliability of classifying tfiermally marked otolitfis 9 Richards, L. J., J, T. Schnute, A. R. Ki-onlund. and K. J. Beamish. 1992. Statistical models for the analysis of ageing error. Can. J. Fish. Aquat. Sci. 49:1801-1815. Regan, W. J., and B. Gladen. 1978. Estimating prevalence from the results of a screening test. Am. J. EpidemiologN' 107:71-76. SAS Institute. 1989. SAS/STAT user's guide, version 6, 4"' ed. SAS Insti- tute, Gary, NC. Statistical Sciences. 1995. S-PLUS guide to statistical and mathematical analy- sis, version 3.3. StatSci. Seattle, WA. Vacek, P. M. 1985. The effect of conditional dependence on the evalua- tion of diagnostic tests. Biometrics 41:959-968. Viana, M. A. G., V. Ramakrishnan, and P. S. Levy. 1993. Bayesian analysis of prevalence from the results of small screening samples. Commun. Statist. Theory Melh. 22:57,5-.585. Volk, E. C., S. L. Schroder, and K. L. PVesh. 1990. Inducement of unique otolith banding patterns as a practical means to mass-markjuvenile Pacific salmon. Am. Fish. Soc. Symp. 7:203-215. Walter, S. D. 1984. Measuring the reliability of clinical data: the case for using three observers. Rev. Epidem. et Sante Publ. 32:206-211. Walter, S. D., and L. M. Ii-wig. 1988. Estimation of test error rates, disease prevalence and relative risk from misclassified data: a review. J. Clin. Epidemiol. 41:923-937. Yang, I., and M. P. Becker 1997. Latent variable modeling of diagnostic accuracy. Bio- metrics 53:948-958. 10 Fishery Bulletin 100(1) Appendix The following SAS (version 6.12) code was used to estimate parameters in the three-reader model discussed above. This program makes use of iteratively reweighted least squares to maximize the likelihood function. Observed values (e.g. the number of HHH) are equated with the corresponding expected value from the model and a weighted least squares fit is computed by using PROC NLIN. This computation is iterated to convergence of the parameter estimates. Weights are inverses of the predicted values at each iteration. Indi- cator variables for each possible outcome are generated so that a model in typical regi'ession form can be written. Bounds on the parameter estimates may be needed to con- strain the estimates to the appropriate intervals. Note that the asymptotic standard errors provided by SAS will be correct if the option SIGSQ=1 is specified. However, the printed degrees of freedom and the associated confidence intei-vals are not correct for this application. The residual weighted sum of squares listed by SAS is the chi-squared goodness-of-fit-statistic. The option, OUTEST, outputs point estimates and the the estimated covariance matrix for the parameters. SAS code for the multistrata model used in the second example is also available from the authors. /* SAS Code for estimating 3-reader, 1 -stratum model 7 data a; array x{8} x1-x8; input y, ntot-i-y. /■ accumulating sample size V if n =8 ttien call symput('ntot'.ntot); /" put total into macro var 7 do 1=1 to 8: if l=_n _ ttien x{i}=1 ; else x{i)=0, /" set up indicator variables 7 end; cards, 406 13 1 1 6 2 6 135 /' H H H 7 /• H H W 7 /• H W H 7 /•WH H 7 /* H W W 7 /• W H W 7 /• W W H 7 /• W W W 7 proc nlin data=a nohalve sigsq=1 outest=esti /' sigsq=1 for correct se's 7 parms a1= 9 a2= 9 a3= 9 b1 = .9 b2= 9 b3= 9 p=,6; /" starting values 7 /* a IS accuracy for H 7 /■ b is accuracy for W 7 el =a1 •a2-a3-p-i-(1 -b1 )71 ■b2)-(1 ■b3)'(1 -p) e2=ara2*(1-a3)'p-i-(1-b1)*(1-b2)*b3*(1-p) e3=a1 •(! -a2)*a3*p+(1 -b1 )*b271 -b3)*(1 -p) e4=(1-a1)*a2'a3-p-i-br(1-b2)*(1-b3)*(1-p) e5=a1 -(1 -a2)-(1 -aSj-p-fll-bl )-b2'b3'(1 -p) e6=( 1 -al )'a2'(1 -a3)'p-fb1 '(1 -b2)*b3'(1 -p) e7=(1-a1)*(1-a2)*a3*p-i-brb2"(1-b3)*{1-p) e8=(1-a1)*(1-a2)'(1-a3)*p-i-b1'b2'b3*(1-p) model y=(e1■x1^-e2■x2-l-e3■x3-^e4■x4+e5*x5+e6■x6-^e7■x7-^e8■x8)■&ntot: bounds 0 5<=a1<=1, 0.5<=a2<=1, 0-5<=a3<=1, 0,5<=b1<=1, 0-5<=b2<=1 0 5<=b3<=1, 0<=p<=1: weigtit_=1/model y; run; Abstract— Distriliution. abundance, and .s('\rral [icipulation features were stud- ied in Ensenada de La Vela (Vene- zuela) between 1993 and 1998 as a first step in the assessment of local fisheries of swimming crabs. Arenaeua cribrarius was the most abundant spe- cies at the marine foreshore. Callinectes danae prevailed at the estuarine loca- tion. Callinectes hocourti was the most abundant species at the offshore. Abun- dances of A. cribrarius and C. danae fluctuated widely and randomly. Oviger- ous females were almost absent. Adults of several species were smaller than pre- viously reported. This study suggests that fisheries based on these swimming crabs probably will be restricted to an artisanal level because abundances appear too low to support industrial exploitation. Local distribution and abundance of swimming crabs (Callinectes spp. and Arenaeus cribrarius) on a tropical arid beach Carlos A. Carmona-Suarez Jesus E. Conde Centre de Ecologia, institute Venezolano de investigaciones Cientificas AP 21827 Caracas 1020-A, Venezuela E mail address (for C. A, Carmona-Suarez) ccarmona(a)oil 6 - n - Id Mean s e Range Dl ssolved oxygen Station 1 7 Bl 0 35 6-10.2 r- Station 2 8 14 0 46 6-12 _ Slot ion 3 7 02 0 43 4 1-95 p - Station 4 7 74 0 42 5-108 : fc i*,tV^ l**Y ~ 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 F M A M JJASONDJF I I 1993 M J J -1994 - a S 0 N D Sampling months Figure 3 Surface temperature, salinity, and dis.solved oxygen at the foreshore and estuarine stations in Ensenada de La Vela (Falcon, Venezuela). at the marine stations was C. danae (Table 2). The highest number of species, six, was recorded at the estuarine site, where C. danae clearly dominated with a relative abun- dance of 75. 29f, followed by C. bocourti ( 14.1%), C. exaspera- tus (4.5%), C. sapidus (2.5%), C. maracaiboensis (2.0%) and C. laruatus (1.5%). Arenaeus cribrarius was absent from the estuarine site. Overall, the highest diversity (Shannon- Weaver index) was registered at the estuarine station, fol- lowed by station 2, station 1, and finally .station 4, the most exposed tract. Hills diversity number 1 (Nl), which indi- cates abundant species, was also highest at the estuarine station 3, followed by stations 2, 1, and 4 (Table 2). In a comparison of the two main biotopes (all three marine sta- tions vs. the estuarine station ) for the most frequent species (A. cribrarius and C. danae ), their abundance was dependent on salinity ( G=306; df=l, P<0.005 ). However, their abundance was independent of wave exposure, when only the stations in the marine biotope were considered (G=5.624; df=2, 0.05 >P>0.1). Offshore A total of 173 swimming crabs were caught witli crab pots. Abundance was highest at the seaward- most station, followed by the inshore and midshore sta- tions (Table 3). The average number of individuals per pot at each site followed a similar sequence (offshore: 5.9 individuals/pot; inshore; 1.75 ind/pot; midshore: 1.5 ind/pot). Differences in abundance between offshore and inshore and between offshore and midshore stations were 16 Fishery Bulletin 100(1) Table 2 Overall abundance (no of crabs) and diversity indexes for swmimmg crabs at seaside in Ensenada de La Vela (Venezuela). Station 1 Station 2 Station 3 Station 4 Totals C*! A. cribrarius 86 70 0 64 220 (46) C. danae 19 22 149 7 197 (41,2) C. bocourti 0 2 28 1 31 (6.5) C. maracaiboensis 0 0 4 0 4 (0.8) C. sapidus 6 1 5 1 13 (2.7) C. exasperalus 0 0 9 1 10 (2.1) C. larvatus 0 0 3 0 3 (0.6) Number of specimens 111 95 198 74 478 Number of species 3 4 6 5 Simpson (A') 0.6292 0.5928 0.5876 0.7542 Shannon-Weaver (W) 0.6580 0.6930 0.8660 0.5230 Hill's numbers Nl 1.9300 2.0000 2.3780 1.6870 N2 1.5894 1.6868 1.7018 1.3260 Table 3 Overall abundance (no. of crabs) and diversity indexes for | swimming crabs captured with crab pots in Ensenada de La Vela (Venezuela). Inshore Midshore Offshore Totals C. bocourti 30 30 69 129 C. maracaiboe/isia 2 4 13 19 C. danae 3 1 8 12 C. oniatus 6 3 2 11 C. sp. (unidentified! 0 2 0 2 Number of specimens 41 40 92 Number of species 4 5 4 Simpson (A') 0.5537 0.5705 0.5860 Shannon-Weaver (W) 0.8485 0.8823 0.7879 Hills numbers Nl 2.3361 2.4164 2.1987 N2 1.8062 1,7528 1.7065 significant, whereas the (difference between inshore and midshore was not. Overall, four species were caught, but C. bocourti prevailed at all the stations with at least 73.2'^f of the total quantity ( Table 3 ). Frequency of crabs by spe- cies varied significantly with the distance of the station to the shore (G=17.024. 0.05 >P>0.01. df=8). C. bocourti maintained a constant presence through the three sta- tions, ranging from 73.2 to 75. 09^ of total crabs at each station, the abundance of C. maracaiboensis increased sea- ward, and the abundance of C. cjrnatus decreased. Calli- nectes danae did not show any trend. In this set of samples, taken at a distance from the shoreline, the highest diversity (Shannon-Weaver index) was registered at the midshore station, closely followed by the inshore station and the offshore station. Hill's diver- sity number 1 (Nl), which indicates abundant species, was also highest at the midshore station, followed by inshore and offshore stations (Table 3). Temporal variability Because data for C. bocourti, C. sapidus, C. e.xasperatus, C. maracaiboensis and C. larvatus were too scarce to allow useful analysis, temporal variability of the abundance at the surf and at the estuarine pond was examined only for A. crihrxii-ius. C. danae, and for total crabs. Temporal variability of the abundances of these species are shown in Figure 4. Abundances fluctuated widely and randomly throughout our study. The density of A. cribrarius peaked in April, July, and October 1993, as well as in February and October 1994 (Fig. 4). In the estuarine site, C. danae abundance peaked in May and October 1993, as well as in February, April, August, and November 1994 (Fig. 4). In the marine sites, C. danae abundance was considerably lower and maxima occurred in June, September, and Octo- ber 1993, and in May and October 1994 (Fig. 4). No sig- nificant correlations were found between abundances of these two species and rainfall, water temperature, salin- ity, and dissolved oxygen (Table 4). However, when total crabs were regressed against rainfall at the estuarine site and oxygen at the foreshore, correlations were significant (Table 4). The negative correlation of this latter factor reached almost significant levels for both species at the marine ecotope. Diel variations Surf zone A total of 196 crabs were caught with hand seines at the foreshore during September 1997-February 1998 samplings: 82 crabs at night and 114 during the day (Table 5). Six species appeared in the diurnal samples (A. cribrarius, C. danae, C. bocourti, C. larvatus, C. mara- caiboensis and C. sapidus), one of which (C. sapidus) did Carmona Suarez and Conde: Distribution and abundance of Callinectes spp and Arenaeus cribrarius 17 Arenaeus cribrarius |/>=I56| T — I — I — I — I — I — I — I — I — I — I — I — V — I — I I I I r JFMAMJJ ASONOJFMAMJJASOND I 1993 II 19 94 1 Sampling months Figure 4 Abundance of Arenaeus cnbranus, Callineclcx danae, and total captured crabs in Ensenada de La Vela (Falcon. Venezuela). not appear at night. Arenaeus cribrarius was the domi- nant species followed by C. danae, during both diurnal and nocturnal samplings, whereas C. maracaiboensis, C. bocour-ti. C. larvatus, and C. sapidus were present in very low numbers. Guild composition did not differ significantly between day and night (G=1.630; 0.90>P>0.50; df=5), nor did the average number of individuals per trawl (0.86 vs. 0.62; <=1.702; 0.10>P>0.05: df=262 ). In both dominant spe- cies, A. crib>-ariu!i and C. danae. the average size of crabs caught during daylight hours (Table 5) did not differ sig- nificantly from those collected at night, nor did the size frequency distributions (G=3.820; df=4; 0.50>P>0.10). Sex ratios of these two species did not show significant diel differences either (G=0.030: 0.90>P>0.50; df=l; G=2.750; 0.50>P>0.10: df=l. respectively). Offshore A total of 64 crab pots were deployed. 32 during each period. Four species were caught during both day and night (C. hocourti, C. maracaiboensis, C. ornatus, and C. danae I. A total of 89 crabs were caught during the day and 84 at night (Table 6). Callinectes bocourti comprised 73.8'7( and 75. 3*/? of the abundance during the day and night, respectively, followed by C. maracaiboensis (15.5% and 6.7%). No differences in guild composition or sex ratios were found between day and night samples at each of the sites (Table 6). Crab species did not show differ- ences in carapace length between day and night captures (C. bocourti, ^=0.704, P>0.05, df=155; C. maracaiboensis, /=1.355, P>0.05, df=13; C. ornatus, ^=0.881, P>0.05, df=9; C. danae, ^=1,811. P>0.05, df=10). Kolmogorov-Smirnov tests run for normality of carapace length distribution for species at the offshore station compared between day and night samples, were statistically nonsignificant. Sex ratios and ovigerous females At the foreshore, all the species had male-biased overall sex ratios (Table 7), although only yi. cribrarius, C. danae 18 Fishery Bulletin 100(1) QJ t— * O o d 1 d A Q, A in o crv C Cfi C C/-J C c c CO d 01 he >, O c: lO lO lO in in O] c^ (N (M C^J +-» <» o lO CO Oi o m IN in in 00 ^ , lO to I— 1 CD CO 1 ■^ ^ o CO .—1 CO d d d d d 1 1 1 "^i o o c u II C ra CO o CO c« Cfi Cfi cfj C :C C c C c C O) 1 tn 3 3 W CJ -M ■c > CI, tie s CO CO CO CO CO CO CO CO CO CO X o Tf ■^ CO 00 ,— < T3 CO t^ 00 t^ CM 0; L. CO c^ c^ !N CO > o o (M o O d d d d d m Xfi -5 0^ -o o J2 C CO < tj cfi m cfi to cfi qJ yn C C c C C tx 3 1 >-. en Ie4 nipe a o "rt rf 'f ^ -* -* i2 *"- C/D " CO CO CO CO CO *J 'c ;r^ ^ 00 E> .-H ^ CO CO o o CO eg t/j L. CD ^ OJ to , 7 O o o o o CO d d d d d c: 1 1 c CO ^ CD -a c C CO CC o d CO Cfl c/: C A tt. CO c c/j OJ A o C 1 to o CO -13 [« M d C 3 _D 'ct3 CO p:^ lO lO in in in J2 CO "- oi O) (M tN CJ t- CJ C 02 c CO c CO c CO '5 'C 3 C 3 u 3 o o '11 o a CO CO o "cO e2 CO 2 Ci> 1 c/i 0.5; df=14). All Kolmogorov-Smirnov tests run for normality of carapace length distribution for species that were compared at the foreshore and estuarine sta- tions, were statistically nonsignificant. Discussion Of the nine species of Callinectes that have been reported for the tropical Western Atlantic (Williams, 1984), seven appeared at the foreshore of Ensenada de La Vela during our study. The species with the widest distributions were C. danae and C. sapiduf;. which were the only ones to appear at all the stations by the sea margin. Callinectes maracai- boensis and C. larvatus had the most restricted distribu- tion, occurring only in the estuarine site, and C. exasperatus was present only in the estuary and at one of the marine stations. At the marine foreshore stations, A. crihrarius was the dominant species, with a share of 78% of the total catch in this ecotope, whereas C. danae (19%) was the second most abundant species. Meanwhile, in the estuarine site, where A. cribrarius was absent, C. danae was the prevail- ing species, followed by C. bocourti. Overall, the highest diversity was registered at the estuarine station, whereas at the foreshore the highest diversity was recorded in the Carmona Suarez and Conde Distribution and abundance of Calllnectes spp and Arenaeus cribmrius 19 Table S Body size carapace le igth in mm ) and species abundance during diel observations at the foreshore of Ensenada de La Vela | (1997-98). Percentages are ?iv en in parentheses in "Abundance ' column. Species Period Abundance Mean body size SE A. cribrarius day 91 (79.8) 19.14 0.94 night 59 (72.0) 19.38 1.24 C. danae day night 17 (14.9) 16 (19.5) 21.92 23.09 1.89 2.45 C. hocourti day night 1 (0.9) 2 (2.4) 21.3 33.83 11.4 C. maracai boensis day night 2 (1.8) 4 (4.9) 47.45 37.28 9.10 6.05 C. larL'otus day night 2 (1.8) 1 (1.2) 30.55 15.65 2.25 C. sapid us day night 1 (0.9) 0 38.4 — Total day night 114 82 Table 6 Distribution of species abundance (no of crabs found) at the offshore ^ tations during diel samplings and comparisons of sex ratios | (all sites poo cdl. C. bncuurti C. maracaiboensis C. danae C. ornatus C sp.' Totals G(df=4) Significance Inshore night 12 2 1 1 0 16 5.238 0.50>P>0.10 day 18 0 2 5 0 25 Midshore night 18 3 0 2 2 25 4.226 0.50>P>0.10 day 12 1 1 1 0 15 Offshore night 32 8 1 2 0 43 8.506 0.10>P>0.05 day 37 5 7 0 0 49 Total night 62 13 2 5 2 84 7.940 0.10>P>0.05 day 67 6 10 6 0 89 Overall total 129 19 12 11 2 173 Sex ratios G Significance C. bocourti 0.01 0.975>P>0.9 C. maracaibo ensis 0.642 0.5>P>0.1 C. danae 0.07 0.9>P>0.5 C. ornatus 2.864 0.1>P>0.05 ' Unidentified species. most protected marine stations (2 and 1) followed by sta- tion 4, which is located at the most exposed tract. The values of Hill's diversity number 1 (Nl) demonstrated a similar pattern and indicated that the number of abundant species was close to two at stations 2 and 3. slightly above this value at the estuarine station, and below at the most exposed station. Offshore guild composition was substan- tially different from that at the sea margin, as shown by pot samplings. Although several species were common to the three biotopes, each habitat had a distinctive dominant species: Aiviiaeus cribrarius at the siu'f zone (stations 1, 2, and 4), C. danae (station 3) in the estuarine pond, and C. bocourti offshore. Because different sampling gears were used at the sea border and offshore because of practical 20 Fishery Bulletin 100(1) reasons, comparisons should be regarded as qualitative. However, artisanal fishermen do hai-vest C. bocoiirti when using beach seines in the areas next to our crab pot stations (senior author, personal obs. ). Inshore-offshore zonations of species at Ensenada de La Vela diverged from the gradients compiled by Norse and Fox-Norse ( 1979) for other areas in the Caribbean. In many localities, C. bocourti, C. sapidiis. and C. maracai- boeusis (the so-called bocourti group) are known to inhabit the waters by the seaside, whereas C. marginatus and C. ornatus are found at the seawardmost zone, and C. daiiae occupies the intermediate area. However, our patterns of Table 7 Sex ratios for portunids captured in Ensenada de La Vela ( 1993- -94). Male:female Ratio G df Significance Marine stations A. crihrarius 155:61 2.5:1 21.607 1 P<0.005 C. clanae 34:11 3.1:1 6.272 1 0.01>P>0.025 C. bocourti 0:1 0:1 — — — C. sapuliis 7:0 7:0 5.232 1 0.01>P>0.025 C. exasjicratiis 1:0 1:0 — — — Estuarine station C. daiiae 79 63 1.3:1 0.900 1 0.5>P>0.1 C. bocourti 13 15 0.9:1 0.070 1 0.9>P>0.5 C. sapidus 2 3 0.7:1 0.088 1 0.9>P> 0.5 C. exaspei-atus 7 2 3.5:1 1.405 1 0.5>P>0.1 C. larvatus 2 1 2:1 — — — C. maracaiboenais 2 2 1:1 — — — Table 8 Carapace length (mm) for the most abundant species at the foreshore and estuarine stat and comparison of carapace sizes, ns = not significant. ions in Ensenada de La Vela (Venezuela), Calliiiectff: daiiac Marine stations Callinectcs danae Estua rine station n Mean Range BE n Mean Range SE Juvenile females Adult females Juvenile males Adult males 8 3 14 20 24.6 39.6 19.5 .39.1 14.92-32.7 36..58-42.0 8.5-32,4 7.42-56.7 2.099 1.619 2 2.869 Juvenile females Adult females Juvenile males Adult males 37 26 43 36 22.6 39.8 15.8 23.7 11.28-35.6 31.45-47.4 7.62-27.4 10.4-48.4 1.069 0.832 0.691 1.897 Arcnacus cribranus Marine stations only Callinectcs bocourti Estuarine station only n Mean Range BE n Mean Range SE Juvenile females Adult females Juvenile males Adult males 61 22 - . Nn 11.3-36.94 0.831 Juvenile females Adult females Juvenile males Adult males 7 10 4 6 23.1 41 19.6 41.8 11.8-34.4 34-45.1 16.6-23 24.4-56.5 3.251 1.126 1.314 5.035 89 66 17.6 21.8 10.25-28.64 9.48-56.55 0.459 1.213 / df Significance C. danae (all crabs) C. danac (adult females) C. danac (adult males) C. danae (foreshore stations) C. danac (estuarine station) foreshore/estuarine foreshore/estuarine foreshore/estuarine females/males females/males 3.799 0.065 4.653 0.766 6.297 185 27 54 43 140 P<0.05 ns P<0.001 ns P<0.001 Caimona Suarez and Conde Distribution and abundance of Callinecles spp and Arenaeus aibraiius 21 abundance for Callincctea species are similar to another Caribbean locality (Buchanan and Sloner. 1988): Laguna Joyuda (Puerto Rico). All the Callinectcs spp. recorded in this coastal estuarine lagoon were also present in the es- tuarine station of Ensenada de La Vela. Callinectcs danae was the dominant species in both sites, whereas C e.v- asperatus and C. larvatus were present in low numbers. Callinecles maracaiboensis was very scarce at Ensenada de La Vela, but it was not reported at all in Lagiina Joyu- da (Buchanan and Stoner, 1988), although Buchanan and Stoner cautioned that specimens of this species might have been misclassified and listed as C. bocoiirti. On the other hand, the high abundance of A. cribrarius at the ma- rine front of Ensenada de La Vela differed from that of other studies in the Caribbean and Gulf of Mexico, where this species has been reported in low numbers. For in- stance, in the SW Gulf of Mexico A. cribrarius was less than 1% of the total portunid community (Garcia-Montes et al., 1988). In Laguna de Terminos (Mexico), a polyha- line coastal lagoon, four species of Callinecles were found in a population sui-\ey conducted during a whole year, but no individuals of Arenaeus were reported (Roman-Contre- ras. 1986). In the same lagoon, Sanchez and Raz-Guzman (1997) caught a single individual of A. cribrarius out of 986 specimens collected over a 17-year span. The differ- ences probably are probably due to the polyhaline con- ditions at these settings, thus restricting the viability of A. cribrarius. However, in temperate sandy beaches, this species can be very common. On Bogue Banks, in North Carolina, Arenoeus cribrarius ranked as the most impor- tant brachyuran in a high-wave-energy sandy beach ( Leb- er, 1982). In the surf zone at Folly Beach, South Carolina, A. cribrarius was one of the dominant brachyuran crabs during the summer and also a key predator of benthic or- ganisms (DeLancey, 1989). A. cribrarius is considered to be well-adapted to marine and slightly hypersaline salinity regimes and to habitats with heavy surf and sand scouring in shallow coastal waters (Fischer, 1978; Williams, 1984). This fact was evident in our study, in which A. cribrarius was abundant and clearly constrained to a narrow strip in the surf zone. Our results suggest the importance of salinity as an ex- eluding axis in the distribution of some species of swim- ming crabs in the surf and estuarine pond of Ensenada de La Vela. In our study, A. ciibrarius was present in salini- ties from SOVcc to 43" i. thus exceeding the upper limits of tolerance commonly reported for this species. The restrict- ed distribution of this species is probably a consequence of its stenohalinity (27.5-36.5"( i (Gunter, 1950; Norse, 1978; Williams, 1984; Pinheiro, 1991; Avila and Branco, 1996), although very occasionally it may show up in estuaries (Williams, 1965) and can tolerate experimental salinities down to 17.25'w (Norse, 1978). This range indicates that A. cribrarius prefers marine or near-marine environments, thus explaining its absence in station 3 (estuarine). In spite of being considered to be well adapted to heavy surf in shallow coastal waters (Fischer, 1978; Williams, 1984), A. cribrarius appeared to be abundant in all three fore- shore stations, independent of water movement, and was most abundant in the more protected stations 1 and 2. Be- cause the salinity did not show any major differences be- tween foreshore and offshore habitats, other factors are at stake in determining the zonation obsei-ved for the oth- er species. One of the main elements to consider is sub- strate composition (Norse and Fox-Norse, 1979; Pinheiro et al, 1997). Pinheiro et al. (1997) stated that distribu- tional patterns of portunids in Fortaleza Bay (Brazil) are driven mainly by the granulometric composition of the sediments. Substrates at the foreshore and estuarine pond differed from offshore bottoms: at the foreshore the sedi- ment was mainly sand; at the estuarine station a muddy bottom prevailed. At the offshore stations, silt was the main substrate. Hence, this difference could influence the distribution of swimming crabs in Ensenada de La Vela. Callinecles danae was found in both biotopes at the fore- shore but was more abundant at the estuarine site. In the marine stations of the surf zone, C. danae appeared more frequently in the most protected areas. The appearance and persistence of this species in both environments probably stems from its euryhalinity. In several Caribbean locations, C. danae has been obsei-ved dwelling in polyhaline environ- ments (Taissoun, 1969; Norse, 1978; Buchanan and Stoner, 1988). Based on this evidence, it is not surprising to find C. danae in the entire range of salinities in Ensenada de La Vela, although it is important to underline that at the estuarine station it appeared when salinity was below the minimum (ll%o) reported by Norse (1978). Also, several of the portunid species in the surf zone in Ensenada de La Vela were found in higher salinities than those reported by Norse ( 1978) in several localities in Jamaica, except C. ma- racaiboensis and C. larvatus. The absence of C. ornatus at the foreshore stations may be due to reasons other than the sampling method, because the same method was used by Carmona-Suarez and Conde (1996), where specimens of C ornatus were frequently captured at different sites in the State of Falcon, Venezuela, including Ensenada de La Vela. Total abundance of all swimming crabs both at the surf zone and at the estuarine station fluctuated widely and randomly through the year. This pattern also emerged when only the temporal abundance variations of the domi- nant species, A. cribr-arius and C. danae, were examined. No significant correlations were found between abundances of these two species and rainfall, dissolved oxygen, water temperature, or salinity fluctuations. However, the inverse correlation of dissolved oxygen and abundance reached al- most significant levels for both species at the marine fore- shore and indeed was significant for the total abundance of crabs in the surf Additionally, there was a positive cor- relation between rainfall and total abundance of crabs in the estuarine zone, possibly due to the increment of organ- ic material washed into this environment from adjacent terrestrial areas. Although bibliogi-aphic evidence supports the adaptation of portunids to low levels of dissolved oxy- gen in their environment (DeFur et al., 1990; Rantin et al., 1996; Manguni, 1997), and the relation between respira- tion rates and salinity in two Callinecles species (Rosas et al., 1989), nothing supports the idea that the increase of swimming crab densities is due to the decrease in dissolved oxygen. It might be possible that augmenting food resourc- es would increase populations of fishes and invertebrates 22 Fishery Bulletin 100(1) or planktonic blooms, which in turn would require a higher oxygen demand in the area, subsequently provoking drops in oxygen and causing mortahties of high-oxygen-demand- ing invertebrates. In any event, these results suggest that fluctuations in oxygen levels might be a key element in regulating portunid populations at Ensenada de La Vela and merit further research efforts. Berried females were remarkably scarce during our study. Only two, both belonging to C danae, were caught throughout the first period at the estuarine site, and none were caught during day and night samplings in the surf zone nor offshore. Nonetheless, scarcity of ovigerous fe- males of swimming crabs in these coasts is not exceptional. During a 2-year survey of crustaceans along 700 km of Fal- con's shoreline, Carmona-Suarez and Conde (19961 caught very few berried females of several of the littoral portunid species. They caught only one berried female of A. cribrari- us and no ovigerous females of C. sapidus, C. larvatiis, C. or- natus, or C. danae. However, in estuarine areas, substantial numbers of berried females of C. bocourti. C. maracaihoen- sis, and C. exasperatus were caught in the tidal zone. The scarcity or sheer absence of egg-bearing females in some species of swimming crabs might be the result of habitat partitioning by sex. Differential distributions by sex have been reported for C. sapidus (Williams, 1965; Perry, 1975; Archambault et al, 1990), C. maracaiboensis (Norse, 1977), and C. bocourti (Taissoun, 1969; Norse, 1978). However, for the dominant species in the surf, A. cribrarius. ovigerous females do not seem to be segregated into deeper waters. In southern Brazil, ovigerous females of this species appeared in shallow waters close to the coast (Pinheiro et al., 1996). Similarly, many ovigerous females were collected in very shallow water, at the sui-fs edge in North Carolina (Wil- liams, 1984). Likewise, for C. bocourti and C. maracaiboen- sis egg-bearing females have also been reported in marine shallow waters (Norse, 1977). Furthermore, adult females of most species inhabiting the surf zone at Ensenada de La Vela were observed in this area year-round. Thus, alter- native explanations should be considered, such as lack of estuarine habitats or sustained harsh environmental con- ditions that do not allow energy to be invested in reproduc- tion. For instance, a highly seasonal reproductive pattern, with periods without berried females, has been observed in populations of the mangrove tree crab, Aratus pisonii, liv- ing in hypersaline lagoons in this area (Conde, 1989); this pattern contrasts with the pattern for populations inhabit- ing other localities, where these crabs reproduce continu- ously throughout the year (Conde and Diaz, 1989a; Diaz and Conde, 1989). Also, undergi-own or stunted specimens of various species of crustaceans have been reported in this area (Conde and Diaz 1989b, 1992a, 1992b; Carmona, 1992; Carmona-Suarez and Conde, 1996). Thus, it is possible that this arid coast lacks the necessary resources for these crabs to reproduce, except in a few estuarine spots. This hypothe- sis is also supported by the fact that the body size of several species of swimming crabs collected in our study was small- er than that reported in other locations (Fischer, 1978; Wil- liams, 1984). The only river near the Ensenada de La Vela is the Coro River, which lies approximately 2 km westwards. Because of current direction (east-west), it cannot influence estua- rine conditions to the sampled area. The small estuary in the Ensenada de la Vela could be a possible local nutrient supplier But its influence is restricted to a few days dur- ing the end of each year, when the estuary opens to the sea. The setup of an untreated sewage discharge in the small estuarine basin at the beginning of 1997 could in fact have a long-term impact, but it is possible that vari- ous species of swimming crabs may not be affected nega- tively, because of their capacity to live in polluted areas. Such is the case with C. bocourti (Taissoun, 1972; Wil- liams, 1974), and C, sapidus, the main species in the Lake Maracaibo crab industry (Oesterling and Petrocci, 1995), where contamination due to several sources (i.e. sewage and oil) has reached high levels (Rodriguez, 2000). Because trawl studies have shown greater abundances of blue crabs (C. sapidus) and, in general, other decapods at night (Wilson et al., 1990), we ran a series of day and night samplings at the marine front over a six-month pe- riod and later also undertook diel offshore pot sampling on several occasions. Although Fischer ( 1978) has stated that A. cribrarius burrows into the bottom during the day and emerges at night, we collected A. cribrarius in the same range of abundances and sizes during both day and night samplings. Similar results were achieved by DeLancey ( 1989) in South Carolina, where no significant differences were obtained from samples collected at day and night. Wilson et al. ( 1990) ascribed the lack of differences in day and night abundances of C. sapidus to the use of more effective sampling devices than previously employed. In our study, no major differences were obsei'ved in diversity, abundance, body size, or sex ratios for most species, even though two kinds of collecting gears were used; thus, it is feasible to consider that if daily cycles exist in the spe- cies, they do not have a significant impact on daily density variations. In turn, these findings may have practical con- sequences for the decisions regarding sampling schemes to assess fisheries in this area. The exploitation of swimming crabs at Ensenada de La Vela must be considered only at the artisanal level be- cause of the low abundance of all species treated in our work and their wide and random density fluctuations. In fact, local fisheries are currently limited to the arti- sanal capture of portunids by hanging nets or hand-driven trawling nets. The most captured species by fishermen is C. bocourti (senior author, personal obs.), but C. danae is also a promising staple to be hai-vested because it appears in all three major biotopes (marine inshore, offshore, and estuarine). Arenaeus cribrarius, a species commercially exploited in Brazil (Pinheiro and Franzoso, 1998) and re- garded to have an excellent fiavor (Fischer, 1978), may al- so be considered a target species because of its great abun- dance, although its small size may make it less desirable commerciallv. Acknowledgments We heartily thank Sebastian Trompiz, whose involvement in most of the phases of the project was instrumental to Carmona Suarez and Conde Distribution and abundance of Calllnectes spp and Arenaeus cribrarius 23 its completion. We thank Oniegar Cespcdes, Angel Lopez, and (Jregorio (lotopo for occasional assistance during field work. Our gratitude extends to Maria Rondon Medicci, and Nicanor Cifuentes for their critical reading of the manuscript and for assistance during field work in August 1998, and to Eloy Conde, Eloina de Conde and Enrique Molina for logistical support. Last but not least, we thank Kate Rodriguez-Clark for her help with the English text. This study was supported in part by grants CTI 90-068 from the UNEFM and BTA-08 from CONICIT-BID. This work was partially undertaken while C. Carmona-Suarez was a staff member of the Centre do Investigaciones Mari- nas (UNEFM I. Literature cited Ai-chanihault. J. A., E. L. Wenner. and .J. D Whitaker. 1990. Life history and abundance of blue crab, Calliiwctefi Kapidus Rathbun, at Charleston Harbor, South Carolina. Bull. Mar Sci.46:145-1.'58. Arnold, W. S. 1984. The effects of prey size, predator size, and sediment composition on the rate of predation of the blue crab, Calll- nectes sapidus Rathbun, on the hard clam, Mercenaria mer- cenaria (Linne). J. Exp. Mar Biol. Ecol. 80:207-219. Avila, M. G., and J. O. Branco. 1996. Aspectos bioecologicos dc Arenaeus cnbrnniis (La- marck) iDocapoda: Portunidael de Praia de Barra da Lagoa, Florianopolis, Santa Catarina, Brasil. Rev. Bras. Zool. 13; 165-174. Buchanan, B. A., and A. W. Stoner 1988. Distributional patterns of blue crabs iCaHinectes sp.) in a tropical estuarine lagoon. Estuaries 11:231-239. Carmona, C. 1992. Interpopulation size variation in the spider crab, Microphrys bieornutus Latreille 1824 (Crustacea, Brachy- ura, Majidael. Crustaccana 63:319-322. Carmona, C, and J. E. Conde. 1989. Caracterizacion de las costas del estado Falcon. Bol. Inst. Oceanog. Venezuela 28:127-133. Carmona-Suarez, C, and J. E. Conde. 1996. Littoral brachyuran crabs (Crustacea: Decapoda) from Falcon, Venezuela, with biogeogi-aphical and ecological remarks. Rev. Bras. Biol. .56:72r3-747. Conde, J. E. 1989. Ecologia poblacional del cangj-ejo de mangle Aratiis pisonii (H. Milne-Edwards) (Brachyura: Grapsidae) en habitats extremes. Ph.Sc. diss.. Institute Venezolano de Investigaciones Cientificas, Caracas, xxxvi -i- 212 p. Conde. J. E., and H. Diaz. 1989a. The mangi"ove tree crab Ara/!/.s p/.s-o/;;/ in a tropical estuarine coastal lagoon. Est. Coast. Shelf Sci. 28:639- 650. 1989b. Productividad del habitat e historias de vida del can- grejo de mangle Aratus pisonii ( H. Milne-Edwards I ( Brach- yura: Grapsidae). Bol. Inst. Oceanog. Venezuela 28:113- 120. 1992a. Variations in intraspecific relative size at onset of maturity (RSOM) in Aratus pisonii (H. Milne Edwards, 1837) (Decapoda, Brachyura, Grapsidae). Crustaceana 62:214-216. 1992b. Extension of the stunting range in ovigerous fe- males of the mangrove crab Aratus pisonii (H. Milne Edwards,1837) (Decapoda, Brachyura, Grapsidae). Crus- taceana 62:319-322. Conde, J. E., and G. Rodriguez. 1999. Integrated coastal zone management in Venezuela: the Maracaibo system. In Integrated coastal zone man- agement: principles and practice (W. Salomons, K. Turner, A. Spoudiolus, L. D. Lacerda, and R. Ramachandram, eds.). Springer- Verlag, Berlin, xviii + 386 p. DeFur, P. L., C. P. Mangum, and J. E. Reese. 1990. Respiratory responses of the blue crab Callinectes sapidus to lonfj-term hypoxia. Biol. Bull. 178:46-54. DeLancey, L. B. 1989. Trophic relationship in the surf zone during the sum- mer at Folly Beach, South Carolina. J. Coast. Res. 5:477- 488. Diaz, H., and J. E. Conde. 1989. Population dynamics and life history of the mangrove crab Aratus pisonii (Brachyura: Grapsidae) in a marine environment. Bull. Mar Sci. 45:148-163. Ferrer-Montano, O. J. 1997. Effectiveness of two pots and other factors for har- vesting hard blue crabs Callinectes sapidus in Lake Mara- caibo, Venezuela. Ciencia, Univ. Zulia 5:111-118. Fischer, W.(ed.) 1978. FAO species identification sheets for fishery purposes. Western Central Atlantic (fishing area 31), vol. VI.. 278 p. FAO, Rome. Garcia-Montes, J. F, L. A. Soto, and A. Gracia. 1988. Cangrejos portunidos del suroeste del golfo de Mexico: Aspectos pesqueros y ecologicos. An. Inst. Cien. Mar y Limnol. UNAM (Universidad Nacional Autonoma de Mexico) 15:135-150. Goddard, D., and X. Picard. 1976. Geomorfologia y sedimentacion en la costa del estado Falcon, cabo San Roman a Chichiriviche. Mem. II Cong. Latin. Geol.,Tomo II. Bol. Geol., Caracas 7:1157-1180. Gunter, G. 1950. Seasonal population changes and distributions as related to salinity, of certain invertebrates of the Texas coast, including commercial shrimp. Publications of the Institute of Marine Science, Univ. Texas 1:7-51. Hsueh, R W., J. B. McClintock, and T S. Hopkins. 1992. Comparative study of the diets of the blue crabs Cal- linectes similis and C. sapidus from a mud-bottom habitat in Mobile Bay Alabama. J. Crust. Biol. 12:615-619. Lahey J. F 1973. On the origin of the dry climate in northern South America and the southern Caribbean. In Coastal deserts: their natural and human environments (D. H. Amiran and A. W. Wilson, eds. ), p. 7.5-90. Univ. Ai'izona Press, Tucson, AZ. Leber, K. M. 1982. Seasonality of macroinvertebrates on a temperate, high wave energy sandy beach. Bull. Mar Sci. 32:86-98. Lin, J. 1991. Predator-prey interactions between blue crabs and ribbed mussels living in clumps. Estuarine Coast. Shelf Sci. 32:61-69. Ludwig, J. A., and J. F Reynolds. 1988. Statistical ecology. Wiley New York, NY, 337 p. Mangum, C. P. 1997. Adaptation of the oxygen transport system to hy- poxia in the blue crab, Callinectes sapidus. Am. Zool. 37:604-611. Moncada, F G., and O. Gomez. 1980. Algunos aspectos bioldgicos de tres especies del gen- 24 Fishery Bulletin 100(1) ero CciUinectey: ( Crustacea, Decapoda ). Rev, Cubana Invest. Pesq, .5:1-3.5, Norse. E. A, 1977. Aspects of the zoogeographic distribution ofCallinectes (Brachyura: Portunidaei. Bull. Mar Sci. 27:440-447. 1978. An experimental gradient analysis: hyposalinity as an "upstress" distributional detei-niinant for Caribbean portu- nid crabs. Biol. Bull. 155:.586-598. Norse, E. A., and M. Estevez. 1977. Studies on portunid crabs from the Eastern Pacific. I. Zonation along environmental stress gradients from the coast of Colombia. Mar Biol. 40:365-373. Norse, E. A., and V, Fox-Norse. 1979. Geographical ecology and evolutionary relationships in Callinectes spp. (Brachyura: Portunidaei. Proceedings of the blue crab colloquium. Gulf States Mar. Fish. Comm. 7 (19821:1-9. Oesterling, M. J. 1984. Manual for handling and shedding blue crabs {Calli- nectes sapidiif: >. Special Report in Applied Marine Science and Oceanography 271, Virginia Inst. Mar Sci., 76 p. Oesterling, M. J., and C. Petrocci. 1995. The crab industry in Venezuela, Ecuador and Mexico. Virginia Sea Grant Resource Advisory 56, 32 p. Orth, R. J., and J. van Montfrans. 1987. Utilization of a seagi-ass meadow and tidal marsh creek by blue crabs Callinectes sapicliis, 2: Seasonal and annual variations in abundance with emphasis on post-set- tlement juveniles. Mar Ecol. Prog. Ser. 41:283-294. Perry, H. 1975. The blue crab fishery in Mississippi. Gulf Res. Rep. 5:39-57. Perry, H. M., and. W. A. Van Engel (eds.). 1979. Proceedings of the blue crab colloquium. Gulf States Mar. Fish. Comm. 7 ( 19821:1-251. Pinheiro, M. A. A. 1991. Distribu(;ao e Biolugia Populacional de Arcnaeus crihranus (Lamarck, 1818) (Crustacea, Brachyura, Portun- idaei, na Ensenada da Fortaleza, Ubatuba, ,SP. Tesis de Maestria en Zoologia. Universidade Estadual Paulista, Botucatu, Brasil, 175 p. Pinheiro, M. A. A., and A. Fransozo. 1993. Relative gi-owth of the speckled crab A/-c()oe;;,s cribrar- ;'us (Lamarck, 1818 1 ( Brachyura, Portunidae I, near Ubatuba, State of Sao Paulo, Brazil. Crustaceana 65:377-389. 1998. Sexual maturity of the speckled swimming crab Are- naeus crihrarius (Lamarck. 18181 (Decapoda, Brachyura, Portunidaei, in the Ubatuba littoral, Sao Paulo State, Brazil. Crustaceana 71:434-452. 1999. Reproductive behavior of the swimming crab Are- naeus crihranus (Lamarck, 18181 (Crustacea, Brachyura, Portunidaei in captivity. Bull. Mar. Sci. 64:243-253. Pinheiro, M. A. A., A. Fransozo, and M. L. Negi-eiros-Fransozo. 1996. Distribution patterns of Arenaeus crihrarius (Lam- arck, 18181 (Crustacea, Portunidae) in Fortaleza Bay, Uba- tuba (SP), Brazil. Rev. Bras. Zool. 56:705-716. 1997. Dimensionamento c sobreposigao de nichos dos por- tunideos (Decapoda, Brachyura), na Enseada da Fortaleza, Ubatuba, Sao Paulo, Brasil. Rev Bras. Zool. 14:371-378. Prager, M. H, J. R. McConaugha, C. M. Jones, and P. J. Geer. 1990, Fecundity of blue crab, Callinectes sapidns, in Chesa- peake Bay: biological, statistical and management consid- erations. Bull. Mar. Sci. 46:170-179. Rantin, F T, A. L. Kalinin, and J. C. de Freitas. 1996. Cardio-respiratory function of swimming blue crab Callinectes danae Smith, during normoxia and graded hypoxia. J. Exp. Mar Biol. Ecol. 198:1-10. Rodriguez, G. 1980. Crustaceos decapodos de Venezuela. Instituto Vene- zolano de Investigaciones Cientificas, Caracas, Venezuela, 444 p. 2000. El manejo de los recursos naturales del sistema de Maracaibo. Chapter 7 in El sistema de Maracaibo (G. Rodriguez, ed.), p. 991-109. Instituto Venezolano de Investigaciones Cientificas, Caracas, Venezuela. Roman-Contreras, R. 1986. Analisis de la poblacion de Callinectes spp. ( Decapoda: Portunidae) en el sector occidental de la lagunadeTerminos, Campeche, Mexico. An. Inst. Cien, Mar. y Limnol. UNAM (Universidad Nacional Autonoma de Mexico) 13:315-322. Rosas, C, G. Barrera, and E. Lazarc-Chavez. 1989. Efecto de las variaciones de la salinidad y de la temperatura estacional sobre el consume de oxigeno de Callinectes rathbunae, Contreras y Callinectes similis (Crustacea: Portunidaei. Trop. Ecol. 30:193-204. Ryer, C. H., J. van Montfrans, and R. J. Orth. 1990. Utilization of a seagrass meadow and tidal marsh creek by blue crabs Callinectes sapidus. II. Spatial and temporal patterns of moulting. Bull. Mar Sci. 46:95-104. Sanchez, A. J., and A. Raz-Guzman. 1997. Distribution patterns of tropical estuarine brachyuran crabs in the Gulf of Mexico. J. Crust. Biol. 17:609-620. .Scelzo, M. A., and R. Varela. 1988. Crustaceos decapodos litorales de la isla de la Blan- quilla, Venezuela. Mem. Soc. Ven Cien. Nat. 47:33-54. Schubart, C. D., J. E. Conde, C. A. Carmona-Suarez, R. Robles, and D. L. Felder. 2001. Lack of divergence between 16S mtDNA sequences of the swimming crabs Callinectes hocourti and C. mara- caiboensis ( Brachyura: Portunidae I from Venezuela. Fish. Bull. 99:475-481. Sholar, T M. 1979. Blue crab fisheries of the Atlantic Coast. Proceedings of the blue crab colloquium. Gulf States Mar. Fish. Comm. 7(19821:111-127. Smith, D. E., R. J. Orth, and .J. R. McConaugha (conference steering committee). 1990. Proceedings of the blue crab conference held in Virginia Beach, Virginia. May 1.5-17, 1988. Bull Mar. Sci. 46:1-251. Sokal, R, R.,andFJ. Rohlf 1995, Biometry. 3''<' ed. Freeman, New York, NY, 887 p. StatSoft. 1992. Statistica/Mac. StatSoft, Tulsa, Oklahoma. Stuck, K. C, and F M. Truesdale. 1988. Larval development of the speckled swimming crab, Arcnaeus crihranus (Decapoda: Brachyura: Portunidae) reared in the laboratory Bull. Mar Sci. 42:101-132, Taissoun, E. 1969. Las especies de cangrejos del genero Callinectes (Brachyura) en el golfo de Venezuela y lago de Maracaibo. Bol. Centro Invest. Biol., Univ Zulia 2:1-103. 1972. Estudio comparative, taxonomico y ecologico entre los cangrejos (Decapoda: Brachyura: Portunidae), Calli- nectes maracaiboensis (nueva especie), C. hocourti (A. Milne Edwards) y C. rathhunae (ContrerasI en el golfo de Ven- ezuela, lago de Maracaibo y golfo de Mexico. Bol. Centro Invest. Biol., LIniv Zulia 6:7-46. 1973a. Biogeogi'afia y ecologia de los cangrejos de la familia "Portunidae" (Crustaceos Decapodos Brachyura) en la costa atlanticade America. Bol. Centro Invest. Biol., LIniv Zulia 7:7-23. Carmona Suarez and Conde: Distribution and abundance of Callinectes spp and Arenaeus cribrarius 25 1973b. Los canffic'jos do hi famili;i Portunidac (Crustaceos Docapodos Brachyura) en i-l Otcidente de Venezuela. Bol. Centro Invest. Biol., Univ Zulia 8:1-78. van Montfrans, J., J. Capelli. R. J. Orth, and C. H. Ryer. 1986. Use of microwiro tags for tagging juvenile blue crab.s {Callinectes sapidi/t^ Rathbunl. J. Crust. Biol. 6:370-376. Warner, G. F. 1977. The biology of crabs. P^lck Science. London, 202 p. Williams, A. B. 196."). Marine decapod crustaceans of the Carolinas. Fish. Bull. 65:1-298. 1974. The swimming crabs of the genus Callinectes (Dccap- oda: Portunidaei. Fish. Bull. 72:685-798. 1984. Shrimps, lobsters, and crabs of the Atlantic Coast of the Ea.stern United States, Maine to Florida. Smithson- ian Institution Press, Washington D.C., 550 p. Wilson. K. A., K. L. Heck Jr., and K. W. Able. 1987. Juvenile blue crab, Callinectes scipidi/s, sui-vival: an evaluation of celgrass, Zostera marina, as refuge. Fish. Bull. 85:5.3-.58. Wilson, K. A.. K Q. Able, and K. L. Heck Jr 1990. Habitat use by juvenile blue crabs: a comparison among habitats in southern New Jersey. Bull. Mar Sci. 46:105-114. 26 Abstract— We examined 536 permit {Tiachinotus fatcatus. 65-916 mm FL) collected from the waters of Florida Keys and from the Tampa Bay area on Florida's Gulf coast to describe their growth and reproduction. Among permit that we sexed, females ranged from 266 to 916 mm in length (mean=617) and males ranged from 274 to 855 mm (mean=601). Ages of 297 permit ranging from 102 to 900 mm FL were estimated from thin-sectioned otoliths (sagittae). The large proportion of oto- liths with an annulus on the margin and an otolith from an OTC-injected fish suggested that a single annulus was formed each year during late spring or early summer Permit reach a maximum age of at least 23 years. Permit gi-ew rap- idly until an age of about five years, and then growth slowed considerably. Male and female von Bertalanffy growth models were not significantly differ- ent, and the sexes-combined growth model was FL=753.1(l-e-" ■'■•»' •^«>'"^s'^' I. Gonad development was seasonal, and spawning occurred during late spring and summer over artificial and natural reefs at depths of 10-30 m. Ovaries that contained oocytes in the final stages of oocyte maturation or postovulatory fol- licles were found during May-July. We estimated that SO'J'r of the females in the population had reached sexual maturity by 547 mm and an age of 3.1 years and that 50% of the males in the population had reached sexual maturity by 486 mm and an age of 2.3 years. Because Florida regulations restrict the maximum size of permit caught in recreational and com- mercial fisheries to 20-inch (508-mml, most fish harvested are sexually imma- ture. With the current size selectivity of the fishery, the spawning stock bio- mass of permit could decrease quickly in response to moderate levels of fish- ing mortality; thus, the regulations in place in Florida to restrict harvest levels appear to be justified. Age, growth, and reproduction of permit (Trachinotus falcatus) in Florida waters Roy E. Crabtree Peter B. Hood Derke Snodgrass Florida Marine Research Institute Florida Fish and Wildlife Consen/ation Commission 100 Eighth Avenue SE St Petersburg, Florida 33701 5095 Present address (for R, E Crabtree) Division of Marine Fisheries Florida Fish and Wildlife Conservation Commission 620 Meridian St Tallahassee, Florida 32399 1600 E mail address (for R E Crabtree) crabtrno'gfc stale fl us Manuscript accepted 19 July 2001. Fish. Bull. 100:26-34 (2002). The family Carangidae supports a di- verse array of economically important fi.sheries in tropical and subtropical waters worldwide. In the southeastern United States, many carangid stocks are managed at both the state and Fed- eral level. Recently, the National Marine Fisheries Service determined that the Gulf of Mexico greater amberjack stock is overfished, but the status of most carangid stocks is unknown (Anony- mous'). For most carangid stocks, no quantitative stock assessments have been completed, in part, because little biological information exists regarding carangid growth rates and reproduc- tion. As a result, the adequacy of cur- rent management measures to prevent overfishing of many carangid stocks is unclear. Permit, Trachinotus falcatus, are the basis of an important recreational fish- ery and a small commercial fishery in Florida. Estimates of Florida recreation- al landings are unreliable but may ex- ceed 100,000 fish per year (Armstrong et al.'-). Commercial landings of permit peaked in 1991 at 200,000 pounds and then decreased to 50,000 pounds in 1995 (Aj-mstrong et al.-). Current reg- ulations in Florida include a 10-inch (254-mm FL) minimum size limit and a 20-inch (508-mm FL) maximum size limit on both the recreational and com- mercial harvest. In addition, recreation- al anglers are permitted daily to take 10 permit per bag of combined permit and Florida pompano {Trachinotus car- olinus). Many anglers pursuing permit do so with professional guides on a char- ter vessel. In addition to being popular in South Florida, permit are targeted by numerous fishing tourists and recre- ational anglers in the Bahamas and at locations throughout the Caribbean. De- spite the economic importance of permit in these regions, there are no published reports describing gi-owth, longevity, or length and age at sexual maturity. Such information is needed to evaluate the ef- fects of fishing mortality on permit pop- ulations. Previous studies of permit life history by Fields (1962) and Finucane ( 1969) were based only on an examina- tion of larvae and young-of-the year per- mit. Our study describes growth, lon- gevity, and the length and age at which fish become sexually mature. In addi- tion, we document spawning of permit in South Florida waters based upon a histological examination of ovaries and seasonal patterns in the abundance of juveniles. ' Anonymous. 2001. Report to Congress: status of fisheries of the United States, 122 p. National Marine Fisheries Sei-vice, 1315 East-west Highway, Silver Spring, MD, 20910. - Armstrong, M. P., P. B, Hood, M. D. Murphy, and R. G. Mullen 1996. A stock assess- ment of permit, Tracltinotiix falcatus. in Florida waters. Unpubl. rep. to the Flor- ida Marine Fisheries Commission. Flor- ida Marine Research Institute, 100 Eighth Avenue SE, St. Petersburg. Florida 33701- 5095. Crabtree et al : Life history of Trachinotus fakatus 27 Methods Collections Permit that we examined were collected from the Florida Keys (/i=308; between 25°40'N. SOnO'W and 24°30'N, 82''20'W) during 1995-97 and the Tampa Bay area (» =228; 27°40'N, 82°45'W) during 1990-95. Most F^lorida Keys permit were caught with hook-and-line gear (;!=215) or speared («=58) over artificial and natural reefs in the waters off the lower and middle Keys in depths ranging from 10 to 30 m. Other, usually smaller, permit were cap- tured with gill nets (/i = 16), seines (/i = 18). and bottom trawls (n=l) over or near shallow banks adjacent to the Keys. Most of the permit sampled in the Tampa Bay area were small (<400 mm FL) and were captured with seines along sandy beaches; some larger Tampa Bay permit were captured with gill nets (/i=53) or trammel nets (/( = 14). Standard length (SL), fork length (FL), and total length (TL) were measured to the nearest millimeter (mm) and weight was measured to the nearest gram. Unless other- wise indicated, all lengths reported in our study are fork lengths. Otoliths (sagittae) were removed, rinsed in water, and stored dry until sectioned; they were later weighed to the nearest 0.01 mg. Gonad weight was recorded to the nearest gram (g), and gonad samples were removed from the fish and preserved in 10'/( buffered formalin; they were later soaked in water for 24 hours and stored in 70^7^ ethanol. Collections of juvenile permit from sandy beaches off Tampa Bay and the Florida Keys were made with a 21.3 x 1.8-m bag seine (6.4-mm mesh in the wings and 3.2-mm mesh in the bag). Seine hauls were made perpendicular to the beach for distances up to 50 m, depending on water depth. Lengths of up to 50 fish from each sample collec- tion were measured to the nearest millimeter. Near Tam- pa Bay, we collected fish at the Gulf of Mexico beaches of Treasure Island (November 1992 -October 1994; 27°46'N, 82°46.5'W) and Indian Shores (August 1993-November 1994; 27°50'N. 82''50'W). Sampling at each site consisted of five seine hauls every two weeks. Six sandy beaches were sampled monthly in the Florida Keys from July 1994 to July 1997: Lower Matecumbe Beach (July 1994-April 1996; 24°50.95'N, 80°4415'W), Coco Plum Beach (July 1994-April 1996; 24°43.65'N, SFOO.lO'Wi. Clarence P Higgs Beach. Key West (July 1994-July 1997; 24°32.79'N, 81°47.26'\V), Bahia Honda State Park (October 1994-May 1997; 24°39.81'N, 81°15.44'W). Boca Chica Beach (Oc- tober 1994-April 1996; 24°33.60'N, 81°41.65'W), and Sugarloaf Beach (January 1995-May 1996; 24°36.57'N, 81°33.49'W). Age and growth The left sagitta was usually used for age estimation; how- ever, if the left otolith was broken, lost, or destroyed during processing, the right otolith was substituted. We prepared otoliths for age estimation by embedding them in Spurr, a high-density plastic medium (Secor et al.. 1992). A 1-mm to 2-mm-thick transverse section containing the otolith core was cut with a Buehler Isomet low-speed saw with a diamond blade. The section was mounted on a microscope slide with thermoplastic glue (CrystalBond 509 adhesive) and was polished with wet or dry sandpaper (grit sizes ranging from 220-2000) until annuli were visible. Sec- tions were then polished on a Buehler polishing cloth with 0.05-gamma alumina powder to remove .scratches. With- out knowledge of fish size or capture date and using a compound microscope equipped with transmitted light, two readers independently counted annuli on each otolith twice. If three of the four readings agreed, then this mode was accepted as the annulus count. If three of the four readings did not agree, each reader again counted annuli independently and without knowledge of previous counts. If three of the resulting six readings agreed, then this mode was accepted as the annulus count. If there were not three readings that agreed, the otolith was excluded from further analysis. In six cases, two sets of three readings that were in agreement occurred. For these six otoliths the two sets of readings differed by only one annulus; there- fore the mean was accepted as the annulus count. The percentage of otoliths with an annulus on the edge was then plotted by month so that we could look for a sea- sonal pattern in annulus formation. We did not attempt to measure marginal increments because the margin of per- mit otoliths is highly sculptured and easily broken; how- ever, we did believe that we could discern the presence of an annulus on the otolith's edge. The von Bertalanffy (1957) growth equation FL, = L,, (1-e '"'"'"') was fitted to observed age-length data with nonlinear regression procedures. Age was esimated as the annulus count because permit both spawn and form annu- li at about the same time of year. Our estimates of length at age include some seasonal growth that occurred after the formation of the final annulus. Length-weight regres- sions were calculated by linear regression of logju-trans- formed data. Sex-specific growth models were compared with an ap- proximate randomization test described by Helser (1996). This test is based on the premise that when the null hy- pothesis of no sex-specific differences in growth is true, a test statistic derived by random assignment of fish to one of two populations will not be different from that observed between sexes. The test statistic is calculated as the re- sidual sums of squares for the sexes-combined von Berta- lanffy growth model minus the residual sums of squares for the two sex-specific models. A probability distribution of the test statistic was generated by a randomization rou- tine with 1000 iterations of the nonlinear models. Only sexed fish were included in the statistical comparison. Age validation Permit used in the age-validation experiments were cap- tured in waters off the Florida Keys with hook-and-line gear After capture, permit were tagged with dart-type tags and injected with Liquamycin LA-200 (200-mg oxy- tetracycline |OTC|/mL) in the dorsal musculature at a dosage of about 100-mg OTC per kg fish weight. Permit were then held in a 33.5-m-long by 5.5-m-wide by 0.75-m- 28 Fishery Bulletin 100(1) deep pond at the Florida Fish and Wildhfe Consei-vation Commission's Keys Marine Laboratory in Long Key. Fish were held at ambient temperatures and were fed frozen shrimp and fish until satiated at least three times a week. Although several permit were injected and held for vari- ous periods, only one fish survived long enough to have formed an annulus after the OTC injection. The otolith section from this fish was examined with a compound microscope (40-lOOx) equipped with ultraviolet light so that the fluorescent OTC mark could be detected. Reproduction Histological sections of gonads were prepared and assessed for reproductive state. Gonad samples were prepared for histological examination with a modification of the peri- odic acid Schiff's (PAS) stain for glycol-methacrylate sec- tions and with Weigerts iron-hematoxylin as a nuclear stain and metanil yellow as a counterstain (Quintero- Hunteret al., 1991). Developmental stages of oocytes were determined and oocytes were counted from histological preparations at lOOx with a compound microscope attached to a digital im- age-processing system. Four oocyte stages were recognized in permit ovaries: primary growth, cortical alveolar, vitel- logenic, and oocvtes in the final stages of maturation (Wal- lace and Selman, 1981). The final stages of oocyte matu- ration (FOM) included yolk coalescence, germinal vesicle migi-ation, germinal vesicle breakdown, and hydration. We also counted postovulatorv follicles (POFs) and PAS-pos- itive melanomacrophage centers (Ravaglia and Maggese, 1995; Crabtree et al., 1997), which were present in many ovaries. When stained with the PAS stain, these PAS-pos- itive structures are brilliant purple. Melanomacrophage centers are thought to be active in degrading atretic oo- cytes, postovulatory follicles, and residual cells of the sper- matogenic cycle (Chan et al., 1967; Ravaglia and Maggese, 1995). The developmental stage of at least 300 oocytes and other structures on each slide was determined and count- ed in arbitrarily chosen fields, and frequencies were ex- pressed as a percentage of the total count. We counted all oocytes that had at least 50*^* of their area visible in a field before moving to the next field. We examined seasonal reproductive patterns by plotting monthly juvenile length frequencies and monthly mean go- nadosomatic indices (GSIs). Gonadosomatic indices were calculated for 129 sexually mature female permit ranging in length from 476 to 916 mm and for 122 sexually mature male permit ranging in length from 449 to 855 mm as GSI = (GW / (7W - GW )) 100, where GW = total gonad weight 0.05). Neither the slopes (P=0.464) nor the elevations (P=0.063) of the length-weight equations for male and female permit were significantly different. The pooled length-weight equation for sexed and unsexed fish was logi„Wr = 2.803 log,,, FL - 4.078, {n=488, 7--=0.996) where WT = weight in grams; and FL - fork length in mm. Age and growth When viewed with transmitted light, permit otoliths have opaque (dark) annuli that alternate with translucent (light) zones (Fig. 2). Proceeding from the otoliths core towards the otoliths proximal margin, annuli are regu- larly spaced along the sulcal ridge. In some individuals, the annuli are indistinct and irregular in appearance, which made age estimation difficult. We considered 51 oto- liths (17.3%) from permit ranging in length from 243 to 916 mm to be unreadable. The length-frequency distri- bution of fish whose otoliths were considered unreadable was not significantly different from that offish whose oto- liths were considered readable (Kolmogorov-Sniirnov two- sample test, Z)=0.144, P=0.32); thus, no particular length Crabtree et a\ Life histoid of Tmchlnotus falcatus 29 200 400 600 800 1000 1200 25^ females 20 ; r^187 ^5'- ■ 10 n ■ -in 5 0 in j L 0 200 400 600 800 1000 1200 Fork length (mm) E 2 20 15 10 5 0 25 20 15 10 5 0 males n=124 10 15 20 25 females n^127 10 15 20 25 Age (yr) Figure 1 Fork lengths (mm) and ages (years) of male and female permit. Trachinotus falcatus, sampled from South Florida waters. group of fish was systematically excluded from the age- and-grovvth analysis. Annulus formation in permit occurs during spring and early summer. The percentage of permit with an annulus on the otolith's margin was greatest during summer and least during October-March, suggesting that annulus for- mation is seasonal and that annuli first become visible during late spring or early summer (Fig. 3). A single OTC-injected permit was successfully held for a sufficient length of time to be useful in age validation. This fish was captured and injected with OTC on 17 June 1993. The fish was sacrificed on 30 January 1996 and measured 600 mm in length. After 31 months in captivity, which included two spring-summer periods, the fish had formed two annuli, a number that is consistent with our hypothesis that a single annual mark forms annually during late spring or early summer. Also visible immediately before the OTC mark was an annulus that was probably formed during late spring of 1993, just prior to capture and OTC injec- tion. Moreover, there was a wide margin subsequent to the last annulus that is consistent with the six or more months of otolith growth after formation of the final an- nulus in late spring or early summer of 1995. Estimated ages of 298 permit ranged from 0 to 23 years for fish 102 to 900 mm long. Permit grew rapidly until about age five, and then growth slowed considerably (Ta- ble 1, Fig. 4). Most of the fish in our sample were less than 10 years old, although fish 10-15 years old were common. The oldest permit examined was a 23-year-old I781-mm) male (Table 1). Estimates of von Bertalanffy growth model parameters are presented in Table 2. The growth models for male and female permit were not significantly differ- ent (approximate randomization test, P=0. 059). Sexual maturation We estimated that 50'7f of the males in the population reached sexual maturity by 486 mm and an age of 2.3 years, and 509^ of the females in the population reached sexual maturity by 547 mm and an age of 3.1 years (Table 3). The smallest sexually mature male in our sample was 449 mm long, and the youngest sexually mature male was 3 years old. Our estimate of the age at 50'7f maturity for males was less than the age of the youngest mature male observed. This knife-edge maturity curve could be an artifact of our small sample size. The smallest sexually mature female in our sample was 476 mm long, and the youngest sexually mature female was 3 years old. All of the ovaries we examined contained primary-growth-stage oocytes. Cortical alveolar-stage oocytes occurred only in ovaries from permit larger than 450 mm and older than 2 years and were common only among permit larger than 500 mm and older than 3 years. Vitellogenic oocytes were found only in ovaries from fish larger than 550 mm and older than 3 years and were common only among permit larger than 600 mm. The length and age at which vitel- logenic oocytes were commonly found agrees well with our estimate of the length and age at which 50% maturity was 30 Fishery Bulletin 100(1) B Figure 2 (A) Sectioned sagitta from a 1-year-old (363-mni-FL) permit. Ti-achinotus falcatiis. collected in the Florida Keys on 27 Fet^ruary 1995. showing the location of the core (white arrow! and the first annulus (black arrow). Scale bar = 200 microns. (B) Sec- tioned sagitta from a 23-year-old male permit (781-mm-FL) collected in the Florida Keys on 4 June 1996. Scale bar = 500 m. reached, suggesting that we misclassified few gonads with regression. Spawning seasonality Permit spawning appeared to be seasonal in the areas we sampled and occurred at least during May^uly. We examined 15 permit ovaries that contained either oocytes in the final stages of maturation or POFs, structures indicative of imminent or recent (<24 h) spawning. We usually did not know the time of day when fish were caught, but all fish were captured during daylight hours (0700-1700 h). Oocytes in the final stages of maturation were found during June and July, and POFs were found Crabtree et a\ : Life histoi"y of Tiachinotus fakotus 31 Table 1 Average obsci-\ed and prediclcd l'( irk lengths (mini of permit, 'I'nn IiiikiIiis fat catus. Till average obsei-ved length at age includes some seasona growtli that occuitl d after the format ion ofthe linal anniilus. \ allies in p; irentheses are standard error and sample | size. Age Sexes combined Females Males Average Average Average (yr) observed Predicted observed Predicted observed Predicted 0 160(12.7:17) 139 277(1) 212 149 1 301 (5.9;56) 319 310(14.5:8) 353 334(12,5:17) 337 2 476(7.6;10) 447 479(6.8:8) 458 465(33.5:2) 464 3 564(10.1:27) 537 555(12.1:13) 538 572(15.9:14) 550 4 612(6.3:28) 601 620(9.8:14) 599 604(7.6:14) 608 5 643 (8,7:29) 645 664(13.8:10) 644 632(10.5:19) 647 6 663 (8.7:26) 677 658(10.7:20) 679 680l9.3;6i 673 i 687(12.0:17) 699 687(10.8:9) 704 687(23.6:81 691 8 703(16.9:12) 715 695(22.7:7) 724 715(27.4:5) 703 9 713(15.1:21) 726 710(26,8,11) 743 717(13.3:10) 711 10 743(30.0:6) 734 754(34.2:5) 750 688(1) 717 11 746(21.8:12) 740 811(25.3:4) 758 714(23,2:8) 721 12 738(35.3;5) 744 797 (0,5:2) 765 698(46,9:3) 723 13 787(19.4:9) 746 803 (26.2:6) 769 754(15,7:31 725 14 762(19.5:13) 748 783(12.8:7) 773 737(38.9:6) 726 1.5 753(13.4:4) 750 737(1) 776 759(17.3:3) 727 16 751 778 727 17 751 779 727 18 745(48.5:2) 752 793(1) 781 696 ( 1 ) 728 19 752 781 728 20 667(1) 753 782 667(11 728 21 687(1) 753 916(1) 783 687(1) 728 22 753 783 728 23 781(1) 753 783 781(1) 728 Table 2 Parameter estimates ofthe von Bertalanffy growth model for permit. Trachinotus falcatus. from South Florida waters. Values in parentheses are standard errors. FL = fork length. Sex n L (mm FL) K 'o adjusted r- Females 127 784.2 0.28 -1.12 0,833 (13.79) (0.027) (0.249) Males 123 728.2 0.39 -0,58 0,855 (9.52) (0.034) (0.168) Combined 297 753.1 0.35 -0.59 0,921 (7.12) (0.015) (0.065) during May-July (Fig 5), Vitellogenic oocytes were most plentiful during March-July and were absent during October-December, No samples were available for histo- logical examination in January or February, but it seems 100 h c h 29 21 O) , 03 E t c 75 o 21 CO 3 8 , 10 1 50 _ ■ ^ 34 5 12 ^ 25 - o 21 o 6 10 °" 0 - * 4 FMAMJ JASOND Month Figure 3 Mean percentage and standard error of permit {Trachi- notus falcatus) otoliths with an annulus on the margin plotted by month. Numbers above the lines are the monthly sample sizes. 32 Fishery Bulletin 100(1) unlikely that spawning occurred during these months. Females with the greatest GSIs (>4%) were captured during March-August, and GSIs were least ( <1.5% ) during October-December (Fig. 6). Male GSIs were generally sim- ilar in magnitude to female GSIs and followed the same pattern. In the Tampa Bay area, small permit (<40 mmi were present from June to November, suggesting that spawning extends into the fall. In the Florida Keys, small fish (<40 mm) were present all year, suggesting an extended spawn- ing season, recruitment from other areas with different seasonal spawning patterns, or variable juvenile growth rates. 1000 750 500 . I 250 >,\ oL 0 xiWW 10 15 All n=297 20 25 1000 750 500 1: i!" ' 250 OL females n=127 0 5 10 15 20 25 Age (yr) Figure 4 Observed and predicted fork lengths (nimi from the von BertalanfTy growth model for sexed and unsexed permit, Trachinotiis falcatus. Discussion We obtained permit from a variety of fishery-dependent and fishery-independent sources; consequently, our sample is biased towards certain size classes, and the bimodal size-frequency distribution of our sample probably does not reflect that of the population or the Florida harvest. All the small fish (<300 mm) we examined were from fishery-inde- pendent sources; most large fish were from fishery-depen- 25^ • 20 ,- Frequency O Ol t 5 ^ 0 ^ • 1 • : : f • M A M J J A S O N D Month Figure 5 The percen t frequency of occurrence of oocytes in the | final stages of oocyte maturation (FOM) and postovu- latory follicles (POF) in individual permit iTrachino- \ tus fa lea tun ) ovaries plotted by month. Table 3 The relationship of percentage mature and fork length (mm I and the relationship of percentage mature and age (years) for permit, Trachiriotus falcatus. from South Florida waters. FL = fork length (mm) and AGS = age (years). Pr,„is the absolute value of ((a +6 )/c). is the inflection point of the curve, and is the length or age predicted by the logistic regression at which 50''r of the permit in our sample were sexually mature. Sex is a dummy variable equal to 1 for males and 0 for females. PD is the adju.sted percentage of deviance explained by the model. Percent female '■'/(1+e" X PD FL 314 -30.41 3.34 0.056 (6.336) (1.087) (0.0114) AGE 233 -6.71 1.71 2.14 (0.878) (0..'576) (0.238) 0.84 0.09 486 mm (males) 547 mm ( females i 2.3 years (males) 3.1 years (females) Crabtree et al,: Life history of Tmchtnotus lakatus 33 dent sources, such as charterboats. Wo did not sample any permit from the commercial fishery, which principally tar- gets smaller fish as a result of the maximum size limit of 20 inches (508 mm FL) for permit caught by commercial ves- sels. Ai'mstrong et al.- reported that most han-ested permit in Florida were <440 mm. In contrast, our sample contained many fish larger than 600 mm. The high proportion of large permit in our sample could reflect a tendency for charter- boats in the Florida Keys to select larger permit than those selected by more typical anglers statewide. Ai-mstrong et al.'s^ assessment was based on more systematic and state- wide sampling than ours, and the differences between their sample and ours probably reflects our attempt to obtain a sample of all available size classes rather than a represen- tative sample of the Florida hai-vest. Age and growth The oldest permit in our sample was estimated to be 23 years old. Although we examined many relatively large permit, larger fish than those we examined have been caught. Robins ( 1992) reported that permit can reach 1100 mm FL and a weight of 23 kg; consequently, permit lon- gevity probably exceeds our estimate of 23 years. There are no other estimates of age and growth of permit for comparison, but our longevity estimates are similar to those determined from sectioned otoliths for other caran- gids. Manooch and Potts (1997) aged greater amberjack and found fish as old as 17 years. The oldest carangid yet studied is the trevally, Caranx georgianus, reported to reach an age of 46 years (James, 1984). The much smaller Florida pompano has been reported to reach an age of 7 years (Hood et al.-^). 8 - females 6 . n=129 4 ! i ■ - ■ 2 0 i 1 Tr i 1 1 V^.-^ M A M J J A S 0 N D 8 males 6 - A • n=122 4 2 0 :/ / 1: i i i s 0 N D ri ■ J ■r t A M A M J Month Figure 6 Gonadosor na tic indice.s i GSI. • anc means (-(-) for sexually mature fer na le and male permit. Trachinotus fatcatus. plot- ted by mor ith. Our estimates of the von Bertalanffy growth model pa- rameters are within the range of those reported for other carangids (James, 1984; Sudekum et al., 1991; Manooch and Potts, 1997). We found no significant differences be- tween male and female von Bertalanffy growth models, but the significance level (P=0.059) was close enough to 0.05 to cause us to suspect that a difference might exist. Hood et al.'^ also found no sex-specific differences in growth models for pompano. Sexual maturation We sampled relatively few permit between 300 and 500 mm long, the size at which sexual maturity is reached. The lack offish in this critical size range resulted in the knife- edge maturity curves. Larger sample sizes are needed to derive more precise estimates of age and size at sexual maturity. In an assessment of the status of permit stocks in Florida, Armstrong et al.- assumed that permit mature at about 440 mm FL on the basis of limited biological data available at the time. Our estimates of length at 50% maturity are larger: 486 mm for males and 547 mm for females. As a consequence of Florida's 20-inch (508-mm) recreational and commercial maximum size limit, most of the permit harvested are sexually immature (Armstrong etal.2). Spawning We believe that permit spawn over artificial and natural reefs in the waters of the middle and lower Florida Keys because ovaries of fish caught over these structures con- tained oocytes in the final stages of maturation and POFs. Other researchers have inferred that permit spawn in nearshore waters from the capture of early-stage lai-vae (Fields. 1962; Finucane, 1969). Permit ovaries that con- tained fresh POFs and oocytes in the final stages of mat- uration also contained clutches of early- and mid-stage vitellogenic oocytes, suggesting that permit are multiple- batch spawners. Spawning occurred at least during May-June in the Florida Keys during 1995-97. Juvenile length frequen- cies in the Keys suggest a more prolonged spawning sea- son— perhaps even year-round spawning; however, the prolonged presence of small juveniles could also be attrib- uted to variable juvenile growth rates rather than extend- ed spawning. This question could be resolved by direct ag- ing of juveniles to evaluate growth rates. Our sample of adult permit may have been too small to reveal low levels of spawning outside of spring and early summer, and no mature permit were collected during January or February. On the basis of seasonal occurrence of juveniles, Finucane ( 1969) suggested that permit spawn during April-June in the Tampa Bay area, but Fields (1962) found juveniles 3 Hood. P. B.. D. T Menyman. and D. J. Harshany 1999. Age, growth, mortality, and reproduction of the Florida pompano, Trachinotus carolinus, from Florida waters. Unpubl. manu- script. Florida Marine Research Institute, 100 Eighth Avenue SE, St. Petersburg, FL. 34 Fishery Bulletin 100(1) year round suggesting a prolonged spawning period. Oth- er carangids spawn during spring and summer: Caranx tgnobilis and Caranx melampygus spawn during May-Au- gust in Hawaii (Sudekum et al., 1991) and T. carolinus spawns during January-August in Florida (Hood et al.-^). Our data suggest that maturation occurs at greater lengths than assumed by Armstrong et al.-; however, even using our maturation data, their observation that most permit landed are sexually immature remains true. With the current selectivity of the fishery, permit spawning stock biomass could decrease quickly in response to mod- erate levels of fishing mortality; thus, the regulations in place in Florida to restrict harvest levels appear to be jus- tified. Significantly better estimates of the magnitude and age structure of the catch would be required to complete a comprehensive age-structured stock assessment. Acknowledgments We thank Capt. J. C. Wells, who provided us with most of the permit examined in this study and whose efforts made this work possible, and Don DeMaria, who also pro- vided specimens. We thank John Swanson, Bill Gibbs, and the staff at the Keys Marine Laboratory for their assis- tance; Jim Colvocoresses, John Hunt, and others at the South Florida Regional Laboratory for their cooperation; and David Harshany, Heather Patterson, Dan Merryman, Graham Gerdeman, and Connie Stevens for their assis- tance. We also thank Jim Colvocoresses, Rich McBride, Jim Quinn, Judy Leiby, and Llyn French for helpful com- ments on the manuscript. This work was supported in part under funding from the Department of the Interior, U.S. Fish and Wildlife Service, Federal Aid for Sportfish Resto- ration F-59. Literature cited Chan, S. T. H., A. Wright, and J. G. Phillips. 1967. The atretic structures in the gonads of the rice-field eel (Monopterus albus) during natural sex-reversal. J. Zool. (Lend.) 153:527-539. Crabtree, R. E., D. Snodgrass, C. W. Harnden. 1997. Maturation and reproductive seasonality in bonefish, Albula vulpes. from the waters of the Florida Keys. Fish. Bull. 95:456-465. Fields. H. M. 1962. Pompanos iTrachinotus spp. ) of south Atlantic coast of the United States. Fish. Bull. 62:189-222. Finucane, J. H. 1969. Ecology of the pompano iTrachinotus carolinus) and the permit (Trachinotiis falcatus) in Florida. Trans. Am. Fish. Soc. 95:478-486. Reiser, T.E. 1996. Growth of silver hake within the U.S. continental shelf ecosystem of the northwest Atlantic Ocean. J. Fish. Biol. 48:1059-1073. 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. NOAA Tech. Rep. NMFS 36. James, G. D. 1984. Trevally, Caranx georgianus Cuvier: age determina- tion, population biology, and the fishery. N. Z. Ministry Agr. Fish. Fish. Res. Bull. 25, 50 p. Manooch, C. S., IH, and J. C. Potts. 1997. Age, growth and mortality of greater amberjack from the southeastern United States. Fish. Res. 30:229-240. Quintero-Hunter, L, H. Grier, and M. Muscato. 1991. Enhancement of histological detail using metanil yellow as counterstain in periodic acid SchifTs hematoxylin staining of glycol methacrylate tissue sections. Biotech- nol. Histochem. 66:169-172. Ravaglia, M. A., and M. C. Maggese. 1995. Melano-macrophage centers in the gonads of the swamp eel, Synbranchus marmoratus Bloch, (Pisces, Syn- branchidae): histological and histochemical characteriza- tion. J. Fish Dis. 18:117-125. Robins, C.R. 1992. American nature guides to saltwater fish. Smith- mark Publ., Inc., New York, NY. 192 p. Secor, D. H., J. M. Dean, and E. L. Laban. 1992. Otolith removal and preparation for microstructural examination. In Otolith microstructure examination and analysis (D. K. Stevenson and S. E. Campana, eds. ), p. 19-57. Can. Spec. Publ. Fish. Aquat. Sci. 117. Sudekum, A. E., J. D. Parrish, R. L. Radtke, and S. Ralston 1991. Life hustory and ecology of large jacks in undisturbed, shallow, oceanic communities. Fish. Bull. 89:493-513. von Bertalanffy, L. 1957. Quantitative laws in metabolism and growth. Q. Rev. Biol. 2:217-231. Wallace. R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte gi-owth in tele- osts. Am. Zool. 21:325-343. 35 Abstract-A total of 1784 legal-size (>35G nun TL) hatchery-produced red drum (Sciaenops ocellatus) were tagged and released to estimate tag-reporting levels of recreational anglers in South Carolina (SC 1 and Georgia ( GAl. Twelve groups of legal-size fish (-150 fish/ group) were released. Half of the fish of each group were tagged with an external tag with the message "reward" and the other half of the fish were implanted with tags with the message "$100 reward."These fish were released into two estuaries in each state (n=4); three replicate groups were released at different sites within each estuary (/i = 12). From results obtained in previ- ous tag return experiments conducted by wildlife and fisheries biologists, it was hypothesized that reporting would be maximized at a reward level of $100/tag. Reporting level for the "reward" tags was estimated by dividing the number of "reward" tags returned by the number of "$100 reward" tags returned. The cumulative return level for both tag messages was 22.7 (±1.9)9; in SC and 25.8 (±4.1)% in GA. These return levels were typical of those recorded by other red drum tagging pro- grams in the region. Return data were partitioned according to verbal survey information obtained from anglers who reported tagged fish. Based on this partitioned data set, 14.3 (±2.1)9; of "reward" tags were returned in SC, and 25.5 (±2.3)9, of "$100 reward" tags were returned. This finding indicates that only 56.79; of the fish captured with "reward" tags were reported in SC. The pattern was similar for GA where 19.1 ( + 10.6)9, of "reward" mes- sage tags were returned as compared with 30.1 (±15.6)9; for "$100 reward" message tags. This difference yielded a reporting level of 639; for "reward" tags in GA. Currently, 509; is used as the estimate for the angler reporting level in population models for red drum and a number of other coastal finfish species in the South Atlantic region of the United States. Based on results of our study, the commonly used reporting estimate may result in an overestimate of angler exploitation for red drum. Tag-reporting levels for red drum (Scioenops ocellatus) caught by anglers In South Carolina and Georgia estuaries* Michael R. Denson Wallace E. Jenkins Marine Resources Research InsKtute South CaroNna Department of Natural Resources 217 Ft Johnson Rd Charleston, South Carolina 29422-2559 E mail address (for W. E Jenkins, contact autlior) lenkinswigimrd dnr.slale sc.us Arnold G. Woodward Coastal Resources Division Georgia Department of Natural Resources 1 Consei^ation Way Brunswick, Georgia 31523 Theodore I. J. Smith Marine Resources Research Institute South Carolina Department of Natural Resources PO Box 12559 217 Ft. Johnson Rd. Charleston, South Carolina 29422-2559 Manuscript accepted 1 August 2001. Fish. Bull. 100:35-41 (2002). There are major marine recreational fisheries along the south Atlantic and Gulf of Mexico coasts of the United States that target red drum, Sciaenops ocellatus (Matlock, 1986a; 1986b). Dur- ing the late 1980s, overexploitation of red drum in many states resulted in the closure of commercial fisheries in most states and in the imposition of creel and size limits on catch of rec- reational anglers (McGurrinM Concur- rently, studies were initiated in a num- ber of coastal states to gain a better understanding of red drum life history and to attempt to estimate exploita- tion rates. These investigations relied heavily on the use of fishery-dependent, mark-recapture studies to obtain the data necessary for creating a robust population model (McGurrin^). Generic population models have been developed by using mark-recapture studies to estimate expected number of animals that survive and are re- captured from a year class within a giv- en year (Brownie et al., 1985). Pollock et al. (1991) emphasized the need to modify tag recovery models in which data from multiyear tagging studies were used and suggested incorporat- ing variables for postmarking survival and for reporting to estimate the re- capture component of the model more accurately. The current model used to estimate recovery (recapture) rates of tagged fish (0) includes a number of variables in an attempt to accurately account for what happens in nature {9=5 km). At each site, fish were released individually approximate- ly every 20 meters along the edge of the salt marsh to min- imize the possibility of schooling behavior and subsequent multiple captures by individual anglers. A total of 1774 fish were tagged and released during the project. Approximately 150 fish were released at each stocking site within each estuary (Table 1 1. Equal num- bers of fish released at each site contained "reward" or "$100 reward" tags. Fish were released into Charleston Harbor, SC, and St Simons Sound, GA, during the fall of 1996 and into Calibogue Sound. SC, and Wassaw Sound, GA, during late spring and early summer 1997 (Table 1, Fig. II. The expiration date for "$100 reward" tags de- ■vy y Charleston Harbor Calibogue Souna Wassaw Sound "^ Atlantic Ocean '\'^' St, Simons Sound L. FL + 40 80 Kilometers Figure 1 Map of coastal South Carolina (.SO, Georgia (GAi, and north Florida (FL) showing the location of each estuary where tagged red drum were released during the reward study. ployed in fall 1996 was 31 March 1997, and for spring and summer 1997 releases, the expiration date was 31 Decem- ber 1997. Neither the study nor the releases were publi- cized in any way other than by the normal information provided by ongoing tagging programs in each state. Cap- tured tagged fish were reported directly to the respective Department of Natural Resources in each state. Partici- pants who returned tags inscribed with "reward" received a prize that would normally be awarded by each agency (e.g. T-shirt or hati and those reporting a "$100 reward" tag received a state-issued check for that amount. Our study was based on two assumptions: 1) $100 was an adequate incentive to maximize reporting (assumed -lOC^'f ) of captured tagged fish; 2) the quotient of returns (the number of "reward"-inscribed tags divided by the re- turns of "$100 reward" tags) would yield the angler report- ing level (A) for the standard "reward" tag. Tags were re- turned in either of two ways; phone message or mail. All anglers who reported tags were later interviewed. During the interviews respondents were asked to confirm their reporting information and to express their attitudes and 38 Fishery Bulletin 100(1) Table 1 Cumulative data on release locations and stocking dat es for fish , and both number of tags released and returned for each reward | message. Release location Stocking date Tag " Reward" "$100 reward" No. released No. returned No released No. returned Charleston Harbor site 1 31 Oct 1996 75 16 75 21 site 2 31 Oct 1996 75 18 75 21 site 3 31 Oct 1996 75 18 75 16 St. Simons Sound site 1 13 Nov 1996 75 10 75 17 site 2 13 Nov 1996 74 11 74 11 site 3 13 Nov 1996 75 10 75 15 Wassaw Sound site 1 8 May 1997 73 31 73 42 site 2 8 May 1997 75 23 75 29 sites 8 May 1997 68 10 68 18 Calibogue Sound site 1 5 Jun 1997 75 9 75 21 site 2 9 Jul 1997 73 19 73 23 site 3 10 Jul 1997 74 9 74 12 opinions about the reporting procedure. All participants were asked the same questions from a standardized sur- vey script. During the interview no information was pro- vided to the anglers about the study design. For statistical analysis each release site was treated as a replicate. By nesting site within estuary, within state, differences associated with each site, estuary, and state could be treated in the analysis to assess influence of the reward messages. The study design was a 2x2 factorial de- sign (state and reward) with three levels of nesting (state, estuary, and site) (Table 1). Owing to differences in growth rates, insufficient numbers of legal-size fish were available to stock all estuaries during the same month. Thus one estuary in each state was stocked in the fall of 1996 and the remaining estuaries were stocked the following spring and summer However, each stocking group was available for capture during the fall season when fishing pressure is heaviest (Wenner'). Percent return data were arcsine square-root transformed prior to analysis. Return data were analyzed by using a two-way analysis of variance ( ANOVA) with significance determined at P<0.05. The ini- tial analysis examined all reported or "cumulative" data. The data were then partitioned in two additional ways: by single returns and survey data. 'Wenner, C. 1997. Personal commun. South Carolina Depart- ment of Natural Resources, 217 Ft. Johnson Rd. Charleston, SC 29422-2559. Single returns This data set was the most restrictive. The assumption was that the partitioned data would be free of any poten- tial bias associated with captures of multiple fish, or with monetary rewards or interactions with project staff Survey data The data were partitioned according to the angler's answers during the interview to determine whether the inducement of a $100 dollar reward changed his or her reporting behavior This data set included all tags reported individually, all tags of the same message reported as mul- tiples, and all $100 tags. However, it excluded "reward" tags in instances where answers during the interview sug- gested that the angler's behavior had been changed by capturing a fish with a "$100 reward" tag. Mean data for each of these analyses were reported with standard errors. Results Nearly 95% of tags that were returned were reported within 160 days after release of fish. More fish with "reward" tags were reported than those with "$100 reward" tags in one of the 12 release sites. Overall in SC, 151 anglers reported capture of 203 fish with tags. Anglers reported capture of 1-9 red drum per trip. One hundred Denson et a\ Tag-reporting levels for Sciaenops ocellatus in Sorith Carolina and Georgia estuaries 39 and nineteen anglers in SC (79.0'7r of total anglers) reported only one tagged fish during the study. In GA, 184 anglers reported capture of 226 tagged fish. Single reports in GA represented 80.4''; (;( = 148i of the total catch of tagged fish. The overall return level for all fish reported in SC (22.7 [±1.8]%) was not significantly different from that in GA (25.8 [±4.1]%) (P=0.8129. F=0.67) (Table 2). For the cumulative data, no significant differences were detected between "$100 reward" (27.8 [±3.3]%l and "reward" tags (20.8 [±2.7]%) (P=0.0724, F=12.33l (Table 2). There were also no statistical differences in the cumulative data among the estuaries within states (P=0.0604. F=4.07) (Table 2) and no detectable interaction between state and reward or reward and estuary within states, from the high variability in the cumulative data among estuaries and sites (52.5% and 47.5% of total variation, respectively). Single returns To further restrict the potential for bias caused by inter- action of different reward messages or caused by the project biologist, capture reports were partitioned to include instances where an angler returned only one tag during the entire study. Overall, no significant differ- ences (P=0.1215,F=6.76) were detected between the single returns of "reward" (11.6 [±1.11% ) and "$100 reward" ( 15.0 [±2.5]%) treatments within SC. This was also the case in GA (P=0.1215, F=6.760 where 15.1(±2.9)% of "reward" tags were returned, as compared with 17.6 (±2.7)% for "$100 reward" tags (Table 3). In addition, when data were compared between states, no differences were detected (P=0.6152, F=0.35). However, when single returns among estuaries were compared, Wassaw Sound in GA (Fig. 1) yielded significantly higher returns (P=0.0126, P=7.95) than any of the other estuaries where fish were released (Table 3 1. Survey data In SC, 52% of respondents indicated that they had previ- ously caught tagged fish. Of those, several anglers admit- ted that they had not routinely reported tags. Additionally, others ( 16% ) indicated that they would not have reported the tag if it had not been worth $100. In one extreme case an angler who reported six "$100 reward" tags and an equal number of "reward" tags at once, indicated that he would not have turned in an individual "$100 reward" tag because in his words "he did not need the money." In GA, 29% of anglers had caught a tagged fish prior to the study; however only 7 ( 5% ) said that they would not turn in tags worth less than $100. In light of this infor- mation, the return data were partitioned to eliminate po- tential bias that would result from encountering a "$100 reward" tag. This partitioned data set revealed that sig- nificantly fewer (P=0.0310, F=30.81) unbiased "reward" tags (14.3 [+2.1]%) were returned in SC than "$100 re- ward" tags (25.5 [±2.3]%) (Table 4). This was also true in GA, where 19.1(±4.3)% of "reward" tags were unbiased re- turns, as compared with 30.1 (±6.4)*^"^ of "$100 reward" tags (P=0.0310, F=30.81) (Table 4). Table 2 Cumulative mean return level C/r) and standard error for red drum tagged with one of two reward messages ("reward" or "$100 reward"). No significant differences were detected between reward message, estuary, or state. Release location Charleston Harbor Calibogue Sound South Carolina (mean) St. Simons Sound Wassaw Sound Georgia (mean) Overall mean Return level "Reward" "$100 reward" (9f) CJl 23.1 ±0.9 25.8 ±2.2 16.7 ±4.7 25.2 ±4.6 19.9 ±2.6 25.5 ±2.3 13.9 ±0.6 19.2 ±2.2 29.3 ±8.0 40.9 ±9.0 21.6 ±5.0 30.1 ±6.4 20.8 ±2.7 27.8 ±3.3 Table 3 Mean tag return level (% ) and standard error for red drum tagged with one of two reward messages ("reward" or "$100 | reward"). There were no significant differe nces in return levels by reward message within or among estuaries with the e.xception of those from Wassaw Sound which were sig- nificantly higher (P<0.05 noted by *) for both reward mes- sages than any other estuary. SC = South Carolina; GA = Georgia. Tag message "Reward" '$100 reward" Release location ('7c) (%) Charleston Harbor, SC 13.3 ±1.6 15.1 ±5.3 Calibogue Sound, SC 9.9 ±1.0 14.8 ±2.0 South Carolina (mean) 11.6 ±1.1 15.0 ±2.9 St. Simons Sound, GA 9.9 ±1.6 13.0 ±1.6 Wassaw Sound, GA 20.2 ±3.8* 22.1 ±3.7* Georgia (mean) 15.1 ±2.9 17.6 ±2.7 Overall mean 13.3+1.6 16.3 ±1.8 Discussion Overall return levels for the tagged fish released in our study were similar to levels of angler return for red drum in each states fishery-dependent tagging programs (Wenner^, Woodward''). Because of high variability within estuaries, there were no significant differences between returns of "reward" and "$100 reward" according to the analysis of cumulative return data. The high variability Woodward. A. G. 1997. Personal commun. Georgia Depart- ment of Natural Resources, 1 Conser\-ation Way, Brunswick, GA 31523. 40 Fishery Bulletin 100(1) Table 4 Mean return level {'?/ ), standard error, and range for un biased data lad ustments based on verbal interviews) fo ■ red drum tagged with one of two reward messages ( "reward" or "$100 rew ard"). Return data for the "$100 rewai d" message were s gnificantly higher (P<0.05 ) for each estuary, state, and overall than those for the "reward' message. Release location Tag message Unbiased reporting Mean level' (rn Range "Reward" Ci I $100 reward"!'*) Charleston Harbor 17.3 ±1.3 2.5.8 ±3.9 67.1 57-78 Calibogue Sound 11.3 ±6.3 25.2 ±8.0 44.8 19-67 South Carolina (mean) 14.3 ±5.2 25.5 ±5.6 56.7 — St. Simons Sound 11.7 ±2.1 19,2 ±3.9 60.9 41-82 Wassaw Sound 26.5 ±10.4 40.9 ±15.7 64.8 56-79 Georgia (mean) 19.1 ±10.6 30.1 ±15.6 63.4 — Overall mean 16.7 ±6.2 27.8 ±11.5 60.1 19-82 ' Example: Charleston Harhnr: -$100 reward" tags reported - 00' r: 17. ■3/2.'") 8 = STl'r reportinj^ level for "rew; ird" ta^s. between sites within the same estuary was unexpected. In addition, variation between estuaries in the same state made comparisons between states difficult. However, "reward" tags were returned less often than "$100 reward" tags from 11 of the release sites in the unpartioned data set. After identifying and excluding possible sources of bias, we found that there were statistically significant dif- ferences between reporting level of "reward" and "$100 reward" tags in all areas (Table 4). The range of 19-82''i in levels of reporting between sites was more variable than anticipated (Table 4). Removal of the suspected biased anglers from the data set resulted in a mean unbiased reporting level of 67.1'~f in Charleston Harbor and 44.8'f in Calibogue Sound (Table 4). Unbiased reporting in GA was somewhat higher than in SC (63.4'7f vs. 56.7'"*). The fact that significant differences were found only after biased angler data were removed from the data set illus- trates that a small number of skilled anglers can have an effect on fisheries-dependent data. Their failure to report tags may be due to a lack of novelty in encountering tagged fish, or to insufficient reward incentives (having already received a number of t-shirts, fishing caps, etc. I. These data suggest that use of noncash rewards is ben- eficial only for the first time an angler catches a tagged fish and decreases as anglers catch additional tagged fish. Further repeated exposure to tagging programs within each state eventually results in angler ambivalence and reduced cooperation. This indifference is of particular con- cern with the use of a constant regional reporting rate as described by Hocnig et al. (1998). A decreasing rate of tag return could be mistaken for lower hai-vest, reduced fish- ing effort, poor survival, or increased population size. Lack of differences in reporting levels between "reward" and "$100 reward" in the single-return (one fish) parti- tion of data (Table 3) confirms that anglers who capture many tagged fish per trip or per season (who were omitted from this data set) significantly influence reporting. Sin- gle return-data also suggest that anglers who catch fewer fish (tagged or not tagged) are more likely to report cap- tures of tagged fish regardless of reward amount. Consid- ering the impact a few skilled anglers can have on tag re- porting estimates, these results demonstrate the need for further evaluation of the interaction between tagging pro- grams and angler behavior. The 50*7^ reporting level cur- rently used by managers is approximately a IT^'i under- estimate (.50/60=0.83) of actual reporting recorded for the red drum fishery in SC and GA. Continuing to use the 50'7( reporting estimate for this fishery will be more conserva- tive than using the actual reporting level (A) to calculate angler recovery rate (H). Reporting was also extremely site specific, and application of data from one site to a broader area may not be appropriate. Ideally tag-recapture models should be weighted by site-specific reporting information to account for this variability which could be accomplished by regular deployment of high value (>$100) reward tags within each system to gauge angler reporting. Even if of- fering a $100 does not result in lOO*^? reporting, as Nichols et al. ( 1991) suggested, it may yield the highest reporting possible with monetary incentives, meaning that our unbi- ased reporting may have been slightly overestimated. Re- gardless, this approach is still more accurate than that of adopting a regional average. Our results emphasize that researchers need to conduct controlled tag reward studies regularly and also to offer sufficient rewards in order to avoid under reporting. Furthermore, tag reports must be followed up with angler interviews to determine attitudes and give managers an opportunity to remove bias from the data (Reinecke et al., 1992. Zale and Bain, 1994, Pegg et al.,1996). Acknowledgments We would like to thank the staff of the Inshore Fisheries Sections of the SCDNR and GADNR for tagging, distri- bution of fish and tag collection and processing. We espe- cially thank John Fortuna and Carolyn Belcher for their assistance with statistical design and data analysis. We Denson et al : Tag reporting levels for Sdaenops ocellatus in South Carolina and Georgia estuaries 41 also thank Charlie Bridghain and Allan Hazel, for produc- tion, maintenance, and transportation offish, and Charlie Wenner, for reviewing this manuscript and providing valu- able insights during the project. The study was funded in part by USDOC, NMFS the Saltonstall-Kennedy Pro- gram gi-ant #A67FD0031 and NA77FD0062 and the state of South Carolina. Literature cited Brownie, C, D. R. Anderson, K. P. Burnham, and D. S. Robson. 1985. Statistical inference from band recovery data: a band- book. 2nd ed. U.S. Fisb and Wildl. Sei-v. Resour. Publ. 156, 305 p.. Butler. L. 1962. Recognition and return of trout tags by California anglers. Oalif Fish Game 48:5-18. Conroy, M. J., and W. W. Blandin. 1984. Geographical and temporal differences in band report- ing rates for American black ducks. J. Wildl. Manage. 48:23-36. Henny. C. J., and K. P. Burnham. 1976. A reward band study of mallards to estimate band reporting rates. J. Wildl. Manage. 40:1-14. Hoenig. J. M., N. J. Barrowman. K. H. Pollock, E. N. Brooks, W. S. Hearn, and T. Polacheck. 1998. Models for tagging data that allow for incomplete mixing of newly tagged animals. Can. J. Fish. Aquat. Sci. 55:1477-1483. Jenkins, W. E., M. R. Denson, and T. I. J. Smith. 2000. Determination of angler reporting level for red drum (Sciaenops ocellattif:) in a South Carolina estuary. Fish. Res. 44:273-277. Matlock, G. C. 1981. Non-reporting of recaptured tagged fisb by saltwater recreational boat anglers in Texas. Trans. Am. Fish. Soc. 110:90-92. 1986a. Estimate of the number of red drum anglers in Texas. N. Am. J. Fish. Manage. 6:292-294. 1986b. Estimating the direct market economic impact of sport angling for red drum in Texas. N. Am. J. Fish. Manage. 6:490-493. Murphy. M. D., and R. G. Taylor 1991 Preliminary study of the effect of reward amount on tag-return rate for red drum in Tampa Hay, Florida. N. Am. J. Fish. Manage. 11:471-474. Nichols. J. D.. R. J. Blohm. R. E. Reynolds, J. E. Mines, and J. P Bladen. 1991. Band reporting rates for mallards with reward bands of different dollar values. J. Wildl. Manage. 55:119-126. Pegg, M. A., J. B. Layzer, and P. W. Bettoli. 1996. Angler exploitation of anchor-tagged saugers in the lower Tennessee River N. Am. J. Fish. Manage. 16:218- 222, Pollock. K. H., J. M. Hoenig and C. M. Jones. 1991. Estimation of fishing and natural mortality when a tagging study is combined with a creel survey or port sam- pling. Am. Fish. Soc. Symp. 12:423-434. Rawstron. R. R. 1971. Non-reporting of tagged white catfish, largemouth bass, and bluegills by anglers at Folsum Lake, California. Calif Fish Game. 57:246-252. Reinecke. K. J., C. W. Shaiffer. and D. Delnicki. 1992. Band reporting rates of mallards in the Mississippi alluvial valley J. Wildl. Manage. 56:526-531. Roberts Jr., D. E., B. V. Harpster. and G. E. Henderson. 1978. Conditioning and induced spawning of the red drum (Sciaenops osellatiis ) under varied conditions of photoperiod and temperature. Proceed. World Aqua. Soc. 9:311-332. Ross. J. L.. T. M. Stevens, and D. S. Vaughan. 1995. Age, growth, and reproductive biology of red drums in North Carolina waters. Trans. Am. Fish. Soc. 124:37-54. Yeager. D. M.. and M. J. Van Den Avyle. 1979. Rates of angler exploitation of largemouth. small- mouth, and spotted bass in Central Hill Reservoir. Ten- nessee. Proc. Annu. Conf Southeast. Assoc. Fish Wildl. Agencies 32:449-458. Zale.A. v., andM.B. Bain. 1994. Estimating tag-reporting rates with postcards as tag surrogates. N. Am. J. Fish. Manage. 14:208-211. 42 Abstract— Mayan cichlids ^Cichlasoma urophthalniiis) were collected monthly from March 1996 to October 1997 with hook-and-line gear at Taylor River. Flor- ida, an area within the Crocodile Sanc- tuary of Everglades National Park, where human activities such as fish- ing are prohibited. Fish were aged by examining thin-sectioned otoliths, and past size-at-age information was gen- erated by using back-calculation tech- niques. Marginal increment analysis showed that opaque gi'owth zones were annuli deposited between January and May The size of age-1 fish was esti- mated to be 33-66 mm standard length (mean=45.5 mm) and was supported by monthly length-frequency data of young-of-year fish collected with drop traps over a seven-year period. Mayan cichlids up to seven years old were observed. Male cichlids grew slower but achieved a larger size than females. Growth was asymptotic and was mod- eled by the von Bertalanffy growth equa- tion L,=263.6( l-exp[-0. 166( ?-0.001 1] ) for males (/•'''=0.82, ;i=581 ) and Z,,=21.5.6 (l-e.\p|-0.197(r-0.058ll I for females !;■-= 0.77, n =639). Separate estimates of total annual mortality were relatively con- sistent 1 0.44-0.60 ( and indicated mod- erate mortality at higher age classes, even in the absence of fishmg mortality. Our data indicated that Mayan cichlids grow slower and live longer in Florida than previously reported from native Mexican habitats. Because the growth of Mayan cichlids in Florida periodi- cally slowed and thus produced visible annuli, it may be possible to age intro- duced populations of other subtropical and tropical cichlids in a similar way. Age, growth, and mortality of the Mayan cichlid (Cichlosoma urophthalmus) from the southeastern Everglades Craig H. Faunce Estuanne and Marine Research Group Tavernier Science Center, Audubon of Florida 115 Indian Mound Trail Tavernier, Florida 33070 E-mail address cfaunceaaudubonorg Heather M. Patterson Florida Manne Research Institute Florida Fish and Wildlife Conservation Commission 100 Eighth Avenue SE St Petersburg, Flonda 33701-5095 Jerome J. Lorenz Estuanne and Manne Research Group Tavernier Science Center, Audubon of Flonda 115 Indian Mound Trail Tavernier. Florida 33070 Manuscript accepted 1 August 2001. Fish. Bull. 100:42-50 (2002). The Mayan cichlid, Cichlasmna uroph- thalmus (Giinther),is native to the fresh and brackish waters of the Atlantic slope of Central America from Mexico to Nicaragua (Miller, 1966), where it is exploited commercially in artesanal fisheries and aquaculture (Martinez- Palacios and Ross, 1992). The first collections of the Mayan cichlid in the United States were made in 1983 from a freshwater habitat and a man- grove creek within Everglades National Park, Florida ( Loftus, 1987 ). Although it remains unknown how or where Mayan cichlids first entered Florida waters, there is evidence that the discovery of this exotic fish was made shortly after their introduction (Loftus, 1987). Since their discovery, Mayan cichlids have expanded their range to include a variety of habitats from Naples (26°05'N, 81°48'W) to West Palm Beach (26°45'N,80''04'W). The species remains abundant in the man-made freshwater canals and estuarine mangrove habi- tats of the region (Trexler et al., 2000). The introduction of the Mayan cich- lid into southern Florida has had both economic and ecological significance. This species supports a small sport fish- ery because it is edible, attractive, and aggressively takes baits and artificial lures (Shafland, 1996). Anglers, howev- er, have mixed feelings towards this fish because it readily takes artificial baits and fights hard on light tackle, and it can interfere with the pursuit of larger gamefishes, such as the common snook (Centropomus undecinialis). In some ar- eas, the Mayan cichlid is the most com- mon fish caught by recreational anglers and is targeted by subsistence anglers. There is concern, however that the in- teraction between Mayan cichlids and native fishes could alter the ecology of the Everglades and Floi'ida Bay region. Although the role of Mayan cichlids as food for higher trophic-level fishes has not been quantified, they themselves are omnivorous and prey upon native fish- es (Martinez-Palacios and Ross, 1988; Howard et al.^). Previous studies of the Mayan cichlid have focused almost entirely on its suit- ability for aquaculture in Mexico (e.g. ' Howard, K. S., W. F Loftus, and J. C. Trexler. 199.5. Seasonal dynamics of fishes in arti- ficial culvert pools in the C-111 basin, Dade County, Florida. Final Rep. CA5280- 2-9024. 34 p. and append. South Florida Research Center, Everglades National Park, Homestead, FL. Faunce et a\ Age, growth, and mortality of Cichlasoma uiophtha/nnis 43 Map of southeastern HC'=Highway Creek i. Martinez-Palacios and Ross, 1986; Flores-Nava et al., 1989; Ross and Bt'veridge, 1995) and on the po- tential for range expansion in the United States I e.g. Stauffer and Boltz, 1994). Few studies have ad- dressed the life history of the Ma- yan cichlid, and only scant infor- mation e.xists on the age structure and growth rate of this species. From the seasonal length-frequen- cy distributions for Celestun La- goon, Mexico, Martinez-Palacios and Ross (1992) concluded that Mayan cichlids from 70 to 130 mm standard length had complet- ed their first spring and were re- productively active, whereas in- dividuals from 131 to 200 mm standard length had entered their second reproductive year. Observ- ing no fish >200 mm, Martinez-Pa- lacios and Ross (1992) concluded that the population of Mayan cich- lids in the lagoon comprised fast- growing fish with one, or two (rarely), reproductive seasons in their lifetimes. Aging of Mayan cichlids using a validat- ed method is needed to determine the accuracy of previ- ously reported age and growth estimates and to compare the age structure between populations from Mexico and Florida. Here we provide a first account of the age, growth, and mortality of the Mavan cichlid from Florida waters. Methods Mayan cichlids were collected from the dwarf mangrove forests of southeastern Florida. This habitat is dominated by small (0.5-2.0 m tall) red mangrove trees (Rhizophora mangle) in an expansive, seasonally inundated wetland of typically shallow water (average maximum depth=30 cm). These mangroves increase in canopy width and height nearer to continuously inundated deeper creeks. The system is inundated mostly by fresh water during July-February but becomes more saline ( 10-35"^^? ) during the dry season (March-June). Cichlids <65 mm standard length (SL) were collected by using drop traps (Lorenz et al., 1997) to determine when Mayan cichlids first recruit. Drop-trap samples were col- lected every six weeks from August 1990 to September 1996 at Highway Creek, Joe Bay, and Taylor River (Fig. 1 ). Larger cichlids (>65 mm SL) were collected by using hook- and-line gear comparable to that used in other studies (Martinez-Palacios and Ross, 1992). Hook-and-line collec- tions were conducted monthly from March 1996 to October 1997 in Taylor River, a major freshwater distributary of the Everglades emptying into northeastern Florida Bay. Each fishing effort continued until approximately 40 fish were obtained. Fish collected by both methods were measured (SL and total length |TL|, mm), weighed to the nearest Figure T Florida showing sampling locations (TR=Taylor River, .JB=Joe Bay, 0.1 gram, and their sex was determined macroscopically when possible (Faunce and Lorenz, 2000). Fish captured during 1994—97 were used for age-and-growth analyses. All lengths reported hereafter are standard lengths. Sagittal otoliths were removed, blotted dry, and stored in vials until they were sectioned. The left sagitta, unless broken, was used for age determination. Otoliths were sec- tioned by using a low-speed Beuhler Isomet saw with dia- mond blade. Three or four 0.5-mm thick transverse sections, one through the core, were cut and mounted on microscope slides with Histomount '■'^' adhesive and allowed to dry. Sag- ittae from fish <100 mm were embedded in Spurr, a high- density plastic medium (Secor et al, 1992) and a 1-2 mm thick transverse section containing the otolith core was then cut. The sections were mounted on a microscope slide with Crystal Bond^-^' 509 adhesive, and polished with wet and dry sandpaper of grit sizes 220-2000 until growth rings were visible. A polishing cloth with 0.05-gamma alu- mina powder was used to remove scratches. A standardized protocol for interpreting otolith growth zones was followed. When viewed with reflected light, the transverse sections of Mayan cichlid otoliths had two dis- tinct regions; 1) an "inner region" extending from the core to the first visible opaque zone (ring), and 2) an "outer re- gion" extending from the first visible opaque zone to the edge of the otolith (Fig. 2). The inner region was typically more opaque than the outer region and sometimes con- tained a visible growth zone or numerous check marks, or both. LTnfortunately, these marks were difficult to inter- pret, inconsistent between sections from individual fish, and in many cases absent altogether Consequently, we did not count any marks from the inner region in our age es- timations. However, the translucent appearance of the out- er region of the otolith made it possible to count distinct, separate, opaque rings when present. The number of rings 44 Fishery Bulletin 100(1) Figure 2 Transverse section of a six-year-old Mayan cichlid tCichlasuma urophthalmiis) otolith showing the outer region (ORl, inner region ( IRi, and five visible annuli ( 1-51. Note that the first annulus (1) corresponds to the fish's second year of growth. A ring correspond- ing to the first year of growth was not consistently visible and was therefore not counted. Measurements for marginal-increment analysis were made on an axis adjacent to the sulcal ridge from the core (C) to the dorsolateral margin (DLM). Scale bar=500 /im. on each otolith section was counted independently by two readers using compound microscopes, and the results were compared. If there was a discrepancy in the counts be- tween readers, the section was re-examined. If a consensus could not be reached between the readers after the third reading, the otolith was excluded from the study. Linear regi-ession was used to determine the relation- ship of otolith radius to standard length and marginal- increment analysis was used to determine the periodicity of ring formation. Distance from the core to the proximal edge of each ring and to the dorsolateral margin of the oto- lith (otolith radius) was measured (Fig. 2). Measurements were made with a digital-image processing system along an axis adjacent to the sulcal ridge. The distance from the outermost ring to the dorso-lateral margin (i.e. mar- ginal increment=MI) was plotted by month (marginal in- crement analysis). Because the majority of Mayan cichlids in Taylor River spawn during May and June (Faunce and Lorenz, 2000), and ring formation occurred during Janu- ary-May, we assigned each fish a biologically realistic me- dian hatching date of 1 June. Fish collected prior to 1 June that had not yet formed a new opaque ring (=high MI), and all fish collected after 1 June, were assigned a yearly age equal to their ring count. Fish collected before 1 June that had already formed a new opaque ring (i.e. an "early" ring) were assigned a yearly age of one less than their ring count. To compare the timing of ring formation between age groups, marginal-increment analysis was performed on pooled ages 0-3 and 4-7 because our monthly sample sizes for individual age classes were insufficient for this analysis. We used linear regression to determine the relationship between standard length and total length for all hook- and-line caught fish. The relationship between standard length and total weight was calculated separately for each sex with logjy-transformed data. Analysis of covariance (ANCOVA) was used to test for significant differences be- tween the slopes and intercepts of male and female length- weight relationships. Length-frequency distributions for males and females caught with hook-and-line were com- pared by using the Mann-Whitney rank sum t-test. Non- linear least squares procedures (SAS, 1989) were per- formed on final obsen'ed age-at-length data to estimate parameters for the von Bertalanffy gi'owth equation L, =L..(l-exp\-Kit-t„)]), where L, = the standard length (mm); L = the asymptotic length; K = the Brody growth coefficient; t = the age (years); and tfy = the age at zero length (von Bertalanffy, 19.57). Faunce et al.: Age, growth, and mortality of Cichlasoma urophthalmus 45 To increase the number of observations used for fitting the prowtli model, we back-calculated past size-al-age information for each sexed fish using the Fraser-Lee melh(ul lollowing Devries and Frie (1996); L, =[(L,, -a)/S,]S, +a, where L, = the back-calculated length of fish when the /"' increment was formed; L^ = the length of fish at capture; S^ = the otolith radius at capture; and S, = the otolith radius at the ;''' increment. The slope, {L^-a)IS^. was calculated for each fish as the slope of a line connecting two points: (S^ , L, ) and (0, a). The \'-intercept parameter a was determined from the relationship between otolith radius and standard length for all fish, and should approximate the fish length at which otolith radius equals zero (Devries and Frie, 1996). Because we could not accu- rately determine the sex of each fish <70 mm. fish whose sex could not be determined were included in the fitting of both male and female growth cui-ves. Catch cui-\'es were analyzed with two methods to determine annual mortality rates for Mayan cichlids. Sun'ival rate (S) and its respective variance were es- timated by using the empirical abundance data (Rob- son and Chapman, 1961) and the regression of the natural logarithm of year-class abundance (Ricker, 1975). The instantaneous rate of mortality (Z) was derived from the relationship Z=-ln(e^). Total annual mortality (A) was computed as A = 1-S. The age at full recruitment to the hook-and-line gear based on our catch cur\'e was determined to be four years. Results The fragile nature of Mayan cichlid otoliths caused a high proportion (54'f ) to be lost during the cutting process. However, only five of the 391 successfully sectioned otoliths were discarded because a consen- sus between readers could not be reached. A newly formed opaque ring was generally obsei'V'ed in fish captured Jan- uary-May, and the mean monthly marginal increment reached a single yearly minimum in June for all age classes examined (Fig. 3). These data indicate that the opaque rings observed were annuli. The growth of young-of-year Mayan cichlids collected with drop traps could be followed by the progression of modal length from monthly length frequencies. Newly re- cruited fish were present in August (mode=10 mm) and grew to a size of 50 mm by June (Fig. 4). An early spawn- ing event in 1993 (senior author, unpubl. data) produced a smaller-size (20 mm) cohort that was obsen'ed in June. Fish with one annulus were much larger (50-149 mm, mean=97 mm) than the size of age-1 fish suggested from our drop-trap data (June mode=50 mm). This information, combined with the presence of marks in the inner region All individuals Figure 3 Monthly mean marginal increment and range for all Mayan cich- lids and pooled age classes 1-3 and 4-7. Note the consistent annual minimum in June. Numbers indicate sample size. of the otolith, led us to conclude that the first annulus in our age estimations was laid down between January and May of the fish's second year of growth, and we added a year to each individual's total age. Length-length and length-weight relationships are giv- en in Table 1. As i-equired by the Fraser-Lee method for back-calculation of size-at-age information, otolith ra- dius and standard length were closely related (SL=131.2x 0R+A.Q2A1, n=37l, r^=0.80). Analysis of covariance did not detect differences between the slopes of length-weight relationships for males and females (F, y.j(,=0.15, P=0.696) but did reveal significant differences between the re- spective intercepts for males and females (F, g3g=4.10, P=0.043). The length of Mayan cichlids at a given age was mod- eled by the von Bertalanffy growth equation (Fig. 5). Pre- dicted lengths fitted well with the final adjusted observed 46 Fishery Bulletin 100(1) 40 30 20 10 0 40 30 20 10 0 '"^f^T^T^T August (n=522) September (n=491) 40 30 - 20 - 10 - 0 I ' I ' I ' I October (0=193) iMMM'!' i^T T-i T 40 30 20 10 - 0 November (n=772) rfMlllf'T'T-T T T 40 -1 30 20 10 0 December (0=714) fiHIv 40 30 20 10 0 m- January (n=597) -T* M*i T-l^ 40-] 30 - 20 10 0 February (n=471) 4llM^M- T-T '!■ 40 -, 30 20 - 10 0 March (n=713) 40 30 - 20- 10 0 April (n = 587) m^^ 40 30 20 10 - 0 I ' I ' I ' I ' T^T'T'I ' I ' I IVIay (n=430) I tMU'I^T I IT 40 -J 30 - 20 - 10 0 June (n=193) iMM'f'H'TT I T 40 -| 30 - 20 - 10 - 0 -■ July No samples I ' I ' I ' I ' I ' I ' I ' I ' I ' I 10 20 30 40 SO 60 70 80 90 100 10 20 30 40 50 60 70 60 90 100 Standard length (mm) Figure 4 Pooled length-frequency histograms for Mayan cichlids collected with drop traps from 1990 to 1996. and back-calculated length-at-age data for males (r'-=0.82, n=581) and females (;'-^=0.77, /!=639). Our observed and back-calculated size of age- 1 fish (mean ±1 standard er- ror=45.5 ±10.11 mm, range=33-68 mm, ti=22} correspond- ed well with the modal length of age-1 fish collected in drop-trap samples (50 mm). Differences in the parameter estimates for the von Bertalanffy growth equation were observed for each sex. Males were larger than females for all ages (Table 2). Although males exhibited a slower growth rate (A") and larger maximum attainable size iL J than females, the von Bertalanffy growth model param- eters were not significantly different between sexes (95% CI, Table 3). Male and female Mayan cichlids up to seven years old were observed. The size of fish examined ranged from 21 to 210 mm (median=127 mm, interquartile range=98 mm, « = 1046). Males ranged from 69 to 210 mm (median=137 mm, inter- quartile range=119 mm, ?;=400), and females ranged from 58 to 190 mm (median=132 mm, interquartile range=115 mm, 71=449) (Fig. 6). The length-frequency distribution for males was significantly larger than that for females (P<0.001). The overall ratio of males to females was 1:1.1. Age-frequency distributions of Mayan cichlids collect- ed by hook-and-line gear suggest that these fish are fully recruited to the fishery at age four (Fig. 7). The majority of males (85.1'7r ) and feinales (81.5%) were 3-5 years old, and there was a significant difference (Mann- Whitney rank sum ^test, P<0.001) in the age-frequency distribution of males (median=3.67 years, interquartile range=2.12) and females (median=4.78 years, interquar- tile range=3.86). Total instantaneous mortality (Z), annu- al survival (S), and annual mortality iA), based on the re- gression of our catch-cui"v'e data, were estimated at 0.57, 0.56, and 0.44, respectively (/■-=0.91, n=3). Robson and Chapman ( 1961) estimates were Z=0.91, S=0.40 (±0.035), andA=0.60. Discussion Transverse otolith sections can be used to precisely age Mayan cichlids from Florida waters. There was a high congruence (98.7%) between the age estimations of each reader Annuli corresponding to years 2-7 were clearly Faiince et a\ Age, growth, and mortality of Cich/nsomn iimphthalrmis 47 Table 1 LeiiKtli-lcngtli. lenKth-vvi'ight, and ololith-r ad us- -standard-len gth regressions for the Mayan cichlid Cichldsonia urophih aintus, from Taylor Slough, Florida. Regi-essions are in the form V = a + bX. TL = = total length (mm); SL = standard length (mm); WT = total weight (g); OR = otolith radius (mini: range = sample standar d length range in r egressions. Values n parentheses are standard | errors. Y a b X n Range (mm) /■- Sexes combined TL 0.6220 1.3067 SL 961 40-210 0.997 (0.2875) (0.00221 SL -0.1281 0.7631 TL 961 40-210 0.997 (0.2202) (0.0013) SL 4.0247 131.2092 OR 371 33-210 0.800 (3.2740) (3.4214) logi„\VT -4.2490 2.9314 log,„SL 847 58-210 0.986 (0.0257) (0.0122) Males log,„UT -9.7958 2.9329 log.oSi 395 84-210 0.986 (0.0874) (0.0178) Females log,„UT -9.7387 2.9232 log„-,SL 444 89-182 0.984 (0,0856) (0.0176) visible on the otoliths. The annulus corresponding to the first year's growth was not consistently clear to the read- ers, which has been observed in thin-sectioned otoliths of other fish species in Florida. Murphy and Taylor (1994) found that the first annulus was visible only in certain individuals of spotted seatrout, Cynoscion nebulosus. Sim- ilarly, Murphy and Taylor (1990) found that the annulus corresponding to the first winter or spring was absent in red drum, Sciaenops ocellatiis. Direct validation of marked otoliths is needed to confirm the presence and location of the first annulus on the otolith of Mayan cichlids. We obsen'ed differences in the growth patterns of males and females that are likely linked to reproduction. Males were larger than females and did not appreciably slow their growth with age. The nearly linear growth of males resulted in a theoretical maximum size (L. i of 263.6 mm, well above the -200 mm commonly observed for this spe- cies (Loftus, 1987; Martinez-Palacios and Ross, 1992; pres- ent study). Larger males are common in riverine and la- goonal populations of tilapias (Cichlidae) and may have a selective advantage during the reproductive season if they can defend a spawning pit or brood against potential pred- ators (Lowe-McConnell, 1982). Because sperm production requires less energy than egg production ( Jalabort and Zo- har, 1982), the slowed growth observed in females com- pared with that for males is likely due to differences in energy budgets during the reproductive season. No significant differences were found by ANCOVA in the slopes of sex-specific length-weight relationships, but there were significant differences in the intercepts of those lines. Because the actual difference between the y-in- tercepts (weight) of each length-weight relationship was <0.001g, we attribute no biological meaning to the statis- Table 2 Average predicted and observed standard lengths ( mn )for male and female Mayan cichlids. Average Standard Age (yrl Predicted observed error n Males 1 40.3 45.5 2.2 22 2 74.4 74.3 1.12 148 3 103.4 102.0 1.24 152 4 127.9 127.1 1.36 139 5 148.7 147.8 1.66 88 6 166.3 173.0 2.56 28 7 181.1 206 1.32 4 Females 1 36.4 45.5 2.16 22 2 68.4 70.6 1.05 156 3 94.7 96.8 1.31 160 4 116.3 119.8 1.4 151 5 134.0 137.0 1.59 102 6 148.6 148.1 1.94 45 7 160.6 151.0 3.22 3 tical difference and consider the length-weight relation- ships for both sexes to be equal. Mayan cichlids in Florida were much smaller at a given age than those reported by Martinez-Palacios and Ross (1992) in Mexico. One-year-olds were 33-66 mm in Florida vs. 70-130 mm in Mexico, and age-2 fish were 44-130 mm 48 Fishery/ Bulletin 100(1) Table 3 Parameter estimates for the von BertalanfTv growth model ( 19571 for Mayan cichlids and associated standard error i SE ) and con- fidence intei'vals (CI). Sex L imm) A' 'i, 'yri U l'~ Males 263.66 0.165 0.001 581 0.82 SE 25.15 0.027 0.124 959, CI 49.28 0.053 0.243 95'^i CI range 214.34-312.95 0.112-0.218 0.242-0.244 Females 215.63 0.197 -0.058 639 0.77 SE 17.33 0.031 0.142 959t CI 33.96 0.061 0.278 95';; CI range 181.67-249.59 0.136-0.258 -0.336-0.220 in Florida and 131-200 mm in Mexico. We found a maxi- mum age of seven yeai's. vvherea.s two years was suggest- ed by Martinez-Palacios and Ross (1992). We offer three explanations for these observed differences in the length- at-age data. First, exploitation rates may differ between 250 Males n = 581 L = 263-6 K = 0 166 'o = 0 0007 (•2 = 0 82 Figure 5 Obsen'ed and predicted lengths at age for male and female Mayan cichlids from the von Bertalanffy gi'owth model. /)=obsei'ved and back-calculated size-at-age information from 148 males and 157 females. study areas. Fish in our study came from the Crocodile Sanctuary within Everglades National Park and had not been exposed to fishing mortality. Fishing for Mayan cich- lids occurs outside of our study area, and heavy exploita- tion can select for faster-growing fish with a shorter life- span (Ricker, 1975 1. Martinez-Palacios and Ross (1992) suggested that their population was over- fished. Second, differences in temperature impact fish growth. Colder winter temperatures in Florida were sufficient to form seasonal marks in the oto- liths of Mayan cichlids and may have caused slower growth than in Mexican populations. Third, the sea- sonal length frequencies of Martinez-Palacios and Ross (1992) were insufficient to accurately identify older year classes. Because growth slows with age, the length-frequency of cohorts corresponding to older age classes can overlap significantly, resulting in erroneously lower age estimates. Future efforts to age Mayan cichlids in Mexico should include thin-sectioned otoliths to evaluate the findings of Martinez-Palacios and Ross (1992). Although the Mayan cichlid has proliferated for over a decade in the natural and man-made habitats surrounding the Everglades, studies are only recent- ly becoming available (e.g. Trexler et al., 2000 1. More introduced fish species are found in Florida than in any other state in the United States, and 13 of the 18 species with established populations are cich- lids (Shafland, 1996). The impact of exotic species on native Florida fishes has been debated: Shafland (1996) proposed no demonstrable effect on native fishes, whereas Courtenay ( 1997 ) argued that lack of available data precludes a determination. Trexler et al. (2000) provided empirical data that support Shafland (1996), concluding that although exotics have been credited with native species extinctions in other ecosystems, native Florida fishes are not specialized or restricted to certain habitats and thus are able to cope with the invasion of exotics. Finding no drastic changes in the native ichthyofauna does not necessarily mean that exotic species do not af- fect indigenous fishes. Exotic species can introduce FaLince et a\ Age, growth, and moitalily of Cichlasoma umphthalmus 49 numerous stresses not easily quantified, e.g. nest prcdation, direct predation, and competition for space (Trexler et al., 2000; senior author, un- puhl. data). These stresses may afTect the pop- ulation dynamics of native fishes by altering their growth rate, increasing mortality, or de- creasing reproductive success. During 1990-99, till' Mayan cichlid population underwent a cycli- cal "boom and bust" pattern of yearly abundance typical of invasive species (Trexler et al., 2000). Why these patterns occur requires a better un- derstanding of the parameters of reproduction, gi'owth. and moi'tality that drive the population dynamics of this species. The presence of the Ma- yan cichlid in the Everglades and Florida Bay es- tuary warrants further research and monitoring efforts in the region to firmly understand the life history of exotics, native fishes, and their role in the ecosystem. Acknowledgments 80 60 40 I 20 n I 20 z 40 Males n = 140 60 • Females n = 449 80 t ^ — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I 0 50 100 150 200 250 Standard length (mm) Figure 6 Length-frt'quency histoprrams for male and female Mayan cichlids col- lected with hook-and-line gear. We would like to thank Roy Crabtree, Daniel Merryman, Connie Stevens, and Rich McBride for their assistance. Ron Taylor and Mike Murphy shared their technical expertise with us, and with their editorial suggestions, Joe Serafy, two anonymous reviewers, and John Merriner greatly improved the manuscript. This project was funded by the U.S. Ai-my Corps of Engineers through cooperative agi-eement 970092 between Everglades National Park and the National Audubon Society. Literature cited Courtenay, W. A., Jr. 1997. Nonindigenous fishes. In Strangers in paradise (D. S. Siberloff, D. C. Schmitz, and T. C. Brown, eds.), p. 109-122. Island Press, Wash- ington, DC. DeVries, D. R.. and R. V. Frie. 1996. Determination of age and growth. In Fisheries techniques, 2nd ed. (B. R. Murphy and D. W. Willis (eds.), p. 483-512. Am. Fish. Soc, Bethesda, MD. Faunae, C. H., and J. J. Lorenz. 2000. Reproductive biology of the introduced Mayan cichlid, Cichlasoma iirophthalmus, in an estuarine mangrove habi- tat of southern Florida. Environ. Biol. Fish. 58:215-22.5. Flores-Nava, A., M. A. Olvera-Novoa, and A. Garcia-Cristiano. 1989. Effects of stocking density on the growth rates of Cichlasoma umphthalmus (Gunther) cultured in floating cages. Aqua. Fish. Manage. 20:73-78. Jalabort, B., and Y. Zohar 1982. Reproductive physiology in cichlid fishes, with par- ticular reference to Tilapia and Sarothe/'odon. In The biology and culture of tilapias (R. S. V. Pullin and R. H. Lowe-McConnell, eds.), p. 129-140. ICLARM Conference Proceedings 7. 80 60 40 "§ 20 0 20 40 60 80 Males n = 152 41 52 1 1 9 I 1 33 ■ 12 4 L 1 Females n = 160 6 3 22 30 45 53 1 (- 1 1 1 1 1- -1 012345678 Age (years) Figure 7 Age-frequency distributions for male and female Mayan cichlids col- lected with hook-and-line gear Numbers indicate sample size. Loftus, W. F. 1987. Possible establishment of the Mayan cichlid, Cichlaso- ma urophthalnius (Gunther) (Pisces: Cichlidae), in Ever- glades National Park, Florida. Florida Scientist 50:1-6. Lorenz, J. J., C. C. Mclvor, G. V. N. Powell, and P C. Frederick. 1997. A drop net and removable walkway used to quantita- tively sample fishes over wetland surfaces in the dwarf man- groves of the southern Everglades. Wetlands 17:346-3.59. Lowe-McConnell, R. H. 1982. Tilapias in fish communities. In The biology and cul- ture of tilapias (R. S. V. Pullin and R. H. Lowe-McConnell, eds.), p. 83-113. ICLARM Conference Proceedings 7. Martinez-Palacios, C. A., and L. G. Ross. 1986. The effects of temperature, body weight and hypoxia on 50 Fishery Bulletin 100(1) the oxygen consumption of the Mexican niojarra, Cich- lasonia urophthcilmus (Giinther). Aqua. Fish. Manag. 17: 243-248. 1988. The feeding ecology of the Central American cichlid Cichlasoma urophthalmus (Giinther). J. Fish. Biol. 33: 665-670. 1992. The reproductive biology and growth of the Central American cichlid Cichlasoma urophthalmus (Giinther). J. Appl. Ichthyol. 8:99-109. Miller, R. R. 1966. Geographical distribution of Central American fresh- water fishes. Copeia 4:773-802. Murphy, M. D.. and R. G. Taylor 1990. Reproduction, gi'owth, and mortality of red drum Sciae- naps ocellatus in Florida waters. Fish. Bull. 88:531-542. 1994. Age, growth, and mortality of spotted seatrout in Flor- ida waters. Trans. Am. Fish. Soc. 123:482-497. Ricker,W. E. 1975. Computation and interpretation of biological statistics offish populations. Bull. Fish. Res. Board Can. 191, 382 p. Robson, D. S., and D. G. Chapman. 1961. Catch curves and mortality rates. Trans. Am. Fish. Soc. 90:181-189. Ross, L. G., and M. C. M. Beveridge. 1995. Is a better strategy necessary for development of native species for aquaculture? A Mexican case study. Aquaculture Res. 26:539-547. SAS Institute Inc. 1989. SAS/STAT users guide, version 6, 4th ed. Gary, NC, 943 p. 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. 19-57. Can. Spec. Publ. Fish. Aquat. Sci. 117. Shafland, P. L. 1996. Exotic fishes of Florida-1994. Rev. Fish. Sci. 4:101- 122. Stauffer, J. R., and S. E. Boltz. 1994. Effect of salinity on the temperature preference and tolerance of age-0 Mayan cichlids. Trans. Am. Fish. Soc. 123:101-107. Trexler, J. C, W. F. Loftus, F Jordan, J. Lorenz, J. Chick, and R. M. Kobza. 2000. Empirical assessment of fish introductions in a sub- tropical wetland: an evaluation of contrasting views. Bio- logical Invasions 2:265-277. von Bertalanffy, L. 1957. Quantitative laws in metabolism and growth. Q. Rev. Biol. 2:217:-231. 51 Abstract— We examinod seasonal ami annual variation in numbers of StcUcr (northern! sea lions iEumetopias juba- tiis) at the South Farallon Islands from counts conducted weekly from 197-4 to 1996. Numbers of adult and sub- adult males peaked during the breeding season (May-July), whereas numbers of adult females and immature indi- viduals peaked during the breeding season and from late fall through early winter (September-December). The seasonal pattern varied signifi- cantly among years for all sexes and age classes. From 1977 to 1996, num- bers present during the breeding season decreased by 5.99r per year for adult females and increased by 1.9% per year for subadult males. No trend in numbers of adult males was detected. Numbers of immature individuals also declined by 4.5'^r per year during the breeding season but increased by S.O't per year from late fall through early winter Max- imum number of pups counted declined significantly through time, although few pups were produced at the South Faral- lon Islands. The ratio of adult females to adult males averaged 5.2:1 and declined significantly with each year, whereas no trend in the ratio of pups to adult females was discernible. Further stud- ies are needed to determine if reduced numbers of adult females in recent years have resulted from reduced sur- vival of juvenile or adult females or from changes in the geographic distri- bution of females. Population status, seasonal variation in abundance, and long-term population trends of Steller sea lions (Eumetopias jubatus) at the South Farallon islands, California* Kelly K. Hastings William J. Sydeman Point Reyes Bird Observatory 4990 Shoreline Highway Stinson Beach, California 94970 Present address (for K K Hastings): Alaska Department of Fish and Game Division of Wildlife Conservation 333 Raspberry Rd Anchorage, Alaska 99518 Email address (for K K Hastings) kellyhaslingsiSfishgame slate ak us Manuscript accepted 1 August 2001. Fish. Bull. 100(11:51-62(20021. Steller sea lions (Eunwtopias jubatus) range from southern California along the West Coast of North America through the Aleutian and Pribilof Islands to the Kuril Islands and Okhotsk Sea, Japan (Kenyon and Rice, 1961). Major haulouts and rookeries have his- torically been centered at the Aleutian Islands and at islands and mainland sites around the Gulf of Alaska, where over 70% of the world population was located in the 1950s and 1960s (Lough- lin et al., 1984). In 1990, the species was listed as threatened throughout its range under the Endangered Spe- cies Act owing to declines of over 50% from an estimated world population of 240.000-300.000 in the early 1960s to 116,000 individuals in 1989 (Loughlin et al., 1992). Numerically the decline was most severe in the western Gulf of Alaska where 50-80'% declines occurred (Loughlin et al., 1992). Reduced juve- nile sui-vival appears to be the prox- imate cause for the decline (York, 1994); ultimate causes, however, are unknown. Effects of long-term environ- mental change and pollutants on Steller sea lions, and interactions or compe- tition of these sea lions with commer- cial fisheries are potential contributing causes of this decline (NMML'). In contrast to rookeries in the west- ern Gulf of Alaska, southeast Alaska rookeries have increased by more than 60% over the past three decades ( Lough- lin et al., 1992). Based on differences in population trends and genetics (Bick- ham et al., 1996), a distinction has been made between two separate stocks: 1) the eastern stock, ranging from south- east Alaska to California, and 2) the western stock, ranging from the Gulf of Alaska, Aleutian Islands, and Prib- ilof Islands to Russia (LIS. Federal Register 62:24345-24355). In 1997, the National Marine Fisheries Service list- ed the western stock as endangered, whereas the eastern stock remained listed as threatened. However, differ- ences in trends between rookeries in southeast Alaska and those in Cana- da, Oregon, and California may indi- cate that these areas deserve separate management considerations. For example, rookeries in Canada and California suffered 40% and 80% declines respectively, from the early 1900s to 1970 (Bigg, 1988; Ainley et al.-); declines continued over the past * Contribution 790 of the Point Reyes Bird Observatory, Stin.son Beach, CA 94970. ' NMML (National Marine Mammal Labora- tory). 1995. Status review of the United .States Steller sea lion [Eumetopias juba- tuf) population. Report of the National Marine Mammal Laboratory. National Marine Fisheries Service, Seattle, WA, 61 p. [Available from National Marine Mammal Laboratory, 7600 Sand Point Way N.E., Seattle, WA 98115-0070.] ' Ainley, D. G., H. R. Huber, R. R Henderson, and T. J. Lewis. 1977. Studies of marine mammals at the Farallon Islands, Califor- nia. 1970-1975. Final report to the Ma- rine Mammal Commission. Washington D.C. I NTIS publication number PB274046. Avail- able from Point Reyes Bird Observatory, 4990 Stinson Beach, CA 94970.1 52 Fishery Bulletin 100(1) 37''42'4.'S'-N ■ .17'42'(l(l" _17"41'I5"- Sugorlodt' Kiel SOUTH FARALLON ISLANDS Nonh Landing. LiyhlhauscHill Lion /■- A .if Cove r^,'?^^^" \jf^.- ■^ 'ir- Y-'A- (»'^***^'- Vl / \ rv Nonh Farjlli.i Isl,.n,i4 ^ Snulh h jullon Klands . Indian Head SOUTHEAST FARALLON ISLAND \ Piicific Occiin 1 1 i:.? IKI'^d" 12.^ IKI'OII"VV Figure 1 Map of the South Farallon Islands, including Southeast F'arallon Island and West End Island. Steller sea lions were counted weekly from 1974 to 1997 from Lighthouse Hill, and several gi'ound areas: North Land- ing. Cormorant Blind Hill, Sewer Gulch, and Garbage Gulch. several decades at several California rookeries (Westlake et al., 1997; Sydeman and Allen. 1999; Le Boeuf et al.'). Wliereas over 2000 Steller sea lions used the Channel Is- lands in the late 1930s, only 50 animals were obsei^ed there in 1959 (Bartholomew and Boolootian, 1960). Pup- ping at San Miguel Island, an historical rookery, has not been observed since 1981 (NMMLM. Therefore to better understand patterns and causes of the population decline, trends and status of the eastern stock at southern rooker- ies deserve further investigation. The Farallon Islands (:37°42'N. 123°00'W). 40 km off the coast of San Francisco, California, are currently one of the most southerly haulout and breeding areas for Steller sea lions; Aiio Nuevo Island is the only rookery farther south. The Farallon Islands consist of three groups of islands: South Farallones (two islands. Southeast Farallon and West End, separated by a small surge channel). Middle Farallon (an intertidal rock), and North Farallones (four large sea stacks; Fig. 1). Although the status of Steller sea lions in California prior to 1800 was poorly documented. Steller sea lions bred at the Farallon Islands in the 1800s 3 Le Boeuf. B. J., K. A. Ono. and J. Reiter. 1991. History of the Steller sea lion population at Ano Nuevo Island. 1961-1991. Final report to National Marine Fisheries Service, Southwest Fisheries Science Center. La Jolla.CA. Admin, report LJ9145C, 9 p. [Available from National Marine Fisheries Service. South- west Fisheries Science Center. P.O. Box 271. La Jolla. CA 92038.) and early 1900s (Allen. 1880; Rowley. 1929) and were the most abundant sea lion in California and at the Farallon Islands from the early to mid 1900s (Rowley, 1929; Bon- not and Ripley, 1948). A large amount of data is now avail- able to examine seasonal variation and long-term trends at the Farallon Islands from historical surveys conduct- ed from 1927 to 1970 by the California Department of Fish and Game (CDFG; Bonnot and Ripley, 1948; Ripley et al., 1962; Carlisle and Aplin, 1971) and from surveys conducted weekly by Point Reyes Bird Obsei-vatory (PR- BO) from 1971 to 1996. Although maximum numbers de- clined significantly from 1974 to 1997 for the total popula- tion ( 1.6% per year) and for adult females (3.6% per year; Sydeman and Allen, 1999), it is unknown whether num- bers of other age classes also declined and in which sea- sons declines occurred. To understand proximate causes and consequences of the decline, several questions have yet to be addressed: have reduced pup production and re- duced reproductive rates also occurred in recent years?; and what effect has the decline had on the adult sex-ra- tio? Finally, seasonal variation in counts for different sex- es and age classes and variation in the seasonal pattern among years also have not been examined in detail at the Farallon Islands. The objectives of our study were to ex- amine 1) seasonal variation in numbers among sexes and age classes; 2) trends in numbers from 1974 to 1996 by age class, sex, and season; and 3) averages and trends in pup production, reproductive rate, and adult sex-ratio. Hastings and Sydeman Population status of Eumetopias jubctus at the South Farallon Islands, California 53 Methods Survey methods PRBO began conducting surveys of pinnipeds at the South Farallon Islands in 1971. In June 1973 surveys were stan- dardized and all Steller sea lions visible on or in the water near the coast of the South Farallon Islands were counted weekly from standard vantage points on Southeast Far- allon Island: 1) atop Lighthouse Hill (110 m) with bin- oculars or a 20-60x spotting scope, 2) from Cormorant Blind Hill (35 m) with binoculars, and 3) from North Land- ing, Sewer Gulch, and Garbage Gulch with no optical aids I Fig. 1). Most surveys were conducted between 1000 and 1800 hours on Thursdays if visibility was adequate. Begin- ning in 1977, animals were classified by age class (adult male, subadult male, adult female, immature, yearling, or pup) when possible, primarily by body size. Adult males were distinctive as very large animals with large muscular necks bearing well-developed manes of long, coarse hair on the chest, shoulders, and back. Subadult males were dis- tinguished from adult males by their smaller size and less developed mane. Immature individuals included animals of distinctly smaller size, such as young-of-the-year (after November) and animals likely one to four years of age. The adult female category included animals smaller than sub- adult males but larger than immature individuals. Pups were distinguishable from June until late November by their thick, dark brown coats, which were later molted and replaced with a lighter brown coat after five to six months of age. Counts were conducted by numerous observers over the years; several observers conducted surveys for over a decade and all observers were trained in identifica- tion of sea lions by age class. Counts represent minimum estimates of numbers of sea lions hauled out because only SS"* to 90'7f of the islands were visible from the study's vantage points. We compared maximum counts taken during the breed- ing season (June-July i in recent years (1974-97) with counts from surveys conducted a single time during the breeding season (once annually) and intermittently over the years by CDFG from 1927 to 1970. From 1927 to 1938, counts of subadult or adult sea lions (i.e. excluding pups) made by at least two obsei-vers from boats were averaged (Bonnot, 1931, 1937: Bonnot et al., 1938). Meth- ods of counting changed after 1938, such that counts af- ter 1938 could only be compared cautiously with earlier years. Surveys were conducted by airplane, blimp, or boat in 1946 and 1947 and by airplane only from 1958 to 1970 (Bonnot and Ripley. 1948: Ripley et al, 1962; Carlisle and Aplin, 1971). Counts from 1946 to 1970 were likely over- estimates because observers assumed that all sea lions north of Point Conception were Steller sea lions (many sea lions may have been California sea lions, Zalophus califoi-niaruis) and because pups were likely included in these counts (Bonnot and Ripley, 1948; Ripley et al., 1962). Counts conducted by PRBO since the 1970s targeted only the South Farallon Islands, whereas CDFG counts includ- ed the South and North Farallon Islands. Monitoring of the North Farallon Islands since 1970. however, has been sparse. Although the North Farallon Islands are a known haulout area for Steller sea lions, pupping rates are un- known. The North Farallon Islands were surveyed during the breeding season by PRBO in 1977 (when 17 adult fe- males and 1 pup were counted) and in 1983 (when 92 adults but no pups were counted; PRBO, unpubl. data^). Because of the exclusion of the North Farallon Islands in recent counts, comparisons with earlier CDFG data were made cautiously. Under the direction of D. G. Ainley and H. R. Huber, pup production and pup mortality were monitored intensively from 1973 to 1986, when animals were breeding in accessi- ble areas. Breeding areas were checked daily for new pups, and prematurely born and dead pups were noted. Breed- ing areas shifted from accessible to inaccessible areas over the years. From 1973 to 1975, all full-term pups were born on the more accessible Saddle Rock, a small islet one- quarter mile offshore (Fig. 1), and a few premature pups were born on the mainland. From 1976 to 1983, females pupped in the equally accessible Sea Lion Cove (Fig. 1), perhaps because of reduced disturbance on Southeast Far- allon Island, although one pup was obser\'ed on Saddle Rock in 1981. Although photogi-aphs from the 1930s show large numbers of Steller sea lions on West End (Huber^), they were not obsei-ved there in recent years until 1983 (one female in the spring). The first pup was born on West End in 1985 ( Huber"' ). Cun-ently, the majority of the popu- lation is found and all pupping occurs at Indian Head and Shell Beach on West End (Fig. 1); both of these areas are inaccessible and difficult to monitor. Statistical analyses Statistical models have been developed that account for effects of obsei-ver, and environmental and survey-related covariates on counts of birds and marine mammals (Link and Sauer, 1997, 1998; Calkins et al., 1999; Frost et al., 1999; Forney, 2000). These models can increase accuracy in estimating and power in detecting population trends by reducing variability in counts and correcting biases in trend that result from methodological changes in sui-vey design over time (such as changes in survey dates), par- ticularly when few surveys are conducted during a stan- dard survey window each year (Calkins et al., 1999; Frost et al., 1999). However, environmental covariates could not be included in the statistical models in our study when the full data set was used because observers recorded the times that sui-veys began and ended on only a few occa- sions prior to 1983 (41 of 569 surveys, or 7.2'^^^f of surveys), such that the majority of data during the first decade of the time series would be excluded. To include the entire •» PRBO (Point Reyes Bird Observatory). 1988. Unpubl. data. (Available from W. J. Sydeman, Point Reyes Bird Observatory, 4990 Shoreline Hwy, Stinson Beach. CA 94970.1 ' Huber, H. R. 1985. Reproduction in northern sea lions on Southeast Farrallon Island, 1973-1985. Final report to the Gulf of the Farallones National Marine Sanctuary, San Fran- cisco. CA, 22 p. [Available from Point Reyes Bird Observatory, 4990 Shoreline Hwy., Stinson Beach, CA 94970.) 54 Fishery Bulletin 100(1) time series, standard regression models, including survey date but excluding effects of environmental covariates, were used to examine seasonal patterns and trends. We believe the exclusion of other covariates during statistical modeling had little effect on trend estimates because sur- veys were conducted consistently over years and over the entire year interval, resulting in large sample sizes (?! = 1134 surveys conducted; range among years 1974- 96: 45 to 52 surveys/year). It is unlikely that population trend estimates were confounded by changes in environ- mental conditions because no obvious annual trends in environmental conditions over the 22 years of the study (weather and tide data were collected daily [at 1000 hours] at Southeast Farallon Island) were apparent, except for a potential increasing annual trend in sea surface tempera- ture (PRBO. unpubl. data^). Seasonal abundance patterns To examine seasonal abun- dance patterns, polynomial regression (Kleinbaum et al.. 1988) was used to fit a cui-ve to counts pooled over years, 1974 to 1996. Data from 1971 to 1973 were excluded because survey methods were not standardized until the end of 1973. We fitted the regression model by first con- verting Julian date to orthogonal polynomial variables (linear combinations of the natural polynomial variables that contain the same information as the natural polyno- mial variables but are uncorrelated to each other) to avoid problems of multicollinearity when using higher-order terms (POeinbaum et al., 1988). Higher-order terms were then added sequentially until the last term was not signif- icant in the model (forward stepwise procedure, P>0.05). We then added year as a variable to the model and tested the year x date interaction to determine if the seasonal pattern varied significantly among years. To examine sea- sonal patterns by sex and age class, polynomial regression curves were fitted separately to counts of adult females, males (adults and subadults pooled), and immature indi- viduals as described above. We excluded surveys in which not all individuals were identified by sex and age class (i.e. all surveys before 1977). Annual abundance trends Because high-order polynomial models were used to address seasonal haulout patterns, annual abundance trends were examined in a separate analysis to simplify results. Seasonal variability in abun- dance was accounted for in annual trend models by using residuals from the regression of Julian date on counts. Assuming e.xponential rates of change, we log-transformed (log^,) the residuals (centered about the mean count) and regressed the transformed residuals against the variable year. Annual rates of change were calculated as el\-^,^^ - 1 X 100%, where Pycar 's ^^^ regression coefficient for annual trend (Caughley, 1977). The following groups were ana- lyzed: 1) all animals, by pooling data over all 12 months and sex and age classes; and 2) each sex and age class, by pooling over a) all months, and b) two periods when peaks in counts were observed for some age classes (the breed- ing [May-July] and late fall through early winter [Sep- tember-December] seasons). Nonlinearity in trend was assessed by using orthogonal polynomials as described earlier in this article. Assumptions of the regression model were verified by visual inspection of residuals. Trends In pup production, reproductive rate, and adult sex ratio during the breeding season We used linear regres- sion to test if the decline in maxnnum pup counts during sui-veys presented in Sydeman and Allen (1999) was sig- nificant. We used only data after 1977, when counts by age class were conducted consistently. Only data from sur- veys conducted from June to July were included because during the fall, the ability to distinguish young-of-the year from immature individuals was difficult and because an influx of nonnative pups may have occurred. For example, in November 1978, five times the number of pups known to have survived the breeding season and an increased number of adult females were observed (PRBO, unpubl. data^). The origin of these young-of-the-year is unknown, but the nearest known pupping areas are Aiio Nuevo Island and the North Farallon Islands. Although Steller sea lions are present at Point Reyes, no pups have been obsei-ved there in the past two decades (Sydeman and Allen, 1999). To examine averages and trends in adult sex ratio and reproductive rate, we used maximum counts of adult fe- males, adult males, and pups during June and July in each year and linear regi'ession to test for annual trends. Re- productive rate was calculated as the maximum count of pups divided by maximum count of adult females. Because not all pups born were observed during surveys, we in- creased the maximum count of pups by 57%, the average amount that maximum pup counts underestimated true pup production from 1973 to 1986 (range: 33-90''^ among years). This average was determined from unpublished data of pup production as determined from daily observa- tions of breeding areas (Huber et al.'^). Results Seasonal abundance patterns When data from all sexes and age classes were pooled, the seasonal abundance pattern was bimodal; one peak in numbers occurred before and during the breeding season (April-July) and another peak occurred from late fall through early winter (October-December; Fig. 2A). The regression model was complex with significant date and higher-order terms (variables date- through date'^); all P<0.001; adjusted r-=0:28. ;i = 1134); the variable date-' was not significant (P>0.65). Counts varied significantly with year (P<0.001) and the seasonal pattern varied sig- nificantly among years (datexyear through date^xyear; P<0.001; adjusted r-=0.61 ). Total numbers during the peak '^ Huber, H. R., D. G. Ainley, R. J. Boekelheide, R. R Henderson, and T. J. Lewis. 1988. Annual and seasonal variation in num- bers of pinnipeds on the Farallon Islands. California (Table 3). Final report to the National Marine Mammal Laboratory, National Marine Fisheries Service. Seattle, WA, 3.5 p. [Avail- able from Point Re.yes Bird Observatory, 4990 Shoreline Hwy., Stinson Beach, CA 94970.1 Hastings and Sydeman Population status of Etimetopias /ubahis at the South Farallon Islands, California 55 JIH) - A ' 200 - ■ '; ■• ■ 100 - 0 - -/.•*-'"v. ■.•>■■. ;■•:•'" ''■.— l-^rf J- '■•,'•■;''■*■■'. ^ f2 0 30 60 W 120 150 180 210 240 270 300 330 360 0 30 60 90 120 150 ISO 210 240 270 300 330 360 Julian J.ile S 100 -- B 0 30 60 90 120 150 180 210 240 270 300 330 360 1 50 100 -- 50 -- D 0 30 60 90 120 150 180 210 240 270 300 330 360 Julian date Figure 2 Seasonal variation in counts of Steller sea lions at the South Farallon Islands for (A) both sexes and all age classes; (B) adult females; (C) subadult and adult males; and (Dl immature mdividuals and yearlings. Data from 1974 or 1977 to 1996 were pooled. Black dots indicate counts: black lines indicate predicted values from the regression model. Divisions on the .v-axis approximate months. The best regression model for each group included the variables date and date'^ through (A) date^ for total counts (adjusted r-=0.28); (B) date^ for adult females (adjusted r-'=0.22); (C) date^'~ for males (adjusted r'^=0.77); and (D) date*^ for immature individuals (adjusted r2=0.13l. breeding season averaged approximately 100 animals, ranging from 50 to 200 animals; whereas numbers from the late fall through early winter peak were more vari- able, averaging slightly less than 100 and ranging from less than 10 to 300 animals (Fig. 2A). Seasonal patterns varied among sexes and age classes (Fig. 2, B-D). Counts of adult and subadult males peaked only during the breeding season (Fig. 2C), whereas counts of adult females and immature sea lions were bimodal (Fig. 2, B and D). When models including the variables date through date'^ were fitted to data for adult females and immature individuals separately, the seasonal pat- tern differed significantly between the two groups (age class, year, and all interaction terms; all P<0.001). Counts of immature sea lions were less peaked during the breed- ing season than those of adult females and, in contrast to the average adult female pattern, numbers during winter peaked on average slightly higher than during the breed- ing season (Fig. 2, B and D). The seasonal pattern varied significantly among years for all age classes (year and yearxdate interactions for adult females and immature individuals; P<0.001, and for subadult and adult males; P<0.05). Variation in seasonal pattern among years was complex but several general pat- terns could be noted. A gradual shift in the peak breeding season count from the beginning of May in 1974 to the be- ginning to middle of June in 1979 was evident (Hastings and Sydeman"). The late fall-winter peak was very pro- nounced from 1984 to 1986, with maximum counts of 200 to 300 animals (Hastings and Sydeman'), most of which were immature individuals. From 1992 to 1996, the sea- sonal abundance pattern was muted with equal or higher numbers in the winter than in the breeding season (Hast- ings and Sydeman'). Hastings, K. K., and W. J. Sydeman. 1998. Status, seasonal variation and long-term trends in numbers of Steller sea lions, Eumetopias jubatus. at the South Farrallon Islands, California: 1927-1996. Final report to the National Marine Fisheries Ser- vice, Southwest Fisheries Science Center, La Jolla, CA, 30 p. (Available from Point Reyes Bird Observatory, 4990 Shoreline Hwy., Stinson Beach, CA 94970.) 56 Fishery Bulletin 100(1) Table 1 Linear rates of change in counts of Steller sea lions on the South P^arallon Islands by season, sex. and age class. Rate of change per year was calculated by 1) removing the effect of date on counts (i.e. by using residuals from regi'ession of date on counts), and 2) log-transforming (log^ ) the sum of the residuals added to the mean count, ■ mdicates significant trends (P<0.05) from regi'essions. ;! = sample size. Age class change /^> SEi/3,,.J All animals All months Adult females All months Breeding season (May-Jul) Late fall through early winter Males (breeding season! All males Bulls Subadult males Immature sea lions All months Breeding season Late fall through early winter -0.44 -0.0044 0.0027 0.03* 1134 3.16 -0.0321 0.0032 <0.001* 866 .5.89 -0.0607 0.0081 <0.001' 217 2..50 0.0247 0.1637 0.80 280 1.12 0.0111 0.0050 0.03* 217 0.16 0.0016 0.0022 0.63 217 1.94 0.0192 0.0086 0.03* 217 0.55 0.0055 0.0019 0.004* 866 4.51 -0.0461 0.0168 0.007* 217 4.97 0.0485 0.0078 0.60; Table 1). Counts of immature indi- viduals also increased slightly (0.6% per year; Table 1) but significantly when counts from all months were pooled (variables yea/- and year-: P<0.01; ;?=866; Fig. 3D). The in- crease was due to the greater numbers of immature in- dividuals from late fall through early winter in recent years (linear trend=5.0%' per year, Table l;yea7- and year^: P<0.01; ;;=280; Fig. 4D). However, numbers of immature individuals present during the breeding season declined at a rate of -4.5% per year (Table l;year: P<0.01; ;;=217; Fig. 3D). Trends in pup production, reproductive rate, and adult sex ratio during the breeding season Maximum pup count from surveys declined significantly from the mid-1970s to the mid-1980s from 15 to 2-4 pups and has remained low in recent years (year and year-: P<0.003; Fig. 4). After adjusting for pups not seen during surveys, reproductive rates of adult females ranged from 2.0% to 21.2'7( among years, with an average rate of 10.7% (Fig. 4). Although reproductive rate appeared to decline in the 1980s and recovered to 1970s levels in the 1990s, no trend was discernible (year and higher-order terms: P>0.20; Fig. 4). The ratio of adult females to adult males during the breeding season ranged among years from 10.3:1 to 1.8:1, with an average of 5.2:1 (Fig. 4). The ratio of adult females to adult males declined significantly and linearly with vear (P<0.001). Discussion Although the Farallon Islands are an important haulout area for Steller sea lions in California, numbers of ani- Hastings and Sydeman Population status of Eumetopim juhahis at tlie South Faiallon Islands, California 57 3 'J "2 -!< ! I : i ! : 1 n h I i 1 Pi I : rprrrrr 73 75 77 79 81 83 85 87 89 91 93 95 97 £ 2 ? 0 76 78 80 82 84 86 88 90 92 94 96 Year 4 - ^ 2 ^ 0 oc .4 B 76 78 80 82 84 86 88 90 92 94 96 6 5 4 -5 •) -4 - D 76 78 82 84 86 88 90 92 94 96 Year Figure 3 Annual trends in counts of Steller sea lions at the South Farallon Islands from 1974 or 1977 to 1996 for (A) both sexes and all age classes; (B) adult females; (C) subadult males; and (D) immature individuals and yearlings. Significant trends in counts, after accounting for survey date ( residual centered about the mean count from Figure 2. square-root transformed ). are shown for; all months (light dashed line); only counts during the breeding season (May - July; solid black line); and only counts from late fall through early winter (September-December; solid light black line). Results of significance tests using square-root and log-transformed counts were identical; Linear rates of change from log.-transformed counts are shown in Table 1. mals at the Farallon Islands are currently lower (0.06 of the 1989 statewide count, 0.09 of the count from four major sites) than at the other three major California sites (Alio Nuevo Island, St. George Reef and Sugarloaf Island) which ranged from 0.16 to 0.18 of the 1989 statewide count, and from 0.26 to 0.37 of the count from four major sites (Loughlin et al., 1992). A smaller proportion of the statewide Steller sea lion population has used the Farallon Islands in recent years, compared with population counts in the 1927-30 data, when Farallon animals accounted for 0.11 to 0.14 of the statewide count (Bonnet and Ripley, 1948). Historical pup production at the Farallon Islands is unknown, but both the Farallon Islands and Ano Nuevo Island were identified as the two largest and most impor- tant Steller sea lion rookeries in the state in the early 1920s (Rowley, 1929). Pup production at the South Faral- lon Islands over the past two decades has been very low at <30 pups per year and in the last 10 years, at <10 pups per year. Pup production since the mid-1980s, how- ever, may be underestimated owing to the reduced prob- ability of sighting pups since 1984 when pupping areas shifted to West End Island, which is farther away from the survey vantage points. Pup production at other major sites in California included 117-137 pups at Sugarloaf Island and Cape Mendocino in the early 1980s, 115 pups at St. George Reef in 1994, and 230-243 pups at Ano Nuevo in 1993-94 (Westlake et al., 1997; NMML'). Reproductive rates of Steller sea lions at the South Far- allon Islands were also low; an average of only 0.11 of fe- males present during the breeding season produced pups. This number may be biased low because some immature males may have been included in the adult female count. This ratio is much lower than that for rookeries in Brit- ish Columbia (>0.70, Pike and Maxwell, 1958), Afio Nue- vo, California (average of 0.40 to 0.50 from 1962-1990; Le Boeuf et al.-') and Ugamak Island, Alaska, where ratio of 58 Fishery Bulletin 100(1) pups to females increased from 0.75 to >1.00 from 1968 to 1986 (Merrick et al., 1987). The South Farallon ratio is more typical of pe- ripheral areas of rookeries in Alaska where only 0.01 to 0.09 of females had pups com- pared with main areas of rookeries where ra- tios averaged 0.63 to 0.74 (Withrow, 1982). Similarly, high pup mortality rates observed at the Farallon Islands (average of 0.49 of pups born from February to August, range of 0.33 to 0.90 among years; Huber et al.^) are more characteristic of peripheral areas of rookeries where pup mortality ranged from 0.30 to 1.00 compared with 6.10 to 0.12 at main rookery sites (Withrow, 19821. Rooker- ies had much lower pup mortality rates dur- ing the first two months of life than those ob- served at the Farallon Islands, including Ario Nuevo Island, California (0.10, Gentry, 1970), and sites in Alaska (0.03-0.14, Merrick et al., 1987). The frequency of premature pupping (0.40 of those born; Huber et al.'') is also very high compared with the frequency at rook- eries in Alaska (0.09; Pitcher and Calkins, 1981), Oregon (0.04; Mate, 1973), at Aho Nue- vo Island (0.02; Gentry, 1970). As at Aho Nue- vo, most premature pups are born from F'eb- ruary to May at the Farallon Islands (0.65 born in April with a range of February to May), whereas full-term pups are born from mid-May to late July (Gentry, 1970; Huber^). Causes of the high rate of premature pupping at the Farallon Islands are unknown but may be due to several factors known to cause reproductive failure in pinnipeds, including disease or exposure to pollutants (Gilmartin et al., 1976; Huber''), or a prevalence of young, inexperienced, or malnourished females (Pitcher et al., 1998). A high frequency of abortions has been observed at haulout sites rather than at rooker- ies in Alaska (Pitcher and Calkins, 1981 ). Low pup produc- tion and reproductive rates, coupled with high pup mortal- ity and premature pupping rates, support characterization of the Farallon Islands in recent years as a haulout site or peripheral rookery for this species. Seasonal patterns in counts Seasonal haulout patterns varied significantly among sexes and age classes. Adult and subadult male attendance was highly seasonal and males were present only during the breeding season. In contrast, adult females and immature individuals were present year-round and their numbers peaked twice (breeding season and from late fall through early winter). Many studies reported the absence of adult and subadult males at California rookeries outside the breeding season, including Ano Nuevo Island (Orr and Poulter, 1967) and San Miguel Island ( Bartholomew, 1967), and the presence of females and immature individuals at rookeries year-round (Rowley, 1929; Bartholomew, 1967). At Canadian rookeries, males were also generally absent in the winter, but small numbers of females and young OReproduclive rale D Sex-ratio • Maximum pup count -1 — ' — 1 — ' — 1 — I — I- 76 78 80 82 84 86 88 90 92 94 96 98 "lear Figure 4 Annual variation in reproductive rate, adult sex ratio, and ma.ximum pup count during tlie breeding season (June-.Julyi at the South Farallon Islands, 1977-97. Reproductive rate was defined as the ma.ximum pup count divided by the maximum count of adult females per year. Adult sex ratio was calculated as the maximum number of adult females divided by the maximum number of adult males (bulls) counted per year Signifi- cant (P<0.05) trends are shown for adult sex ratio (dashed black line) and maximum pup count (bold black line). of the year usually remained at rookeries throughout the year (Bigg, 1988). Circumstantial evidence suggests males from California migrate northward or males from South- east Alaska move southward in winter, or both movements take place. Large numbers of males have been seen outside the breeding season off northern California (Fry, 1939), Oregon, Washington (Mate, 1973), and southern Vancouver Island (Bigg, 1988). Total numbers of Steller sea lions are also higher in the winter than in the summer off the Cana- dian coast (Bigg, 1988); some winter haulouts in Canada consist almost exclusively of males (Bigg, 1988). The earli- est evidence for sea lion migrations was provided by the recovery of north-coast native American spearheads from several sea lions killed off southern California in the late 1800s; and in June 1870, a spearhead used by native Alas- kans was found in a large male sea lion at Point Arena, California (Scammon, 1874). Seasonal northward move- ment has also been documented in male California sea lions, which were similarly absent from southern sites out- side the breeding season but which ranged up into Wash- ington and British Columbia during winter (Starks, 1921; Fry, 1939; reviewed by Bartholomew, 1967). In contrast to animals on the Farallon Islands, animals of all age classes and both sexes on Aho Nuevo Island were present in significant numbers only during the breeding season from 1967 to 1990 (Le Boeuf and Bonnell, 1980; Le Boeuf et al.^). Data from 1962 and 1963 indicated a sub- stantial presence of Steller sea lions at Aho Nuevo through the fall and winter (Orr and Poulter, 1965) and therefore the lower numbers and, more recently, near absence of all Hastings and Sydeman Population status of Eumetopias jubatus at the South Farallon Islands, California 59 2()()() -1 1 750 c 3 1500 •r. = I2?0 i 1000 i 750 I 500 - o 250 0 • Siellcis - hiskinca! icnint-. O Slcllcrs iiiui Cajitoinuins ciinilimcJ - hislonciil coiinls ■ Slcllciv I'RBOcounls D Slclk-rs Miul C.ililomums coinhincd I'RHC.) cmmls - LiniitcJ lumrsr 1 1920 1930 1940 1950 19(i0 1970 19X0 1990 2000 \c.,r Figure 5 Counts of sea lions at the Farallon Islands during the breeding seasons, from 1927 to 1997. Historical counts, 1927-.38: total count from single census per year conducted by boat; includes North and South Farallones and adults and sub- adults only (Bonnet et al., 1938). Historical counts, 1946-70: total count from a single census conducted each year by airplane. Wimp, or boat; includes North and South Farallon Islands and may include pups, subadults and adults. Steller and California sea lions were not distinguished during these surveys. Instead all sea lions north of Point Conception were considered Steller sea lions and those south of Point Conception were considered California .sea lions (Bonnot and Ripley. 1948; Ripley et al., 1962; Carlisle and Aplin, 1971 ). Point Reyes Bird Obsei-vatory counts, 1974-97: maximum total counts during June and July from weekly censuses at South Farallon Islands only (North Farallones excluded); includes pups, immature individuals, subadults, and adults. Means or trends over years are shown for Steller counts only (solid black lines) and for counts of Steller and California sea lions combined (dashed linesA age classes after the breeding season may be a recent phe- nomenon. Similarly. Steller sea lions of various sexes and age classes were present off Humboldt County, California, only from mid-April to September (Sullivan, 1980). Diverse seasonal patterns among sites were also evi- dent in Canada and Alaska. In Canada, animals were usually present year-round on rookeries and numbers peaked during July, whereas year-round haulouts showed no marked seasonal variation and a variety of sexes and age classes were present in winter (Bigg. 1988). Winter haulouts were occupied only in the winter and consisted of either only males or a variety of sexes and age classes (Bigg, 1988). In Alaska, many rookeries were abandoned and some haulouts were occupied only in winter; other haulouts and rookeries were occupied year-round (Ken- yon and Rice, 1961; NMMLM. Major seasonal shifts in distribution were not evident in Alaska, although winter counts were substantially lower than summer counts and there was a greater proportion of animals at haulouts than at rookeries in winter ( NMML' ). The diversity in sea- sonal patterns observed among sites (including rookeries and haulouts) in California and elsewhere has confounded generalizations concerning seasonal haulout patterns, al- though a general shift from rookeries to haulouts in win- ter seems to occur throughout most of the species range. Population status of Steller sea lions in southern and central California Decline from historical numbers Substantial declines in Steller sea lions at the Farallon Islands have been evident since the 19'20s and in recent decades. Numbers declined approximately 75-809( from an average of 600-790 ani- mals from 1927 to 1947 to an average of 150 animals (maximum count) from 1974 to 1997 (Fig. 5). This decline may be overestimated because animals on the North Far- allon Islands have not been included in sui-veys since 1970 and because more animals are likely visible by boat or air than from island-based vantage points (Westlake et al., 1997). However, 85% to 90% of the island is visible from vantage points and therefore effects of incomplete cover- age should be small. Although the decline in numbers was severe between 1938 and 1974, the rate of decline cannot be determined for this period because surveys from this period did not distinguish Steller from California sea lions (Fig. 5). These surveys assumed that all sea lions north of Point Conception were Steller sea lions and that all sea lions south of Point Conception were California sea lions (Carlisle and Aplin, 1971). Assessing the status of Steller sea lions from the 1946-70 CDFG counts has been con- founded by growth in the California sea lion population 60 Fishei7 Bulletin 100(1) over the same period. For example, California sea lions made up only 20% of the total sea hon count at the Faral- lon Islands in 1938 (Bonnot and Ripley, 1948); but by the mid 1970s, California sea lions were twice as numerous as Steller sea lions during June and July (Fig. 51. The role of commercial hai-vest and direct take or ha- rassment of sea lions by humans in this decline is un- certain. Large numbers of sea lions were hunted in Cal- ifornia in the late 1800s for oil, hides, and "trimmings" (which included the whiskers, genitalia, and gall bladder of adult males) that were sold to Chinese markets (Scam- mon, 1874). Hunting sea lions for oil became unprofitable around 1900 because of the reduction in sea lion numbers and the wide-spread availability of petroleum products (Rowley, 1929). A reduced sea lion hai-vest for hides, trim- mings, and (in Mexican waters) pet food, continued until the end of the 1930s when Chinese markets disappeared with the onset of the Japanese-Chinese war and protests were successful in stopping Mexican harvests (Bonnot, 1951). During the same period, although fewer sea lions were taken by sportsman, fisherman, and collectors for museums and zoos, rookery abandonments and population declines still persisted in Oregon and southern California (Rowley, 1929; Bonnot, 1931). An additional cause for these population declines may have been the sea lion hunts that were introduced by commercial fisheries around 1900 to reduce competition for fish (Bonnot, 1937). For example, a bounty was offered for Steller sea lions in the early 1900s in areas north of California (Rowley, 1929; Bonnot, 1931; Bonnot, 1951). Although numbers hai-vested in California are not well documented and the role of harvest in the decline is not obvious, several arguments can be made that declines in Steller sea lions from the 1940s to 1970s were likely not due to effects of hai-vest alone. During the period of com- mercial harvest, Steller numbers appeared stable (Bonnot and Ripley, 1948), whereas the 75-80% decline was evi- dent after 1947, after commercial hunting and collections had ended, although harassment by fisherman continued. After 1947, the California sea lion population increased exponentially throughout the state from 3050 in 1947 to a minimum of 18,047 in 1970 (Bonnot and Ripley, 1948; Carlisle and Aplin, 1971), whereas numbers of Steller sea lions on the Channel Islands and at the Farallon Islands declined from 80% to 100% during this period. Large in- creases in California sea lions were evident after commer- cial hai-vesting ended, even though many more California than Steller sea lions were likely hunted commercially, poached, or captured because of difficulty hunting in the steep, rocky intertidal areas frequented by Steller sea li- ons (Rowley, 1929; Bonnot, 1951). This reasoning suggests that factors in addition to hai-vest have influenced the population decline. Proposed causes include reduction of the prey base due to overexploitation by commercial fish- eries (Ainley and Lewis, 1974), shifts in prey composition due to ocean warming, and competition for food with grow- ing numbers of California sea lions (Bartholomew, 1967). Human disturbance, however, likely played some role in the decline, in respect of which Steller sea lions may be more affected by human disturbance than California sea lions. For example, the large Steller sea lion rookeries at San Miguel Island and at Seal Rocks, just off San Fran- cisco, were abandoned permanently because of harassment and shooting by hunters for sea lion trimmings or by fisher- man (Rowley, 1929). Southeast Farallon Island was inhab- ited by fair numbers of lighthouse keepers and their fami- lies (since the mid- 1800s) and egg hunters (men collecting seabird eggs for sale in commercial markets for human consumption) from the mid-1800s to the mid-1900s. High- est human occupancy occurred during World War II, when over 50 military personnel wore added to the island's pop- ulation (Ainley and Lewis, 1974). Families were removed in 1965 and the lighthouse was automated in 1972, after which time only PRBO researchers remained on the island (Ainley and Lewis, 1974). Despite the designation of the North and Middle Farallon Islands in 1909 and the South Farallon Islands in 1969 as a national wildlife refuge, ha- rassment by fisherman and disturbance from low-flying helicopters was common into the 1970s (Ainley and Lewis, 1974). Heightened human presence in the mid-1900s likely increased the abandonment of Steller sea lions from the is- lands during the period of dramatic decline. Recent population trends Over the last 20 years, the numbers of Steller sea lions on the South Farallon Islands has continued to decline significantly. Numbers of adult females present during the breeding season declined by 5.9% per year from 1977 to 1996, although the rate of decline has lessened since the mid to late 1980s (Fig. 3B). This rate of decline is much higher than the 3.6% per year decline reported for adult females by Sydeman and Allen (1999), who used maximum counts and data from all sea- sons, although rates are similar between the two studies when similar data were used (3.2% per year estimated from our study, when data from all seasons were pooled). These findings demonstrate the importance of accounting for seasonal effects when investigating population trends. The rate of decline of 5.9% per year is similar to the rate of decline obsei-ved during the breeding season in the area of greatest decline in Alaska (from Kiska Island to the Kenai Peninsula), where rates of decline varied from approxi- mately 5% (1975-85 and 1990-94) to 16% (1985-90; York etal., 1996). Numbers of immature individuals present during the breeding season have also declined by 4.5% per year over the past several decades, but an overall net increase in immature individuals on the islands has been apparent owing to increased numbers in the late fall and early win- ter. Numbers of immature individuals on the Farallon Is- lands in the winter were particularly high from 1984 to 1986. Immature individuals have continued to be present in significant numbers during winter in recent years. It is uncertain where these young animals originated from, but overall declines in juvenile counts, coupled with sig- nificant declines in juvenile counts during the breeding season, suggest that increased numbers in winter may represent changes in movement and haulout patterns of juveniles rather than improved juvenile sui-vival in recent years. Increased numbers of subadult males hauled out on the South Farallon Islands during the breeding season in Hastings and Sydeman Population status of Eumetopias jubatus at the South Farallon Islands, California 61 recent years may have resulted from increased emigration or movement of subadult males from Ano Nuevo Island due to increased competition for the declining number of females there. A stable number of adult males, couplfxl with declines in numbers of adult females, has resulted in a significant reduction in the adult male-to-female ratio on the South Farallon Islands during the breeding season in recent years. These results demonstrate that reduced numbers of Steller sea lions on the Farallon Islands in recent years have been driven by reduced numbers of adult females during the breeding season, although reproductive rate and pup mortality rate were stable at this peripheral rook- ery. Patterns were similar at Ano Nuevo, where there were sharp declines in numbers of females and pups during the breeding season but where no trend in reproductive rate w-as apparent from 1962 to 1990 (Le Boeuf et al.'l. How- ever, unlike the Farallon Islands, number of males at Ano Nuevo during the breeding season also declined sharply during the same time period (Le Boeuf et al.'). Although the rate of decline at the Farallon Islands has lessened in recent years, large declines of 9.9*^^ per year for pups and 31.5'~r per year for older animals may have occurred at Ano Nuevo from 1990 to 1993. when negative effects of the 1992 El Nino may have affected estimates from this short time series (Westlake et al., 1997). It is unknown whether reduced numbers of adult fe- males and immature individuals present during the breed- ing season have resulted from reduced survival or chang- es in geographic distribution. Because significant declines in Steller sea lions from historical numbers and over the past several decades have occurred at San Miguel Island. Alio Nuevo Island, and the South Farallon Islands, gi-eater monitoring and protection by state or federal agencies of the southern populations are warranted. Estimates of age- class specific sui-\'ival rates of females are needed to deter- mine if reduced numbers of females are due to increased juvenile or adult mortality. More intensive studies track- ing individual Steller sea lions in California are required to determine if declining numbers indicate a northward shift in the breeding range and to document migratory movements of males and females. Population dynamics and movements of prey of Steller sea lions, dietary overlap with California sea lions, and interactions of sea lions with commercial fisheries in California must be examined to determine natural and anthropogenic causes for changes in sea lion numbers or distribution. Acknowledgments H. R. Huber deserves special recognition for her contri- butions during the early years of our study. Financial support for manuscript preparation was provided by the National Oceanic and Atmospheric Administration. National Marine Fisheries Ser\ice. Southwest Fisheries Science Center under contract 40JGNF600336 to W. J. Sydeman. The Friends of the Farallones. Homeland Foun- dation. Roberts Foundation, Bradford Foundation, and Exxon Corporation also provided funds for data prepara- tion and fieldwork. We are particularly grateful to D. G. Ainley for initiating pinniped studies on the Farallon Islands in 1971. We also sincerely thank Nadav Nur and (irey Pendleton for statistical advice and reviews of the manuscript. We also thank the many obsei-\-ers who have conducted surveys over the past three decades: D. Ainley, G. Ballard, B. Boekelheide, H. Carter, S. Emslie, P. Hen- derson, M. Hester, H. Huber, S. Johnston, J. Lewis, E. McLaren, S. Morrel, J. Nusbaum, J. and T. Penniman, P. Pyle, T. Schuster, J. Walsh, and others. General studies of marine mammals at the South Farallon Islands have been graciously supported over the years by the Marine Mammal Commission. LI.S. Fish and Wildlife Service (USFWS), and the Gulf of the Farallones National Marine Sanctuary. In particular. USFWS and the San Francisco Bay National Wildlife Refuge have provided 28 years of financial, logistical, and moral support; to those involved, we offer sincere gratitude. We also thank the Farallon Patrol for transport to and from Southeast Farallon Island. Michael Rehburg assisted with creating the Far- allon Island map. This manuscript benefited greatly by suggestions from Andrew Trites and several anonymous reviewers. Literature cited Ainley, D. G.. and T. J. Lewis. 1974. The history of Farallon Island marine bird popula- tions. 1854-1972. The Condor 76:432-446. Allen, J. A. 1880. History of the North American pinnipeds. A mono- graph of the walruses, sea lions, sea bears, and seals of North American. Publ. U.S. Geol. Geogr. Surv. 12. 785 p. Bartholomew. G. A. 1967. Seal and sea lion populations of the California Islands. In Proceedings from the symposium on the biolog>' of the California Islands ( R. N. Pliilbrick, ed. i, p. 229-244. Santa Barbara Botanic Garden. Santa Barbara, CA. Bartholomew, G. A., and R. A. Boolootian. 1960. Numbers and population structure of the pinnipeds on the California Channel Islands. J. Mammal. 41:366-375. 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. Bigg, M. A, 1988. Status of the Steller sea lion, Eumetopias jubatus. in Canada. Can. Field-Nat. 102: 315-336. Bonnot, P. 1931. The California sea lion census for 1930. Calif Fish Game 17:150-1.55. 1937. California sea lion census for 1936. Calif Fish Game 23:108-112. 1951. The sea lions, seals and sea otter of the California coast. Calif Fish Game 37:371-389. Bonnot, R, G. H. Clark, and S. R. Hatton. 1938. California sea lion census for 1938. Calif Fish Game 24:415-419. Bonnot. P.. and W. E. Ripley. 1948. The California sea lion census for 1947. Calif. Fish Game 34:89-92. 62 Fishery Bulletin 100(1) Calkins, D. G., D. C. McAllister, K. W. Pitcher, and G.W.Pendleton. 1999. Stellar sea lion status and trend in Southeast Alaska: 1979-1997. Mar Mammal Sci. 15:462^77. Carlisle, J. G., and J. A. Aplin. 1971. Sea lion census for 1970, including counts of other California pinnipeds. Calif Fish Game 57:124-126. Caughley, G. 1977. Analysis of vertebrate populations. John Wiley and Sons, London, England, 2.34 p. Forney, K. A. 2000. Environmental models of cetacean abundance: reduc- ing uncertainty in population trends. Conserv. Biol. 14: 1271-1286. Frost, K. J., L. F. Lowry, and J. M. Ver Hoef 1999. Monitoring the trend of harbor seals in Prince Wil- liam Sound, Alaska, after the Exxon Vatdez oil spill. Mar Mammal Sci. 15:494-506. Fry, D. H. 1939. A winter influx of sea lions from lower California. Calif Fish Game 25:245-250. Gentry, R. L. 1970. Social behavior of the Steller sea lion. Ph.D. diss., Univ. California, Santa Cruz, CA, 113 p. Gilmartin, W. G., R. L. Delong, A. W. Smith, J. C. Sweeney, B. W. De Lappe, R. W. Risenbrough, L. A. Griner, M. D. Dailey, and D. B. Peakall. 1976. Premature parturition in the California sea lion. J. Wildl. Dis, 12:104-115. Kenyon, K. W., and D. W. Rice. 1961. Abundance and distribution of the Steller sea lion. J. Mammal. 42:223-234. Kleinbaum, D. G., L. L. Kupper, and K. E. Muller 1988. Applied regression analysis and other multivariable methods. 2nd ed. Duxbury Press, Belmont, CA, 718 p. Le Boeuf B. J., and M. L. Bonnell. 1980. Pinnipeds of the California Islands: abundance and distribution. In The California Islands: proceedings of a multidisciplinary symposium (D. M. Power, ed. I, p. 475-493. Santa Barbara Museum of Natural History Publications, Santa Barbara, CA. Link, W. A., and J. R. Sauer 1997. Estimation of population trajectories from count data. Biometrics 53:488^97. 1998. Estimating population change from count data: appli- cation to the North American breeding bird survey. Ecol. Appl. 8:258-268. Loughlin, T. R., A. S. Perlov, and V. A. Vladimu-ov. 1992. Range-wide survey and estimation of total number of Steller sea lions in 1989. Mar Mammal Sci. 8:220-239. Loughlin, T. R., D. J, Rugh, and C. H. Fiscus. 1984. Northern sea lion distribution and abundance: 1956-80. J. Wildl. Manag. 48:729-740. Mate, B. R. 1973. Population kinetics and related ecology of the northern sea lion, Eumetopias jubatus. and the California sea lion. Zalophus califormanus, along the Oregon coast. Ph.D. diss., LTniv. Oregon, Eugene, OR. 94 p. Men-ick. R. L., T. R. Loughlin. and D. G. Calkins. 1987. Decline in abundance of the northern sea lion.i?(/me/o- pws jubatus, in Alaska, 19,56-86. Fish. Bull. 85:351-365. Orr, R. T., and T. C. Poulter. 1965. The pinniped population of Aiio Nuevo Island. Cali- fornia. Proc. Cal. Acad. Sci. 32:377-404. 1967. Some observations on reproduction, growth, and social behavior in the Steller sea lion. Proc. Cal. Acad. Sci. 35: 193-226. Pike, G. C, and B. E. Maxwell. 1958. The abundance and distribution of the northern sea lion {Eumetopias jubatus) on the coast of British Columbia. J. Fish. Res. Board Can. 15:5-17. Pitcher, K. W., and D. G. Calkins. 1981. Reproductive biology of Steller sea lions in the Gulf of Alaska. J. Mammal. 62:599-605. Pitcher, K. W., D. G. Calkins, and G. W. Pendleton. 1998. Reproductive performance of female Steller sea lions: an energetics-based reproductive strategy? Can. J. Zool. 76:2075-2083. Ripley W. E., K. W. Cox, and J. L. Baxter 1962. California sea lion census for 1958, 1960 and 1961. Calif Fish Game 48:228-231. Rowley, J. 1929. Lifehistoryofthe sea-lions on the California coast. J. Mammal. 10:1-36. Scammon, C. M. 1874. The marine mammals of the north-western coast of North America. Dover Publications, Inc., New York, NY ( re- print!, 319 p. Starks, E. C. 1921. Notes on the sea lions. Calif Fish Game 7:250-253. Sullivan, R. M. 1980. Seasonal occurrence and haulout use in pinnipeds along Humboldt County, California. J. Mammal. 61:754- 760. Sydeman, W J., and S. G. Allen. 1999. Pinniped population dynamics in Central California: correlations with sea surface temperature and upwelling indices. Mar Mammal Sci. 15:446-461. Westlake, R. L., W. L. Perryman, and K. A. Ono. 1997. Comparison of vertical aerial photographic and gi'ound censuses of Steller sea lions at Ano Nuevo Island, July 1990-1993. Mar Mammal Sci. 13:207-218. Withrow, D. E. 1982. Using aerial surveys, ground truth methodology, and haul out behavior to census Steller sea lions, Eumeto- pias jubatus. M. S. thesis, Univ. Washington, Seattle, WA, 102 p. York. A. E. 1994. The population dynamics of northern sea lions, 1975- 85. Mar Mammal Sci. 10:38-51. York. A. E., R. L. Merrick, and T. R. Loughlin. 1996. An analysis of the Steller sea lion metapopulation in Alaska. In Metapopulations and wildlife conservation (D. R. McCullough, ed.), p. 259-292. Island Press, Washing- ton D.C. 63 Abstract— Tins study rcporis new nilormatioEi about soarobiii iPnoiKitiis spp. ) early life history from samples col- lected with a Tucker trawl (for plank- tonic stat;es) and a beam trawl (for newly settled fishi from the coastal waters of New^ Jersey. Northern scaro- bin, Prionoliis caroliiuis. were much more numerous than striped searobin, P. evotans, often by an order of mag- nitude. Larval Prionotus were collected during the period July-October and their densities peaked during Septem- ber For both species, notochord fle.\ion was complete at 6-7 mm standard length (SLi and individuals settled at 8-9 mm SL. Flexion occurred as early as 13 days after hatching and set- tlement occurred as late as 25 days after hatching, according to ages esti- mated from sagittal microincrements. Both species settled directly in conti- nental shelf habitats without evidence of delayed metamorphosis. Spawning, larval dispersal, or settlement may have occurred within certain estuar- ies, particularly for P. evolans; thus col- lections from shelf areas alone do not permit estimates of total larval produc- tion or settlement rates. Reproductive seasonality of P carolinus and P. evo- tans may vary with respect to latitude and coastal depth. In this study, hatch- ing dates and sizes of age-0 P. caro- linus varied with respect to depth or distance from the New Jersey shore. Older and larger age-0 individuals were found in deeper waters. These varia- tions in searobin age and size appear to be the combined result of intraspecific variations in searobin reproductive sea- sonality and the limited capability of searobin eggs and larvae to disperse. Larval and settlement periods of the northern searobin (Prionotus carolinus) and the striped searobin {P. evolansY Richard S. McBride Marine Field Station Institute of Marine and Coastal Sciences Rutgers University 800 Greal Bay Blvd Tuckerton, New Jersey 08087 Present address: Flonda Marine Research Institute 100 Eighth Avenue SE St Petersburg, Florida 33701 5095 E-mail address rictiard mcbnden fwc stale 11 us Michael P. Fahay Sandy Hook Laboratory Northeast Fisheries Science Center National Manne Fisheries Service, NCAA Highlands, New Jersey 07732 Kenneth W. Able Marine Field Station Institute of Marine and Coastal Sciences Rutgers University 800 Greal Bay Blvd Tuckerton, New Jersey 08087 Manuscript accepted 30 July (2001). Fish. Bull. 100:6.3-73 (2002). Although adult fish assemblages off- shore of the middle Atlantic states are fairly well known (e.g. Edwards, 1976; Colvocoresses and Musick, 1984; Gabriel, 1992), the early life history of many of these same species and the function of shelf habitats as nurs- ery grounds are poorly understood (e.g. Fahay, 1983, 1993; Able and Fahay, 1998). Because year-class strength is believed to stabilize prior to the early juvenile stage, information about the transition from the plankton to ben- thic (i.e. settlement) habitats should contribute to our understanding of the population processes of benthic fishes (Gushing and Harris, 1973; Gampana et al., 1989; Myers and Cadigan, 1993). Settlement is regarded as a dynamic period of early development because mortality rates can differ between pre- and postsettlement life stages (Sale and Ferrell, 1988), dramatic morpho- logical and physiological transforma- tions occur ( Youson, 1988; Markle et al., 1992; McCormick, 1993), and behav- iors become evident that allow for delay- ing settlement until suitable juvenile habitat is found (Cowen, 1991; Sponau- gle and Gowen, 1994). Ultimately, an understanding of the life cycle of any benthic species is constrained if the set- tlement period is not viewed as an inte- gral transition from the planktonic to the adult period. Our study contributes to an under- standing of how fishes use continental shelf habitats as nurseries with an ex- amination of the early life history of the northern searobin, Prionotus caro- linus, and the striped searobin, P. evo- lans. Both are common species in the coastal region between Gape God and Gape Hatteras, but relatively little is known about their early life history ow- ing largely to their low economic impor- tance in relation to the heavily exploit- ed fisheries of this region (McBride et * Contribution 2001-28 of the Institute of Marine and Coastal Sciences, Rutgers Uni- versity, New Brunswick, NJ 08901. 64 Fishery Bulletin 100(1) al., 1998). Both species are known to begin spawning as early as May and to continue spawning into Octo- ber as determined by maturity indices (e.g. Richards et al., 1979; Wilk et al., 1990). Prionotus spp. eggs and larvae are known to be seasonally abundant above the continental shelf and within some estuaries (e.g. Rich- ards et al., 1979; McBnde and Able, 1994) but eggs and lai-vae are difficult to identify to species on a rou- tine basis. Therefore we took advantage of recently reported morphological information (Able and Fahay, 1998) to examine ichthyoplankton collections. Our study was designed to examine how spawning patterns varied between two congeners, but intraspe- cific spawning variation also became evident. A sec- ond goal of our study was to examine settlement — to date not reported for either species. Both species un- dergo flexion and complete fin-ray development at about 6-8 mm SL ( Yuschak and Lund, 1984; Yuschak, 1985; Able and Fahay, 1998). Separation of prehensile rays on the pectoral fin. a major adaptation for ben- thic feeding (Morrill, 1895; Bardach and Case, 1965; Finger and Kakil, 1985), occurs in fish as small as 12 mm SL (Yuschak, 1985). Yet settled juveniles <25 mm SL are rare (Lux and Nichy, 1971; Richards et al., 1979; McBride and Able, 1994), which raises the question of whether Prionotus spp. are competent to settle after completing fin-ray development or whether they common- ly delay settlement. Using a novel combination of sam- pling gears, we collected a continuum of late lai-val and early juvenile Prionotus spp. to examine settlement di- rectly. We report for the first time species-specific larval abundances, distributions, ages, sizes, growth rates, and descriptions of early benthic existence. Materials and methods Collections were made in coastal waters of New Jersey, specifically near Beach Haven Ridge (Fig. 1, Table 1), a prominent sand ridge formation that rises to about 8 m depth and is surrounded by depths of 14-16 m (Stahl et al., 1974). Sampling frequency at two stations, one land- ward and the other seaward of the ridge, was every two to six weeks from July 1991 to November 1992. Two tows of a Tucker trawl ( 1 m-i were made at each station in a double, stepped-oblique fashion. One tow was made from the sur- face to the bottom (three minutes duration) and the other tow was fished from the bottom back to the surface (six minutes). Newly settled juveniles and older fishes were sampled with a 2-m beam trawl in Great Bay estuary, near Beach Haven Ridge, as well as in other habitats (Fig. 1). The data from these stations were arranged in the follow- ing gi'oups: 1) the two principal ridge stations (described above); 2) miscellaneous stations scattered on top of and around the ridge; 3) stations along a transect leading directly offshore from the ridge; and 4) a cluster of stations within nearby Great Bay. Generally, three tows were com- pleted at stations immediately landward and seaward of the ridge, but only two tows were completed at other sta- tions. Beam trawl tows offshore of Little Egg Inlet took 0 1 2 U4=J kilometers -/ MULLICA "" ,5' RIVER £*■ i\ -, Areata / A' ^ BAY A LITTLE / lil A A EGG / / / BEACH HAVEN/ ^ ridge/ v./ / 39 30' - / O^ JL. A. / / A 39 25' - 74 20' / ' 74" 10' Figure 1 Map of sampling station locations in southern New Jersey, including the main stations at Beach Haven Ridge (landward and seaward: filled circles), other ridge stations (open circles), continental shelf transect stations (filled triangles), and estuarine stations (open triangles). The state of New Jersey, and the study location, are shown in the inset. one minute to complete, but estuarine tows were reduced to 20 or 30 seconds to avoid collecting large volumes of macroalgae, detritus, shell, etc. Sampling occurred during daylight unless otherwise stated. Details of sampling pro- cedures are provided by Hales et al.' Volume or area sampled was calculated by using a flow-meter for ichthy- oplankton collections or a meter wheel for beam trawl collections. Larval density is presented as the geometric mean number of fislVm' for Tucker trawl collections. Juve- nile density is presented as the geometric mean number of fislVm- of sea bottom. Calculations of geometric means follow Sokal and Rohlf ( 1981). The standard length (SL) of all, or at least 20 fish per tow, was measured after the fish were presei-ved in 95'^?- ETOH. The term "lai'va" was used in reference to individu- als collected in Tucker trawl tows. Preflexion larvae were distinguished from flexion lai-vae by the absence or pres- ence, respectively, of cartilaginous urals on the ventral edge of the notochord tip; the development of these urals accompanied flexion of the notochord tip (Kendall et al., 1984 ). Larvae were characterized as postflexion stage once the notochord tip moved anterior to the posterior edge of the hypurals. Daily age was estimated from counts of sagittal otolith mici'oincrements, which were validated as daily by Mc- Bride.- Otoliths with a maximum length less than about Hales. L. S., Jr., R. S. McBride, E. A. Bender, R. L. Hoden, and K.W.Abie. 1995. Characterization of non-target inverte- brates and substrates from trawl collections during 1991-1992 at Beach Haven Ridge (LEO-1.5) and adjacent sites in Great Bay and on the inner continental shelf ofT New Jersey. Techni- cal report (contribution 95-09). 34 p. Institute of Marine and Coastal Sciences, Rutgers, The State University of New Jersey, New Brunswick, NJ. ' McBride, R. S. In review. Spawning, growth, and ovei-wintering size of searobins (Triglidae: Prionotus carolinuK and P. evolans). McBiide et al Ldivdl dnd settlement peiiods of Pnonotus catolinus and P cvolans 65 500 \\n\ were removed and mounted whole on glass slides in immersion oil. Otoliths longer than about 500 pm were mounted in nail polish on a glass slide, sanded with 1500 grit sandpaper along the sagittal plane, and polished with 0.3-pm grinding powder. Immersion oil was used liberally to enhance the clarity of all otoliths, and polarized light aided the viewing of microincrement structure. Micro- increment counts were made with a compound microscope, typically at 400x. Slides were coded and microincrements were counted by one reader on three separate occasions. A constant of 4 days, representing the period between hatch- ing and deposition of the first ring, was added to the mean microincrement count to estimate age since hatching (Mc- Bride2). Preserved (95% ETOH) P. carolinus and P. evo- lans were selected in a stratified i0.5-mm intervals), ran- dom manner to compare ages and lengths. Microincrement counts from this comparative material ranged, based on all individuals, between 07i and 32% of the mean micro- increment count for each otolith (mean=12.0'~f ; 11=41). Pnonotus carolinus were collected in far greater num- bers than P. evolans and they were examined in greater detail. Size and age distributions were initially defined from collections made during the period of peak seasonal abundance (i.e. late September 1991), when a random sample of 34 larvae was selected from a Tucker trawl sam- ple for 23 September. Another sample of juvenile P. caro- linus was selected from a 2-m beam trawl tow on 23 Sep- tember 1991 at a station near the above plankton tow (Table 2). Four final samples were selected from 2-m beam trawl tows set one month later (21-22 October) at four sta- tions along a transect of varying depths. Otoliths from all juveniles collected at these stations were analyzed (i.e. on- ly fish that were mutilated or that had cracked otoliths or otoliths sectioned beyond the core were e.xcluded). Gener- al methods of measuring and staging individual fish, and preparing otoliths, followed that described above. Sagittal microincrements were counted on two (for lai-vae) or three (for juveniles) separate dates by one reader. The range of these microincrement counts, for all individuals, was from 0.0% to 25.0% (mean=9.6%; n = 127) of each mean count. Results Interspecific comparisons Prionotus carolinus were more numerous and occurred more frequently than P. evolans in nearly all collections, typically by an order of magnitude (Table 1). Spawning by both species occurred from at least July to October off- shore of southern New Jersey (Fig. 2). Modal size of larvae generally increased with time, but there were exceptions that indicated a pattern of multiple spawning events. For example in August 1991, modal size for Prionotus spp. and P. carolinus was notably smaller (3-4 mm) than the pre- vious month (5-6 mm) (Fig. 3). Peak larval abundances varied somewhat between years but were highest from July to September. Prionotus carolinus was the smaller but older congener at each developmental stage. Size and stage were com- ■- £ d. U] a ° c- (ft §3 hi CO 3 -r I- o o o o o o O .S ''- O X ^^ T3 Oj T3 o a o O i _C "^ -i X c Cj > P u y-j !-§ o =; a OJ 5 _ C a-, -a o 00 o CO o CD 00 CO '^ o _ o 5 t^ lO CM 00 CO t-- >= !- rM 't (M •II a; 0; p =C S CD ^ '5- Z i CTi c "^ ^ to CT> CM CO c- 00 CO a " 3 2 C^ lO CD 00 C^ Tf IM CO >•! •z !- rt ra o ^ G a> a > > > .,-> o o Q> o CJ o o u 2 t x: Q 2 Q z o Z Z o O C ^ a -^ c "3 c bfi c *-> ^ M .2.S o •^ CO CO 3 CO -3 O < 3 < ^=S 7 « - t) a ^ ^ r- £ o c i_ ,—1 0-1 ,-H (N ,—1 C-1 ,—1 CM o-l r- o c rt (71 (35 (35 05 Oi 05 Ci (T> 05 :2i (35 Oi OJ (T> Ol 05 O^ 05 (35 S3 o 3 rt i 2 « 3 5 -J -^ -2- = 2: 3 X 1 CO _ J ci _ 1 _ C3 - ^ O g £ CO (ft £ u *-• £ (ft £ £ -C Cfi E i3 ^ 1 e g 3. .s g ^ g J £ ^ s ■5 ^ X o g c £ g £ £ e Qj Qj [/} lO (N CO CN CO iTJ CD ci ? C J2 rt e = 1 X tUD n m C ~ data gener ampli c lO ^ I> -+ 00 D -C OQ 1 f- -K 1 1 1 1 ^ bjD C cfl o Ol t— ' t- * ■S s^ ^^ CD *"* 1 o fc B S C t. *j S *" a; M ^ f: ^ t- C 0) o >-, 0) Tj n and b arating Figure ] ~ Si. CO c o 3 9- o ^ Oj n, -a (ft -a _o C3 lis of planl acters for s ntheses). S be c a; > C3 c CO CO -a c b£ u u ■s (LI X c CO i- Qi >> CO m TO t_ ni s CJ CO 0) bo CO t^ ra ^ CO -C T3 p U-1 O O. « ZJ ■1.^ i- c/: CQ o S a 66 Fishery Bulletin 100(1) Table 2 Daily age. size, and hatching dates for planktonic (flexion and postflexion stages) larvae and benthic (settled stage) juveniles of P. caroHnus collected in September and October 1991, offshore of southern New Jersey (see Fig. 8 for station locations). Data are presented as means (±1 standard error), and the range of values is given in parentheses. Larvae were collected with a 1 « 1-m Tucker trawl (0.505-mm mesh) and juveniles with a 2-m beam trawl (6-mm mesh). Date Stage Station n Age (days) Length (mm) Hatching date 23 Sep flexion 0T5 9 15.4 ±0.73 (12-17.5) 5.3 ±0.20 (4.1-6.2) 8 Sep ±0.73 (6Sep-ll Sep) 23 Sep postflexion OT5 25 17.7 ±0.53 (12-23.0) 7.0 ±0.20 (5.7-9.5) 5 Sep ±0.53 (31 Aug-U Sep) 23 Sep settled OT5 23 37.4 ±1.91 24-61.3) 12.2 ±0.44 (8.5-15.8) 17 Aug ±1.91 (24 Jul-30Aug) 21 Oct settled OT2 15 60.9 ±3.06 (46-94.7) 17.5 ±1.42 (12.8-30.4) 19 Aug ±3.06 (18Jul-5Sep) 21 Oct settled 0T5 13 62.2 ±2.81 (52-90.7) 16.8 ±1.. 58 (12.8-35.3) 20 Aug ±2.81 (22Jul-30Aug) 22 Oct settled Sta. C 29 75.1 ±2.71 (54-134.0) 21.9 ±1.44 (13.1-59.4) 8 Aug ±2.71 (10Jun-29Augl 22 Oct settled Sta. E 13 90.0 ±3.20 (69.3-105.3) 26.3 ±0.96 (20.2-33.7) 24 Jul ±3.20 (8 Jul-13Aug) 1 g '*^ ^ '2^ ^ -73.2) 35 30 - 2.5 20 - 15 10 0,5 0.0 ■X - Prionolus spp. — P evolans — P carolinus ■ ■ I ■ • I — ■ ■ I ■ — '—] — — — r-" — • 1 Jul 1 Oct 1 Jan 1 Apr 1 Jul 1 Oct Figure 2 Density (geometric mean number of larvae per 100 m^ |±1 standard error, SE]) of Pnonotiis spp.. P. carolinus, and P. evolans larvae for each cruise near Beach Haven Ridge, based on daylight tows of a Tucker trawl at the land- ward and seaward stations (see Fig. 1). Note break in scale (range of SE bars are given in parentheses). Pnonotus carolinus 50 40 30 20 50] 40 30 20 10 0 50 40 30 20 10 0 50 40 30 20 10 0 July n=64 Pnonotus evolans Pnonotus spp. July ffeSI \L^ 10 0 100 August 80 n=46 60 EtU 20 0 September 50 f7=553 40 30 20 10 0 11^ 2 4 6 8 1012 September rtll 8 10 12 Standard length (mm) Figure 3 Size frequency of Prionolus carolinus, P. evolans, and Pri- onotus spp. from Tucker trawl collections near Beach Haven Ridge during July-September 1991. n = total number of larvae collected. pared for 534 P. carolinus and 81 P. evolans collected with the Tucker trawl during both day and night. Flexion was complete at a larger size for P. evolans than for P. ca/-- olinus (range: 6.7-7.5 mm versus 5.4-6.8 mm SL), and planktonic postflexion P. evolans larvae were captured at larger sizes than postflexion P. carolinus (range: 6.7-11.9 versus 5.4-9.8 mm SL). Pnonotus evolans completed flex- ion at a younger age than P. carolinus (approximately 13 McBnde et a\ Larval and settlement periods of Pnonotus carolinus and P cvolans 67 versus hS days after hatching) \V\g. 4). Both species set- tled as early as 18-19 days after hatching, but this was more characteristic of P. cvolans: most P. canilinus did not settle until 24-25 days old. Both species grew relatively slowly, and approximately linearly, during the larval and early juvenile period (i.e. <0.3 mm/d; Fig. 4, and next subsection). These slow growth rates, combined with the late peak in spawning (i.e. around August), resulted in small body sizes by the onset of win- ter. These smaller body sizes were particularly true for P. carolinus, for which the most pronounced size mode was 10-15 mm SL in autumn 1991 and 1992 (Fig. 5). At this time (i.e. September-December), individuals <50 mm SL constituted 85% of P. carolinus and 52% of P. evolans from all beam trawl tows combined; during autumn a majority of Prionotus spp. were <25 mm SL. At beam trawl stations, densities of P. carolinus were consistently higher than those for P. evolans in both 1991 and 1992 (Fig. 6). Geometric mean densities of age-0 P. carolinus during the peak period of settlement (Septem- ber-October) were much higher in 1991 (8.98 fish 100/m-) than in 1992 (1.56 fish 100/m-'). Geometric mean densities of age-0 P. evolans during September-October were also higher in 1991 (0.32 fish 100/m2) than in 1992 (0.09 fish 100/m-'). These interannual differences were consistent with higher larval densities of both species in 1991 versus 1992 (Fig. 2). Maximum densities of age-0 searobins at a single station reached 28.9 P. carolinus 100/m- and 3.2 P. evolans lOO/m^, both in September 1991. Searobins larger than 150 mm SL were collected infre- quently from June to October; occasionally they were found together with age-0 conspecifics in the same beam trawl tows. Age-0 searobins of both species were collected primarily in continental shelf versus estua- rine habitats during July-December (Fig. 7). Settlement of Prionotus carolinus The seasonality of settlement by P. carolinus, although lasting from at least July to October, 1991, was punc- tuated by a 2-3 week period in September when the vast majority of larvae appeared to settle near Beach Haven Ridge (Fig. 8). Densities of age-0 P. carolinus near Beach Haven Ridge were very low during both July and August (geometric means ranging from 0.0 to 1.1 fish 100/m-). During September, densities increased dra- matically (range: 0.8-7.3 and 0.8-28.9 fish lOO/m^ on September 12 and 23-24, respectively). Individuals were collected at all stations along a depth transect, from 6 to 16 m, in late September Settled, age-0 P. carolinus were still widespread and abundant in late October (0.0-13.3 fish lOO/m^), but they were not collected on 2 December 1991, and on 28 January and 10 March 1992. Collections for 23 September 1991 demonstrate a wide range of P. carolinus developmental stages and ages present at Beach Haven Ridge (Fig. 9A). All flex- ion stages were present (6.5% preflexion, 26.0% flex- ion, and 67.4% postflexion; /? = 169). Planktonic larvae subsampled randomly from a Tucker trawl tow (;!=34; Table 2) had hatched during a two-week period from 9 preflexion/flexion A pelagic/postflexion a settled luveniles ■ □ 15. □ ° - ■ 0 10 5- ■ - D a -"^--'i-a ° - tU'^^ • CP 6«* 0- 0 5 10 15 20 25 30 35 40 45 Age (days after tiatctiing) Figure 4 Relationship between daily age and length for Pnon- otus carolinus (open symbols) and P. evolans (filled symbols) for preflexion and flexion stages collected with a Tucker trawl (circles), postflexion stages col- lected with a Tucker trawl (triangles), and settled juveniles collected with a beam trawl (squares). The upper dashed line indicates the approximate size at settlement, and the lower dashed line indicates size at completion of flexion for both species. 10 Pnonotus carolinus Summer. 1991 r7=63 .rprm Pnonotus evolans 20 10 69.7 Autumn, 1991 n=267 Summer, 1991 n=^ Autumn, 1991 n=39 Q- 10 Summer, 1992 n=52 T1 , ,n r|ff][|l1lr]lln}i 85.7 10 Autumn, 1992 n=42 0 50 100 150 200 0 50 Standard length (mm) Figure 5 Size frequency of Prionotus carolinus and P. evolans for beam trawl collections near Beach Haven Ridge (daylight tows at the landward and seaward stations (Fig. 11). Data were pooled by season: summer (May-August) and autumn (September-Decem- ber), n = total number of fish collected. No data for January- April 1992 are shown because only a single fish (P. carolinus; 4.5 mm SL) was collected at these stations during this period. 68 Fishery Bulletin 100(1) August 31 to September 11. Individuals collected by beam trawl on the same day (23 September 1991) had hatched about 2 weeks earlier (from 24 July to 30 August) than the above larvae (Fig. 9). These juveniles appeared to settle as young as 24 days after hatching and at sizes as small as 8.5 mm SL (Table 2). The total hatchmg date distribu- tion for both larvae and newly settled juveniles collected on September 23 reflected a spawning period that ranged from late July to early September and that peaked in late August and early September. Settled juveniles with a similar hatching date distribu- tion were identifiable one month later at stations near Beach Haven Ridge, but not at stations farther offshore (Fig. 9, B and C). Fish collected near Beach Haven Ridge on 21 October 1991 had a hatching date distribution with a mode from late August through early September and the overall distribution was skewed to the left. This period was similar to the hatching date distributions for larvae and newly settled fish collected on 23 September 1991. In contrast, fish collected from offshore stations (i.e. stations C and E) on 22 October 1991 were 2-4 weeks older and 5-10 mm larger on average (Table 2, Fig. 9). Plots of P. carolinus size versus age did not indicate any abrupt change at settlement, specifically for postflex- ion lai^vae and settled juveniles collected on 23 Septem- ber 1991 (Fig. 10). Growth rates for this September collec- tion fitted a hnear model (SL=3. 24-1-0. 229[age|; r2=0.77). Because Prionotus lai-vae hatch at about 3 mm SL (Yus- chak, 1985), this model's y-intercept is biologically realis- tic. Growth rates offish collected in October did not differ significantly between stations (ANCOVA:p7-o6.^, .^=0.13, pro6.,,,,p ,=0.51); therefore the data were pooled. Linear, least squares regression of all data produced an unreal- istic y-intercept (SL=-7.01-H0.382lage]: SE^=2.0; ;--=0.74). This model was rerun after restricting the y-intercept to 3 mm and the resulting equation indicated that age-0 P. ca/'olinus continued to grow at about 1 mm every 4 days (SL=3 +0:25l[age\: ;-=0.65) as they had during the lai-val and settlement period. Size and age of P. caro/i>H^s juveniles varied significantly along a 12-km transect ( 12-20 m depths; Fig. 1 ). The linear relationship: Hatching age = 17.8 -i- 3.43 x depth; r-=0.35, P<0.01; ri=69) showed that for every two meters change in depth offshore the fish collected were about one week older on average (Fig. 11). Sampling in both 1991 and 1992 showed a consistent trend for larger (and presumably old- er) fish to be collected in deeper water in October and No- vember (Fig. 12). After accounting for the effects of depth, or possibly the distance from shore, it appeared that fish reached a larger size in October of 1991 than in 1992 or that larger fish in 1992 were not found in the sampling area. Discussion Spawning grounds and seasonality of spawning Prionotus caro/inus are more abundant than P. evolans in continental shelf habitats whether they are measured as eggs, larvae, juveniles, or adults (Keirans et al., 1986; 1601 120- 8,0 4.0 00 Prionotus carolinus Age-0 Age-1 + ^ u - Prionotus evolans r —•- Age-0 1 5- —A — Age-1 + T 1 0- <► \ ' T 0,5- J n m UU- •I* 1 ' ' 1 • ■ 1 ■ • 1 • ■ 1 ■ ■ 1 Jul 1 Oct Figure 6 Density (geometric mean number of fish per 100 m- |±1 standard en-or | ) of different cohorts of postsettlcnient Prionotus carolinus and P. evolans during daylight tows at the landward and seaward stations near Beach Haven Ridge. Note scale differences for each species. McBride and Able, 1994; Able and Fahay, 1998; McBride et al., 1998; our present study). The low numbers of P. evo- lans observed in our study may be biased somewhat by our focused effort to sample the continental shelf rather than estuaries. Prionotus evolans reside in shallower, warmer habitats than do P. carolinus during the spawning season (McBride and Able, 1994). If P. evolans spawn to some degi'ee in shallower waters or estuarine habitats, then this would at least partly explain the generally low abundance of P. evolans early life stages in our collections. In general, we expect that larval distributions are good predictors of spawning locations for both Prionotus spe- cies because of the short (i.e. about three weeks) lai-val dispersal periods of these species (e.g. Houde and Zastrow 11993] reported several shelf species with planktonic du- ration >100 days). In some coastal areas, the distribution of Prionotus spp. eggs and larvae indicates that spawning may be limited to estuaries; however, the abundance of Pri- onotus larvae offshore of New Jersey suggests that spawn- ing by these species occurs outside estuaries as well. For example, Merriman and Sclar ( 1952) did not find Prionotus McBride et al : Larval and settlement periods of PnonottK camlinus and P evolans 69 03 3.0 20 1 0 0.0 100- 90 80' 70 60 50 40 30 2.0 1 0 00 Jul-Aug 1991 _Qd_ -OiL Sep-Dec 1991 i 30 2,0 1,0 00 100 90- BO- ZO 60 50 40 30 20 1 0 00 Main ridge Other ridge Transect Estuary stations stations liw. Mam ridge May-Aug 1992 0 0 Sep-Nov 1992 K. Ottier ridge Transect Estuary stations stations Figure 7 Density (geometric mean number offish per 100 m^ |±1 standard error]) of age-0 Pnonotus carnlinuf: (open bars) and age-0 P. evolans (filled bars) collected with a beam trawl from four major station groups ( nd= no data; 0=sampling occurred but no Pnonotus were collected). See Figure 1 and Table 1 for station groupings and locations. 74°20' / 74°10' 1 24 July 1991 '/I \ - 39°30' / E 20. V ^/ - 39 25' / o o • 10. 10 1 /.- 74°20' # / ! 1 / 74°10' 1 14-15 August 1991 / / / - 39°30' / / / / / ./ - 39 25' / - 39°30' 1 — /T- 74°20' /74°10' 21-22 October 1991 ^ fl 's 1 w / 1 1 74°20 / 74 "1 0' 12 September 199/ / - 39° -■ i / / y>sf ^ /■ / 'K 1 # / «t ; / ^ / J -39° 25' / / Figure 8 Densities (geometric mean [±1 standard error) ) of agc-0 Pnonotus carolmus collected with a beam trawl during six consecutive cruises (July-December 1991 1. Number of stations varied between cruises. The scale bar indicates 10 fish/100 m-. 70 Fishery Bulletin 100(1) 100 80. 60 40 20 0 \ Pelagic larvae & benthic luveniles (at OT5) 23 September 1991 ■ flexion larva (n=9) E3 postflexion larvae (n=25) n benttiic luveniles (n=23) n n n Jl _ ■ 1 ■ ■ ' f ^° -■ B Benttiic juveniles - near Beacti Haven Ridge 21 October 1991 ^ ■ Station 0T2(n=1 5) nStationOTS (n=^3) 40 30- 10. 50-, 40 30 20. 10. 0 Q Benttiic juveniles - offstiore of Beacti Haven Ridge 22 October 1991 ■ Station C (n=29) D Station E{n=13) H m Jun 1 Jul 1 Aug 1 Sep 1 Ocl 1 Hatctiing date Figure 9 Hatching-date distributions for larval and newly settled juvenile Pnonotus carolinus collected 23 September 1991 (A) and for juve- niles collected on 21 October 1991 (B) and 22 October 1991 IC). See Figures 1 and 8 for locations of sampling stations. 60-, + Sept - Flexion larvae » Sept - Postflexion larvae 50- • Sept - Benthic luveniles A Oct -Benthic juveniles 1 40. i 30. ■D en 1 20. CO 10- 0- ^ A . .v.. ■• ( ) 20 40 60 80 100 120 140 Age (days after hatching) Figure 10 Age (days after hatching) and standard length (mm) for flexion, post- flexion, and juvenile Pnonotuf; carolinus collected seaward of Beach Haven Ridge on 23 September 1991 and 21-22 October 1991. spp. eggs or larvae in Block Island Sound, and Able and Fahay ( 1998) did not observe Prionotus larvae above the continental shelf north or east of Hud- son Canyon, New York. Instead there are many reports of Prionotus eggs, larvae, and juveniles in southern New England estuaries, specifically in Long Island Sound (Wheatland, 1956; Richards, 1959; Williams, 1968; Richards et al., 1979) and Narragansett Bay (Herman, 1963; Bourne and Go- voni, 1988; Keller et al, 1999). Thus, the relative importance of estuaries versus shelf habitats as spawning grounds for Prionotus may vary in other regions compared with our results for New Jersey. Nonetheless, Prionotus spawning seasonality ap- pears to follow a pattern similar to that of other species with a wide latitudinal range that have a shorter spawning season at higher latitudes (e.g. Conover, 1992) and that spawn later in the south (e.g. Barbieri et al., 1994). An important departure from this general trend is that Prionotus repro- ductive seasonality may vary not only with respect to latitude but along an estuary-shelf gradient as well. Because adults of both Prionotus species enter estuaries early in the spring and migrate back out to the shelf in summer (McBride and Able, 1994), we postulate that spawning occurs first in estuaries at a given latitude. In support of this hypothesis are the collective results from our study and other published reports. After the summer spawning peak within estuaries such as Chesapeake Bay and Long Island Sound, Priono- tus spawn during August and September offshore of Chesapeake Bay and New Jersey. In contrast, spawning does not continue into late summer off- shore of southern New England (Pearson, 1941; Richards et al., 1979; Able and Fahay 1998). To explain this potentially novel spawning pat- tern does not require any new controlling mecha- nism other than that used to explain spawning by other coastal fishes of the region. Temperature and photoperiod are known to influence spawn- ing activity in fishes (Burger, 1939) and may in- fluence spawning seasonality of searobins. Water temperatures offshore of the middle Atlantic sea- board are known to fluctuate widely both tem- porally and spatially (Colvocoresses and Musick, 1984) and this fluctuation affects the spawning pattern of many species. For example, a simple south to north progression of spawning activity above the shelf is evident for Centropristis stri- ata (Able et al., 1995) and Scophthalmus aquo- sus (Morse and Able, 1995). For Prionotus, how- ever, we propose that spawning seasonality is controlled by an interaction between latitudinal and estuarine gradients of temperature (i.e. earli- er spawning in estuaries occurs because of earlier warming of these shallow embayments). Temper- ature has already been shown to affect the distri- bution of Pr-ionotus adults along both latitudinal and estuarine gi-adients (McBride and Able, 1994; McBride et a\ Larval and settlement periods of Pnonotiis carolinwi and P evolam McBride et al.. 1998). Because few other .species use both estuarine and shelf habitats lor spavvninj^. such pallcrns arc not commonly obser\i'(l. Linking metamorphosis and settlement Prionotus carollniis are otten ranked as among the most abundant species in regional trawling surveys for adult fish or plankton sui-veys for larval fish (McBride and Able, 1994). The results from the small-mesh beam trawl used in our experiment demonstrate that the juvenile stages of P. caroliniift are also very abundant offshore. Age-0 Pri- onotus spp. are found in estuaries, as discussed above for the southern New England region, but our observation of high densities of juvenile Prionotus in shelf habitats off- shore of New Jersey suggest that neither species requires estuarine nursery habitats during their life cycle. Most searobins complete their life cycle in continental shelf hab- itats (Hoff, 1992), with the notable exception of P. scitulus whose young are concentrated in lower salinity, estuarine habitats (Ross, 1978). Our findings of age and size at settlement largely agree with Yuschak and Lund's (1984) and Yuschak's (1985) de- scriptions of early development of cultured specimens. The developmental rate of cultured specimens of P. carolinus and P. evolans-^ did not differ notably from our obsei-va- tions of field-collected individuals, which further supports our conclusion that neither species delays settlement. Pri- onotus evolans. cultured at 20°C, were all at prefiexion and flexion stages after 11-13 days; they were at flexion and newly postflexion stages after 18-20 days; and all were at the postflexion stage at 25 days. All available P. carolinus specimens, cultured at 15°C, were less than 20 days old; they followed a similar, if slightly slower, rate of development compared with P. evolans. Fin-ray development, which greatly facilitates locomotion, is complete in postflexion individuals. Prehensile, chemosen- sory pectoral rays, which would facilitate benthic feeding, are completely separated by 11.5 mm SL. Thus, on the basis of cultured and field-caught specimens, both species are well developed (i.e. are similar to adults) and well-suited for a bottom- feeding and swimming life style as they complete flexion. Other species delay metamorphosis to set- tle during favorable lunar phases (Sponaugle and Cowen, 1994), but settlement by Prionotus was so concentrated in a single month (i.e. September) that we could not test spawning or settlement cy- cles in more than one month. Nevertheless, be- cause larvae o{ Prionotus species commonly bury themselves in loose substrate (Bardach and Case, 1965), and this material was common to all our sampling stations (Hales et al.M, competent lar- vae are likely not habitat limited (one mechanism identified with the delay of settlement). The variation of P. carolinus sizes and ages along a depth gradient could be caused by one or a combination of three processes. There could be differential larval survi- vorship, juvenile movements, or adult reproduction rates a 100- c u 90 Z 70. S 60. A i < 50- 12 13 14 15 16 17 18 19 20 30-, Standard length (mm) B I 1 1 12 13 14 15 16 17 18 19 20 Depth (m) Figure 11 Ago (A) and size (mean ±1 standard error) (B) of juvenile Prionotus carolinus plotted in relation to depth. Data are for fish collected on 21-22 October 1991 (near and offshore of Beach Haven Ridge: sta- tions 0T2, 0T5, C, and E) and aged by using sagittal microincrements (see Table 2 and Fig. 9 for sample sizes). 30-, E 25. o September 23-24, 1991 o October 21-22, 1992 ■ October 27, 1992 ▲ November 10, 1992 , 1 i en .§ 20. ■D ro •o |1 T [ ro or, 15. 10 0 « 0 (-. 1 ) 10 15 20 25 Depth (m Figure 12 Standard length (mm; mean [±1 standard error]) of juvenile Pnono- | tus carolinus in relation to depth for four separate cruises in 1991 (open symbols) and 1992 (filled symbols). AH benthic juveniles col- lected were included in these calculations This material was cultured by P. Yuschak (see Yuschak and Lund 119841 and Yuschak 11985]) and has been examined by R.S.M. 72 Fishery Bulletin 100(1) across the shelf to account for this spatial pattern of older and larger juveniles farther offshore. The last process was identified earlier as potentially important. Testing and eliminating these hypotheses, however, requires spatially explicit lan'al distribution data, in addition to the benthic data that we collected, which would allow a comparison of pre-and postsettlement distributions with abundance of Prionotus propagules. Spatially explicit environmental data would also be useful because we obsei-ved dynamic changes in the physical parameters in our sampling area, and we suspect that these could affect Prionotus sui-vival. Vertical stratification of the water column was noted near Beach Haven Ridge on 14 August 1991, but not during cruises in September or October 1991. Low dissolved oxy- gen levels near the bottom of the ridge, at about 3 ppm (in contrast to >8 ppm in the upper water column) could have negatively affected settlement rates near the ridge in 1991. Stratification near the ridge was also noted in 1992 with similarly depressed levels of dissolved oxygen (Hales et al.^). Low dissolved oxygen offshore of New Jersey is not uncommon (Falkowski et al., 1980; Glenn et al., 1996) and may be another process that can contribute to geographic variations in size and age of Prionotus species offshore of the middle Atlantic states. We postulate that spatially ex- plicit patterns of reproductive seasonality and age-0 fish size for P. carolinus and P. evolans within coastal waters offshore of the middle Atlantic states are related to each other because the short planktonic larval durations for both species limit larval dispersal. Interannual variations in water temperature or vertical stratification of oxygen concentrations may be proximate causes for these geo- graphic variations of reproductive seasonality and age-0 size. These patterns could be somewhat unique to searo- bins, compared with other regional fishes, because searo- bins use both estuarine and shelf habitats for spawning. Acknowledgments R. Cowen, J. Hare, J. P. Grassle, R. Loveland, and C. L. Smith contributed thoughtful discussions and helpful com- ments on earlier drafts. S. Richards provided cultured specimens of searobins and miscellaneous data that had been used in P. Yuschak's research. This study was part of a doctoral dissertation (R. S. M.) and was supported by the Institute of Marine and Coastal Sciences (IMCS), Rutgers University Marine Field Station, Anne B. and James H. Leathem Fund, Manasquan Marlin and Tuna Club, NOAA National Undersea Research Program for the Middle Atlantic Bight, and NOAA Sea Grant Program. Final preparation and revision of this manuscript were made while the senior author was employed by the Florida Marine Research Institute. We thank all of the above. Literature cited Able, K. W.. and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the Middle Atlantic Bight. Rutgers Univ. Press, New Bruns- wick, NJ, 342 p. Able. K. W, M. P Fahay, and G. R. Shepherd. 1995. Early life history of black sea baas, Centropristis stri- ata, in the mid-Atlantic Bight and a New Jersey estuary. Fish. Bull. 93:429-445. Barbieri, L. R., M. E. Chittenden Jr., and S. K. Lowerre-Barbieri. 1994. Maturity, spawning, and ovarian cycle of Atlantic croaker, Micropof;oniaf> undulatus. in the Chesapeake Bay and adjacent coa.stal waters. Fish. Bull. 92:671-685. Bardach, J. E., and J. Case. 1965. Sensory capabilities of the modified fins of squirrel hake (Urophycis chuss) and searobins iPnoiidtus carolinus and P. evolans). Copeia 1965:194-206. Bourne, D. W., and J. J. Govoni. 1988. Distribution of fish eggs and lai-vae and patterns of water circulation in Narragansett Bay, 1972-1973. Am. Fish. Soc. Symp. 3:132-148. Burger, J. W. 1939. Some experiments on the relationship of the external environment to the spermatogenic cycle of Fiinduhis het- eroclttus. Biol. Bull. 77:96-103. Campana, S. E.. K. T Frank. P. C. F Hurley. R A, Koeller, F H. Page, and P C. Smith 1989. Survival and abundance of young Atlantic cod ( Gadus morhua ) and haddock (Melanogrammus aeglefinus) as indi- cators of year-class strength. Can. J. Fish. Aquat. Sci. 46(suppl. 11:171-182. Colvocoresses. J. A., and J. A. Musick. 1984. Species associations and community composition of Middle Atlantic Bight continental shelf demersal fishes. Fish. Bull. 82:29.5-313, Conover, D. O. 1992. Seasonality and the scheduling of life history at dif- ferent latitudes J. Fish Biol. 41lsuppl. B):161-178. Cowen, R. K. 1991. Variation in the planktonic lai-val duration of the tem- perate wrasse Semicossyphus pulchcr. Mar. Ecol. Prog. Ser 69:9-15. Cushing, D. H., and J. C. K. Hams. 1973. Stock and recruitment and the problem of density de- pendence. Rapp. P.-V. Reun. Cons. Perm. Int. Explor Mer 164:142-155. Edwards, R. L 1976. Middle Atlantic Fisheries: recent changes in popula- tions and outlook. In Middle Atlantic continental shelf and the New York Bight: proceedings of the symposium, 3-5 November 1975. Special symposium, vol. 2 (M. G. Gross. Lawrence, ed.), p. 302-311. Am. Soc. Lininol. Oceanogr. Lawrence, KS Fahay, M. P 1983. Guide to the early stages of marine fishes occun-ing in the western North Atlantic Ocean, Cape Hatteras to the southern Scotian Shelf J. Northwest Atl. Fish. Sci. 4:1-423. 1993. The early life history stages of New Jersey's saltwa- ter fishes: sources of information. Bull. N.J. Acad. Sci. 38:1-16. Falkowski, R G., T S. Hopkins, and J. J. Walsh. 1980. An analysis of factors affecting o.\ygen depletion in the New York Bight. J. Mar. Res. 38:479-506. Finger, T. E., and K. Kakil. 1985. Organization of motorneuronal pools in the rostral spinal cord of the sea robin, Prionotus carolinus. J. Comp. Neurol. 239:384-390. Gabriel, W.L. 1992. Persistence of demersal fish assemblages between Cape Hatteras and Nova Scotia, Northwest Atlantic. J. Northwest Atl. Fish. Sci. 14:29-46. McBride et a\ : Larval and settlement periods of Pnonotus carollnus and P evolans 73 Glenn. S. M., M. F. Crowley, D. B. Haidvogel, and Y. T. Song. 1996. Underwater observatory captures coastal upwclling events off New Jersey. Eos 77:233-236. Herman. S. S. 1963. Planktonic eggs and larvae of Narragansett Bay. Lininol. Occanog. 8:103-109. Hoff.J. G..Jr. 1992. Comparative biology and population dynamics of scarobins (Genus Prionotus) with emphasis on populations in the northwestern Gulf of Me.xico. Ph.D. diss.. Virginia Institute of Marine Science. College of William and Mary, Gloucester, VA, 229 p. Houde, E. D., and C. E. Zastrow. 1993. Ecosystem- and taxon-specific dynamic and energet- ics properties of larval fish assemblages. Bull. Mar. Sci. 53:290-335. Keirans. W. J.. Jr., S. S. Herman, and R. G. Malsberger 1986. Differentiation of Prionotus carolinus and Prionotus evolans eggs in Hereford Inlet Estuary, southern New Jersey, using immunodiffusion. Fish. Bull. 84:63-68. Keller, A. A., G. Klein-MacPhee, and J. St. Onge Burns. 1999. Abundance and distribution of ichthyoplankton in Narragansett Bay Rhode Island, 1989-1990. Estuaries 22:149-163. Kendall, A. W. Jr., E. H. Ahl.strom, and H. G. Moser 1984. Early life history stages of fishes and their character- istics. In Ontogeny and systematics of fishes (H. G. Moser. ed.), 11-22 p. Am. Soc. Ichthyol. Herpetol. Lux, F. E., and F. E. Nichy 1971. Number and lengths, by season, of fishes caught with an otter trawl near Woods Hole. Massachusetts. September 1961 to December 1962. Spec. Sci. Rep. Fisheries 622. Wash- ington. D.C., National Marine Fisheries Service. NOAA. Markle. D. F. P. M. Harris, and C. L. Toole. 1992. Metamorphosis and an ovei-view of early-life-history stages in Dover sole Microstomus pacificus. Fish. Bull. 90:285-301. McBride. R. S., and K. W. Able. 1994. Reproductive seasonality, distribution, and abundance ofPriono/u.s caro/(n us and P. er!o/ans( Pisces: Triglidae) in the New York Bight. Estuarine Coastal Shelf Sci. 38:173-188. McBride. R. S.. J. B. O'Goi-man. and K. W. Able. 1998. Seasonal movements, size-structur'e. and interannual abundance of searobins (Triglidae: Pnonotus) in the tem- perate, northwestern Atlantic. Fish. Bull. 96:303-14. McCormick, M. I. 1993. Development and changes at settlement in the barbel structure of the reef fish. Upeneus tragula (Mulli- dae). Environ. Biol. Fishes 37:269-282. Merriman. D., and R. C. Sclar. 1952. The pelagic fish eggs and larvae of Block Island Sound. Bull. Bingham Oceanog. Coll. 13:16.5-219. Morrill. A. D. 1895. The pectoral appendages of Pnonotus and their innci'- vation. J. Morphol. 11:177-192. Morse. W. W.. and K. W. Able. 1995. Distribution and life history of windowpane. Scoph- thalmus aquosus. olf the noi'theastern United States. Fish. Bull. 93:675-693. Myers. R. A., and N. G. Cadigan. 1993. Density-dependent juvenile mortality in marine de- mersal fish. Canadian J. Fish. Aquat. Sci. 50:1576-1590. Pearson, J. C. 1941, The young of some marine fishes taken in lower Ches- apeake Bay, Virginia, with special reference to the gray sea ti-Qut, Cynoscion regal is iBloch). Fish. Bull. 50:97. Richards, S. W. 1959. Pelagic fish eggs and larvae of Long Island Sound. Bull. Bingham Oceanogr. Coll. 17:95-124. Richards, S. W., J. M. Mann, and J. A. Walker. 1979. Comparison of spawning seasons, age, growth rates, and food of two sympati'ic species of searobins, Prionotus carolinus and Prionotus evolans, from Long Island Sound. Estuaries 2:255-268. Ross, S. T 1978. Trophic ontogeny of the leopard searobin, Prionotus scitulus (Pisces: Triglidael. Fish. Bull. 76:225-234. Sale, P F., and D. J. Ferrell. 1988. Early survivorship of juvenile coral reef fishes. Cor- al Reefs 7:117-124. Sokal. R. R..andRJ. Rohlf 1981. Biometry W.H. Freeman and Co., New York, NY, 859 p. Sponaugle, S.. and R. K. Cowen. 1994. Larval durations and recruitment patterns of two Caribbean gobies (Gobiidae): contrasting early life histo- ries in demersal spawners. Mar Biol. 120:13.3-143. Stahl. L.. J. Koczan. and D. Swift. 1974. Anatomy of a shoreface-connected sand ridge on the New Jersey shelf implications for the genesis of the shelf surficial sand sheet. Geology. 2:117-120. Wheatland. S. B. 1956. Oceanography of Long Island Sound. 1952-54. VII. Pelagic fish eggs and larvae. Bull. Bingham Oceanogr Coll. 15:234-314. Wilk. S. J., W. W. Morse, and L. L. Stehlik. 1990. Annual cycles of gonad-somatic indices as indicators of spawning activity for selected species of finfish collected fi-om the New York Bight. Fish. Bull. 88:775-786. Williams. G. C. 1968. Bathymetric distribution of planktonic fish eggs in Long Island .Sound. Limnol. Oceanogr. 13:382-385. Youson, J. H. 1988. First metamorphosis. In Fish physiology, vol. lib (W. S. Hoar and D J. Randall, eds.), p. 13.5-196. Academic Press, San Diego, CA. Yuschak, P. 1985. Fecundity, eggs, larvae and osteological development of the striped searobin, Prionotus evolans (Pisces, Trigli- dael. J. Northwest Atl. Fish. Sci. 6:65-85. Yuschak, P., and W. A. Lund. 1984. Eggs, larvae and osteological development of the northern searobin, Prionotus carolinus (Pisces, Triglidae). J. Northwest Atl. Fush. Sci. 5:1-15. 74 Abstract— In trawl surveys a duster of fish are caught at each station, and fish caught together tend to have more similar characteristics, such as length, age, stomach contents etc., than those in the entire population. When this is the case, the effective sample size for estimates of the frequency distri- bution of a population characteristic can, therefore, be much smaller than the number of fish sampled during a survey. As examples, it is shown that the effective sample size for estimates of length-frequency distributions gen- erated by trawl surveys conducted in the Barents Sea, off Namibia, and off South Africa is on average approxi- mately one fish per tow. Thus many more fish than necessary are measured at each station (location). One way to increase the effective sample size for these sui"veys and, hence, increase the precision of the length-frequency esti- mates, is to reduce tow duration and use the time saved to collect samples at more stations. Assessing the precision of frequency distributions estimated from trawl-survey samples Michael Pennington Institute of Marine Research Department ol Marine Resources Nordnesgaten 33 N-5005 Bergen, Norway E mail address michaeliaimrno Liza-Mare Burmeister Ministry of Fishenes and Marine Resources of Namibia, NatMIRC PO Box 912 Swakopmund, Namibia Vidar Hjellvik Institute ol Manne Research Department of Marine Resources Nordnesgaten 33 N-5005 Bergen, Norway Manuscript accepted 21 August 2001. Fish. Bull. 100:74-80 (2002). Survey-based assessments often appear to provide a more accurate prognosis of the status of a fish stock than catch- based assessments (Nakken, 1998; Pen- nington and Stromme, 1998: Kors- brekke et al., 2001). Aji advantage that sui-vey-based assessments have over those based on commercial catch sta- tistics is that the uncertainties asso- ciated with survey estimates can be studied and quantified, and based on such research, survey methods, and ultimately stock assessments, can be improved (Godo, 1994). In contrast, it is generally difficult to determine either the accuracy or the precision of esti- mates based on commercial catch data, and it is not clear how to improve, at a reasonable cost, the collection of catch data so that these data would more accurately reflect the mortality caused by fishing (Christensen, 1996). Trawl surveys provide estimates of the abundance or relative abundance of a fish stock and estimates of the relative frequency of various population charac- teristics, such as length, age, and stom- ach contents. In our study we examined the precision of survey-based estimates of the length-frequency distributions of cod and haddock in the Barents Sea, hake off South Africa, and hake off Namibia. The focus was on length, but the results are relevant for es- timating the frequency distribution of other population characteristics. Materials and methods Survey length data Bottom trawl survey length data for Northeast Artie cod ^Gadun mohua) and Northeast Arctic haddock (Mela- nogrammus aeglefinusY were collected during the Institute of Marine Research (Norway) winter and summer surveys in the Barents Sea. The surveys were stratified systematic surveys and at each station the trawl was towed for 30 minutes. - Also known as "Atlantic cod" and "had- dock," respectively, according to Common and scientific names of fishes from the United States and Canada. 1991. Am. Fish. Soc. Spec. publ. 20. Besthesda, MD, 183 p. ■ Aglen, A. 1999. Report on the demersal fish sui-veys in the Barents Sea and Sval- bard area during summer/autumn 1996 and 1997. Unpubl. manuscr. Fisket og Havet NR. 7-1999. Institute of Marine Research. PO Box 1870 Nordnes. N-5817 Bergen. Norway. Pennington et a\ Assessing the precision of frequency distributions from trawl survey samples 75 The data for the Naniihian deepwater hake iMt'lurciiis paradoxus) were collected during bottom trawl sui-\-cys off Namibia conducted by the Ministry of Fisheries and Ma- rine Resources of Namibia in conjunctioTi with the Noi-we- gian Agency for Foreign Aid (NORAD). For these sui-v-eys, tows of 30-minute duration were made at stations along transects perpendicular to the coast. ' The data for the deepwater hake for South Africa, were collected from during bottom trawl surveys off the west coast of South Africa. The sui-veys were conducted by the Marine and Coastal Management Centre, South Africa, by using a stratified random design. Tows of 30-minute dura- tion were made at each station (see Payne et al., 1985). Assessing the precision of length-frequency estimates The sample offish of a particular species measured during a survey is not a random sample of individual fish from the entire population but a sample of?! clusters, one cluster from each station. Because fish caught together are usually more similar than those in the general population, a total of M fish collected in 7i clusters will contain less informa- tion about the population length distribution than M fish sampled randomly. One way to measure the information contained in a sample of length measurements is to esti- mate the number of fish that one would need to sample at random (the effective sample size) to obtain the same infor- mation on length contained in the cluster samples. The effective sample size for cluster sampling can be defined and calculated as follows (Pennington and Vols- tad, 1994: Folmer and Pennington. 2000). First estimate the population mean fish length and its variance based on the clusters of fish caught at n stations. Because both the lengths and the number of fish at a station are ran- dom variables, a ratio estimator is appropriate (Cochran, 1977). The ratio estimator, R. of the mean length is given by R = ^, (1) where M, = the number of fish caught (either actual or estimated) at station /; and fi, = an estimate of the average length of fish at station i. var( «-I (Af, Mrit-t, -/?) nUi-1) (2) ^■here M = M. In. Next estimate the variance, a'~, of the population length distribution. If/?;, fish are randomly selected at each sta- tion (or if all fish are measured), then y V 0 li.e. fish of similar length tend to be caught together), then the terms in the parentheses can greatly increase the variance and thus drastically reduce the effective size. In particular, the term ap M is relatively large for trawl surveys. Finally, if p < 0, which is rarely if ever the case for trawl surveys, then the effec- tive sample size will be larger than M. The precision of estimates of other population charac- teristics, such as age distribution, can also be relatively low compared with the number of fish sampled if the par- ticular attribute or measurement is more similar for fish caught together than for those in the general population. For example, the precision of estimates of mean stomach contents (Bogstad et al. , 1995) or diet composition (Tirasin and Jorgensen, 1999) can be relatively low because of in- trahaul correlation. An effective sample size of one fish per tow does not mean only one fish should be measured at each station, but it implies that the only way to improve survey pre- cision significantly is to increase the number of stations, i.e. to sample fish from as many locations as possible. The bootstrapped estimates of precision and the sampling sim- ulations showed that reducing or increasing the number of Winter 1995 ll Jiilii, Winter 1999 n. ~^si«....._. 20 40 60 100 120 140 20 40 60 80 100 120 140 Lengtti (cm) Figure 3 Bootstrapped estimates of the 95'7r confidence intervals for the proportion of cod in the Barents Sea in each 5-cm length bin. for winter 1995 and for winter 1999. The inner brackets denote the confidence intervals if the estimates are based on all the cod measured during the surveys and the outer brackets denote the confidence intervals if 10 fish arc measured for each subsample. fish measured (or caught and measured) at a station will not significantly affect the precision of length-distribution estimates. In general, if intracluster correlation is positive for an attribute, then it is usually best to take a small sample from as many locations as possible (e.g. Bogstad et al.. 199.5; McGarvey and Pennington, 2001 ). It has been shown that tows of short duration are in general more efficient for estimating stock abundance than long tows (Godo et al., 1990; Pennington and Vols- tad, 1991; Gunderson. 1993; Carlsson et al.. 2000). There- fore one way to collect samples from more locations and improve overall survey efficiency without increasing sur- vey cost is to reduce tow duration and use the time saved to increase the number of survey stations (Pennington and Volstad, 1994). For example, if tow duration were re- duced from 60 minutes to 15 minutes for a trawl survey of shrimp off West Greenland, then 44^^^ more stations could be surveyed (Carlsson et al., 2000). Likewise, a reduction in tow duration from 30 minutes to 10 minutes for a trawl survey on Georges Bank would increase the number of survey stations by about 30*7^ (Pennington and Volstad, 1994). The total number of fish caught would be fewer, on av- erage, if tow duration was reduced, but estimates of fish density would be more precise and the resulting sample of individuals would be more representative of the entire population (Pennington and Volstad. 1994). 80 Fishery Bulletin 100(1) Acknowledgments We thank Rob Leslie (Marine and Coastal Management Centre, South Africa) for providing us with the South Afri- can survey data, and Jon HelgeVolstad ( Versar, Inc., USA) and two anonymous referees for their constructive com- ments and suggestions. Literature cited Bhattacharyya, G. K., and R. A. Johnson. 1977. Statistical concepts and methods. John Wiley and Sons, New York, NY, 639 p. Bogstad, B., M. Pennington, and J. H. Volstad. 1995. Cost-efficient sui-vey designs for estimating food con- sumption by fish. Fish. Res. 23:.37-46. Carlsson, D., R Kanneworff, O. Folmer, M. Kingsley, and M. Pennington. 2000. Improving the West Greenland trawl sui-vey for shrimp iPandalus borealis). J. Northwest Atl. Fish. Sci. 27:151-160. Christensen, V. 1996. Virtual population reality. Rev. Fish Biol Fish 6: 243-247. Cochran, W. G. 1977. Sampling techniques, 3'''' ed. John Wiley and Sons, New York, NY, 428 p. Efron, B. 1982. The jackknife, the bootstrap, and other resampling plans. Society for Industrial and Applied Mathematicians (SLAM), Conference Board of the Mathematical Sciences iCBMSi- National Science Foundation ( NSF i regional conference series in applied mathematics 38. Philadelphia, PA, 92 p. Folmer O., and M. Pennington. 2000. A statistical evaluation of the design and precision of the shrimp sui-vey off West Greenland. Fish. Res. 45: 16.5-178. Godo, O. R. 1994. Factors affecting the reliability of gi-oundfish abun- dance estimates from bottom trawl surveys. In Marine fish behaviour in capture and abundance estimation (A. Feme and S. Olsen, eds. ), p. 166-199. Fishing News Books, Farn- ham, UK. Godo, O. R.. M. Pennington, and JH. Volstad, 1990. Effect of tow duration on length composition of trawl catches. Fish. Res. 9:165-179. Gunderson, D. R. 1993. Surveys of fisheries resources. John Wiley and Sons, New York. NY, 248 p. Korsbrckke, K., S. Mehl, O. Nakken, and M. Pennington. 2001. A sui-vey-based assessment of the Northeast Ai'ctic cod stock. ICES J. Mar Sci. 58:76.3-769. McGai-vey, R., and M. Pennington. 2001. Designing and evaluating length-frequency surveys for trap fisheries with application to the southern rock lob- ster Can. .J. Fish. Aquat. Sci. .58:254-261. Nakken, O. 1998. Past, present and future exploitation and manage- ment of marine resources in the Barents Sea and adjacent areas. Fish. Res. 37:23-35. Payne, A. I. L., C. J. Augustyn. and R. W. Leslie. 1985. Biomass index and catch of Cape hake from random stratified sampling cruises in division 1.6 during 1984. Colin. Scient. Pap. Int. Com. SE Atl. Fish. 12:99-123. Pennington, M., and T. Stromme. 1998. Surveys as a research tool for managing dynamic stocks. Fish. Res. 37:97-106. Pennington. M., and J. H. Volstad. 1991. Optimum size of sampling unit for estimating the density of marine populations. Biometrics 47:717-723. 1994. Assessing the effect of intra-haul correlation and vari- able density on estimates of population characteristics from marine surveys. Biometrics 50:725-732. Research Triangle Institute. 2001. SUDAAN users manual, release 8.0. Research Tri- angle Institute, Research Triangle Park, NC, 886 p. Tirasin, E. M., and T. Jorgensen. 1999. An evaluation of the precision of diet description. Mar Ecol. Prog. Ser 182:243-252. 81 Abstract— Tlie red porgy, Pagrus pag- Ills. IS ;in important roof fish in sovoral offshore fisheries along the southeastern United States. We examined samples from North Cai-olina through south- east Florida from recreational i head- boat) and commercial (hook and line) fisheries, as well as samples from a fishery-independent source. Red porgy attain a maximum age of at least 18 years and 733 mm total length. The weight-length relationship is repre- sented by the In-ln transformed equa- tion: VV = 8.85 X 10-"(L)-!'"^, where W = whole weight in gi'ams, and L = total length in mm. The von Bertalanffy growth equation fitted to the most recent, back -calculated lengths from all the samples is L, = 644( 1 - e-"'^-'" * "-'S'). Our study revealed a difference in mean length at age of red porgy from the three sources. Red porg\' in fishery- independent collections were smaller at age than specimens examined from fishery-dependent sources. The differ- ence in length-at-age may be related to gear selectivity and have important consequences in the assessment of fish stocks. Estimated ages of red porgy (Pagrus pagrus) from fishery-dependent and fishery-independent data and a comparison of growth parameters Jennifer C. Potts Charles 5. Manooch III Center for Coastal Fisheries and Habitat Research Beaufort Laboratory National Manne Fishenes Service, NOAA 101 Pivers Island Road Beaufort, North Carolina 28516 9722 Email address (lor J C Potts) Jennifer pottsidinoaa gov Manuscript accepted 20 August 2001. Fish. Bull. 100:81-89(2002). Red porgy, Pagrus pagrus, inhabit con- tinental shelves in temperate and trop- ical waters throughout the Atlantic Ocean and Mediterranean Sea. The spe- cies supports fisheries in many coun- tries and is heavily exploited. Since 1992, red porgy has ranked relatively high (38 of 200) in value among all fin- fish landed commercially in the south- eastern United States.' Red porgy form a substantial part of overall reef fish landings, especially in North Carolina and South Carolina, although there is little directed fishing for the species. Commercial landings of red porgy from the southeastern U.S. peaked in 1982 at 535 metric tons (t) and declined to 134 t in 1993 (Potts and Burton-). Red porgy ranked second by weight for reef fish landed by recreational headboat^ anglers through the early 1980s. Since then, headboat landings of red porgy have declined, and landings of vermil- ion snapper, Rhomhoplites aurorubens, which are also declining, have now sur- passed red porgy. White grunt, Hae- mulon plumieri, and gray triggerfish, Balistes capriscus, which were less pre- ferred than other members of the snap- per grouper complex, have increased in landings and now surpass red porgy.^ Mean weight of red porgy from the commercial and recreational fisheries has declined from 1.06 kg in the 1970s to 0.66 kg in 1997.- Minimum size regu- lations ( 305 mm total length ) for recre- ational and commercial fisheries enact- ed in 1992 did little to increase mean weight in catches, although the head- boat fishery did show a slight increase from 0.48 kg in 1991 and 1992 to 0.60 kg in 1997. Additionally, population bio- mass estimates for red porgy in the southeastern United States have plum- meted from a peak of 3.27x10*" kg in 1978 to 0.43xl0« kg in 1992 (Huntsman et al.''). These trends suggest that red porgy stocks are being overexploited. Age determination studies have been conducted throughout the range of red porgy. Manooch and Huntsman (1977) conducted the first comprehen- sive study using scales (n = 1777) and whole otoliths {n=222) to age red porgy that were caught by recreational fisher- men using hook-and-line gear off North Carolina and South Carolina when the species was lightly exploited ( 1972-74). Harris and McGovern (1997) aged red porgy from whole otoliths (;!=4281) of ' General canvas. 1998. Unpubl. data. Miami Laboratory, National Marine Fish- eries Sei-vice, 75 Virginia Beach Dr, Miami, Florida 33149. - Potts, J. C, and M. L. Burton. 1999. Trends in catch data for fifteen species of reef fish landed along the southeastern United States. Unpubl. data. South At- lantic Fishery Management Council, 1 Southpark Circle, Charleston, SC 29407. ' A "headboat" is a fishing vessel that car- ries more than six passengers who pay per person lor by the "head") to go offshore fishing. ■• Headboat annual summaries. 1998. Un- publ. data. Center for Coastal Fisheries and Habitat Research. Beaufort Labora- torv, 101 Pivers Island Rd.. Beaufort. NC 28516-9722. '■ Huntsman, G. R., D. S. Vaughan, and J. C. Potts. 1994. Trends in population status of red porgy, Pagrus pagrus. in the Atlantic Ocean of North Carolina and South Carolina, USA, 1971-1992. Unpubl. data. SouthAt- lantic Fisherv Management Council, 1 South- park Circle, Charieston, SC 29407. 82 Fishery Bulletin 100(1) fish caught from North Carolina to Florida with fishery- independent gear during 1979-81 and 1988-94. Nelson ( 1988) aged red porgy with scales (?! = 126) from fish caught in the northwestern Gulf of Mexico with fishery-indepen- dent hook-and-line gear and trap gear during 1980-82. In the eastern Gulf of Mexico, Hood and Johnson (2000) aged red porgy from sectioned otoliths (^=852) collected from headboat and commercial catches during 1995-96. Vassi- lopoulou and Papaconstantinou (1992) used scales from 138 red porgy that were taken with fishery-independent hook and line and trammel nets in the Mediterranean Sea during 1985-86, and Serafim and Krug (1995) aged red porgy from whole otoliths (?!=358) that were collected by using commercial longlines and fishery-independent gear in the Azores during 1991-93. Researchers in the Canary Islands aged 1505 red porgy from commercial trap and longline samples during 1985-86 and 1991-93 (Pajuelo and Lorenzo, 1996), and researchers off the Argentinian coast used trawl-caught samples during 1972-81 to obtain 5859 red porgy that were aged from scales (Cotrina and Raimondo, 1997). Predictions offish populations from models rely heavily on input data sets, including age and growth. If samples used in the aging study are not representative of the en- tire population (i.e. the entire geogi'aphic range of the stock, full range of fish size, and different gear types), model predictions (e.g. spawning potential ratio |SPR|) can mislead management decisions. A comparative stock assessment of red porgy was done by using growth pa- rameters and age-length keys generated from two stud- ies: 1) fishery-independent data (Harris and McGovern, 1997) and 2) fishery-dependent data (Manooch and Hunts- man, 1977; Potts et al.'M. Each set of age and growth data was applied to fishery-dependent landings and length fre- quencies. The fishery-independent age and growth data produced a static SPR of 46^^^, which is well above the overfished definition (SPR<30'7f) as set forth by the South Atlantic Fishery Management Council (SAFMC). The fish- ery-dependent age and growth data produced a static SPR of 19*7^ (Potts et al.''), which makes red porgy, by definition, overfished and which necessitates that stringent manage- ment measures be put in place to protect the stock. The purpose of our study was to update the age and growth information on red porgy caught in the recreation- al and commercial fisheries operating along the southeast- ern United States. We present the von Bertalanffy growth model, weight-length relationship, and age-length keys for red porgy collected from the headboat hook-and-line fishery, commercial hook-and-line fishery, and fishery-in- dependent samples. We also compare mean age at length of red porgy collected from recreational fisheries, commer- cial fisheries, and fishery-independent sources. We discuss how data source selection affects the growth parameters. Materials and methods Sagittal otoliths were collected from red porgy landed by hook-and-line fishermen from the headboat (recreational) fishery («=249) between 1989 and 1998 (59% from 1996 to 1998) and the commercial fishery (n=264) between 1997 and 1998 operating from North Carolina to southeast Flor- ida. From the two fisheries, 64% of the samples came from North Carolina, 14% from South Carolina, and 22% from the east coast of Florida. Because of minimum size limit regulations (305 mm total length), the South Caro- lina Department of Natural Resources (SCDNR) Marine Monitoring and Prediction (MARMAP) Program supplied us with otoliths from red porgy that were smaller than those available from the fisheries (n=59) and an additional 62 samples ranging from 300 to 425 mm total length. These fish were caught primarily with Chevron traps off South Carolina during 1996 and 1997. Total length, whole weight, port of landing, and date of capture were recorded for each sample. Tlie otoliths were stored dry in coin envelopes. For age analysis, three transverse (dorsoventral) sec- tions from the left otolith of each fish were taken by us- ing a low-speed saw. One section was made on either side of the core, and the other encompassed the core. The sections were mounted on glass slides with thermal ce- ment, and examined through a microscope at 80x and illuminated with reflected light. Clove oil was applied to each section to enhance the legibility of the growth zones on the section. The samples were put in sequential order from smallest to largest, and one reader counted the number of opaque zones in the otolith section. A sec- ond reader examined a random sample of the otoliths. If the readers disagreed on the age of a sample, they exam- ined it again. If consensus was reached, the sample was retained; otherwise, the sample was discarded. Measure- ments from the core to the outer edge of each successive opaque zone and the otolith margin were taken along the lateral plane on the dorsal lobe of the section by using an ocular micrometer. Analysis of the marginal increment (the distance be- tween the last opaque zone and otolith margin) was used to validate the annual deposition of the opaque zones in the otoliths. For each age and month, the mean of the rela- tive marginal increment, the ratio of the marginal incre- ment to the distance between the last two opaque zones, was plotted. An opaque zone was considered an annulus if a minimum ratio was recorded for one month or season. The relationship of fish length and otolith radius was described by regi'essing the obsei-ved total length on oto- lith radius (/?(.). The linear equation was L = a +blR^.). where L = total length in mm. 6 Potts, J. C, M. L. Burton, and C. S. Manooch, III. 1998. Trends in catch data and estimated static SPR values for fifteen spe- cies of reef fish landed along the southeastern United States. Unpubl. data. South Atlantic Fishery Management Council, 1 Southpark Circle, Charleston, SC 29417. The back-calculated total lengths at each age were deter- mined from the body proportional equation (Francis, 1990): L^ =[(a+hR^)/{a + bRc)]Lc, Potts and Manooch Estimated ages of Pagrus pagtvs 83 where L^ = back-calculated total length to aniuilusA; a = intercept from the linear total lengtii-otohth radius regression; b = slope from the linear total length-otolith radius regression: L(. = total length at time of capture; /?, = otolith radius to annulus A; and /?(. = total otolith radius at time of capture. The von Bertalanffy ecjuation. L, = L |1 - e.\p(-A'(^-/(,)|, was fitted to back-calculated lengths-at-ages for the most recently formed annuli (Ricker, 1975; Everhart et al., 1981; Vaughan and Burton, 1994). Growth parameters were es- timated by using SAS PROC NLIN with the Marquardt Option (SAS Institute, 1982) for all aged fish and for fish obtained from fishery-dependent sampling. Differences in mean back-calculated length at age for the most recently formed annulus for the three sample sources, i.e. recreational, commercial, and fishery-indepen- dent, were tested by using the general linear model analy- sis of variance. To estimate the whole weight of gutted red porgy landed in the commercial fishery and to estimate stock biomass from assessment models, a regression of hii fisli tveif^ht) on \n(fish length ) was performed and transformed to W = aiL)^, where W = weight in g, and L = total length in mm. Age-length keys were constructed from observed age at length by sample source in which the ages were unadjust- ed for time of year. Fish that were aged were assigned to 25-mm length intei-vals. Results Red porgy sampled for our study ranged from 176 to 733 mm TL and from 1 17 to 5895 g in whole weight. Ages were determined for 631 of 634 (99'7f) sectioned sagittal oto- liths. Of those aged, 603 idd'v'c) otoliths were considered legible to record measurements from the core to each suc- cessive opaque zone and the otolith margin. On twenty additional samples, we were able to measure only the oto- lith radius. Sectioned otoliths exhibited a recurrent pat- tern of alternating wide translucent zones and thin opaque zones. Estimated ages ranged from 1 to 18 years. Analyses of marginal increment data indicated that the opaque zones were annular in nature and were formed in the spring (Fig. 1). Mean relative marginal increments for ages 2 through 8 were lowest in March through May and were the only months that had marginal increments equal to zero. They then steadily increased from June through October and remained high through February. Back-calculated total lengths at age of red porgy were estimated from the parameters from the regi-ession equa- tion of total length (L) on otolith radius (R^.). The plot of length on radius was linear, and the linear regression equation that best fitted the data was L = -132.84 -i- 10.87(fl ) (;--=0.91, «=623). Using the Francis (1990) body proportional hypothesis, we found that weighted mean back-calculated lengths ranged from 103 mm for age-1 fish to 721 mm for age-18 fish (Table 1). 1.2 iOBp\ .b^^ CD ^ 0.6 0.2 0 Mean n porgy fi w^ ^ - • + - ■ Age 7 ■-Q--Age8 2 3 4 5 6 7 8 9 10 11 12 IVlonth Figure 1 lonthly relative marginal increment (MI) of red om the southeastern United States plotted by age. The back-calculated lengths at the last annulus forma- tion were used to estimate the von Bertalanffy equation. The equation parameters (±1 SE) were L„ = 644.72 ±17.93, A' = 0. 15 ±0.01, and /„ = -0.76 ±0. 10. The theoretical lengths at age ranged from 149 mm at age 1 to 605 mm at age 18. Theoretical lengths closely fitted the observed and back- calculated lengths through age 14 (Table 1). When we used fishery-dependent samples only to generate the von Berta- lanffy growth equation, the resulting parameters (±1 SE) were L,= 773.73 ±39.49, A' = 0.09 ±0.01 and t„ = -1.96 ±0.21. The fishery-dependent theoretical lengths at age ranged from 181 mm at age 1 to 646 mm at age 18. We used ages 2 through 6 and data years 1996 through 1998 to compare length at age of the three data sources be- cause the three sets overlapped for those ages and years. The ANOVA on the mean back-calculated length at age of the most recently formed annulus between sample sources indicated a significant difference in age at size between the MARMAP, headboat, and commercial red porgy sam- ples (r-=0.88; F-value=522.21; P=0.0001 for all combina- tions) and were represented by the model TL = a„ + Ycc + y,,h + ^ /3, A, , where c = 1 if fishery = commercial, or c = 0 if fishery i^ commercial; h = 1 if fishery = headboat, or /i = 0 if fishery ^ headboat; A, = 1 if age = 2, or A, =0 if age ^ 2, etc.; and J = age categories. Average TL = a^ for fishery = fishery-independent and age = 6; average TL = a,, -i- y^.c for fishery = commercial and age = 6; etc. The model indicated no year effect, and no interaction between fishery and age. MARMAP (fishery- independent) samples were smaller at age than those from the commercial and headboat fisheries. Although mean back-calculated lengths at age between headboat and com- mercial data sources were statistically different, the dif- ferences were slight (<15 mm) (Fig. 2). 84 Fishery Bulletin 100(1) Table 1 Mean back-calculated. mean observed, and theoretical total lengths Immt of red porgy from the sou theastern United States. Age (yrl 11 An nulus numl er 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1 15 134 2 33 96 206 3 136 98 208 274 4 158 104 212 274 327 5 136 105 212 274 326 363 6 56 104 215 275 329 367 399 7 31 107 216 284 342 381 413 441 8 26 100 213 278 332 371 400 427 449 9 6 106 221 283 337 377 410 438 461 483 10 1 121 221 278 334 379 412 457 491 513 536 11 2 112 215 275 329 362 395 427 449 476 498 520 12 1 108 219 274 330 374 407 440 462 485 507 529 551 14 1 111 226 294 351 385 419 442 465 487 510 533 556 567 579 18 1 116 247 306 365 413 448 484 506 531 555 567 591 614 638 662 686 709 721 Total 603 Weighted mean TL 103 211 275 329 368 404 436 455 489 517 534 566 591 608 662 686 709 721 Incremental growth 103 108 64 54 39 36 32 19 34 28 17 32 25 17 54 24 24 12 Observed TL 198 236 303 350 386 423 459 470 508 547 536 562 590 733 Theoretical TL 149 218 278 329 374 410 443 471 495 516 534 549 562 574 583 592 599 605 0 — 1 2 3 4 5 6 7 8 9 10 11 12 13 14 IE Age (yr) -Fishery-Independent — »— Headboat -h- Commercial Figure 2 Mean back-calculated length at age of red porgy obtained from a fishery-independent source ( MAR- MAPl, headboat operations, and commercial fish- ery operations between 1996 and 1998. The weight-length relationship for red porgy in the southeastern United States was best described by the con- verted In-ln regression equation of IV = 8.85 x 10^''(L)''^"' (r-=0.96, /!=230, MSE=0.01l. Age-length keys by sample source are presented in Table 2. The fishery-independent key is appropriate for fishery-independent length data only The headboat and commercial keys can be used to convert unaged length sam- ples of red porgy from the fisheries operating in the south- eastern United States to aged ones. Annual keys were not available owing to the small sample size from each year. Discussion In two previous aging studies on red porgy from the south- eastern United States, populations were examined at two different levels of exploitation and different structures were used to determine age. Manooch and Huntsman (1977) sampled recreationally caught fish from an almost virgin stock off North Carolina and South Carolina during 1972-74. Although they used scales and whole otoliths, the main focus of their study and the analysis were on ages determined from scales. Of the 3278 scales analyzed, only 54'^'f (1901) were legible enough to record ages. The main problem of aging red porgy with scales was the large number of regenerated scales." Harris and McGov- ern (1997) used whole otoliths (?7=4281) to age fish col- lected between 1979 and 1981 and between 1988 and 1994 from the MARMAP survey, a fishery-independent source. The 1988-94 samples in their study came from the area off North Carolina through northeast Florida, although 73% were collected in the area off Charleston, SC, between 32°N and 33°N. The samples were limited mainly to indi- viduals below 450 mm TL (less than l'~f of samples were Manooch. C. S. 1998. Personal conimun. NOAA. Center for Coastal Fisheries and Habitat Research. 101 Fivers Island Road, Beaufort, NO 28516. Potts and Manooch; Estimated ages of Pagrus pagrus 85 Table 2 Age-lt>ngth keys for red porgy fron the southeastern United States by sample source; fishery- independent headboat. and com- morcial. Total length classes are in 25-mm intervals (i.e. 175 = 17.5-199). TL class n Age (yr) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Fishery-independent 175 11 10 1 200 21 5 16 225 7 1 6 250 9 9 275 10 10 300 16 8 6 2 325 20 11 9 350 11 5 5 1 375 14 6 8 400 1 1 425 450 475 500 525 550 575 725 Total 120 Headboat 175 200 225 3 3 250 6 5 1 275 28 5 23 300 44 36 7 1 325 48 15 31 2 350 41 2 23 16 375 37 11 21 5 400 19 11 5 2 1 425 11 2 5 2 2 450 5 1 3 1 475 2 2 500 3 1 2 525 2 1 1 550 575 725 Total 249 continued >450 mm). The data Harris and McGovern presented for the period between 1979 and 1981 showed that 89; of the samples for the reproductive study were greater than 450 mm TL. We sampled from a heavily exploited stock, and fish ranged up to 733 mm TL, and 14'7r of the fishery- dependent samples were greater than 4.50 mm TL. Addi- tionally, our samples of red porgy were from a broader geographic range (529f from North Carolina, 309^ from South Carolina, assuming all MARMAP samples were from South Carolina, and 18% from east coast Florida) than that of previous studies. Our distribution of samples more closely reflected the landings of red porgy in the southeastern United States. Also, we used sectioned oto- liths, which have been determined to be the best struc- 86 Fishery Bulletin 100(1) Table 2 (continued) TL class n Age (yr) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Commercial 175 200 225 250 275 5 2 3 300 25 22 3 325 34 4 27 3 350 28 24 4 375 37 8 24 5 400 44 29 14 1 425 27 6 17 3 1 450 31 7 16 8 475 20 5 15 500 7 1 6 525 1 1 550 1 1 575 1 1 725 1 1 Total 262 Table 3 Mean observed total I ength (mm) of red porgy from this study and others: 1 = our stud V (all data combmed); 2 = 1 Manooch and Huntsman (1977: scale data); 3 = Manooch and Huntsman (1977: atolith data); 4 = Harris and McGov- | ern (1997: 1994 data year); 5 = Harris and McGovern | (1997: 1988 data year) Age (yr) Study 1 2 3 4 5 1 198 238 229 218 224 2 236 290 288 294 297 3 303 342 331 306 329 4 350 382 374 328 379 5 386 419 402 344 445 6 423 451 425 362 399 7 459 483 453 386 431 8 470 505 474 374 406 9 508 527 496 448 10 547 543 534 449 11 536 558 557 12 562 604 13 595 14 590 15 694 16 17 18 733 ture to age fish in many studies (Beamish, 1979; Boehlert, 1985; Smale and Punt, 1991). Red porgy form their opaque zones during the spring along the southeastern United States. Both Manooch and Huntsman (1977) and Harris and McGovern (1997) re- ported that annulus formation occurred in red porgy dur- ing March and April. We found that the opaque zone formed from March through May Springtime formation has also been reported from the Gulf of Mexico (Nelson, 1988; Hood and Johnson. 2000), from the Azores (May; Se- rafim and Krug, 1995) and from Argentinian waters (Oc- tober; Cotrina and Raimondo, 1997). Pajuelo and Lorenzo, in their (1996) study of red porgy off the Canary Islands, found that the opaque zone was formed during the sum- mer (June through October). A comparison of the mean length at age from our study and that from Harris and McGovern's ( 1997 ) study from 1994 (Table 3) clearly reveals that red porgy caught with fishery-independent gear are much smaller at age for fish 3 years and older than fish caught with fishery-dependent gear. Mean size at age in our study was similar to data for fish 5 years and older reported by Manooch and Hunts- man (1977) using otoliths, although our mean sizes were smaller that those reported for their scale data. Ages 1-5 of our study were a mix of MARMAP samples and fishery- dependent samples, which may explain why red porgy were smaller for those ages than those reported by Manooch and Huntsman (1977). The mean obsei-ved length at age from our study were on average 23 mm smaller than the corre- sponding lengths from Manooch and Huntsman (1977) for ages 6 through 12 as assigned by scales. The differences may Potts and Manooch Estimated ages of Pagnis pagnis 87 -M&H: Southeastern US ■ H&M 79-81 Soulheaslern US • HSM ee 90 Soulheaslern US ■ H&M 91-94 Soulheaslern US P&M Soulheaslern US P&L Canary Islands V&P Easlern Medilerranean S&K Azores C&R North Buenos Aires C&R South Buenos Aires H&J Eastern Gull ol Mexico N: Western Gull ol Mexico 1 3 5 7 9 11 13 15 17 Age (yr) Figure 3 Comparison of von BertalanfTy growth curves from various locations in the Atlantic Ocean: M&H = Manooch and Huntsman 1 1977); H&M = Harris and McGovern (1997) from three separate data year sets; P&M = Potts and Manooch (this studyi; P&L = Pajuelo and Lorenzo 11996); V&P = Vassilopoulou and Papaconstantinou (1992); S&K = Serafim and Ki'ug ( 1995); C&R = Cotrina and Raimondo ( 19971 from two study areas; H&J = Hood and Johnson (2000); N = Nelson (1988). 100 1 3 5 7 9 11 13 15 17 Age (yr) -»-M&H -iic-P&M - Fishery-dependent data Figure 4 Comparison of the von Bertalanffy growth curve from Manooch and Huntsman (1977) using headboat data versus the resulting growth curves from headboat and commercial fisheries data from this study. have been due to heavy fishing on the population, the small sample size of older age fish, or differences in assigned ages due to the structure used for aging, because the comparison of ages determined from otoliths was very close. Harris and McGovern (1997) and Hood and Johnson (2000) have put forth the theory that heavy fishing pres- sure may cause a shift in the size and age structure of a population to smaller, slower growing fish. Our study does not support this theory. Cotrina and Raimondo ( 1997 ) dem- onstrated the differences in growth of red porgy caught from two areas off Argentina. Because we feel that our samples encompassed the full range of red porgy along the southeastern United States., our data more truly rep- resents that population than the data from Harris and McGovern's (1997) study. We also feel that the perceived changes in length at age reported in Harris and McGov- ern's (1977) study may be confounded by the changes in sampling strategy of the MARMAP program between 1979 and 1994 (e.g. sampling gear used, locations of sampling, personnel, etc). Differences in growth of red porgy in the Gulf of Mexico between Hood and Johnson's (2000) study and Nelson's (1988) study may certainly be affected by the different locations the samples came from for the two studies. We suggest that further investigation into sample design and effects of habitat, temperature, fishing pres- sure, weather, fishing gear, etc. on fish populations is need- ed to help resolve differences between these studies. The von Bertalanffy growth parameters L . and K are integral components of stock assessment models. Samples for age determination should be representative of the tar- get population (i.e. entire geographic range, all habitats, and all gear types used in the fisheries). Because back- calculated length-at-age information was unavailable to us from all previous studies, we compared the various von Bertalanffy growth curves with direct comparisons of length at age (Fig. 3). The growth curve of red porgy es- timated by Manooch and Huntsman (1977), L, = 763(1 - g-uo96i( -1- 1 88i)_ jg similar to our equation for all samples. Our calculated von Bertalanffy equation with fishery-de- pendent samples only was almost identical to that of Ma- nooch and Huntsman (1977): L, = 774(1 - g-ooaa/ * i96i) (Fig. 4). Although aging structures for the two studies were very different, ages were validated in both studies, and overall size ranges were similar In comparison with our study and that of Manooch and Huntsman (1977), the growth curves estimated by Harris and McGovern ( 1997) from their 1979-94 data had L, val- ues ranging from 411 to 451 mm TL. Red porgy in their study exhibited theoretical growth only from 58 to 69 mm, between ages 5 and 11. Red porgy from our study (all data combined) and Manooch and Huntsman's ( 1977) study the- oretically grew from 164 mm to 172 mm, between ages 5 and 11. Also, the growth coefficients (0.27 to 0.34: 1979-94 data) from Harris and McGovern's ( 1997) data seemed high for red porgy in relation to those reported in other red por- gy age and growth studies (Table 4) (Manooch and Hunts- man, 1977; Vassilopoulou and Papaconstantinou, 1992; Se- rafim and Krug, 1995; Pajuelo and Lorenzo, 1996; Cotrina and Raimondo, 1997; Hood and Johnson, 2000). Studies using fishery-independent sources for red porgy showed smaller L^. and higher A' values than those from most studies where a combination of commercial, recre- ational, or fishery-independent samples were used (Table 4). The differences in growth curves are likely a result of smaller fish in the fishery-independent samples and a con- sequent truncated upper-length range. Although Hood and 88 Fishery Bulletin 100(1) n u a e CO 5 r ^ (^ s: rfi i- K 6- Si " II ■n tL. < w z a, II E H -n o !^ X o 3 o [/J C/J II •n 3 1.; C/D w X ffl :S ^ fci rt T) o C X C8 o o T3 3 3 o 3 O CO CO 3 CO 3 CO -a o CTj ^ 3 03 o o •c CO 02 1 en _o 'cfi CO 11 S > 1.0 ,_ o Ol 00 CO 1 1 o K o CO ^ CO ^ e:^ c — ti. 5 ^ Potts and Manooch Estimated ages of Pagrus pagrus 89 .Johnson (2000) used fishery-dependent samples, the larg- est fish in that study was 489 mm TL, and their samples came from primarily one location. Selectivity of fishing gear may also explain differences in growth parameters. For example, hook-and-line gear may catch faster growing and more aggi'essive fish, whereas traps and trawls may catch slower growing fish and overall smaller fish. We rec- ommend that in future age and growth studies of any fish species, samples represent the full range of fish sizes in a population, including fish caught by many different gear types, and are obtained from the entire geographic range of the stock. Where practical, landings should be stratified and sampled accordingly. Acknowledgments We would like to thank the port agents of the National Marine Fisheries Service and the South Carolina Depart- ment of Natural Resources MARMAP personnel for pro- viding us with otolith samples. Jim Waters and Doug Vaughan, NOAA's Center for Coastal Fisheries and Habi- tat Research, were instrumental in the statistical analysis of the length-at-age data and provided thoughtful com- ments on the manuscript. Dean Ahrenholz and Joe Smith also of NOAA's Center for Coastal Fisheries and Habitat Research provided a critical review of the manuscript. Literature cited Beamish. R. J. 1979. Differences in the age of Pacific Hake (Mcrluccius productus) using whole otoliths and sections of otoliths. J. Fish. Res. Board Can. 36: 141- 151. Boehlert, G. W. 1985. Using objective criteria and multiple regression models for age determination in fishes. Fish. Bull. 83:103-117. Cotrina, C. P.. and M. C. Raimondo. 1997. Study on the age and gi'owth of the red porgy Pagrus pagrus from the Buenos Aires coa.stal shelf Rev. Invest. DesaiT. Pesq. 11:95-118. Everhart. W. H.. A. W. Eipper, and W. D. Youngs. 1981. Principles of fishery science, 2nd ed. Cornell Univ. Press, Ithaca, NY. 288 p. Francis, K. 1. C. C. 1990. Back-calculation offish Iciigtlis: a critical review. J. Fish. Biol. 36:883-902. Harris. P. J., and J. C. McGovern. 1997. Changes in the life history of red porg>-. Pagrus pagrus. from the southeastern United States. Fish. Bull. 95:732-747. Hood, P. B., and A. K. Johnson. 2000. Age, growth, mortality, and reproduction of red porgy, Pagrus pagrus, from the eastern Gulf of Mexcio. Fish. Bull. 98:723-7.35. Manooch, C. S., Ill, and G. R. Huntsman. 1977. Age, growth, and mortality of the red porgy. Pagrus pagrus. Trans. Am. Fish. Soc. 106:26-33. Nelson, R. S. 1988. A study of the life history, ecology, and population dynamics of four sympatric reef predators iRhombnplites aurnrube/is. Lutjanus campechanus, Lutjanidae; Haeniu/on luctanurum. Haemulidae: and Pagrus pagrus. Sparidae) on the East and West Flower Garden Banks, northwestern Gulf of Mexico. Ph.D. diss.. North Carolina State Univ, Raleigh, NC, 197 p. Pajuelo, J. G., and J. M. Lorenzo. 1996. Life history of the red porgy, Pagrus pagrus (Teleostei: Sparidae I. off the Canary Islands, central east Atlantic. Fish. Res. 28:163-177. Ricker. W. E. 1975. Computations and interpretations of biological sta- tistics of fish populations. Bull. Fish. Res. Board Can. 191:l-.382. SAS Institute, Inc. 1982. SAS user's guide: statistics. SAS Institute, Gary, NC, 1028 p. Serafim, M. P R. and H. M. Ki'ug. 1995. Age and growth of the red porgy, Pagrus pagrus (Lin- naeus, 1758) (Pisces. Sparidae). in Azorean waters. Arqui- pelago (Life Mar Sci.) 13A:ll-20. Smale. M. J., and A. E. Punt. 1991. Age and growth of the red steenbras Pctrus rupcslns I Pisces: Sparidae) on the southe-east coast of South Africa. S. Afr J. Mar .Sci. 10:131-139. Vassilopoulou, v., and C Papaconstantinou. 1992. Age, growth and mortality of the red porgy. Pagrus pagrus. in the eastern Mediterranean Sea (Dodecanese, Greece). Vie Milieu 42:51-55. Vaughan, D. S., and M. L. Burton. 1994. Estimation of von Bertalanffy gi-owth parameters in the presence of size-selective mortality: a simulation exam- ple with red grouper Trans. Am. Fi.sh. Soc. 123:1-8. 90 Abstract— Bycatch taken by the tuna purse-seine fishery from the Indian Ocean pelagic ecosystem was estimated from data collected by scientific observ- ers aboard Soviet purse seiners in the western Indian Ocean (WIO) during 1986-92. A total of 494 sets on free- swimming schools, whale-shark-associ- ated schools, whale-associated schools. and log-associated schools were ana- lyzed. More than 40 fish species and other marine animals were recorded. Among them only two species, yellow- fin and skipjack tunas, were target spe- cies. Average levels of bycatch were 0.518 metric tons (t) per set, and 27.1 t per 1000 t of target species. The total annual purse-seine catch of yellowfin and skipjack tunas by principal fishing nations in the WIO during 1985-94 was 118.000-277,000 t. Nonrecorded annual bycatch for this period was estimated at 944-2270 t of pelagic oce- anic sharks, 720-1877 t of rainbow runners, 705-1836 t of dolphinfishes, 507-1322 1 of triggerfishes, 1 13-294 t of wahoo, 104-251 t of billfishes, .53-112 t of mobulas and manias, 35-89 t of mackerel scad, 9-24 t of barracudas, and 67-174 t of other fishes. In addi- tion, turtle bycatch and whale mortal- ities may have occurred. Because the bycatches were not recorded by some purse-seine vessels, it was not possible to assess the full impact of the fish- eries on the pelagic ecosystem of the Indian Ocean. The first step to solving this problem is for the Indian Ocean Tuna Commission to establish a pro- gram in which scientific observers are placed on board tuna purse-seine and longline vessels fishing in the WIO. Bycatch in the tuna purse-seine fisheries of the western Indian Ocean Evgeny V. Romanov Southern Scientific Research Institute of Marine Fisheries and Oceanography (YugNIRO) 2, Sverdlov St 98300, Kerch, Crimea, Ukraine E mail address islande'cnmeacom Manuscript accepted 20 March 2001. Fish. Bull. 100(1): 90-105 (2002). One of the most inipoitaiit require- ments of the UN Convention on the Law of the Sea of 1982, which determines strategies for exploitation of marine living resources (Article 119, b), is to take into account the impact of fish- eries on ". . . species associated with or dependent upon harvested species with a view to maintaining or restor- ing populations of such associated or dependent species above levels at which their reproduction may become seri- ously threatened. . ." (United Nations, 1983). Estimating the magnitude of bycatch is one of the first steps to deter- mine the impact of fisheries on associ- ated species. Tuna purse-seine fisheries probably apply the most intensive direct htnnan impact on the tropical epipelagic eccsys- tems in all oceans. Because of the world- wide scale of purse-seine fisheries, an assessment of their impact on associat- ed and dependent species is essential. Two tunas, yellowfin Thunnus alba- cares (Bonnaterre, 1788) and skipjack Katsiiwonus pe/amis (Linnaeus, 1758), are the target species of most purse- seine fisheries. In this study bycatch is defined as the fraction of the catch that consists of nontarget species (including other species of tuna) that are encircled by the fishing gear and are unable to escape by themselves. Bycatch of asso- ciated and nonassociated species dur- ing purse-seine fishing for tropical tu- nas may be rather high, and generally depends on fishing tactics. The species composition of bycatch in purse-seine fisheries depends on the structure, behavior, and spatial organi- zation of siu'face multispecies aggrega- tions. Schools of different tuna species and other pelagic fishes, marine mam- mals, and other marine animals have aggregated distributions. From our ob- servations and in the opinion of other researchers (Au and Ferryman, 1985; Au and Pitman, 1986; Au, 1991; Cort, 1992), marine birds arc also an inte- gral component of the majority of these multispecies groups. The tunas, as a rule, prevail by bio- mass and abundance in such groups. Tuna schools are traditionally classi- fied by the visually distinctive part of the group or by whether they associate with floating objects or marine mam- mals (Scott, 1969; Petit and Stretta, 1989). "Free-swimming schools" may include associations between different species of tuna. For each type of school, its various components occur in differ- ent ratios. Some epipelagic species that occur in the purse-seine bycatches are not mem- bers of multispecies aggregations. They, instead, may comprise members of the flotsam community or are tuna forage. Several associated components, such as whales and birds, usually escape or avoid the nets and do not become by- catch. Therefore, the composition of the catch often does not represent the actu- al species composition of the multispe- cies associations. Assessments of bycatches have been made for the eastern Pacific Ocean purse-seine tuna fishery (Joseph, 1994; Garcia and Hall, 1995; Hall, 1996, 1998; Anonymous, 1997, 1998, 1999). where the bycatch problem attracted attention because of dolphin mortality during sets on dolphin-associated tuna schools. The economic, political, and ecological implications of this problem produced wide international attention (Charat- Levy, 1991; Jcseph, 1991, 1994; Hall, 1998). Bycatch estimates for the west- ern Pacific purse-seine tuna fisheries have been published also (Bailey et al., 1996). Romanov Bycatch in the tuna purse seine fisheries of the western Indian Ocean 91 111 th(_' \v(-stern Indian Ocean iW'IOi. lima-clolpliin as- sociations are well known in coastal pelagic zones, e.g. Gulf of Aden (Deniidov') and Sri-Lanka (do Silva and Bon- iface-). They are often used in small-scale troll and pole- and-line fisheries for locating yellowfin tuna. In offshore regions of the WIO tuna-dolphin associations are rare, purse seining for them is not practiced, and there is no dol- phin bycatch problem. Perhaps for this reason, the magni- tude of bycatch in the WIO is unknown, except for recent information on species composition (Santana et al.. 1998). Bycatches are not recorded for tuna seiners operating in the WIO, except bycatches of nontarget tuna species. This paper represents a first attempt to estimate catches of as- sociated species by tuna purse seiners in the WIO, based on scarce information collected bv scientific obsen'ers. Materials and methods Bycatch assessments were based on data collected by Yug- NIRO scientific observers aboard Soviet (since 1992 — Rus- sian) tuna purse seiners in the WIO, during 1987, and 1990-91. The vessels were the "Rodina" type.-^ In addition, observer data collected in the same area aboard sister- ships by AtlantNIRO^ and "Zaprybpromrazvedka""' during 1986-90 and data by TINRO'^ and TURNIF' during 1990 and 1992 were used. The fishing vessels all used purse seines of 1800 m in length, 250-280 m in depth, and 90-100 mm mesh size in the bunt. The principal goal of the observer sampling program was an estimation of the species composition of catches in this fisheries, biological analysis of the principal species, and estimates of the length and weight compositions of these principal species in the catches. The observers were placed on board opportunistically (i.e. if a vessel had a free sleeping bed and if there was available funding), without a sampling scheme and without preference to any vessel type. Thus, the sampling could be considered as random. ' Demidov, V. F. 1998. Personal commun. Southern Scien- tific Research Institute of Marine Fisheries and Oceanography (YugNIRO), 2. Sverdlov St., 98300, Kerch, Crimea Ukraine. -' de Silva, J, and B. Boniface. 1991. The study of the handline fishery on the west coast of Sri Lanka with special reference to the use of dolphin for locating yellowfin tuna I Thimnuti albacares I. /;i Indo-Pacific Tuna Development and Management Pi'ogramnie (IPTPl Coll. Vol. Work. Doc TWS/90/18., Vol. 4, p. 314-324. Food and Aginculture Organization of the United Nations (FAO), Viale delle Terme di Caracalla, 00100, Rome, Italy. ■'* Length overall: 85 m; CRT (gross tonnage): 2634; carrying capacity: -1600 mV ■> AtlantNIRO— The Atlantic Scientific Research Institute of Marine Fisheries and Oceanography, 5 Dmitry Donskoi St., 2.36000 Kaliningrad. Russia. ■'' The Department of Searching and Scientific Research Fleet of the Western Basin "Zaprybpromrazvedka," ^" Dmitry Donskoi St., 236000 Kaliningi-ad. Russia. " TINRO— The Pacific Scientific Research Institute of Marine Fisheries and Oceanography, 1 Shevchenko Alley, 690600 VHad- ivostok, Russia. ' TLIRNIF — The Pacific Department of Fish Searching and Sci- entific Research Fleet, 2 Pervogo Maya St., 690600 Vladivostok, Russia. Two other types of Soviet fishing vessels, "Tibiya"*' and "Kauri,""' which took part in the Indian Ocean fisheries during 1985-87 and since 1991 (under the Liberian fiag), were not sampled. In this study coverage rate was esti- mated as percentage of sampled catch to total catch. The obsei-vers recorded the results of each set. The type of school, according to Scott ( 1969) and Petit and Stretta ( 1989), of each set was recorded. I considered sets ftir which an ob- server recorded catch in any quantity as positive sets. The average bycatch level was estimated for all positive sets. For the positive sets, species composition, total weights, and numbers of each species in the catch were recorded. In the vessels of the "Rodina" type, the retained catch was frozen and stored separately. The retained catch was weighed after freezing while being moved to the ship's holds. In nine cases, the weight of some of the catch was es- timated by the ship masters because the holds were over- loaded and some catch was stored in the freezers till land- ing. Therfore estimates of retained catch are presented in this study as frozen weights rather than wet weights. The bycatch was estimated as wet weight. CJnly bycatch taken on board was sampled. The sets when bycatch was not taken onboard but discarded alive (usually with negligible target species catch) and malfunction sets, which do not produce any catch, were not analyzed in this study. Large species, sharks and billfishes generally, were weighed and counted. The weights of specimens heavier than 200 kg (i.e. Mobulidae) were estimated. When the bycatch was more than 200-300 kg, species composition and weight were estimated by using representative samples. Sometimes the obsei-ver recorded the bycatch in num- bers. In these rare cases, the total weights of the fishes were estimated from the average weights of these species in previous catches. The obsei-vers had free access to every fish in the catch. Nevertheless, some obsei-vers had difficulties identifying some billfishes, sharks, and Mobulidae species. Therefore, I pooled the records with doubtful species identification into these three groups for my analysis. These are marked by "?" in the tables. The data were gi'ouped and analyzed by free-swimming schools (including associations between schools of differ- ent species of tuna) and associated schools. The latter in- cluded whale-associated schools and log-associated schools (associated with floating objects). Schools caught in the area of seamounts and shoals — at the peaks of the Equator Seamount and at Saya-de-Malha bank — were considered free-swimming schools. Some ob- servers did not record the type of floating objects that were set on; therefore the sets on natural floating objects (509f to 90% of the log sets sampled) and on fish aggregation devices (FADs) (10-50%) were grouped. Several log sets were made in areas with surface evidence of water masses or current interactions (rips). A set that could not be clearly identified as to set tjqpe was made in such an area and was treated as a log set because of the species composition of the catch and the occurrence of small scattered debris in the rips. ^ Length overall: 55.5 m, GRT: 736, carrying capacity: -361 m-'. '' Length overall: 79.8 m. GRT: 2100. carrying capacity: -1200 m-'. 92 Fisher-y Bulletin 100(1) Table 1 Numbers of sets sampled by year. Positive sets are sets m which an obsei-ver registered catch in any quantity . 1986 1987 1988 1989 1990 1991 1992 Total Total number of sets Number of positive sets Percentage of sets with catch 115 102 30 41 113 54 39 494 68 62 28 41 92 53 33 377 59* 63% 93% lOO'-i 81-"/ 98^; 85'-'f 76 Table 2 Numbers of sets sampled by season and type of school. Type of school Seasoi s Total/positive Winter Spring Summer Autumn Free-swimming 136 35 27 8 206/121 Whale-shark-associated 2 0 0 0 2/2 Whale-associated 23 21 1 0 45/37 Log-associated 46 50 80 65 241/217 Total 207 106 108 73 494/377 Because tuna purse-seine fishing in the WIO is clearly seasonal (monsoons governing fishing techniques and op- erations), the data were analyzed by season. I followed Romanov's (1982) seasonal divisions, in accordance with long-term average seasonal variations in the monsoon atmospheric circulation for the WIO. The winter season (northeastern monsoon) lasts from December to March. the spring intermonsoon period falls during April and May, the summer (southwestern monsoon) lasts from June to August, and the autumn intermonsoon period lasts from September through November. The wind regime de- termines the onset and duration of the hydrological sea- sons, which do not quite coincide with seasons of atmos- pheric circulation owing to a considerable time lag of the processes occurring in the ocean. However, the wind re- gime is instrumental in determining the tactics of purse seining for tuna; therefore I used seasonal strata based on atmospheric rather than on hydrological processes. The spatial and temporal distribution of catch and ef- fort for the Soviet tuna purse-seine fishery in the Indian Ocean was determined from data in the YugNIRO data- base, a collection of daily radio reports from vessels fishing in the area from 1983 until the mid- 1990s. i" The catches reported by the author's estimates varied by 96-99'7f dur- ing 1985-91, decreasing to 71% in 1992. This study did not take into account reflagging of some Soviet (from 1992 — Russian) vessels with the Liberian flag, and the vessels' nationality was defined in this study by the loca- tion of their shipowners. Analysis of fleet activity and ex- '" Daily information on fishing activity of these vessels in the Indian Ocean in 1983-84 and since 1995 is not available. trapolations of results were made on the assumption that the operations and procedures on vessels that did not car- ry observers did not differ from the operations and proce- dures on vessels with an obsei"ver aboard; similarly it was assumed that the species composition of the catch from these vessels did not differ. Some of the bycatch was retained on board the fishing vessels. Unused bycatch was discarded in the ocean. The observers usually did not record the levels of discards, and it was not possible to assess quantitatively the discards of tuna and associated species. Average values are presented as arithmetic means, plus or minus 95% confidence intervals for estimated values. Estimates of unrecorded bycatches for all fishes, except tu- nas, are provided in numbers and metric tons per positive set and per 1000 t of target species. Results Primary data and adequacy of samples A total of 494 purse-seine sets were sampled and 377 posi- tive sets were analyzed. The total catch in the sets that were sampled amounted to 7713 t. The distribution of sets sampled by years, seasons, and the types of schools is given in Tables 1 and 2. The catch sampled by type of school is presented in Table 3. The obsei-ver coverage rate varied from 0% (no obsei-v- ers at sea) to 75% and averaged 14% during 1986-92. Dur- ing the periods when observers were on board, the cover- age rate averaged 30% and varied from 5% to 75%. The spatial distribution of sampled sets agi-eed quite well with Romanov Bycatch In the tuna purse seme fisheries of the western Indian Ocean 93 S20 35E 40E 45E 50E 55E 60E 65E 70E 35E 40E 45E 50E 55E 60E 65E 70E Figure 1 (A) Fishing effort distribution I0=noon positions of vessels on fishing days with sets) of the Soviet tuna purse- seine fishery in 1985-94; (B-Di sampled set positions: (B) on free-swimming schools that were sampled; (C) on whale-shark (A) and whale-associated schools (x); (D) on log-associated schools. The shaded area represents the region of the main international tuna purse-seine fishing activity in the WIO. according to Ardill." Table 3 Sampled catch I metric tons) by season and type of school. Type of school Seasons Total Winter Spring Summer Autumn Free-swimming 1884 249 73 24 2230 Whale-shark-associated 28 0 0 0 28 Whale-associated 584 467 4 0 1055 Log-associated 925 785 1156 1534 4400 Total 3421 1501 1213 1558 7713 the distribution of the total fishing effort of the Soviet fleet in the WIO (Fig. 11. Sampled sets were distributed throughout the region of the principal international tuna purse-seine fishing activity in the WIO (Aj-dill"). Thus, I Ardill, J. D. 1995. Atlas of industrial tuna fisheries in the Indian Ocean ( IPTP/95/AT/3 ). IPTP, Colombo, Sri Lanka, 138 p. FAO. Viale delle Terme di Caracalla, 00100, Rome. Italy 94 Fishery Bulletin 100(1) Total number of sampled sets and average annual fishing effort by seasons 250 200 150 - Two whale-shark assocrated sets ] Log-associaled ] Whale-associated I Whale-shark-associated 9 Ffee-swimming - Average effort (fishing days) - Average effort (sets) E 100 Winter Spring Summer Autumn Total sampled catch and average annual catch (t) by seasons Winter Spnng Summer Autumn 500 450 400 350 300 250 200 150 -I- 100 50 0 B 3 500 -T 1 - 4 000 \ 1 1 Log-associated 1 1 Whale-associated ■■ Whale-shark-associated BIB Free-swimming „,4.„ Average catch - 3 000 - 2 500 - 28 1 caught in whale- shark-asscx;iated sets - 3 500 - 3 000 sz u S 2 000 - ^V < - - 2 500 > ■ \ t caught in whale- associated set * - - 2 000 :uras iixyniuhu.'i Rafinesque, 1809 + hiiriit: spp. 9 Carcharhinidae Cai-charhmiix falcifnrmis {Bihron. 1839) + + + C. longimaniis (Poey, 1861 ) + + + r'C. obsciirus (LeSueur, 1818) + ? +7 Ccirchaihinus spp. 9 9 Sphyrnidae Sphyrna lewini i Griffith & Smith. 1834) -h Sphyrna spp. + Exocoetidae sp. + Belonidae sp. + TylosuruK cruciidilus (Peron & LeSueur. 1821) + Lampidae Lcimpris giittatiis (Brunnich, 1788) + SphjTaenidae Sphyraena barracuda (Walbaum, 1792) + Sphyraena spp. + Carangidae Caranx spp. ■f Decapterus macarelliis Cuvier, 1833 + Decapterus spp. + Elagatis bipmnulata (Quoy & Ganiiard, 1824) + + Seriola spp. + + Naucratea ductor (Linnaeus. 1758) + Coryphaenidae Corypliaena hippuriis Linnaeus, 1758 + + Coryphaena spp. + Kyphosidae Kyphofiiis cinerasccns (Forsskal, 1775) + Gempylidae Gempy/iis serpens Cuvier, 1829 + Ruvettus pretiosus Cocco, 1829 + Ephippididae Platax spp. + + Scomberomoridae Scomberomortisconiiiicrsoii (Lacepedc, 1800) + Scomberomonis spp. ■^ Scombridae AcanthocybiiiDi solaitdri (Cuvier. 1831) + continued Romanov: Bycatch in the tuna purse-seine fisheries of the western Indian Ocean 97 Table 4 (continued) Family and species School type Free-swimmins Whale-associated Log-associated Pisces 1 continued 1 Aiixis rochvi (Risso, 1810) + Aiixis thazard (Lacepede, 1800) + -1- + Euthynnus affinis (Cantor, 1849) + Katsiiironiix pplamis (Linnaeus, 1758) + + + Tluiiinut: alalunga (Bonnaterre, 1788) + -f- Thuiiniis albavarea (Bonnaterre. 1788) + + + Tht/nrius obcsiis (Lowe, 1839) + + + Istiophoridae laliophorus platypterus iShaw & Nodder. 179 2) + Makaira indica (Cuvier, 1832) + + M. mazara (Jordan et Snyder, 1901) + + Makaira spp. + + Tctrapti/riis audax (Philippi, 1887) + Xiphiidae Xiphias g/adius (Linnaeus, 1758) + Nomeidae Cuhiceps paiuiradiatus Gunter, 1872 + Balistidae Canthidennis inaculatus (Bluch, 1786) + + Monacanthidae Alutcnis monoceros (Linnaeus, 1758) + Alutcnit: spp. + Diodontidae Diudon spp. + + > Mammalia Balaenopteridae Balacnoptera borealis Lesson, 1828 + Salpae + Ctenophora + Chelonidea ^■ Number of species (taxa) 19 17 45 ' Recorded in whale-shark-a.ssociated schools. Table 5 Average tuna catch per positive set (t) by "Rodina -type Soviet vessels in the western Indian Ocean (total and by species). YFT = yellowfin tuna, SKJ = skipjack tuna, BET = bigeye tuna, ALB =albacore, FRI = frigate tuna. KAW = kawakawa. + = catch was <0.001 t. Type of school Total Species YFT SKJ BET ALB FRI KAW Free-swimming 18.4 ±5.2 14.7+4.9 2.8 ±1.7 0.8 ±1.0 0.03 ±0.03 0.05 ±0.06 — Whale-associated 31.0 ±9.3 9.8 ±4.3 18.3 ±8.5 2.0 ±2.4 — 0.2+0.2 — Log-as.sociated 20.6 ±3.2 4.9+0.9 13.9 ±2.7 0.6 ±0.2 0.04 ±0.04 0.3 ±0.3 0.001 +0.001 Total 20.6 ±2.7 8.6 ±1.8 10.5 ±1.9 0.8 ±0.4 0.03 ±0.03 0.2 ±0.2 + 98 Fishery Bulletin 100(1) Table 6 Estimates of the bycatch (t) of various species (groups) of marine animals by school type The numerator is the average values per a positive set, the denominator is the average values per lOOOtoftai get species. + = catch was <0.001 t. School type' Free- Whale- Log- All types Species or group of species swimming associated associated of schools Billfishes (Istiophoridae, Xiphiidae) 0.016/0.89.5 0.006/0.218 0.019/1.008 0.017/0.880 Wahoo (A. solandn) — — 0.031/1.621 0.018/0.934 Sharks (Lamnidae, Carcharhinidae, Sphyrnidael 0,02.3/1.296 0.289/10.302 0.17.5/9.288 0.151/7.938 Rainbow runner (£. hipinnulata) 0.001/0.0.54 — 0.19.5/10.314 0.114/5.962 Dolphinfishes (C. hippurus) +/0.027 0.001/0.0.51 0.191/10.098 0.111/5.836 Barracuda (S. barracuda) — — 0.002/0.132 0.001/0.076 Triggerfishes (C. maculatua.Alutcrus spp.l +/+ — 0.137/7.277 0.080/4.195 Mackerel scad (£). macarf/lus) — — 0.0093/0.491 0.00.5/0.283 Mantas, mobulas (Mobulidae) 0.020/1.128 0.009/0.318 0,002/0.126 0.009/0.455 Sea turtles — — +/0.025 +/0.014 Other bycatch +/0.002 +/0.003 0.018/0.958 0.011/0.553 r For positive set I For 1000 t of target species 0.060+0.031 3.403 +2.770 0.306 ±0.344 10.891 ±15.787 0.780 ±0.144 41.337 ±14.281 0.518 ±0.099 27.127 ±8.869 ' Because of the small sample size, estimates of bycatch for whale-shark-associated schools are not presented in the Table, 70 -| 60 - 50 - 40 ■ I 30 - 20 ■ 10 ■ 1 i Free- Whale- Log- swimming assoc ;iated assoc ;iated All types Figure 4 Bycatch (t) per 1000 t of target species by school type for the Soviet tuna purse-seine fishery. Dots are means, bars are 95'>'f confidence intervals. sible. Whales often remain in the net until the end of purs- ing and then escape from the purse seine by either diving under the purse line, by ramming through the net wall, or by sinking the corkline (a rare occurrence). Observers registered a single case of entanglement in the net and subsequent death of a young sei whale about 10 m in length and about 12 t in weight. The dead animal was taken up on the vessel's deck, released from the purse seine, and discarded into the ocean. It is not possible to as- sess the frequency and probability of whale mortality by the purse-seine fishery in the WIO. There were 1 7 species ( or groups ) of marine animals iden- tified in the catches of whale-associated schools (Table 4). Salps, ctenophores, and batfish (Platax spp.) were consid- ered accidental bycatch, whereas long-finned fathead (Cu- biceps pauciradiatus) was a prey item of both tunas and whales. Nontuna bycatch in this type of association aver- aged 0.306 ±0.344 t for a positive set or 10.891 ±15.787 t per 1000 t of target species (Figs. 3 and 4). Sharks of the genus Cairharhiniis and Isui-iis made up the bulk of the bycatch in whale-associated school .sets (0,289 t/10.302 t) (Tables 4 and 6). Log-associated schools Log-associated schools are one of the predominant school types found in the WIO all year round (Table 2, Fig, 2, A and B). Sets on log-associated schools were made through- out the sampling area as far south as 15°S (Fig. ID). In log-associated schools the bulk of the catch were skipjack, yellowfin, and bigeye tunas — 6Ty( , 2A'7i . and y?( . respec- tively (Table 5). Log-associated schools in all cases con- sisted of several fish species. Bycatch was found in 93% of the sets, and nontuna bycatch in 87%. The absence of bycatch was rare, observed only during successive sets on the same floating object. The species composition associated with floating objects was the most diverse of any set type and included 45 spe- cies (or higher taxa of fishes) (Table 4). Nontuna bycatch was at its highest in log-associated sets, as much as 0.780 ±0.144 t per positive set or 41.337 ±14.281 t per 1000 t of target species (Figs. 3 and 4). The bulk of the bycatch in sets on log-associated schools was made up of rainbow runner. Romanov: Bycatch in the tLina purse seine fisheries of the western Indian Ocean 99 Elcii^atis hipinnulata (0.195 t/10.314 t), common dolphin- fish. Coryphaena hippiirus (0.191 1/10.098 t), triggerfish of the genus Canthidermis (0.137 t/7.277 t), sharks of the ge- nus Carcliarhinits (O.n.'i t/9.28cS t), wahoo. Acaiithocyhi- uni solandri (0.031 t/1.621 t), billfishes of the genera AUik- aira and Tetrapturus (0.019 t/1.008 t), and mackerel scad, Decapterus macarclliis (0.0093 t/0.491 kg). One capture of a sea turtle (unknown species) was recorded (Tables 4 and 6). All types of schools Considering all school types in the aggregate, skipjack, yel- lowfin, and bigeye tuna prevailed in the catch — Sf/r , 429r, and 4'?; by weight, respectively (Table 5). Albacore repre- sented a mere 0.2'7(, frigate tuna 0.9%, and kawakawa, Etithyninis affinls. less than 0.1%. Nontuna bycatch accounted for less than 3'^f of the catch. On the average, there was 0.518 ±0.099 t of nontuna by- catch caught per positive set, or 27.127 ±8.869 t per 1000 t of target species (Fig. 3). Bycatch levels by species (groups) are given in Table 6. Discussion The lowest fish bycatch in the WIO tuna purse-seine fish- ery was taken from free schools (mainly carcharhinid sharks and Mobulidae rays) (Figs. 3 and 4, Tables 4 and 6). Bycatch of fishes was highest and most diverse from catches on log-associated schools. Rainbow runner, common dolphinfish. triggerfish. carcharhinid sharks, wahoo, bill- fishes, and mackerel scad were predominant. Whale-asso- ciated schools were characterized by an intermediate level of bycatch (mainly carcharhinid and lamnid sharks) (Figs. 3 and 4, Tables 4 and 6 1. It is interesting to compare the bycatch rates obtained in this study with those published for other regions. The principal bycatch fishes in the Pacific (Bailey et al., 1996; Hall, 1996, 1998; Anonymous, 1997) are the same as those presented here. Bycatch levels are known to vary consid- erably by year, area, fleet (Bailey et al, 1996; Hall, 1996; Anon., 1997), and school type; this variability hampered direct comparisons of the results from the present study with those from published data. However, for the purpose of comparison, I pooled my estimates by gi'oups in accor- dance with the published data (Bailey et al., 1996; Hall, 1996, 1998; Anonymous, 1997). Bycatch levels per set and per 1000 t of target species for various regions of the Pa- cific and my estimates for the Indian Ocean are on the same order of magnitude for most groups in similar types of associations (Figs. 5 and 6). I also attempted to estimate the unrecorded bycatch by the purse-seine fleets of the principal fishing nations of the WIO by a comparison of fishing tactics. The Soviet fleet in the WIO made an equal proportion of sets on free-swim- ming schools and on log-associated schools during the year (Table 2). Seasonally they switched effort from sets on free-swimming schools to those on log-associated schools (Fig. 7, A and B).The fishing practices of French and Span- ish tuna seiners showed similar seasonality until the niid- 1990s (Anonymous;'"'"' 1*^ Planet;'^'" Moron'-'). The fishing tactics of the Japanese (Hallier;^'' Okamoto and Miyabe-') and Mauritian (Norungee et al.;-- No,.yn. gee and Lim Shung-') purse-seine fleets differed consider- ably from that described above. Japanese and Mauritian vessels made sets on log-associated schools all year round, with single instances of sets on other schools types. Only two school types (log schools and free schools) have been described by Hallier;-" Hallier;-^ Parajua Aranda;^'' 'Anonymous. 1992. Report of the workshop on stock assess- ment of yellowfin tuna in the Indian Ocean, Colombo, Sri Lanka. 7-12 October 1991, 90 p. |IPTP/91/GEN/20.| FAG, Viale delle Termc di Caracalla, 00100, Rome, Italy. • Anonymous. 1994a. Report of the expert consultation on Indian Ocean tunas, .5th session. Mahe, Seychelles, 4-8 Octo- ber 199.3, 32 p. IIPTP/94/GEN/22.1 FAO. Viale delle Terme di Caracalla. 00100, Rome, Italy. ' Anonymous. 1994b. National report of Spain. In Proceed- ings of the expert consultation on Indian Ocean tunas, 4-8 October. 1993 (J. D. Ardill. ed. i. p. 44-47. IPTP Coll. Vol. 8., TWS/93/1/14. FAO. Viale delle Terme di Caracalla. 00100, Rome. Italy. ' Pianet. R. 1994a. Purse seine fishery trends in the western Indian Ocean from data collected in Victoria (Seychelles), 1984-1992. //( Proceedings of the expert consultation on Indian Ocean tunas, 4-8 October, 1993 (J. D. Ardill. ed.). p. 41-44. IPTP Coll. Vol. 8.. TWS/93/1/13. FAO. Viale delle Terme di Caracalla, 00100. Rome. Italy. ' Pianet. R. 1994b. National report of France. //! Proceedings of the expert consultation on Indian Ocean tunas. 4-8 October. 1993 (J.D.Ai-dill,ed.i.p.48-.52. IPTP Coll. Vol. 8.TWS/93/1/16. FAO, Viale delle Terme di Caracalla, 00100, Rome. Italy ' Moron. J. 1996. National report of Spain. In Proceedings of the expert consultation on Indian Ocean tunas. 6th session, Colombo. Sri Lanka. 2.5-29 September. 1995 (A. A. Anagnuzzi, K. A. Stobberup, N. J. Webb, eds. I, p. 63-69. IPTP Coll. Vol. 9. FAO, Viale delle Terme di Caracalla, 00100, Rome, Italy ' Hallier. J.-P. 1991. Tuna fishing on log associated schools in the Western Indian Ocean: an aggregation behaviour. /;; IPTP Coll. Vol. Work. Doc, Vol. 4. p. 325-342 [TWS/90/66.1 FAO, Viale delle Terme di Caracalla. 00100. Rome. Italy. Okamoto. H., and N. Miyabe. 1996. Review of Japanese tuna fisheries in the Indian Ocean. In Proceedings of the expert consultation on Indian Ocean tunas. 6th session, Colombo. Sri Lanka. 25-29 September. 1995 (A. A. Anagnuzzi, K. A. Stobb- erup, N.J. Webb, eds.), p. 15-21. IPTP Coll. Vol. 9. FAO, Viale delle Terme di Caracalla, 00100, Rome, Italy ' Norungee, D., A. Venkatasami, and C. Lim Shung. 1994. Catch and landing statistics of the Mauritian tuna fisheries (1987-1992) and an analysis of the skipjack tuna catch of the Mauritian purse seine fishery (1987-1993). In Proceed- ings of the expert consultation on Indian Ocean tunas. 5th ses- sion. Mahe. Seychelles. 4-8 October. 1993 (J. D. Ai'dill. ed.), p. 266-273. IPTP Coll. Vol, 8. TWS/93/4/5. FAO. Viale delle Terme di Caracalla. 00100. Rome. Italy. ' Norungee. D.. and C. Lim Shung. 1996. Analysis of the purse seine fishery of Mauritius, 1990-1994, and comparison of catch rate and species composition of catches of Mauritian purse seiners to those of French fleet. In Proceedings of the expert consultation on Indian Ocean tunas, 6th session, Colombo, Sri Lanka. 25-29 September, 1995 (A. A. Anagnuzzi, K. A. Stobb- erup. N.J. Webb. eds.). p. 1.5-21. IPTP Coll. Vol. 9. FAO. Viale delle Terme di Caracalla, 00100. Rome. Italy Hallier. J.-P 1994. Purse seine fishery on floating objects: What kind of fishing effort? Wliat kind of abundance indices? In continued 100 Fishery Bulletin 100(1) Table 7 Bycatch estimates in tons in the western Indian Ocean pu ■se-SfUie fisheries during 1985-94. MIX = fleets targe ted all t\ pes of schools (France Spain. USSR ,LOG = fleets targeted log-associated schools (Japan an d Mauriti us). Species, a groui: of species 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 MIX 913 1047 1257 1674 1622 1503 1471 1793 1796 1925 Pelagic oceanic sharks LOG 31 30 61 81 108 ISO 278 477 451 143 Total 944 1077 1318 1755 1730 1683 1749 2270 2247 2068 MIX 686 786 944 1257 1218 1129 1105 1347 1349 1446 Rainbow runners LOG 34 33 68 90 120 199 309 530 500 159 Total 720 819 1012 1347 1338 1328 1414 1877 1849 1605 MIX 671 770 925 1231 1193 1105 1082 1318 1320 1415 Dolphinfishes LOG 34 33 67 88 117 195 303 518 490 1.56 Total 705 803 992 1319 1310 1300 1385 1836 1810 1571 MIX 483 554 665 885 857 794 778 948 949 1017 Triggerfishes LOG 24 24 48 64 84 141 218 374 353 113 Total 507 578 713 949 941 935 996 1322 1302 1130 MIX 108 123 148 197 191 177 173 211 211 227 Wahoo LOG 5 5 11 14 19 31 49 83 79 25 Total 113 128 159 211 210 208 222 294 290 252 MIX 101 116 139 185 180 167 163 199 199 213 Billfishes LOG 3 3 7 9 12 20 30 52 49 16 Total 104 119 146 194 192 187 193 251 248 229 MIX 52 60 72 96 93 86 84 103 103 110 Mobulas and in intas LOG <1 <1 1 1 1 2 4 6 6 2 Total 53 60 73 97 94 88 88 109 109 112 MIX 33 37 45 60 58 54 53 64 64 69 Mackerel scad LOG 2 2 3 4 6 10 15 25 24 8 Total 35 39 48 64 64 64 68 89 88 77 MIX 9 10 12 16 16 14 14 17 17 19 Barracudas LOG <1 <1 1 1 1 3 4 7 6 2 Total 9 11 13 17 17 17 18 24 23 21 MIX 64 73 88 117 113 105 102 125 125 134 Other fishes LOG 3 3 6 8 11 18 29 49 47 15 Total 67 76 94 125 124 123 131 174 172 149 MIX 3120 3576 4295 5718 5541 5134 5025 6125 6135 6574 Total nontuna bycatch LOG 137 134 273 360 479 799 1239 2121 2004 638 Total 3257 3710 4568 6078 6020 5933 6264 8246 8139 7212 Anonymous;'^ ^'' "' Planet;'"- 1- Hastings and Domingue;-'' and Moron'" for the tuna purse-seine fishery in the In- 24 i""itinui'di Proceedings of the expert consultation on Indian Ocean tunas, 5th session, Mahe, Seichelles, 4-8 October. 1993 (J. D. Ai-dill,ed.), p. 192-198. IPTP Coll. Vol. 8.,TWS/9.3/2/25, FAO. Viale delle Terme di Caracalla. 00100, Rome, Italy 2= ParajuaAranda.J. I. 1991. Spanish status report of vellowfin tuna fishery 1984-1990. In IPTP Coll. Vol. Work. Doc, Vol. 6, TWS/91/13, p. 99-130. FAO, Viale delle Terme di Caracalla, 00100, Rome, Italy '^^ Hastings, R. E., and G. Domingue. 1996. Recent trends in the Seychelles industrial fishery. In Proceedings of the expert consultation on Indian Ocean tunas, 6th session, Colombo, Sri Lanka, 25-29 September, 1995 (A. A. Anagnuzzi, K. A. Stob- berup, N. J. Webb, eds), p. 97-109. IPTP Coll. Vol. 9. FAO, Viale delle Terme di Caracalla, 00100, Rome, Italy. dian Ocean. Free schools in these analyses included all types of associations with marine animals. The propor- tion of sets of the French fleet on other types of schools and on resulting catches is not known. Cort (1992) pre- sented such data for Spanish vessels, based on fishing logbooks. Therefore, I used the observers data of the Sey- chelles Fishing Authority (SFA) (Cort, 1992) for the ves- sels of France, Spain, Japan, and USSR to assess these values in the WIO. The percentage of sets on whale-asso- ciated schools varied from 1.7% to 8.8% in 1986-90, the percentage among positive sets was from 1.2% to 9.1%, and the catch from such schools was 1.6% to 7.8% (cited from Cort, 1992). These values are slightly lower than the observer data I report in the present study (9%, 10%, and Romanov Bycatch in the tnna purse seine fisheiies of the western Indian Ocean 101 1,000 1 900 800 - 700 S 600 f 500 z 400 300 200 100 0 Billfisties A DThis study ■ I-ATTC1993 Dl-ATTC 1994 Free-swimming Log-associated Marine mammals Small fishes 0 0 2 8 DThis study ■ l-ATTC 1993 Dl-ATTC 1994 00 42 24 Free-swimming Log-associated H/larine mammals 450 400 350 ■ 300 250 200 150 100 50 0 Stiarl^s Free-swimming Log-associated IVIarine mammals Dolptiinfisties, wahoo, rainbow runners D 07 DThis study ■ l-ATTC 1993 Dl-ATTC 1994 02 0 1 Of Free-swimming Log-associated Manne mammals Sea tunies DThis study ■ l-ATTC 1993 Dl-ATTC 1994 Free-swimming Log-associated t^anne mammals Figure 5 Bycatch levels in numbei's per set by groups of species and by types of schools in the western Indian Ocean and eastern Tropical Pacific (Anonymous, 1997). 14^'r . respectively), which is explained by the fact that the SFA data included Japanese vessels known to fish on log- associated schools only. Nevertheless, the SFA values and those from our obser\'ers were on the same order of mag- nitude. Proceeding from this, I estimated the ratio of sets on various school types and the magnitude and species composition of bycatch by the French and Spanish ves- sels. These values were close to those for the Soviet fleet employing similar fishing tactics.-^ Thus, the average bycatch estimates presented in this study can be extrapolated for this period to the total WIO purse-seine catch of principal fishing nations targeting all types of schools.-^** Estimates of bycatch from log-associat- ed schools, I believe, can be extended, with some caution, to the pooled purse-seine catch of Japan and Mauritius in the WIO. The annual purse-seine catches of yellowfin and skip- jack tunas by fleets targeting all types of schools (France, Data from logbooks (Cort, 1992) show a lower proportion of sets and of catches on whale-associated .schools for Spanish vessels, but in the author's view a comparison of data collected in the same way (by observers) is preferable. France and Spain (along with catch from the vessels from these two countries flying "flags of convenience" [Panama, Cote d'lvoire, and recently Belize] and applying the same fishing tactics), and USSR (recently Russia or Liberia). 102 Fishery Bulletin 100(1) Billfishes 0 1 ^ 1 — 1 ^ 1 — DThis study ■ Hall, 1996 Sharks and rays 700 -, 600 500 300 - 200 100 B J DThis study [■^Hall, 1996 I Free-swimming Log-associated Marine mammals Dolphinfishes Free-swimming Log-associated Marine mammals Wahoo 5,000 4,000 23,000 a) n E ^2,000 1,000 0 2,500 2,000 2! 1,500 OJ E z 1,000 500 0 165 nThis study ■ Hall, 1996 76 24 Free-swimming Log-associated Marine mammals Rainbow runners E DThis study m^all, 1996 24 5 36 5 00 00 Qi 2,500 2,000 >. 1,500 ) ) ' 1,000 500 0 10,000 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 D 0 0 26 7 DThis study ■ Hall, 1996 1 00 06 Free-swimming Log-associated Marine mammals Tnggerfishes F 0 5 75 6 DThis study ■ Hall, 1996 1 00 74 Free-swimming Log-associated Manne mammals Sea turtles Free-swimming Log-associated Marine mammals Total by-catch Free-swimming Log-associated Marine mammals Free-swimming Log-associated Marine mammals Figure 6 • A-G), Bycatch levels, in numbers per 1000 t of target species, by groups of species and by types of schools in the western Indian Ocean and eastern Tropical Pacific; (H) bycatch levels, in tons per set by types of schools in the western Indian Ocean and western Pacific Ocean. Romanov: Bycatch in the tuna purse-seine fisheries of the western Indian Ocean 103 Spain, and I'SSRV-''' In the WIO ranged between 115,000 and 242.000 t in 1985-94 (Anonymous'"). Japanese and Mauritian catches varied from 3000 to about 51,000 t. Based on these vakies, the estimated bycatcli was 3257 to 8246 t of various fishes during the same period (Table 7), These fishes could serve as food for the coastal countries of the area. Estimat- ed bycatch in numbers is presented in Table 8. Turtle bycatch and whale mortality in purse seines are also possible in the WIO, but the probability of the latter is very low. No instances of whale mortal ity have been recorded earlier for tuna purse-seine fisheries in other areas (Northridge, 1984, 1991a. 1991b: Medina-Gaertner and Gaertner, 1991; San- tana et al., 1991; Cort, 1992; Cayre et al.. 1993; Bai- ley et al.. 1996). No avian mortality by the Soviet tuna purse-seine fishery has been noted by observ- ers. A similar fact was reported for the western Pa- cific (Bailey et al.. 1996). Target fishing for rainbow runner, dolphinfish, triggerfishes, wahoo, mackerel scad, and barracuda is not conducted in the WIO, and these fish are taken only as bycatch. Their bycatch levels, estimated in this study, do not seem to endanger the populations of these species. Estimated bycatch of billfishes ( 104-251 1 annual- ly) was less than I'Ti of the total catch for these spe- cies (14,000-33,000 t during 1985-94) in the WIO (Anonymous^"). The bycatch by the purse-seine fish- ery was unlikely to substantially affect the billfish stocks. Many pelagic sharks are taken as bycatch by the longline, trawl, coastal driftnet, and other fisheries, but are not recorded. The total shark catch by all fisheries may be considerable. Many shark species are characterized by low abundance, low fecundity, long life span, and conse- quently, by high vulnerability to overfishing. Underesti- mation of the removal through fisheries of a number of pe- lagic shark species, and the impact of the fisheries on their populations, may lead to a reduction in their abundance to critical levels, diminishing the biodiversity of the pelagic ecosystem of the Indian Ocean. Some part of the bycatch is released into the ocean alive, although subsequent survival rates are unknown. The lack of bycatch and discard records and estimates of survival rates of discarded animals prevents assessment of the impact of the fishery on the Indian Ocean pelagic ecosystem. Fishing tactics in the WIO have changed considerably by all principal purse seine fleets toward the extensive use of FADs in recent years (generally from 1995). The majority of Japanese vessels have left the area and have moved to the eastern Indian Ocean. Therefore estimates presented here for total WIO purse-seine fisheries are ap- 100 1 , ^ —•— Free swimming 75- -e- Log associated ; ^,_^ Percentage U1 o y<:^^^Z 0 ■ ^"~~~* Winter Spring Summer Autumn lOOn B —•—Free swimming Jd- 75- 0) en m 1 50- o Q) CL 25- -e- Log associated y^ Xl 0- ^~~~~~~~-~~-~,»-________ Winter Spring Summer Autumn Figure 7 (A) Percentages of free-school and log-schools sets; (B) percent- ages of free-school and log-school catch in the .Soviet tuna purse- seine fishery. plicable for a limited time span only (pre- 1995). Recent de- velopment of the WIO fisheries warrants further investi- gation of bycatches through extensive observer sampling by time-area strata. Establishing a scientific program by the Indian Ocean Tuna Commission to monitor the principal tuna fisheries in the region, by placing international scientific observers on purse-seine and longline vessels, might be the first step to- ward a more accurate assessment of the impact of bycatch- es on the epipelagic ecosystem of the Indian Ocean. This program might also lead to developing technical and man- agement measures to reduce the bycatches or to use them. The solution to the bycatch problem should take two di- rections: 1 1 an effort to reduce or eliminate bycatches of un- desired species; or 2) to use bycatch animals to make them target species. The former involves developing gear modi- fications or changes in fishing tactics. The latter involves management regulation of the fishery so that bycatch spe- cies are treated in the same way as other target species. Acknowledgments '^ Including vessels flying flags of convenience. ■"' Anonymous. 1998. Indian Ocean tuna fisheries data sum- mary, 1986-1996. Indian Ocean Tuna Commission (lOTC) data summary 18, 180 p. lOTC, P.O. Box 1011, Victoria, Seychelles. I am giateful to AtlantNIRO scientists V. F Bashmakov, G. A. Budylenko, V. Z. Gaikov, M. E. Grudtsev, to TINRO sci- entist K. A. Karyakin for their data made available to the author and to V. F. Bashinakov and G. A. Budylenko for their personal sampling efforts. I sincerely thank masters of the 104 Fishen/ Bulletin 100(1) Table 8 Bycatch estimates in numbe ■s in the western Ir dian Ocean pursu-seine fisheries during 1985-94. Codes are same as table 7. MIX = fleets targeted all types of schools (Fr ance, Spain. USSR ; LOG = fleets targeted log-associated schools (Japan and Mauri iusi. Species, a gi-oup of species 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 MIX 4.X600 52,273 62,780 83,581 80,993 75,052 73,455 89,529 89,676 96,094 Pelagic oceanic sharks LOG 2161 2100 4280 5653 7531 12,.546 19,456 33,320 31,488 10,030 Total 47,761 54,373 67.060 89,234 88,524 87,598 92,911 122,849 121,164 106,124 MIX 162,4.57 186,232 223.664 297,770 288,5,50 267,386 261,694 318,961 319,485 342,351 Rainbow runners LOG 8112 7883 16.065 21,218 28.267 47,090 73,029 125,065 118,190 37,646 Total 170,.569 194,115 239,729 318,988 316.817 314.476 334,723 444,026 437,675 379,997 MIX 107,711 123,473 148,291 197,424 191.312 177,280 173,.505 211,474 211,821 226,982 Dolphinfishes LOG 5373 5221 10,641 14,0.53 18,723 31,190 48,370 82,835 78,282 24,934 Total 113.084 128.694 158,932 211,477 210.035 208,470 221,875 294,309 290,103 251,916 MIX 621.823 712,823 856,096 1.139.747 1.104.4.58 1,023,4.50 1,001,661 1.220.857 1,222,862 1,310,387 Triggerfishes LOG 31.215 30.334 61,820 81,646 108.774 181,205 281.018 481.252 4.54,799 144,863 Total 653.038 743.156 917.916 1.221. .393 1,213,232 1,204,655 1.282.679 1.702.109 1,677,661 1,4.55,2.50 MIX 17.444 19,996 24.016 31.973 30.983 28.710 28,099 34.248 34,304 36,760 Wahoo LOG 876 851 17.34 2290 3051 5083 7883 13..501 12,7.59 4064 Total 18.320 20,847 25,750 34,263 34.034 33,793 35,982 47.749 47,063 40,824 MIX 750 859 1032 1374 1332 1234 1208 1472 1474 1580 Billfishes LOG 26 25 51 68 90 151 233 400 378 120 Total 776 884 1083 1442 1422 1385 1441 1872 1852 1700 MIX 250 286 344 458 444 411 403 491 491 527 Mobulas and manias LOG 3 2 5 7 9 15 23 39 37 12 Total 253 288 349 465 453 426 426 530 528 539 MIX 45,1.34 51,739 62,138 82.726 80.164 74,285 72.703 88,613 88,758 95.111 Mackerel scad LOG 2266 2202 4487 5926 7895 13,1.53 20,398 34,931 33.011 10,515 Total 47.340 53,941 66,625 88,652 88,059 87,438 93,101 123,.544 121.769 105,626 MIX 1350 1.547 1858 2474 2397 2221 2174 2650 2654 2844 Barracudas LOG 68 66 1.34 177 236 393 610 1044 987 314 Total 1418 1613 1992 2651 2633 2614 2784 3694 .3641 31.58 KUTF tuna seiners A. G. Burlyko. V. N. Volvach, A. A. Kiry- anov for their assistance rendered to observers in sampling. The author is grateful to V. F. Demidov, N. N. Kukharev, M, A. Pinchukov, L. K. Pshenichnov. S, T, Rebik, B, G, Trot- senko for useful discussions when preparing the manu- script and to two anonymous reviewers for their comments and suggestions. The author wishes to thank L V. Charova for translating the paper into English. Revisions and an edition of the pa- per by R. J. Olson (I-ATTC) and his corrections of English were extremely valuable. Literature cited Anonymous. 1997. Annual report of the Inter-Ainerican Tropical Tuna Commission. 1995. lATTC, La Jolla, CA. 334 p. 1998. Annual report of the Inter-American Tropical Tuna Commission. 1996. lATTC. La Jolla. CA. 306 p. 1999. Annual report of the Inter-American Tropical Tuna Commission. 1995. lATTC, La Jolla, CA, 310 p. Au. D. W. K. 1991. Polyspecific nature of tuna schools: shark, dolphin and seabird associations. Fish. Bull. 89:34.3-354. Au. D. W. K., and W. L. Ferryman. 1985. Dolphin habitats in the Eastern Tropical Pacific. Fish. Bull. 83(4 ):62,3-643. Au. D. W. K.. and R. L. Pitman. 1986. Seabird interactions with dolphins and tuna in the Eastern Tropical Pacific. The Condor 88:304-317. Bailey. K., P. G. Williams, and D. Itano. 1996. Bycatch and discards in Western Pacific tuna fisher- ies: a review of SPC data holdings and literature. South Pacific Comm. Tech. Rep. 34. Noumea, New Caledonia. 171 p. Cayre. P. J. B. Anion Kothias, T. Diouf and J. M. Stretta. 1993. Biology of tuna. In Resources, fishing and biology of the tropical tunas of the Eastern Central Atlantic, p. 147-244. FAO Fish. Tech. Pap. 292. FAO, Rome. Charat-Levy, F. 1991. The consequences of the tun;Vdolphin issue in the Romanov Bycdlch in the tLina pLiise seme fishenes of the western Indian Ocean 105 Eastern Pacific. In Tiiiia 91 Bali papers of the '2'"' world tuna trade conference Bali. Indonesia, l,'!-!!) May, 1991 iHenri dc Saram, cd.), p. 19-22. INKOFISH, Kuala Lumpur, Malaysia. Cort, J. L. 1992. Estudio de las asociaciones de tunidos, en especial la denominada "atun-delfin." Su integracion en la biologia de estos peces migi'adores. //) International Commission for the Consei-\-ation of Atlatic Tunas (ICCAT) Coll. Vol. Sci. Pap. 39(1 ):358-384. Garcia. M.. and M Hall. 1995. Spatial and temporal distribution of bycatches of yel- lowfin, skipjack, mahi-mahi and wahoo in the eastern Trop- ical Pacific's purse seine tuna fishery. In Proceedings of the 46th annual tuna conference. (A. J. Mullen, and J. Suter, eds.), p. 54. lATTC, La JoUa. CA. Hall, M. A. 1996. On bycatches. Rev. Fish Biol. Fish.. 6:319-352. 1998. An ecological view of the tuna-dolphin problem: im- pacts and tradeoffs. Rev. Fish Biol. Fish. 8:1-34. Joseph. J. 1991. The consei"vation ethic and its impact on tuna fisher- ies. /'( Tuna 91 Bali papers of the 2'"' world tuna trade con- ference Bah, Indonesia. 13-15 May. 1991 (Henri de .Saram. ed.), p. 12-18. INFOFISH, Kuala Lumpur. Malaysia. 1994. The tuna-dolphin controversy in the Eastern Tropical Pacific Ocean: biological, economic, and political impacts. Ocean Development and International Law 25:1-30. Medina-Gaertner, M. and D. Gaertner. 1991. Factores ambientales y pesca atunera de superficie en el Mar Caribe. ICCAT Coll. Vol. Sci. Pap. 36:523-550. Northridge. S. P. 1984. World review of interactions between marine mam- mals and fisheries. FAG Fish. Tech. I'a[). 251, 190 p. FAO, Rome. 1991a. An updated world review of interactions between marine mammals and fisheries. FAO Fish. Tech. Pap. 251, suppl. 1, 58 p. V\0, Rome. 1991b. Driftnet fisheries and their impact on non-target species: a worldwide review. FAO Fish. Tech. Pap. 320, 115 p. FAO, Rome. Petit M., and J. M. Stretta. 1989. Sur le comportement des bancs de thons observers par avion. ICCAT Coll. Vol. Sci. Pap. 30( 1 1:488-490. Romanov, Yu. A. 1982. Climate features. In The Indian Ocean (series: the world ocean geography) (V. G. Kort and S. S. Salnikov, eds.), p. 43-62. Nauka, Leningrad Santana, J. C, J. Ariz, and A. Delgadu de Molina. 1991. Nota sobre la presencia de mamiferos marinos en la pesquera de tunidos al cerco en el Atlantico este intertropical. ICCAT Coll. Vol. Sci. Pap. 35(1):196-198. Santana. J. C, A. Delgado de Molina, R. Delgado de Molina. J. Ariz, J. M. Stretta, and G. Domalain. 1998. Lista faunistica de las espccics asociados a las capturas de atun de las flotas de cerco comunitarias que faenan en las zonas tropicales de los oceanos Atlantico e Indico. ICCAT Coll. Vol. Sci. Pap. 48(3):129-137. Scott, J. M. 1969. Tuna .schooling terminology. Calif Fish Game 55(2): 136-140. LInited Nations. 1983. The law of the sea. Official text of the United Nations convention on the law of the .sea with annexes and index/ final act of the third United Nations conference on the law of the sea/introductory material on the convention and the conference. United Nations, New York, NY, 224 p. 106 Abstract-Thf natural diet of 506 American lobsters iHomarus america- niis) ranging from instar V (4 mm cephalothorax length. CLi to the adult stage (112 mm CD was determined by stomach content analysis for a site in the Magdalen Islands, Gulf of St. Lawrence, eastern Canada. Cluster and factor analyses determined four size groupings of lobsters based on their diet: <7.5 mm, 7.5 to <22.5 mm, 22.5 to <62.5 mm, and >62.5 mm CL. The onto- genetic shift in diet with increasing size of lobsters was especially appar- ent for the three dominant food items: the contribution of bivalves and animal tissue (flesh) to volume of stomach con- tents decreased from the smallest lob- sters (2871 and 399r, respectively) to the largest lobsters (2'* and 11'^, respec- tively), whereas the reverse trend was seen for rock crab Cancer irroratus il'.i in smallest lobsters to 539* in largest lobsters). Large lobsters also ate larger rock crabs than did small lobsters. This study is the first to examine the natural diet of shelter-restricted juveniles (SP{Js, <14.5 mm CLi, which were thought to be principally suspension feeders and to a lesser degree browsers or ambush pred- ators in or near their shelter However, at our study site no planktonic organ- isms were identified from the stom- achs of SRJs, whereas formaniferans, crustacean meiofauna, and macroalgal debris that could be derived by brows- ing, together represented only 10-14'(^ by volume of stomach contents. We infer that SRJs obtained bivalves by pre- dation and flesh by exploiting larger lobsters' meal scraps or food resei-ves. Some implications of these findings for lobster arti,ficial reef programs and for the conservation of lobster stocks are discussed. Ontogenetic shifts in natural diet during benthic stages of American lobster iHomarus amen'canus), off the Magdalen Islands Bernard Sainte-Marie Denis Chabot Division des invertebres et de la biologie expenmentale Institut Maurice-Lamontagne Peches et Oceans Canada 850 route de la Mer Mont-Joh (Qc), G5H 3Z4 Canada E-mail address (for B Sainte Mane) Sainte Maneadfo mpo gc ca Manuscript accepted 3 October 2001. Fish. Bull. 100(l):106-116i2002). American lob.ster, Homaiiis amcrica- niis, is a long-lived, dominant predator in temperate coastal waters of eastern North America (Elner and Campbell, 1991; Ojeda and Dearborn, 1991). After the lar\'al phase, lobsters settle and spend much of their time in burrows or natural shelters (Cobb, 1971; Lawton, 1987; Barshaw and Bryant-Rich, 1988). However, laboratory and in situ obser- vations indicate that benthic lobsters pass through successive life-history phases as they grow in size, changing from a shelter-restricted habit to a more overt lifestyle involving daily forays and seasonal migrations away from shelter (Cooper and Uzmann, 1977; Cobb and Wahle, 1994). A variety of classifica- tions have been proposed for these suc- cessive ontogenetic phases. The latest scheme, by Lawton and Lavalli ( 1995), recognizes five life-history phases: shel- ter-restricted juvenile (SRJ, -4-14 mm cephalothorax length, CL), emergent juvenile (-15-25 mm CL), vagile juve- nile (-25 mm CL to size of physiological maturity), adolescent, and adult. In several decapod crustaceans, diet changes as individuals grow and be- come more mobile and their chela size and strength increases (e.g. Lee and Seed, 1992; Freire et al., 1996). Such dietary shifts should occur in the lob- ster as well, especially considering this species' changing dependency on shel- ter which, in turn, has implications for foraging range and accessibility of prey types (Elner and Campbell, 1987; Lawton, 1987). Some studies of the nat- ural diet of lobsters 12-125 mm CL have found little or no differences in the identity or in the frequency of food items that were ingested by different size groups (Weiss, 1970; Ennis, 1973; Hudon and Lamarche, 1987). However, other studies have pointed to changes in the identity and especially in the fre- quency of food items ingested by dif- ferent lobster size groups. Carter and Steele ( 1982b). using their own results and data from nonconcomitant studies conducted at different sites in New- foundland (Squires, 1970; Ennis, 1973), have suggested that lobsters of 12-73 mm CL consume sea urchins, ophi- uroids, and mussels more frequently than larger (adult) lobsters. Scarratt ( 1980) reported that lobsters consumed more crabs, mussels, and fish, but fewer echinoderms. as they grew in size and approached maturity. This trend was attributed to differential accessibility of prey. Elner and Campbell (1987) in- dicated that the stronger chelae of larg- er lobsters would enable them to crush prey that are protected by heavy shells, such as gastropods and bivalves, more so than the chelae of smaller lobsters. The natural diet of SRJ lobsters has not been examined to date (Lawton and Lavalli, 1995). e.xcepting rare spec- imens of 12-14 mm CL. The feeding appendages of SRJs are capable of capturing and processing both plank- tonic and benthic organisms ( Lavalli and Factor, 1995). From laboratory ob- servations, several authors have pro- posed that SRJs may live primarily as suspension-feeders, and to a lesser de- gree as browsers, within the shelter or as ambush predators at the shelter's entrance (Barshaw and Brvant-Rich, Sainte Mane and Chabot Natural diet of Homarus americanus off the Magdalen Islands 107 1988; Barshaw, 1989; Lavalli and Barshaw, 1989; Lawton and Lavalli, 1995 ). Wahle ( 1992 ) offered a conceptual mod- el suggesting that lobsters shift from a cryptic to a wide- roaming behavior as predation risk becomes offset by the need for a high-energ\' diet that cannot be satisfied through shelter-restricted feeding. Our study was conducted at the Magdalen Islands, east- ern Canada, to resolve the natural diet of SRJ lobsters and to compare it with that of larger lobsters by using stomach content analysis. We found a gradual ontogenetic shift in lobster diet over the size range of 4 to 112 mm CL. SRJs were carnivorous and probably derived their meals main- ly through predation and scavenging. We also determined the predator-prey size relationship for one of the lobster's preferred and most important prey, i.e. Atlantic rock crab. Cancer irroratus (Reddin, 1973; Evans and Mann, 1977; Carter and Steele, 1982a I. Materials and methods The study site was a narrow 2-km rocky section (47°14.5'N. 6r50..5' to erSl.S'W) of the south shore of Baie de Plai- sance, Magdalen Islands, eastern Canada. This site corre- sponds to the Butte-a-la-Croix location that Hudon (1987) determined to be a settlement ground for lobster Divers collected lobsters by hand or by suction-sampling at depths of 1 to 7 m. Lobsters were processed live usually within minutes and at most two hours after collection. The sex of collected specimens was determined and their CL was measured to the nearest 0.1 mm with a vernier caliper. Lobsters that were not berried and that were judged to be intermolt, based on criteria of shell hardness, coloration, and fouling in Aiken (1980), were dissected to remove the stomach which was preserved in buffered formalin diluted to 4'7(- in seawater Stomachs with calcified gastroliths were subsequently disregarded, thereby effectively eliminating from the present study all premolt lobsters from stage D'-5 (=Dq) on (Aiken, 1980). The resulting sample con- sisted of 471 stomachs from lobsters of 7-112 mm cepha- lothorax length (CL) collected from 24 July to 31 October 1996, and of 35 stomachs from lobsters of 4—12 mm CL col- lected between 4 August and 13 September 1997. The 1997 lobsters were added to improve coverage of stomach con- tents of the early juveniles because very little settlement occurred in 1996 iSainte-Marie et al., 2001). There was no commercial fishery during the sampling periods; therefore items in lobster stomachs were not discards or bait. In the laboratory, stomachs were opened and their con- tent was emptied into dishes for examination under a Wild M8 compound microscope (10-50x). Identity of food items was determined to the lowest taxonomic level possible, based on comparisons with illustrations in literature and samples of benthic and pelagic fauna from our study site. Particular care was taken when examining the stomach contents of lobsters <12 mm CL; for these stomach con- tents we often resorted to higher magnification (>100x) with a Leitz Dialux 20 microscope. The contribution of each food item, exclusive of miner- als and nylon debris, to the volume of stomach contents of each lobster was visually scored from 0 to 10, by 10% increments (0=0% of volume, 1=1-10%, 2=11-20%, etc.). The total for all food items could exceed 10, for example, if more than two minor food items each were scored 1 in addition to one predominant food item that was scored 8. In such cases, the corrected contribution of each food item was obtained by dividing its score by the sum of scores for all organic food items in a given stomach. Corrected volu- metric contribution of each food item was expressed as a proportion of stomach content volume. To obtain information on the size spectrum of rock crab consumed by lobsters, we established predictive (least squares) linear regressions (Sokal and Rohlf 1995) be- tween 30 measurements of distinctive hard body parts and cephalothorax width (CW) of 26 crabs ranging from 7 to 62 mm CW (following the approach in Lovrich and Sainte-Marie, 1997). All the predictive regressions were highly significant (/■-=0.970-0.999. P<0.001). When rock crab remains were encountered in lobster stomachs, dis- tinctive hard body parts were measured with an eyepiece micrometer to estimate crab CW from predictive regres- sions. When more than one body part could be measured, the crab's CW was determined as the mean of the various estimates unless it was obvious that multiple crabs had been ingested. Such was considered to be the case when more than two similar fragments of a paired structure (e.g. eyes or claws) were found in one lobster stomach or when there was considerable divergence among crab CW estimates based on different body parts. The functional re- lationship between the CW of rock crab prey and the CL of lobster predators was established with a model II regi'es- sion (Laws and Archie, 1981; Sokal and Rohlf 1995). The stomach contents, once identified and scored for volume, were transferred separately to preweighed trays, dried to constant mass at 60°C, and weighed to the near- est mg. Dry mass was not obtained for eight stomach con- tents because of manipulation errors. The allometric rela- tionship between the dry mass of stomach contents and lobster CL was established by least squares linear regres- sion, after logarithmic transformation of both variables. Diet was described by occurrence, volumetric contribu- tion, and the specific abundance of food items in the stom- achs of lobsters grouped into 5-mm CL size classes (2.5 to <7.5 mm, 7.5 to <12.5 mm, etc.). Percent occurrence (PO) was the percentage of stomachs in one size class that con- tained a given food item. Volumetric contribution ( VC ) was the average of corrected contributions of each food item to the stomachs of all lobsters in a given size class. Spe- cific abundance (SA) was the average volumetric contribu- tion of a food item determined only for lobsters that had this food item in their stomach. This index is useful for food items with a low average volumetric contribution be- cause it allows the distinction between the case when few animals consume large quantities of a given food item or when many animals consume small quantities of the same food item (Amundsen et al., 1996). The mathematical rela- tionship of the three indices is SA = VC x 100/PO. To assess how the overall diet varied with lobster size, and thus whether or not there were size-related shifts in diet supporting the ontogenetic phases of lobster, we 108 Fishei-y Bulletin 100(1) performed a cluster analysis (Ward's minimum variance method) on the volumetric contribution of food items per lobster size class, after standardization. A sudden increase in the joining distance of the clustering sequence repre- sented by the dendrogram represents a natural cutting point for the determination of meaningful clusters (SAS Institute. 1995). In addition, a factor analysis (VARIMAX rotation of the first three principal components) was per- formed on the correlation matiix of the volumetric contri- bution of food items for each 5-mm-CL size class of lob- sters. Cluster and factor analyses were done with JMP statistical software (SAS Institute, 1995). Relationships between volumetric contribution and lob- ster CL were described by least squares linear regression for bivalves, rock crab, and flesh. Relationships between percent occurrence of bivalves and rock crab were described by locally weighted (lowess) regi"ession with a SOf smootii- ing factor, and by least-squares regression for flesh. Results Sample composition, stomach fullness, and types of food items The 506 lobsters retained for analyses varied in size from 4.3 to 112.4 mm CL (median=35.6 mm CL). Most size classes contained more than 25 lobsters (Table 1). The smallest size class (2.5 to <7.5 mm CL) contained only 16 lobsters with a median of 7.0 mm CL; therefore we refer to this group of lobsters as the 7-mm-CL size class. The 21 lobsters >67.5 mm CL were pooled together into a single size class, which we refer to as the 77-mni-CL size class in reflection of their median CL. Females and males accounted respectively for 43.2'^fi and 44.1''r of all lobsters examined; the remainder were too small to deter- mine sex. Lobsters were pooled for analyses irrespective of sex because Weiss ( 1970) and Ennis ( 1973) concluded that diet was the same for both sexes. Only two lobsters had empty stomachs and they be- longed to the 10-mm size class. With these two empty stomachs excluded, there was a highly significant relation- ship between the dry mass of stomach contents and lob- ster CL (Fig. 1). Identifiable food items included macroal- gae or benthos that were grouped into broad taxonomic or ecological categories (Table 2). No planktonic organisms were identified from the stomachs, even of the smallest lobsters. However, the crustacean meiofauna group includ- ed the remains of very small crustaceans, some like the harpacticoids and ostracods, known to be bottom-dwell- ing, whereas unidentified minute crustacean lemains may have originated from holo- or mero-planktonic forms or from juvenile amphipods, isopods. or carideans. Sand, silt, and infrequently bits of nylon rope were also found in the stomachs. "Flesh" refers to tissue bolus composed of an- imal soft parts that could not be attributed to a taxon, generally because no distinctive part was found in the stomach along with the tissue or less commonly because distinctive parts from several prey types were present in the stomach but none was attached to the tissue. Table 1 Nunibci nf lobster stom achs sample d by classes r f cepha- lothorax length (CL, in iim). Size cl asses represent 5-nim groupin ?s except the smallest i7 mm CL) and lai gest (77 mm CL , which include all 1 obsters <7.5 mm CL and all lobsters >67.5 mm CL, respectively. Numbei of stom achs Cfphalo thorax leii Kth isizc classes) 1996 1997 Total 7 1 15 16 10 17 20 37 1.5 28 28 20 38 38 ■'F, 56 56 30 45 45 35 52 52 40 51 51 45 45 45 50 45 45 55 31 31 60 25 25 65 16 16 77 21 21 Total 471 35 506 Ontogenetic shifts in diet A cluster analysis on the volumetric contribution of food items to lobsters by size class yielded four groups: 7 mm, 10-20 mm, 25-60 mm, and 65-77 mm CL lobsters (Fig. 2). These same groups could be seen on a plot of the fii'st three factors of a factor analysis of the correla- tion matrix of the volumetric contribution of food items (Fig. 3). The three factors explained 68. 87^ of the vari- ance (39.9%, 18.2%, and 10.7% for factors 1, 2, and 3). The first factor had strong loadings for crustacean meiofauna (0.96), foraminiferans (0.96), bivalves (0.84), macroalgae (0.82), amphipods (0.78). and rock crab (-0.71). Because lobsters in the 7-mm-CL size class had little rock crab in their stomachs, but relatively high proportions of the other food items, they stood out with a very large score (3.1) on this factor. The next two size classes, 10 and 15 mm CL, scored 0.8 and 0.6. respectively. All other size classes scored between 0 and -0.6 on the first factor. The second factor had strong loadings for flesh (0.73), lobster (-0.82), and barnacles (-0.73). Lobsters of the two largest size classes (65 and 77 mm CL) had strong negative scores on this factor (-2.5 and -1.5, respectively), whereas lob- sters of the 10-35 mm size classes scored between 0.5 and 1.1. The smallest size class (7 mm CL) and size classes of 40-60 mm CL had scores close to 0. Finally, the third factor had a high loading for carideans (0.74) and some- what smaller loadings for isopods (0.67), coralline algae (-0.57), and pagurids (-0.54). This third factor separated Sainte Mane and Chabot: Natural diet of Homarus amencanus off the Magdalen Islands 109 10' O °°° / O) 10" ° ,rS&?r c ^.^^^W% c tomach co o if) o 10'^ iy^^ "° if) i ' V** °° Q 10' /tf "o o o IC o o 4 10 100 200 Carapace length (mm) Figure 1 Relationship of dry mass of stomacli contents (DMi to cephalothorax length (CL) of lobsters from the Magdalen Islands. Two lobsters of the 10-nim class had an empty stomach and are not shown. Model II regression: DM - 7.9fi7 . 10 «xCZ.-^''-'^ |;-'=0,51.P<0.001|. the 25- and 35-mni-CL size classe,s from the 10-20 mm CL size classes. For each grouping, Figure 4 shows the specific abun- dance of each food item plotted against its percent occur- rence. Bivalves and flesh accounted for a large proportion of stomach contents of the smallest lobsters (7-mm-CL size class) and were found in >751 of stomachs, making them the most important food items for this grouping. Rock crabs, amphipods, and polychaetes contributed 0.2 to 0.4 of stomach volume when they were ingested, but were found in fewer than 30'* of the stomachs. Macroalgae and gastropods, on the other hand, were eaten by >50'; of small lobsters but were ingested in small volumes. All other prey categories contributed little to stomach volume and were found in a small proportion of stomachs. Flesh and bivalves were also the most important food items for the 10-20 mm CL lobster grouping (Fig. 4). They accounted for 0.46 and 0.22 of stomach volume, respec- tively, when they were ingested, and were found in 90' ^ of stomachs. Rock crab was another important prey, with a specific abundance of 0.32 and an occurrence of 41'r. Pagurids, carideans, and echinoderms had high specific abundances but were found in less than 5'^r of stomachs. Gastropods and polychaetes were found in about 40'5c of stomachs, but accounted for a small fraction of stomach volume. All other prey categories constituted a small frac- tion of the volume of very few stomachs. The two main food items of lobsters measuring 25-60 mm CL were rock crab and flesh: specific abundance was high (0.34 and 0.38, respectively) and these food items Table 2 Major categories of liiod ilc nis, divic ed into specific food items when possible, and t heir overall volumetric contribu- tion (total=l ) to stomach coi tents of; ill examined lobsters from Baie de Plaisance, Ma gdalen Islands. Abbreviations | for major categories of food i tems are shown in brackets. Volumetric Categories of food items contribution Formaniferans |For| 0.0031 Macroalgae (Algl 0.0394 Coralline algae iCiirnllinn o fi(in(dis 1 ICorl 0,0178 Hydrozoans |Hyd| 0.0207 Bivalves [Biv| 0.1657 Mytihis ediilis 0,0202 Modiolus modiolus 0.0992 Unidentified Pelecypoda 0.0463 Gastropods |Gasl 0.0.585 Lacuna vincta 0.0028 Unidentified tiastropoda 0.0057 Polychaetes |Pol| 0.0597 Ncreidae 0.0318 Polynoidae 0.0271 Unidentified Polychaeta 0.0008 Barnacles iBalanuN sp.l [Bar] 0.0012 Crustacean meiofauna ICru 0.0053 Harpacticoida 0.0003 Ostracoda 0.0021 Unidentified minute Crus t acea 0.0029 Amphipods lAmpl 0.0054 Coropltium sp. 0.0004 Gamniarus sp. 0.0003 Caprellidea 0.0004 Gammaridae 0.0016 Unidentified amphipods 0.0027 Isopods llsoj 0.0067 Idotea sp. 0.0013 Idoteidae 0.0019 Unidentified valvif'eran isopods 0.0034 Carideans [Carl 0.0024 Crangon septemspinoaa 0.0010 Unidentified carideans 0.0013 Pagurids [Pag] 0.0416 Pagurus acadianus 0.0051 Paguridae 0.0365 Rock crab 'Cancer irroratufi [Cral 0.2637 American lobster iHuniarus anicricanuK) [Lobl 0.0076 Echinoderms [Ech] 0.0222 Strongylocentrotus drueha chiensis 0.0102 Ophiuroidea 0.0012 Unidentified echinoderms 0.0109 Fish [Fisl 0.0066 Flesh [Flel 0.2724 no Fishen/ Bulletin 100(1) Figure 2 Dendrogj-am resulting from a cluster analysis on the mean volumetric contribution of major food categories by size class of lobsters from the Magdalen Islands. The bottom graph shows the joining distance at each step. The vertical dashed line indicates the cut-off value for clusters, selected because of the sudden increase in joining distance. were found in more than TCS of stomachs (Fig. 4). Bi- valves were still found in a large proportion of stomachs {8T7c) but accounted for a low proportion (0.18) of volume. Gastropods, polychaetes, and macroalgae also occurred frequently but accounted for only a small fraction of stom- ach volume. Pagurids and lobsters were found in few stom- achs but contributed >0.'2 of stomach volume. The grouping of the largest lobsters, 65-77 mivi CL, had rock crab as the most important food item (specific abun- dance=0.55; occurrence=86' 7 ). Lobsters, pagurids and fish contributed a large proportion of stomach volume when they were eaten, but these prey were ingested by <20'^( of lobsters. Gastropods, flesh, bivalves, polychaetes, and macroalgae were found in a large proportion of stomachs but occupied a small proportion of the volume of these stomachs. Overall, bivalves, rock crab, and flesh were the only food items that each accounted for >0.1 of stomach volume for the whole sample (Table 2). For these food items, a signifi- cant linear relationship existed between volumetric con- tribution and lobster CL, the latter explaining 68*7? to 929c of the variability in volume (Fig. 5). Regi-ession of volumet- ric contribution on lobster CL produced a negative slope for bivalves and flesh, and a positive slope for rock crab. Similarly, strong linear or nonlinear relationships existed between percent occurrence of these three food items and lobster CL (Fig. 5). Furthermore, large lobsters tended to eat larger rock crabs than small lobsters, as evidenced by the significant positive linear relationship between the CW of rock crabs found in lobster stomachs and lobster CL (Fig. 6). Figure 3 Results of the factor analysis on the correlation matrix ol volumetric contribution of major food categories by size class of lobsters from the Magdalen Islands. See text for factor loadings. Three clusters identified in Figure 2 are shown inside ellipses; the other size classes constitute the fourth cluster. Discussion Data Stomach content analysis is a useful method for the inves- tigation of the natural diet of animals, even though the lack of distinctive hard parts in some prey and differential digestibility of soft and hard body parts limits the spec- trum of food items that can be recognized and can lead to biased perception of the relative importance of the food items. We took care to process lobsters as quickly as pos- sible after collection, thus attenuating the effects of differ- ential digestibility, and we examined only intermolt and nonovigerous lobsters, thus reducing sources of diet vari- ability associated with molt cycle and female reproduc- tive status (e.g. Weiss, 1970: Ennis, 1973). In addition, our study was conducted over a small area where the various lobster size classes were evenly distributed; therefore all lobsters potentially could access the same food. We rec- ognize that our volumetric contribution index underesti- mates the importance of predominant food items, owing to correction for stomachs with multiple food items and total scores >10. However, this was a minor problem because analyses using uncorrected values revealed that the vol- umetric contribution of the three main food items was underestimated by no more than 2-5'^< and that relation- ships to lobster size class were unchanged. Therefore, we are confident that the dietary differences among the lob- ster size classes that we detected are real and that they Sainte Marie and Chabot NatLiial diet of Haniaiui ameiicaniis off tlie Magdalen Islands 111 1 0 A 7 mm 1 0 H Id 20 mm 08 08 ■.,>- 06 08 i» ** / '.-' ..-^- 04 - jC- «> 04 <.-» ^°'~ J^ .-^^ .f- • .o'- <^ 02 ■ f ^<^ 02 ■ .o^S? .s • bundance o o Bal Ech ISO 0 ^ .c^ ,CarF,s LobV O* .'^ ||CorHydPag • 9 . . . . 00 /•' .*" 0 20 40 60 80 100 0 20 40 60 80 100 3Cific a o C 25-(i() mm 1 0 D 65 -77 mm Q. (/2 08 08 J' 06 06 r"" .o"^ 04 04 . [Amp • Bal J Car 02 «^ 0 2 \Cru .<^» .^<^:> .o^^ 00 ^1^ t 1 1 1 00 1 ■t 0 20 40 60 80 100 0 20 40 60 80 100 Occurrence (%) Figure 4 Relationship between specific abundance and percent occurrence for the major food catego- | ries in relation to clusters for size classes ofthe(Al 7 mm.(Bl 10- -20 mm. (C) 25-60 mm, and (D) 65-77 mm CL (see Fig. 2) for lobste ■s from the Magdalen 1 slands. Refer to Table 2 for abbreviations of major food categories. reflect mainly changing lobster preferences and differen- tial accessibility of prey types. Ontogenetic shifts in diet There was clear evidence of a progressive dietary shift with increasing lobster size at our study site. Smaller lobsters relied to a greater extent than larger lobsters on soft or easily acquired food items (flesh, sessile juvenile bivalves, macroalgae, meiobenthic crustaceans, and foraminiferans). Larger lobsters fed on bigger, more mobile and also more nutritious prey, including crustaceans that were protected by heavy shells, and fish. Fishes were probably taken by predation (see Weiss, 1970) because there was no fishing activity at or near our study site that might have provided lobsters with fish bait or discards. The most striking ontogenetic changes in volumetric contribution of prey types occurred for rock crab and bi- valves, the former increasing from 0.07 to 0.53 and the lat- ter decreasing from 0.28 to 0.02 from the smallest to the largest lobster size class, respectively (Fig. 5). Only lim- ited comparisons with other studies are possible, given the differences in methods and in the size range of lobsters examined. However, the observed trends of increasing im- portance of rock crab and of decreasing importance of bi- valves with increasing lobster size were consistent with the analyses of Scarratt (1980) and of Carter and Steele ( 1982b), and they suggest that lobsters are not simply op- portunistic or unspecialized feeders (see Elner and Camp- bell, 1987). Multivariate analysis of lobster diet resulted in size groupings that are quite consistent with Lawton and La- valli's (1995) size classification of the early life-history phases based on a broad set of behavioral and ecological criteria. Major shifts in diet in the present study occurred at about 7.5, 22.5, and 62.5 mm CL (Fig. 2). The two clas- sifications differ in the smaller size for the transition from the first to second group (7.5 mm in our diet-based classifi- cation compared with 14.5 mm CL in Lawton and Lavalli's scheme), but the size for transition from the second to the 112 Fishery Bulletin 100(1) 1 0 A bi\al\Os Q 100 08 80 06 \ ■^ 60 04 • l/C = 0 304-0 004 CL r- = 0 72 40 02 "^^^^^^^*^^^^^_ 20 00 10 *~~ —-* . 0 100 ' B Hcsh □ ^ ^^ ^ ^ PO = 90 506 - 0 467CL c 08 ~"~--_n r^ = 0 59 - 80 o ~~" -— S" ~ 06 O^ ^ - ^ ^ ^ n CD 60 3 o o o o o • • o c S 04 — ^_____^ 40 g E " ~~— i^ • 13 13 O o • *^-— -__• (D > 02 • """^ — •- — •-_ \/C = 0 441 -0 004CL ^"— » ^ r- = 0 68 20 00 1,0 1 .... 1 ..,, 1 .... 1 .... 1 .... 1 .... ~ 0 100 C iiick crab ^ ^ n___,_n-- ,, 08 80 0,6 60 04 40 0.2 " / _»,-— r"'*'^^ ;/c = 0 02i +0 007C/. .X-"*"^ /-' = 0 92 20 00 • 0 c 10 20 30 40 50 60 70 8 0 Cephalothorax length (mm) Figure 5 Relation between percent occurrence IPO, Di or volumetric contribution (VC.») and lobster cephalothorax length for the three main food items of lobsters from Magdalen Islands: (Al bivalves, (Bi flesh bolus, and (C) rock cral). All linear regi-essions are highly significant (P<0.001 1. third group is the same in both studies (22.5 and -25.0 mm CD. Comparison of the size threshold for transition from the third to the fourth group is less appropriate be- cause Lawton and LavaUi (1995) considered this thresh- old to be determined by physiological maturity, which is a temperature-dependent trait that varies among regions. Natural diet of shelter- restricted juveniles This first investigation of the diet of SRJ lobsters does not support the view that these juveniles derive a substan- tial portion of their diet by suspension feeding and brows- ing in their shelters, at least at our study site and during the two years we sampled. With respect to suspension feeding, there was no evidence of planktonic organisms in stomachs, although some of the unidentified prey of the crustacean meiofauna category may have been planktonic. Foraminiferans, harpacticoids, ostracods, and macroalgal debris represented food items that potentially could be browsed within shelters. However, these taxa together contributed relatively little to stomach volume of lobsters in the 7-mm size class (0.14 for the combined categories. Sainte Mane and Chabot Natural diet of Homanis amencamis o\\ the Magdalen Islands 113 60 r / 50 / D n / D / E E a / — 40 / ^ S On/ ° S X 2 30 □ /o° ° o o o. 20 0) o °d/ d n " 10 a /na4D ct, cP 7?^ D □/ Oan o d/ Q 0 20 40 60 80 100 120 Lobster cephalothorax length (mm) Figure 6 Relationship of predator (lobster) size to prey (rock crab) size, based on rock crab cephalothorax width (CWl esti- mated from measurements of indicator fragments and lob- ster cephalothorax length (CL), for the Magdalen Islands. Model II regression: CW = -12.341 -i- 0.677CL |/-=0.34, /'<(). 001 1. in spite of the fact that one or the other category occurred in S8'/( of stomachs) and even less to stomach volume of lobsters in the 10-mm size class (0.10, 86'^). During our study, lobsters settled in August at sizes of 4.3-5.2 mm CL and grew to 12-14.5 mm CL by October (Sainte-Marie et al., 2001). Thus, we sampled the lobster population during the only period of time when SRJs were present and sea- sonal sampling bias cannot be invoked to explain the lack of plankton in their diet. The other food items in the stomachs of SRJs, and es- pecially the predominant bivalves and flesh (Figs. 4 and 5 ), probably were derived by predation and scavenging. Bi- valves in the stomachs of SR.J lobsters were represented by recently settled Modiolus and Myfilus. Mussel spat may settle aggregatively and quite synchronously, forming dense patches that can provide a short-term prey pool requiring little or no search time (e.g. Auster, 1988). Furthermore, be- cause mussel spat often settle in crevices or under rocks (e.g. Nair et al., 1975), SRJs could access them with little or no risk of exposure to predators. Lawton ( 1987 ) argued that dominance and territoriality were likely to exist early in the ontogeny of lobsters, as demonstrated subsequently ( James- Pirri and Cobb. 1999; Paille and Sainte-Marie, 2001). and that prolonged occupation and defense of shelters located close to a food patch would be advantageous for juveniles. Exploitation of mussel patches, inferred from the present study, is consistent with that hypothesis. Flesh (tissue boluses) that could not be attributed to a particular animal for lack of indicator fragments was a very important food item in the diet of SRIs, both in terms of percent occurrence and of volumetric contribution (Figs. 4 and 5). Elner and Campbell ( 1987) also found that uniden- tified animal tissue was one of the most frequent and most volumetrically important foods in the stomachs, however, of larger lobsters. Weiss (1970) observed that adolescent and adult lobsters often captured crabs or other shelled prey, cracked them open, and then selectively ingested only soft tissue. Interestingly, the percent occurrence and volu- metric contribution of flesh to diet was greater in lobsters of size classes <30 mm CL (i.e. SRJs and emergent juve- niles) than in larger lobsters (Fig. 2). It is unlikely that the smallest of lobsters could find (within the confines of their shelter) and subdue prey sufficiently large to provide tis- sue boluses devoid of hard parts. Furthermore, claws are not differentiated into cutter and crusher forms in SRJs (Govind and Lang, 1978; Costello and Govind, 1984) and early juveniles may be incapable of breaking open shelled prey (Costello and Lang, 1979; Lawton and Lavalli, 1995). Therefore, flesh ingested by SRJs and emergent juveniles probably was obtained by scavenging animal remains. Con- sidering that larger lobsters may hoard and bury food in or nearby their dens (Herrick, 1895; Smith, 1976; Lawton, 1987; Wickins et al., 1996), we propose that early juveniles exploit the meal scraps or food resei-ves of larger lobsters. Indeed, we obsei-ved that small lobsters often occupied gal- eries beneath, or in rock pilings nearby, the dens of larger lobsters. This is consistent with reports that odor from con- specific adults is a proximate cue for lobster settlement (Boudreau et al.. 1993). Cohabitation of small lobsters with large lobsters would offer the former protection from pred- ators and a potentially abundant, high-quality, sheltered food source, and would therefore represent a form of com- mensalism. The risk of cannibalism for small lobsters liv- ing in the vicinity of larger lobsters probably does not off- set the benefits. Few lobster remains were found in lobster stomachs in this (Fig. 4) as in other studies (Weiss, 1970; Carter and Steele, 1982b; Elner and Campbell, 1987), and an unknown proportion of those remains may have been exuviae. Some other rarer food items found in the stomachs of SRJ lobsters were probably taken by predation, possibly within, but more likely in the neighborhood of, the lob- sters" shelters. The most important of these prey by volu- metric contribution were polychaetes, comprising juvenile nereids and polynoids that are frequently found in soft sediment or on the underside of rocks, and recently settled rock crab. Similarly, amphipods and gastropods found in the stomachs of SRJs were juveniles or small species that may abound in crevices and in spaces beneath rocks. A carnivorous, high-energy diet such as the one demon- strated for SRJs in our study would promote growth from settlement time. By contrast, Lavalli ( 1991 ) demonstrated that a diet of only diatomous algae was insufficient for ex- tended growth and sui-vival of early juvenile lobster A diet of mesozooplankton sustained growth of juvenile lobsters, at least for some time after settlement (e.g. Daniel et al., 1985; Barshaw, 1989; Lavalli, 1991). However, Lawton and Lavalli ( 1995) pointed out that intermolt periods tended to be longer and molt increments smaller in laboratory-held. 114 Fishery Bulletin 100(1) juvenile lobsters reared on mesozooplankton than in wild lobsters, suggesting that the latter incorporated more nu- tritious foods into their diet. The finding that early juvenile lobsters are primarily predators or scavengers, if confirmed by studies at other sites, has implications for the development and implemen- tation of artificial reefs. Such structures are increasingly being considered as a means to enhance lobster produc- tivity on traditional grounds or to expand lobster habitat onto less hospitable grounds (e.g. Gendron, 1998). The car- nivorous benthic feeding mode of SRJs and of emergent ju- veniles at our site implies that successful reefs will have to be designed, localized, and weathered so that they are ini- tially well colonized and subsequently regularly colonized by benthic prey that are easily accessible and of high nu- tritional value to juvenile lobsters. Additionally if SRJs and emergent juveniles derive some protective and nutri- tional benefits from the presence of larger conspecifics, reefs designed to offer shelter to a full suite of lobster sizes may prove to be more productive in the long term than reefs offering shelter only to small lobsters. Importance of rock crab to lobster Several previous studies have noted the importance of rock crab in the diet of lobster (Reddin, 1973; Evans and Mann, 1977; Carter and Steele, 1982al. Boghen et al. (1982) found that juvenile lobsters survived and grew better on a diet containing crab protein alone than on a diet of live brine shrimp iAiienua salina) or of protein extracts from urchin iStrongylocentrotus droebachiensi.';). mussel (Mytilus ediilis). or shrimp iPenaeus sp.). Gendron et al. (2001) found that condition, somatic gi-owth, and gonadal development of lobster increased with increasing amount of rock crab in diet. In nature, even SRJs may ben- efit from a diet including large amounts of rock crab pro- tein because they preyed directly on very small rock crabs (Figs. 4 and 5), and the tissue boluses they contained may have been that of rock crab (see above). We were able to establish a positive size relationship for lobster preying on rock crab (Fig. 6). The smallest rock crab prey were 2-6 mm CW and belonged to the first ben- thic instars of this species. In our study, apparently no rock crabs larger than 50 mm CW were consumed by lob- sters, and the maximum ratio of crab CW over lobster CL was 0.90, even though rock crabs up to 120 mm CW were seen (own personal diving obsei-vations). In the labo- ratory Weiss (1970) observed that lobsters of 60-80 mm CL attacked crabs offered in the size range of 62-78 mm CW. Lawton and Lavalli (1995) reported that juvenile lob- sters can subdue juvenile intermolt rock crabs up to ap- proximately 0.40 times their own body size. Their obser- vation was based on the comparison of predator and prey wet masses; when expressed in terms of crab CW over lob- ster CL, the maximum ratio was equivalent to about 1.27.' This ratio of prey CW to predator CL is somewhat larger than that derived from our stomach analvses. Because lob- Lawton, P. 2000. Personal comniun. Fisheries and Oceans Canada, St. Andrews, New Brunswick, Canada. sters probably ingest only soft tissue when the prey-pred- ator size ratio is sufficiently high (Weiss, 1970; and see above ), our analysis of rock crab prey-size frequencies may correctly estimate the minimum prey size but underesti- mate the maximum prey size and the volumetric contribu- tion and occurrence of rock crab in the diet of any given lobster size class. Nevertheless, the present study clearly shows that all lobster size classes rely on rock crab as food and that the size spectrum of rock crab that is used by lob- sters is broad and includes even those at the settlement stage. Given the much greater economic value of lobster in relation to rock crab, and the trophic dependency of the former on the latter, caution should be exercised in devel- oping rock crab fisheries (Gendron and Fradette, 1995). Acknowledgments We thank our diving partners F. Hazel. J.-G. Rondeau, J. A. Gagne, K. Gravel, R. Larocque, J.-F. Lussier, N. Faille, and A. Rondeau. We are particularly gi-ateful to J. Hudon for her major contribution to the identification of stomach contents and to three anonymous reviewers for construc- tive comments. This is a contribution to the Canadian Atlantic-Wide Lobster Studies (CLAWS) research initia- tive of Fisheries and Oceans Canada. Literature cited Aiken, D. E. 1980. Molting and gi-owth. In The biology and manage- ment of lobsters iJ. S. Cobb, and B. F. Phillips, eds. ), p. 91-163. Academic Press, New York. NY. Amundsen, P. -A., H.-M. Gabler, and F. J. Staldvik. 1996. A new approach to graphical analysis of feeding strategy from stomach contents data — modification of the Costello (19901 method. J. Fish Biol. 48:607-614. Auster, P. J. 1988. Response of megafaunal predators to synchronous settlement of sessile prey. J. Shellfish Res. 7:108-109. Barshaw, D. E. 1989. Growth and survival of early juvenile American lob- sters. Homarua ODiencanuK. on a diet of plankton. Fish. Bull. 87:366-370. Barshaw, D. E., and D. R. Bryant-Rich. 1988. A long-term study on the behavior and survival of early juvenile American lobster, Homarus americantis, in three naturalistic substrates: eelgrass, mud, and rock. Fish. Bull. 86:789-796. Boghen, A. D., J. D. Castell, and D. E. Conklin. 1982. In search of a reference protein to replace "vitamin- free casein" in lobster nutrition studies. Can, J. Zool. 2:2033-2038. Boudreau, B., E. Bourget, and Y. Simard. 1993. Behavioural responses of competent lobster postlar- vae to odor plumes. Mar. Biol. 117:63-69. Carter, J. A., and D, H. Steele. 1982a. Attraction to and selection of prey by immature lob- sters iHomai'us omencaniis). Can. J. Zool. 60:326-336. 1982b. Stomach contents of immature lobsters (Homarus americanus'i from Placentia Bay, Newfoundland. Can. J. Zool. 60:337-347. Sainte Marie and Chabot Natural diet of Homaiw, ameiicanus off the Magdalen Islands 115 C.hh. .1 s 1971. 'Plu' slu'ltiT-ii'lati'tl bchavHir (il'thc l(il)sl('i', Hnniinus anicnviiniis. F>ology 52:108-115. Cnl)l).-I. S..and K. A. Wahliv 1994. Karly life history and rccniilnirnt |>roc(.'.ssi.'s olclauud lobsters. Crustaceana 67:1-25. Cooper, R. A., and J. R. Uzmann. 1977. Ecolofiy of'juvenile and adult clawed lobsters, Homa- rus aniericaniis. Hiiniarus ganimaruti. and ^Vcp/i/op.s nor- wgiciis. Div. Fish. OceanogT. Circ. ( Aust. C.S.I.R.O. I 7: 187- 208. Costello, W. J., and t'. K. (lovind. 1984. Contractile proteins of fast and slow fibers during dif- ferentiation of lobster claw muscle. Dev. Biol. 104:434- 440. Costello, W. J., and F. Lang. 1979. Development of the dimorphic claw closer muscles of the lobster Hoinarus americanus. IV. Changes in functional morphology during growth. Biol. Bull. iWoods Hole, MAi 156:179-195. Daniel, P. C. R. C. Bayer, and S. Chapman. 1985. Barnacle larvae (Batanus spp. ) as a potential diet for juvenile lobsters {Homarus americantis). Aquaculture 46:67-70. Elner. R. \V., and A. Campbell. 1987. Natural diets of lobster Homarus americamia from barren gi-ound and macroalgal habitats off southwestern Nova Scotia, Canada. Mar Ecol. Prog. Ser 37:131-140. 1991. Spatial and temporal patterns in recruitment for American lobster, Homarus americanus, in the northwest- ern Atlantic. Mem. Queensl. Mus. 31:349-363. Ennis, G. P 1973. Food, feeding and condition of lobsters, Homarus amer- icanus. throughout the seasonal cycle in Bonavista Bay, Newfoundland. J. Fish. Res. Board Can. 30:190.5-1909. Evans, P D., and K. H. Mann. 1977. Selection of prey by American lobsters [Homarus amencanus) when offered a choice between sea urchins and crabs. J. Fish. Res. Board Can. 34:2203-2207. Freire, J., M. Paz Sampedro, and E. Gonzalez-Gurriaran. 1996. Influence of morphometry and biomechanics on diet selection in three portunid crabs. Mar Ecol. Prog. Ser 137:111-121. Gendron, L. 1998. Proceedings of a workshop on lobster stock enhance- ment held in the Magdalen Islands (Quebec) from October 29 to 31, 1997. Can. Ind. Rep. Fish. Aquat. Sci. 244. Gendron, L., and P. Fradette. 199.5. Revue des interactions entre lecrabecommun [Cancer irroratus) et le homard americain (Homarus americanus), dans le contexte du developpenient d'une peche au crabe commun au Quebec. Can. MS Rep. Fish. Aquat. Sci. 2306. Gendron, L., Fradette, P., and G. Godbout. 2001. The importance of rock crab iCancer irroratus) for growth, ovary development and condition of adult Amer- ican lobster (Homarus americanus). J. Exp. Mar. Biol. Ecol. 262:221-241. (iovind, C. K., and F. Lang. 1978. Development of the dimorphic claw closer muscles of the lobster Homarus americanus. III. Transformation to dimorphic mu.scles in juveniles. Biol. Bull. (Woods Hole, MAl 1.54:5,5-67. Herrick, F. H. 1895. The American lobster: a study of its habits and devel- opment. Bull. U.S. Fish. Comm. 15:1-252. lliidon, C. 1987. Ecology and growth of postlarval and juvenile lobster, Homarus americanus, off lies de la Madeleine (Quebec). Can. J. Fish. Aquat. Sci. 44:185.5-1869. Hudon, C, and G. Lamarche. 1987. Niche segi-egation between American lobster Homa- rus americanus and rock crab Cancer irroratus. Mar. Ecol. Prog. Ser 52:155-168. James-Pirri, M. J., and S. J. Cobb. 1999. Behavioral interactions of postlarval and fifth instar lobsters {Homarus americanus) in a simulated cobble envi- ronment. Mar Freshwater Behav. Physiol. 32: 207-222. LavalH, K. L. 1991. Sui-vival and growth of early-juvenile American lob- sters Homarus americanus through their first season while fed diets of mesoplankt.on, microplankton, and frozen brine shrimp. Fish. Bull. 89: 61-68. Lavalli, K. L., and D. E. Barshaw. 1989. Post-larval American lobsters (Homarus americanus) living in burrows may be suspension feeding. Mar. Behav. Physiol. 15:255-264. Lavalli, K. L., and J. R. Factor. 1995. The feeding appendages. In Biology of the lobster Homarus americanus (J. R. Factor, ed.), p. 349-393. Aca- demic Press, San Diego, CA. Laws, E. A., and .J. W. Archie. 1981. Appropriate use of regression analysis in marine biol- ogy. Mar. Biol. 65:13-16. Lawton, P. 1987. Diel activity and foraging behavior of juvenile Ameri- can lobsters. Homarus americanus. Can. J. Fish. Acjuat. Sci. 44:119.5-1205. Lawton, P.. and K. L. Lavalli. 1995. Postlai-val, juvenile, adolescent, and adult ecology. In Biology of the lobster Homarus americanus (J. R. Factor, ed. ), p. 47-88. Acad. Press, San Diego. CA. Lee, S. Y., and R. Seed. 1992. Ecological implications of cheliped size in crabs: some data from Carcinus maenas and Liocarcinus holsatus. Mar. Ecol. Prog. Ser. 84:151-160. Lovrich, G. A., and B. Sainte-Marie. 1997. Cannibalism in the snow crab, Chionoecetes opilio (Brachyura: Majidac), and its potential importance to recruitment. J. Exp. Mar Biol. Ecol. 211:225-245. Nair, R. V., K. N. Nayar, and S. Mahadevan. 1975. On the large-scale colonisation of the spat mussel, Mv^ ilus viridis, in Cochin region. Indian J. Fish. 22:236-242. Ojeda, F. P.. and J. H. Dearborn. 1991. Feeding ecology of benthic mobile predators: experi- mental analyses of their influence in rocky subtidal com- munities of the Gulf of Maine. J. Exp. Mar. Biol. Ecol. 149:1.3-44. Paille, N., and B. Sainte-Marie. 2001. Effects of crowding and shelter limitation on the behaviour and survival of the first benthic stage of lobster. In Proceedings of the Canadian lob.ster Atlantic wide stud- ies (CLAWS) symposium, Moncton, March 2000 (M. J. Tremblay and B. Sainte-Marie, eds. ), p. 30-33. Can. Tech. Rep. Fish. Aquat. Sci. 2328. Reddin, D. 1973. The ecology and feeding habits of the American lobster (Homarus americanus) (Milne-Edwards, 1837) in Newfoundland. M.Sc. thesis. Memorial LIniv., St. John's, Newfoundland, 101 p. Sainte-Marie, B., D. Chabot, F. Hazel, and L. Gendron. 2001. Preliminary analysis of settlement intensity and 116 Fishery Bulletin 100(1) growth oljuvenile lobster in the shallows of Baie de Plai- sance, Maplalon Islands. In Proceedings of the Canadian lobster Atlantic wide studies i CLAWS) symposium, Monc- ton, March 2000 (M.J. Tremblay and B. Sainte-Marie, eds. I. p. 27-29. Can. Tech. Rep. Fish. Aquat. Sci. 2328. SAS Institute. 1995. JMP statistics and graphics guide, version 3. SAS Institute Inc., Cary NC. 593 p. Scarratt,D. J. 1980. The food of the lobster. //( Proceedings of the work- shop on the relationship between sea urchin gi-azing and commercial plant/animal hai-vesting (J. D. Pringle, G. J. Sharp, and J. F. Caddy eds.), p. 66-91. Can. Tech. Rep. Fish. Aquat. Sci. 954. Smith, E. M. 1976. Food burial behavior of the American lobster, Hn/ua- rus americanus. M.Sc. thesis, Univ. Connecticut, Storrs, CT, 52 p. SokalR. R.and F J. Rohlf 1995. Biometry; the principles and practice of statistics in biolog- ical research, 3"' ed. W.H. Freeman, New York, NY, 887 p. Squires, H .J. 1970. Lobster iHomarus cinwricaniis) fishery and ecology in Port au Port Bay Newfoundland, 1960-65. Proc. Natl. Shellfish Assoc. 60:22-39. Wahle, R.A. 1992. Body-size dependent anti-predator mechanisms of the American lobster. Oikos 65:52-60. Weiss, H. M. 1970. The diet and feeding behavior of the lobster, Honiarus americanus, in Long Island Sound. Ph.D. diss., Univ. Con- necticut, Storrs, CT, 80 p. Wickins, J. F, J. C. Roberts, and M. S. Heasman. 1996. Within-burrow behaviour of juvenile european lob- sters Homariis gammaruft (L.). Mar Freshwater Behav. Phvsiol. 28:229-253. 117 Abstract— Skeletochronological data on fjniu ih changes in humerus diam- eter were used to estimate the age nf Hawaiian gi'een seaturtlos ranging from 28.7 to 96.0 cm straight carapace length. Two age estimation methods, correction factor and spUne integration, were compared, giving age estimates ranging from 4.1 to 34.6 and from 3.3 to 49.4 yr, respectively, for the sample data. Mean growth rates of Hawaiian green seaturtles are 4-5 cm/yi- in early juveniles, decline to a relatively con- stant rate of about 2 cm/yr by age 10 yr. then decline again to less than 1 cm/yr as turtles near age 30 yr. On average, age estimates from the two techniques differed by just a few years for juvenile turtles, but by wider mar- gins for mature turtles. The spline-inte- gration method models the curvilinear relationship between humerus diame- ter and the width of periosteal gi'owth increments within the humerus, and offers several advantages over the cor- rection-factor approach. Age and growth of Hawaiian green seaturtles (Chelonia mydas): an analysis based on skeletochronology George R. Zug Division of Amphibians and Reptiles Department of Systematic Biology National Museum of Natural History Washington, DC. 20560-0162 E-mail address: zuggeorgeiSnmnhsiedu George H. Balazs Jerry A. Wetherall Honolulu Laboratory Southwest Fisheries Science Center National Manne Fishenes Service, NOAA 2570 Dole St, Honolulu Hawaii 96822-2396 DenJse M. Parker Shawn K. K. Murakawa Joint Institute for Manne and Atmosphenc Research 2570 Dole St Honolulu, Hawaii 96822-2396 Manuscript accepted 20 August 2001. Fish. Bull, 100:117-127 (20021. The Hawaiian population of the green seaturtle {Chclonia mydas) provided some of the first published gi-owth data (Balazs, 1979, 1980. 1982) for this spe- cies. These early data showed how slowly seaturtles grow and how long a female must survive simply to lay her first clutch of eggs. Twenty or more years to reach sexual maturity seemed biologi- cally unrealistic, yet slow growth and late maturity has been repeatedly con- firmed for some seaturtle species, e.g. Bahamian C. mydas (Bolten et al,, 19921, West Atlantic Caretta caretta (Parham and Zug. 1998). Slow growth and the resulting delayed maturity greatly affect the demography of a population (Grouse et al., 1987; Chaloupka and Musick, 19961. An understanding of growth and gi-owth-pattern variation in seaturtles is an important prerequisite to the devel- opment of population models that are required to guide seaturtle population recovery and conservation. Green seaturtles within the Hawai- ian Islands contribute to the larger In- do-Pacific C. mydas gene pool, yet the nesting females of the Hawaiian popu- lation comprise a distinct genetic unit and contain a unique intDNA haplotype (Bowen et al., 1992), Except during their posthatching pelagic phase, the gi'eat majority of Hawaiian green seaturtles reside in coastal waters, primarily around Hawaii. Kauai, Maui, Molokai, Oahu, and other islands in the south- eastern part of the Hawaiian chain. Most reproduction takes place at French Frigate Shoals in the Northwestern Ha- waiian Islands (Balazs, 1980; Wether- all et al,. 1999), The population of Ha- waiian Chelonia mydas has benefited from over two decades of intense con- servation management (Balazs, 1998). Despite important early research on the Hawaiian population of C. mydas (Balazs, 1982), the patterns of growth within and among the geographic habi- tat components of this population have remained incompletely documented ( Balazs et al„ 1994, 2000 ), The chief rea- son for this has been a lack of methods to age sea turtles. This deficiency has been overcome recently by the devel- opment of skeletochronological tech- niques that estimate age from the num- ber of growth increments formed on the humerus (Parham and Zug, 1998), 118 Fisher/ Bulletin 100(1) Our goal in this study was twofold. First, we used humerus growth-increment data to estimate ages of a sample of Hawaiian green seaturtles from various locations in the ar- chipelago and developed a growth model for the general Hawaiian population; geograph- ic variation in growth will be addressed in a subsequent paper. Second, we compared two different methods of deriving the age esti- mates, the so-called "correction-factor" meth- od described by Parham and Zug ( 1998) and a newer approach, the "spline-integi'ation" method, introduced in the present study. Materials and methods Our sample consisted of 104 individuals of C. mydas, collected from the islands of Hawaii, Kauai, Lanai, Maui, Oahu, and the Northwestern Hawaiian Islands; the Oahu sample predominated with 64 individuals. All individuals were measured to the near- est 0.1 cm straight carapace length (SCL). The smallest individual was a 5.3-cm-SCL hatchling. The smallest posthatchling was a pelagic juvenile (of assumed Hawaiian origin) recovered from the former squid driftnet fishery north of the island chain. All other posthatchling turtles were from coastal Hawaiian waters, found stranded dead and re- trieved by the National Marine Fisheries Service's Hawai- ian Islands Seaturtle Stranding Network. The salvaged turtles ranged from 28.7 to 96.0 cm SCL. The sample was divided into eight 10-cm size classes; representation was roughly equivalent for the middle six classes (Fig. 1). The 30-39 cm sample contained only turtles in the upper quar- tile of this size class. Each turtle was necropsied and its right humerus removed for skeletochronological examina- tion. The necropsy data included a complete set of carapace measurements, organ condition evaluations, and informa- tion on fibropapilloma tumor occurrence and severity; see Work and Balazs (1999a, 1999b) for details on the entire data set. In addition to the skeletochronological data, we used only the SCL measurements and tumor-evaluation observations in the present study. Where possible, we selected tumor-free individuals for the present analysis because our goal was to examine the overall growth pattern for normal Hawaiian Chelonia my- das. Because the prevalence of fibropapillomatosis is high in the wild Hawaiian population (Murakawa et al., 2000), we included in our sample individuals with fibropapillo- mas. but otherwise appearing normal, in order to ensure adequate representation in the larger size classes. Healthy animals were those showing no evidence of weight loss or other indicators of illness and no evidence of disruption or retardation of normal growth. Individuals with tumors represented 27'7r of the 50-59 cm, 507^ of 60-69 cm, 76% of 70-79 cm, 73% of 80-89 cm, and 17% of 90-99 cm SCL size classes (Fig. 1). 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99 Straight carapace lengtti (cm) Figure 1 Size (straight carapace length) distribution of the Hawaiian Chelonia mydas skeletochronological sample. The members of each class are segre- gated into individuals without (shaded bar) and with (black bar) fibropapil- loma tumors. Tumors in our sample are present only in larger turtles. Our skeletochronological data derived from cross-sec- tions (0.6-0.8 mm thick) from the middle of the humeral shaft just distal to the deltopectoral crest and at the nar- rowest diameter of the diaphysis (Zug et al., 1986). On each specimen, we counted the number of visible growth layers and measured the widths (long-axis diameters) of the humerus at each successive growth cycle and the width of the resorption core. Bone sections were taken from mid-shaft, the narrowest location of the bone, be- cause the humerus retains the gi'eatest number of perios- teal growth layers there, and hence this location permits the most accurate estimation of the number of growth cy- cles (periosteal layers) and the relative rates of growth (=successive humerus diameters). We used two procedures for estimating the total num- ber of growth layers, and hence age, of each turtle. In the correction-factor (CF) method, as described in Parham and Zug ( 1998), the turtle's age is estimated as the number of growth layers observed in the outer region of the humerus section plus the predicted number of resorbed growth lay- ers represented in the remodeled core of the humerus. The latter, unobservable component is estimated as C (i? - R[^), where R is the radius of the absorption core, Rf^ is the ra- dius of a hatchling's humerus (before the beginning of in- crement formation), and C is the so-called correction fac- tor The correction factor is a constant "aging rate" (yr/mm) assumed to apply to the resorption core, and calculated as the reciprocal of the mean growth layer width in small turtles. The mean growth layer width was estimated from 129 periosteal growth layer widths observed in 34 turtles Zug et al Age and growth of Hawaiian Chelonia 119 witli S(^Ls <60 cm and resorption core diameters <19.0 mm. Selecting only small turtles with minimum core di- ameters reduces the frequency of the narrower periosteal layers found in the outer margin of the humerus in larger turtles; hence, it reduces the possibility of overestimating the number of layers in the resorption core. The resulting correction factor was used to estimate the number of re- sorbed periosteal layers. A second method, spline integration (SI), is introduced here. The SI method uses a scatterplot smoothing spline (Hardle, 1990; Hastie and Tibshirani, 1990) to model the relationship between the aging rate and humerus diam- eter. Once the aging function is estimated, a turtle's age is estimated by integrating the spline over the total diam- eter of the turtle's humerus section. The method of model- ing increment width patterns in hard parts and of estimat- ing age by the integration of the resulting aging function was first formalized by Ralston and Miyamoto ( 1983) for a Hawaiian snapper and first applied to seaturtles by Zug et al. ( 1995). In those applications, the aging rate was a para- metric function of size. In our analysis, we modeled the ag- ing rate nonparametrically by fitting a smoothing spline to pairs of obsei-\-ations of growth-layer width and humer- us diameter The SI approach uses the same source of data as the CF method but without selection. Lines of arrested growth (LAG) delimit each observable growth layer Incre- ment width is measured as the difference between the hu- merus diameters at the outer LAG and the inner LAG. As- suming an increment represents one year of growth, each increment width measurement provides a measure of the humerus growth rate (nini/yr) and its reciprocal, a mea- sure of the aging rate (yr/mm) at the obsein-'ed humerus di- ameter (the mean diameter of the pair of LAGs). The skel- etochronological sample yielded 269 such observations of aging rate and humerus diameter. The aging rates were grouped in 1-mm intei-vals of humerus diameter and aver- aged. A cubic smoothing spline was fitted to the mean ag- ing rates by usnig S-PLUS (MathSoft, Inc., 1999). The age (yr) of each turtle was estimated by integrating the aging spline from its origin to the observed outside diameter of the humerus section. To assess the effect of estimation method on age esti- mates, the data were divided into 10-cm SCL groups. With- in each gi'oup a Student's t statistic was used to test the hypothesis that the two methods give equal age estimates. Nonparametric growth models were estimated based on the CF- and Sl-derived age estimates and associated carapace lengths, by using the same S-PLUS procedure employed for the Sl-method aging spline. The validity of the growth models was judged qualitatively by comparing growth predicted by the models with obsei"ved giowth in a sample of 171 Hawaiian gi-een turtles tagged and recap- tured in waters around Molokai (Balazs et al, 1999). To assess uncertainty in the Sl-based growth curve, the 269 pairs of aging rate and mean humerus diameter data were resampled 100 times, and the SI procedure applied to each bootstrap replicate data set. The 100 aging curves derived in this manner generated a bootstrap distribution of estimated age for each turtle. Nonparametric growth curves were then fitted to each derived data set, produc- ing bootstrap distributions of predicted mean length at age. Empirical confidence intervals for the predicted mean length at age were approximated by using percentiles of the latter bootstrap distributions. A linear regression of S('L on outside humerus diam- eter was estimated for the 104 sample turtles. The slope of the linear predictor was applied to the 269 humeral in- crements to estimate a corresponding set of carapace in- crements, presumed to represent annual growth. These growth rate estimates were summarized in box plots over 10-cm intervals of SCL. Mean growth rate as a function of estimated age was also estimated by computing finite dif- ferences of the Sl-based growth model. Results Patterns in humerus growth and aging Carapace length has a strong linear association with humerus diameter (7=0.643 + 2.326A: (where y=SCL cm; A''=humerus diameter mm), r- =0.98, P<0. 001, « = 104 includ- ing the hatchlingl. Humerus growth-increment width, on the other hand, is nonlinearly associated with humerus diameter at the point of growth (Fig. 2 ). Specifically, growth increments tend to be larger when the turtles are smaller (i.e. at smaller humerus diameters) and decline as the tur- tles grow. Variation in humerus increment width (growth rate) shows the same pattern. The estimated aging rate, as the reciprocal of growth rate, increases as the turtles grow. The aging rate does not increase uniformly (Fig. 3). Rather, it increases gradually in small turtles, plateaus over a broad range of length for mid-size turtles, increases abruptly as turtles approach maturity, and maintains an increased rate as the mature turtles grow. Age and growth-rate estimates In the CF-method analysis, the correction factor, C, was estimated as 1.14 yr/mm. The resulting age estimates range from 4.1 to 34.6 yr (/;=70; excluduig the hatchling with age zero). The smallest turtle in the sample had the lowest age estimate and the two largest turtles, the high- est estimates. Only 68% could be aged by the CF method. Skeletochronology requires a pattern of distinct layering within the bony element examined. Such patterns are most evident in the smaller, presumably younger, individ- uals, and the frequency of individuals with distinct perios- teal layers decreases as body size increases. In selecting specimens for the CF analysis, growth layers were suffi- ciently distinct to estimate the number of resorbed layers, and hence the age, in decreasingly fewer turtles: 899c of turtles in the 30-69 cm SCL group, 729;^ in the 70-79 cm group, 18% in the 80-89 cm group, and 29% in the >89 cm gi'oup were used in the CF analysis. Of the individ- uals for which we were unable to obtain an estimate of resorbed layers, a nearly equal number (48%) had fibro- papillomas. The prevalence of tumors for the CF-aged sub- sample (31%) was somewhat less than in the total sample (37%). Importantly, the tumor prevalence in the subsam- 120 Fishei-y Bulletin 100(1) 5-| ■ 4- ■ ■ width (mm) CJ 1 ■ Increment 1 1 ■ ■ ■ ■ ■■ ■ m ■■■§■ ■■ ^m m m ^ 0 10 20 30 40 50 Humerus diameter (mm) Figure 2 Humerus growth increment width (the increase in humerus diameter) in relation to humerus diameter (mean of the inner and outer diameters i. Mean diameter better reflects the size of the humerus during the entire growth inten'al than does outer diameter e-. 5- ■ / f '- ■ / ■>^^ / aging rate CO 1 / ■D CD E 7 ■ ■ . ■ / ■ 111 >^-^*^,«>^ 1 - ^r^^^^ " ^■^''^mm n ^^-^ ^ ! - 1 1 ' 1 1 0 10 20 30 40 50 Humerus diameter (mm) Figure 3 Estimated "aging rate" (average of the reciprocals of the humerus growth- increment widths) in relation to humerus diameter and a fitted smoothing spline. The expected age at a given humerus diameter is obtained by inte- grating the spline up to the specified diameter Not all members of the sample could be aged by the CF method; sec explanation in ■■^hlterials and methods" section. pie retained for CF analysis and the sub- sample e.xcludecl from analysis were not significantly different ichi-square test of homogeneity Z"=--80. 1 df). A nonparamet- ric growth model was estimated by fitting a scatterplot smooth to the carapace length and estimated age data (Fig. 4A). Because the CF method could not be applied to the largest turtles in the sample, the model provided no information about growth in mature turtles. Age estimates for posthatchling turtles based on the SI method range from 3.3 to 49.4 yr (n = 103). The nonparametric growth model for these data fitted very well (Fig. 4B ), a result iiot unexpected giv- en the dependence of the SI age estimate on humerus diameter and the tight linear relationship between humerus diameter and carapace length. Variation in carapace length around the SI growth model reflect- ed only variance in the linear relationship between carapace length and humerus di- ameter: it did not incorporate variance in the age of turtles at a fi.xed humerus di- ameter or estimates of their age. The boot- strap distributions of age estimates for the SI method suggested that age estimates for a turtle can vary by up to 10 years I Fig. 5 A I. Nevertheless, the nonparamet- ric confidence inten'als for predicted mean length at age were narrow (Fig. 5B). Although the CF and SI methods yield- ed different estimates of age iFig. 4). over a wide range of carapace lengths, the dif- ferences were not striking. The magnitude and direction of differences in age esti- mates depend on SCL. Judging from the fitted smooth curves, for turtles up to about 81 cm SCL. the correction factor method was expected to give estimates of age up to 2 years higher than those produced by the spline-integration meth- od. For larger turtles, however, age esti- mates derived from the correction-factor method were predicted to be as many as 10 years (or more) lower than those generat- ed by spline integration iFig. 6). The /-tests (Table 1) indicated there were highly sig- nificant differences in age estimates be- tween methods within the 40-50 cm size group (/=4.36, 9 df) and the 50-60 cm group ( f =4.66, 12 df). Differences for the 30-40 cm. 60-70 cm, and 70-80 cm size groups were nonsignificant; samples were too small for comparisons in other size groups. The Molokai mark-recapture data al- lowed a visual comparison of observed growth and growth predicted by the CF and SI models derived from skeletochro- Zug el a\ Age and growth of Hawaiian Chelonia 121 - lOOi w 80 ■^J^T-"-^ ■ Spline integration 20 30 Estimated age (yr) 40 50 Figure 4 Relationship betwoen observed carapace length and estimated age (squares) for the two methods of age determination: correction factor lAi and spline integration (B). Fitted gi'owth models (cui-ves) are cubic smoothing splines. Table 1 Mean age estimation for correction-factor and splmc-integi-ation method significantly between estimation methods for turtles in the middle length s for 10-cm length groups, groups. * " indicates a sign Estimates of mean age vary ficant difference. Length group (SCL.cm) Sample size Mean age estimate lyri / Idfl.P Correction-factor method I OF I Spline-integration method (SI) 30-40 14 7.3 6.9 1.09 1131,0.296 40-50 10 12.9 11.0 4.36 19). 0.002 50-60 13 19.1 15.8 4.66 1121,0.001 60-70 14 22.1 22.1 -0.06 1131.0.952 70-80 13 25.3 25.6 -0.67 1121,0.514 nological data (Fig. 7). In plotting the growth vectors for marked (tagged) turtles, we fixed the origin of each vector by assuming that the carapace length at time of first capture was given exactly by the CF- or Sl-based growth curve. The obsei-ved length at recapture and time at liberty then determined the cndpoint coordinates of the growth vector. Despite considerable variation in the ob- served gi'owth of the marked turtles, the growth vectors were generally concordant with the growth model predic- tions (Fig. 7). 122 Fishery Bulletin 100(1) 100 100 10 20 30 40 Estimated age (yr) 50 Figure 5 (A) Bootstrap distributions of estimated age at observed SCL generated by resam- pling the data for aging rate to humerus diameter 100 times and by applying the spline-integration method to each bootstrap replicate. iBi Corresponding distribu- tion of growth curves, described by the bootstrap mean growth cui-ve (line) and the approximate 50'^f and 95^c confidence intervals for predicted mean length at age (outer and inner edges of solid region). Age- and length-specific growth rates The box plots of the Sl-estimated carapace growth rates (Fig. 8) indicated relatively fast growth for smaller tur- tles, a reduced growth rate remaining fairly constant over intermediate length classes, and declining growth rates in larger, mature turtles. Mean carapace gi'owth rate declined from 4.4 cm/yr for turtles in the 20-30 cm SCL group to less than 1 cm/yr for mature turtles in the 90-100 cm group, and remained around 2.0-2.5 cm/yr for imma- ture juveniles in the intermediate length groups (Table 2, Fig. 8). Differences in mean growth rate among the intermediate length groups were not significant. The aver- age growth rates predicted by first differences of the SI- based growth model (Fig. 9) indicated a similar pattern, as expected, showing a decrease in growth rate during the first decade of life (when Hawaiian gi-een turtles are still foraging in the open ocean or in the early years of their residence in inshore habitats), relatively constant growth during the next 15-20 year interval, and a further decline in growth rate as the turtles approach 30 years of age. The gi-owth rate appears to remain low in older turtles. Discussion Age and growth-rate estimates Age estimates by both methods indicate an age range of 4-49 years for the Hawaiian green seaturtles in their coastal habitats. The pelagic juvenile (28.7 cm SCL) in our sample was about four years old (4.1 and 3.3 yr by CF and SI methods, respectively). The smallest juveniles (35-37 cm SCL) in coastal waters were 6 to 9 years old by the CF Ziig et al Age and growth of Hawaiian Chclonia 123 40 60 Straight carapace length (cm) 80 100 Figure 6 Difference in estimated age (yr) between the correction-factor method and the spline-integration method within the comparable range of carapace lengths (SCL). For positive values (black region!, the CF method gives older ages and for negative values (striped region), younger ages. methoci and 4 to 10 years old by the SI method. These age estimates for Hawaiian greens in the last years of their pelagic developmental stage are similar to those reported for C. mydoK populations in the southern Great Barrier Reef (SGBR) (5-6 yr; Chaloupka et al., in press) and for the Atlantic coast of central Florida (3-6 yr; Zug and Glor, 1999). The smallest C. mydas turtle in the Florida sample was 28 cm SCL (several others were less than 35 cm), whereas the smallest SGBR turtle was 38.5 cm CCL (Limpus and Chaloupka, 1997) and the smallest Hawai- ian specimen was 34.8 cm SCL, indicating an earlier shift from pelagic to benthic life in Florida greens. The growth of the juvenile turtles predicted by both CF and SI models is consistent with the growth observed in the Molokai mark-recapture sample, but predictions of the CF model depart from the tag-recapture results in older turtles (Fig. 6). Mean growth rates for smaller (30-60 cm) turtles estimated from our transformed humerus incre- ment data (Table 2) were about half as high as growth rates reported for turtles of the same size in most Atlan- tic and Caribbean locales (based on tagging and skeleto- chronology: Tables 2 and 3 in Zug and Glor, 1999). Our estimates were similar to the observed growth rates of tagged turtles in Kiholo Bay, Hawai'i (Balazs et al., 2000) and nearly twice as high as rates observed in some other Pacific samples (Galapagos. Heron Island; Tables 2 and 3 in Zug and Glor, 1999). Subsequent studies of Australian populations (Limpus and Chaloupka, 1997; Chaloupka et. al., in press) have shown a mid-juvenile growth rate more similar to our estimates; however, growth rate is associated with a gi-owth surge in the Australian turtles over a narrow mid-juvenile length range (50-60 cm SCL). Such a spurt Table 2 Growth rate f tati ^tics by length c ass Rates were derived from spline-in tegi ation data. Grow th ■ate (cm/yr) Length gi'oup (SCL. cm) Sample size Mean SD 20-30 9 4.4 2.2 30-40 37 3..5 2.7 40-.50 67 2.1 1.2 .50-60 53 2.3 1.0 60-70 62 2.2 0.9 70-80 21 2.1 1.0 80-90 12 1.3 0..5 90-100 7 0.6 0.3 in growth was not evident in our Sl-based growth cui"ve, and growth rates were fairly constant in the 40-80 cm size classes (Fig. 8). Our age and growth estimates pertain to the Hawaiian population as a whole, because the sampled turtles orig- inated from locations throughout the archipelago. Some variation in age and growth between island foraging groups is likely, given the extensive latitudinal range of the habitats and the associated variation in physical and biological parameters affecting growth (Balazs, 1982). A geographic analysis of age and growth will be the subject of a future study. Future study will also investigate the 124 Fisheiy Bulletin 100(1) UU ' A .l^r 80- ^^ 60- ^ 40- Correction factor 20- n- 1 1 10 20 30 40 50 UU - B > 80 ■ ^^ 60- 7\^^^P wr^ 40- ^ Spline integration 20 - / 0 - 1 1 ; 10 20 30 Estimated age (yr) 40 50 Figure 7 Obsei-i'ed gi'owth vectors of marked Molokai green seaturtles between release and recapture (vectors) in relation to the gr-owth predicted by non- parametric models of SCL and age (cui-ves); age was estimated by the two methods: correction factor (A) and spline integration (Bl. The turtle's length at release is assumed to fall on the growth cur\'e. effects of fibropapillomatosis and gender on growth rates among the Hawaiian population. Age-estimation methods A key assumption of both the CF and SI methods is that each estimated humerus growth layer represents 1 year of gi-owth. This assumption has been validated only recently (Hohn and Snover' ). Hawaiian turtles tagged and injected with tetracycline have been recaptured, and these turtles show the appropriate number of LAGs for the years since their receipt of tetracycline. Furthermore, strong support is provided by the consistency of the growth model pre- dictions with obsei-ved gi-owth in tagged Molokai turtles. Additional justifications have been advanced in other stud- ies of seaturtle humerus LAG formation (e.g. Zug and Glor, 1998; Coles etal., 2001). ' Hohn, A., and M. Snover. 2001. Personal commun. Beaufort Laboratory, Southeast Fisheries Science Center. Beaufort, NC. Both the CF and SI methods require histological prepa- ration and analysis of humerus sections. But they use the same skeletochronological data in independent and differ- ent ways to estimate the total number of humerus growth layers. The CF method assumes that a constant humerus growth rate (the "correction factor") applies to the resorp- tion core regardless of the diameter of the core and despite the fact that periosteal increment width decreases with length of the turtle (Fig. 2). The correction factor, C. is esti- mated from a subset of the skeletochronological data tak- en from juvenile turtles only, i.e. excluding larger turtles likely to have narrower increments in the outer region of the humerus. Even so, the CF estimates of age for juve- nile turtles appear to be biased upward (Fig. 5), suggest- ing that the constant correction factor also failed to reflect the effect of wider increments deposited in the early years of life. In age estimation, the CF method can be applied only to turtles displaying a complete set of periosteal lay- ers, i.e. distinct LAGs, from the resorption core to the outer margin of the humerus. Zug el a\ Age and growth of Hawaiian Chelonia 125 20-30 30-40 40-50 50-60 60-70 70-80 80-90 90-100 Straight carapace length (cm) Figure 8 Variation of the growth rate estimates (Sl-based) by carapace length class. Each box plot shows median (white bar), limits of 2'"' and 3"' quartiles (solid notched box), range (brackets), and outliers (solid lines). The SI metho(3 models increment-width variation over the entire distance from humerus center to outer margin by using a nonlinear, nonparametric smoother. The model is estimated from all available sample data with two or more LAGs and associated diameter measurements with- out regard to size of the turtle. The SI model can be ap- plied to age all turtles for which the outside diameter of the humerus section has been measured. In the skeletochronological sample we studied, the CF method gave age estimates significantly higher than the SI method for turtles of intermediate length (Table 1), but expected differences for such turtles were no greater than about 2 years. On the other hand, based on current data the CF-based model gives much lower age estimates than the Sl-based model for turtles longer than about 86 cm SCL (Fig. 5). Thus although either method may suffice for aging juvenile Hawaiian green seaturtles, only the SI method appears to provide support for inferences about growth in turtles larger than about 80 cm. The CF method is computationally simpler, because it involves only linear regression rather than fitting and integrating a nonpara- metric smoother, and this consideration may recommend it to some users. Based on our experience with Hawaiian green seaturtle data, the main issue in judging the two techniques ap- pears to be the assumption with the CF method that hu- merus growth is linear. In reality, it is curs-ilinear, and the SI method explicitly models this cun'ilinearitv. Moreover, the SI method makes fuller use of available humerus in- crement data than the CF method. Further comparisons of the methods with additional skeletochronological data sets are recommended. Ecological precis 1 Chelonia myclas within the Hawaiian Islands is a com- ponent of the larger Indo-Pacific C. inydas gene pool, yet the nesting females of the Hawaiian population comprise a distinct genetic unit and contain a unique mtDNA haplotype (Bowen et al., 1992). 2 Except for the posthatching pelagic phase, the green seaturtles of the Hawaiian coastal waters are year- round residents, and all known individuals reproduce within the Hawaiian island chain, predominantly on the beaches of the Northwestern Islands at French Frigate Shoals (Balazs, 1998; Wetherall et al, 1999), 3 Skeletochronological age estimates indicate that Hawai- ian juveniles exit the pelagic phase between the ages of 4 to 10 years. 4 In coastal waters, juveniles 10 years and older possess a relatively constant growth rate until about 28 to 30 years (approximately 80 cm SCL), then growth begins to slow as individuals attain sexual maturity. 5 The mean SCL of nesting females is 92 cm (range 81- 106 cm; Balazs, 1980), suggesting ages of 30 or more years at first nesting for some individuals. 126 Fishery Bulletin 100(1) 6 5 4H 3 2 1 20 30 Estimated age (yr) 0 20 40 60 80 100 Straight carapace length (cm) Figure 9 Growth rates of Hawaiian Clielonia inydas in relation to age and carapace length, estimated by taking first differences of the Sl-based growth model. Acknowledgment We thank the following individuals and organizations for their valuable contributions to this research; A. Aguirre, D. Akaka, G. Antonelis, S. Eames, W. Gilmartin. S. Hau. D. Heacock, J. Kendig, R. Morris, W. Puleloa, L. Hallacher, W. Dudley, J. Coney, S. Patton, R. Sparks, M. Rice, G. Watson, M. Wisner, T. Work, the State of Hawaii Depart- ment of Land and Natural Resources, the Hawaii Prepara- tory Academy, Makai Animal Clinic, and the Marine Option Program of the University of Hawaii (Manoa and Hilo). George Zug thanks the NOAA-NMFS-SWFSC Honolulu Laboratory and SI/NMNH Department of Systematic Biol- ogy for encouraging and supporting his skeletochrono- logical research. Literature cited Balazs, G. H. 1979. Growth, food sources and migrations of immature Hawaiian Chelonia. Marine Turtle Newsl. 10:1-1.3. 1980. Synopsis of biological data on the gi-een turtles ni the Hawaiian Islands. U.S. Dep. Commer., NOAA Tech. Memo. NOAA-TM-NMFS-SWFC-7, p. i-ix, 1-141. 1982. Growth rates of nnmature gi-een turtles in the Hawai- ian Archipelago. In Biology and conservation of sea tur- tles (K. A. Bjorndal, ed.). p. 117-125. Smithsonian Inst. Press, Washington, D.C. 1998. Sea turtles, 3'"'' ed. /;; Atlas of Hawaii (S. P. Juvik and J. O. Juvik, eds.), p. 115. Univ. Hawaii Press, Honolulu, HI. Balazs, G. H., W. C. Dudley, L. E, Hallacher, J. P. Coney, and S. K. Koga. 1994. Ecology and cultural significance of sea turtles at Punalu'u, Hawaii. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-35 1:10-13. Balazs, G. H., W. Puleloa, E. Medeiros, S. K. K. Murakawa. and D. M. Elhs. 1999. Growth rates and incidence of fibropapillomatosis in Hawaiian green turtles utilizing coastal foraging pas- tures at Palaau, Molokai. U.S. Dep. Commer, NOAA Tech. Memo. [1998] NMFS-SEFSC-415:130-132. Balazs. G. H., M. Rice, S. K. K. Murakawa, and G. Watson. 2000 Growth rates and residency of immature green tur- tles at Kiholo Bay. Hawaii. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-436:283-285. Bolten, A. B., K. A. Bjorndal. J. S. Grumbles, and D. W. Owens. 1992. Sex ratio and sex-specific growth rates of immature green turtles, Chelonia mydas. in the southern Bahamas. Copeia 1992:1098-1103. Bowcn, B. W., A. B. Meylan, -J. R Ross, C. J. Limpus. G. H. Balazs, and J. C. Aviso. 1992. Global population structure and natural history of Zug et a\ Age and growth of Hawaiian Chclonia 127 the p'een turtle [Chclonia mydcia) in terms of matriarchal phylogeny. Evolution 46:865-881. ( 'haloupka, M. Y., C. J. Limpus, and J. D. Miller. In press. Sea turtle growth dynamics in a spatially struc- tured population. Can. J. Zool. < 'halou|)ka, M. Y., and .J. A. Musick. 1996. Age. growth, and population ilynamus, /;; The liiol og>' of sea turtles (P. L. Lutz and J. A. Musick. eds.). p. 233- 276. CRC Press, Boca Raton. Fl. Coles. W. C, J. A. Musick, and L. A. Williamson 2001. Skeletochronology validation from an adult logger- head {Caretta caretta). Copeia 2001:240-242. Cnnise, D. T., L. B. Crowdcr, and H. Caswell. 1987. A stage-based population model for loggerhead sea tur- tles and implications for conservation. Ecology 68:1412- 1423. Hardle, W. 1990. Applied non-parametric regi'ession. O.xford Univ. Press. New York. NY. 333 p. Hastie. T. J., and R. J. Tibshirani. 1990. Generalized additive models. Chapman and Hall. New York, NY, 335 p. Limpus, C. J., and M. Y. Chaloupka. 1997. Nonparametric regi-ession modelling of green sea turtle growth rates (southern Great Barrier Reef). Mar Ecol. Prog. Ser 149:23-34. MathSoft. Inc. 1999. S-PLUS 2000 professional: modern statistics and advanced graphics [software]. MathSoft. Seattle. WA. Murakawa. S. K. K., G. H. Balazs, D. M. Ellis. S. Hau. and S. M. Eames. 2000. Trends in fibropapillomatosis among gi"een turtles stranded in the Hawaiian islands, 1982-98. U.S. Dep. Conimer., NOAATech. Memo. NMFS-SEFSC-443:239-242. Parham.J. F.andG. R. Zug 1998. Age and growth of loggerhead sea turtles iCaivtta caretta) of coastal Georgia: an assessment of skeletochrono- logical age-estimates. Bull. Mar. Sci. 11997] 61:287-304. Ralston. S.. and G. Miyamoto. 1983. Analyzing the width of daily otolith increments to age the Hawaiian snapper. PristiponiDttlcn filamcntosus. Fish. Bull.81:.52.3-535. Wetherall, J. A.. G. H. Balazs. and M. Y Y Yong, 1999. Statistical methods for gi'ccn turtle nesting surveys in the Hawaiian Islands. U.S. Dep. Commer., NOAATech. Memo. 11998] NMFS-SEFSC-415:278-280. Work. T. M., and G. H. Balazs. 1999a. Causes of gi'een turtle iChclnnin niydafi) morbidity and mortality in Hawaii, U.S. Dep. Commer. NOAA Tech. Memo. 119981 NMFS-SEFSC-415:291-292. 1999b. Relating tumor score to hematology in green turtles with fibropapillomatosis in Hawaii. J. Wildl. Disease 35: 804-807. Zug. G. R.. G. H. Balazs, and J. A. Wetherall. 1995. Growth in juvenile loggerhead seaturtles iCaretta caretta ) in the North Pacific pelagic habitat. Copeia 1995: 484-487. Zug. G. R.,andR. E. Glor 1999. Estimates of age and gi-owth in a population of green sea turtles tChelouia mydas) from the Indian River lagoon system. Florida: a skeletochronological analysis. Can. J. Zool. 11998] 76:1497-1506. Zug, G. R., A. H. Wynn. and C. Ruckdeschel. 1986. Age determination of loggerhead sea turtles, Caretta caretta, by incremental growth marks in the skeleton. Smithson. Contrib. Zool. (427):l-34. 128 Estimates of lobster-handling mortality associated with the Northwestern Hawaiian Islands lobster-trap fishery Gerard T DiNardo Edward E. DeMartini Honolulu Laboratory, Southwest Fisheries Science Center National Marine Fisheries SeiA/ice, NOAA 2570 Dole Street Honolulu, Hawaii 96822 E-mail address (for Gerard T DiNardo) gdinardoShonlab nmfs hawaiiedu Wayne R. Haight Joint Institute of Marine and Atmosphenc Research School of Ocean and Earth Science and Technology University of Hawaii, 1000 Pope Road Honolulu, Hawaii 96822 The commercial lobster fishery in the Northwestern Hawaiian Islands ( NWHI ) is a distant-water trap fishery that tar- gets the Hawaiian spiny lobster (Pan- ulirus marginatus) and slipper lobster (Scyllarides squammosus). The ISTWHI are an isolated group of islands, atolls, islets, reefs, and banks that extend 1500 nmi west-northwest of the main Hawai- ian Islands from Nihoa Island to Kure Atoll (Fig, 1). Reported landings in the NWHI peaked at about 2,000.000 lob- sters (spiny and slipper combined) in 1985, and then declined to about 38.000 lobsters from 1986 to 1995 (Fig, 2). The NWHI lobster fishery is man- aged under the Fishery Management Plan for the Crustaceans of the West- ern Pacific Region (Crustaceans FMP) implemented in 1983 and developed by the Western Pacific Regional Fish- ery Management Council (WPRFMC), The National Marine Fisheries Sei-vice (NMFS) is responsible for stewardship of the resource and review and im- plementation of proposed management measures, A variety of management measures have been adopted in re- sponse to declining catches: a limited- entry fishing regime that limited the number of permit holders to 15; a prohi- bition on fishing from January through June when lobsters spawn; an annual catch quota system; a minimum legal tail width (TW) of 50 mm for spiny lob- ster and 56 mm TW for slipper lobster. which are close to the sizes at first ma- turity for these species in the NWHI; a prohibition on landing berried (ovig- erous) females; and a requirement that traps be equipped with escape vents to reduce capture of undersize lobsters (WPRFMCM, Prior to 1996, fishermen were required to discard all berried and undersize lobsters, which were not counted against the catch quota. The management plan assumed that escape vents allowed substantial num- bers of undersize lobster to escape cap- ture and that undersize and berried lobsters do not die during the discard process. Although research on lobster fisheries has found that escape vents effectively reduce the capture of un- dersize lobsters (Ki'ouse. 1978; Fogarty and Borden, 1980; Harris, 1980; Ever- son et al.. 1992; Skillman et al.-), con- siderable numbers of undersize (hence- forth termed "sublegal") and berried lobsters are caught in the NWHI lob- ster fishery Between 1983 and 1995 the reported lobster discard rate in- creased from 2Q"c to 62^?^ (Fig. 3), re- sulting from changes in the size- and age-structures of the populations and in the areas fished. The average size of spiny lobsters generally increased northwestward from Nihoa along the Hawaiian Archipelago (Uchida et al., 1980). Although as many as 16 banks within the NWHI have been fished, the spatial distribution of fishing effort has shifted to banks in the southeast of the Hawaiian Archipelago where there is a higher concentration of spiny lobsters. Qualitative data collected during the early days of the fishery suggested that mortality associated with the handling and discarding practices of the NWHI commercial lobster-trap fishery might be high (Gooding. 1985; Gooding'^). Un- less discard mortality is explicitly con- sidered, fishing policy decisions can be suboptimal, or worse. Where catch quo- tas are used, the total fishing-induced mortality of the population is greater than expected and can even result in recruitment overfishing. Using an equi- librium yield-per-recruit (YPR) model, Kobayashi^ found that the reproduc- tive potential of the NWHI lobster population more than doubled, and mean weight per individual increased by 22% in a retain-all fishery (all lob- sters brought on deck were retained as catch) if the mortality rate of discard- ed lobsters was high (>75'7f ). Based on these results, the observed high discard rate of sublegal and berried lobsters (62%), and the presumption that the 1 Western Pacific Regional Management Council. 199.5. Fisheiy management plan for the crustacean fisheries of the Western Pacific region, amendment 9. Western Pa- cific Regional Fishery Management Coun- cil. Honolulu. Hawaii. 227 p. - Skillman. R. A.. A. R. Everson, and G. L. Ki-amer. 1984. Prospectus escape vent experimental procedure for the spiny lob- ster fishery under management of the Magnuson Fishery Conservation and Man- agement Act. Southwest Fish. Sci. Cent, Admin. Rep. H-84-1.3, unpubl. report, lip. Honolulu Lab.. Southwest Fish. Sci. Cent., Natl. Mar Fish Serv., NOAA. Honolulu. HI 96822-2396. * Gooding, R. M. 1979. Obsorv-ations on surface-released, sublegal spiny lobsters, and potential spiny lobster predators near Necker and Nihoa. Southwest Fish. Sci. Cent. Admin. Rep. H-79-16. unpubl. report, 8 p. Honolulu Lab.. Southwest Fish. Sci. Cent., Natl. Mar Fish Sei-v., NOAA, Hono- lulu, HI 96822-2396. ■t Kobayaslii, D. R. 2001. Southwest Fish. Sci. Cent. Admin. Rep., in prep. Simu- lated effects of discard mortality on spiny lobster iPanulirus mai-ginatus) sustain- able yield and spawning stock biomass per recruit in the Northwestern Hawaiian Islands. Honolulu Lab., Southwest Fish. Sci. Cent., Natl. Mar Fish Serv., NOAA. Honolulu. HI 96822-2396. Manuscript accepted 11 Mav 2001. Fish. Bull. 100:128-133 (2002). NOTE DiNardo et al.: Estimates of lobster mortality in the Northwestern Hawaiian Islands 129 Hancock . Bank Kure Atoll '^"^"ay Island Salmon Bank ' Pearl & Hermes Reef Laysan Island Lisianski ^ Island Maro Reef Northwestern Hawaiian Islands Ralta Bank Gardner '-■' Pinnacles Necker Island Nihoa 180 Pacific Oci'ciu 175 30 25' Kauai A Molokai Oahu .;'J^^Maui 170" 165 Mam Hawaiian Islands 160 155" Figure 1 Map of the Hawaiian Archipelago. 2500 2000 ^ r-. i / \ 1 1500 / \ ■o (U T3 C / \ ^^\ ^ 1000 J \ /*^ \ Numbe / \ 500 / \ 0 ^^^^^^ ^ 1982 1984 1986 1988 1990 1992 1994 1996 Year Figure 2 Annual commercial landings for the NWHI lobster fishery, 1983-96. mortality rate of discarded lobsters is greater than 75"*, the WPRFMC amended the Crustaceans FMP in 1996 to allow the retention of all lobsters caught in the NWHI com- mercial lobster-trap fishery subject to the quota on total catch (WPRFMC). The WPRFMC also recommended that NMFS conduct experiments to assess mortality associated with possible handling and discarding practices. This study reviews research conducted in the NWHI on fishery-induced mortality of sublegal and berried lobsters. The authors examined on-deck mortality of sublegal and 130 Fishery Bulletin 100(1) 70 60 50 ? "D S 30 a 20 10 1982 1984 1986 1988 1990 Year 1992 1994 1996 Figure 3 Annual estimates of the reported discard rate of lobsters in the NWHl lobster fishery. 1983-95 berried spiny and slipper lobsters that occurred within two days after capture under handling methods known to have been used in the NWHI lobster-trap fishery. Methods Studies of handling-induced mortahty ( referred to as "han- dling mortahty" in the remainder of this note) for spiny lobster were conducted at Necker Island onboard the NOAA ship Towitsend Cromwell during the NMFS Hono- lulu Laboratory's 1996 NWHI lobster sui-vey from 21 to 26 June 1996 and for slipper lobster at Maro Reef from 4 to 11 July 1996 (see Fig. 1). These locations were chosen because of their historically high lobster catches and their overall importance to the commercial lobster fishery over the past 13 years. Handling mortalities were estimated for two handling methods: "dry" and "wet." The effects of on-deck exposure time on handling mortality were also estimated for spiny lobsters handled by the dry method. We defined a particu- lar combination of factors (method and exposure time) as an experimental treatment. For dry treatments, lobsters were held on deck in 30-gallon containers without water and in direct sunlight for 1, 2, or 3 hours. In the wet treat- ment, lobsters were held on deck for 3 hours in shaded 30-gallon containers with circulating seawater. The 2- and 3-h dry treatments represent prevailing commercial fish- ing practices (Anderson'^), whereas the 1-h dry treatment ^Anderson,?. 1997. Personal commun. University of Hawaii, Marine Option Program, 1000 Pope Road, Honolulu. HI 96822. Kazama, T. 1997. Personal commun. Honolulu Laboratory, Southwest Fisheries Science Center, 2.570 Dole Street, Hono- lulu, HI 96822-2396. and the 3-h wet treatments represent possible handling alternatives or mitigative measures. Spiny and slipper lobsters were collected in baited com- mercial lobster traps and sorted (legal, sublegal, and ber- ried). All sublegal and berried lobsters were held in tanks of circulating water until a sufficiently large paired exper- imental treatment sample («=200 lobsters) was collected. Experiments with the 3-h wet and dry treatments were done first, followed by experiments involving 1- and 2-h dry treatments. For each treatment, 100 lobsters (sublegal and berried) were randomly chosen from the tank and placed into two 30-gaUon treatment containers (dry or wet). 50 lobsters/container After the exposure time (1, 2, or 3 hours), lobsters were removed from the treatment containers, their condition recorded as active (i.e. lobsters were capable of tail flexion), weak (incapable of tail flexion but able to move ap- pendages when prodded), or dead. Five lobsters were placed in each of 20 lobster-holding traps (commercial lobster traps with sealed entrances) and held in two 1320-gallon bait- wells with recirculating seawater (21 gallons/minute) for 2 days. It was assumed that the 2-d holding period would al- low ample time for latent effects from a treatment to man- ifest and that each lobster holding trap was a replicate (7! =20 replicates/treatment). To check for possible baitwell effects, five control traps, each containing five previously untreated lobsters, were placed in the baitwells (two traps in the port baitwell and three traps in the starboard bait- well) at the beginning of the holding phase, and their condi- tion was recorded at the beginning and end of the holding phase. At the conclusion of the holding period, the condition of the treated lobsters was again assessed. Handling mortality for each treatment was computed as the arithmetic mean of the percent mortality (dead/ total or (dead -i- weak)/total — see below) observed in the 20 NOTE DiNardo et al Estimates of lobster mortality in the Northwestern Hawaiian Islands 131 100 80- ? 60 ..■■' ■^^^^ ^' 03 t: o 5 40 20 1 1 2 3 Exposure lime (hours) Figure 4 Handling mortality (dead/total) estimates for spiny lobster subjected to 1-, 2 - . and .3-h drv treatments. Dashed lines indicate the 95'^^ confidence limits. treatment replicates. Randomization resampling was used to evaluate estimates of handling mortality. Approximate lower and upper QS'/f confidence limits for handling mor- tality were computed as the 2.5'7f and 97 .59i percentiles of the bootstrap distributions by using the computer pro- gram RT (Manly, 1994). Results Spiny lobster Spiny lobster sample sizes by discard category were suble- gal («=84-88) and berried (n = l2-16) (Table 1 ). All spiny lob- sters subjected to the .3-h wet treatment were active at the conclusion of the 2-d baitwell holding period. For dry treat- ments, the number of active spiny lobsters decreased with increasing exposure time, whereas the number of dead lob- sters increased with increasing exposure time. All control lobsters were active, indicating no baitwell holding effects. Handling mortality (dead/total) for the dry treatment ranged from 12'^ for the 1-h treatment to 707r for the 3-h treatment (Fig. 4) and did not differ between berried and unberried lobsters (P>0.05). Pooled dead and weak lobsters resulted in a handling mortality that ranged from 16% for the 1-h treatment to 77% for the 3-h treatment (Fig. 5). Slipper lobster Handling mortalities for slipper lobster were estimated only for the 3-h dry and wet treatments. Mechanical failure of the baitwell recirculating water pumps during the first 3-h paired dry and wet experiments forced us to reduce the baitwell holding time to 1 day. After the baitwell pumps Table 1 Spiny and slipper lobster sample sizes by discard category for each treatment. NB = nonberried and sublegal; B = berried. Treatment Spiny lobster Slipper lobster NB B NB B 1 hour dry 84 16 — 2 hours dry 88 12 — — .3 hours dry 87 13 87-89 11-13 3 hours wet 88 12 81 19 failed, we repeated the 3-h dry treatment with a 1-d hold- ing period, suspending the holding and control traps 3 m below the sea surface from the Townsend Cromwell. Slipper lobster sample sizes by discard category (suble- gal and berried) for each treatment are shown in Table 1. Most slipper lobsters subjected to the 3-h wet treatment were active at the end of the 1-d holding period. For the 3-h dry treatment, the number of dead lobsters ranged from 14 to 39, and the number of active lobsters ranged from 56 to 83. All control lobsters were active, indicating no baitwell holding effects. Slipper lobster handling mortalities for the 3-h dry treat- ment ranged from 14% to 39% with an average estimate of 27% and were unrelated to berried condition (P>0.05). Esti- mates of handling mortalities with weak and dead lobsters combined ranged from 17% to 44% , with an average of 31%. Handling mortality for the 3-h wet treatment was 1%. 132 Fishery Bulletin 100(1) 100 1 80 „.......-. ■■■;^ S" 60 ....-•■■■" ^^^^^^^'^^ _.■■■■■" 2 40 ,..•••■■' ^^^^'^^^ ...■■••■' 20 .:,--^.-----""" 0 ^ 1 2 3 Exposure time (hours) Figure 5 Pooled handling mortality ((dead + weakl/total) estimates for spiny lobster subjected to 1-, 2-, and 3-h dry treatments. Dashed lines indicate the 95''; confidence limits. Discussion Many factors affect the survival of lobsters discarded in commercial trap fisheries, including capture, handling, and discarding processes. In our experiment we focused on handling factors to assess their impacts on mortality in the NWHI commercial lobster fishery. If the 2- and 3-h dry experimental treatments typify commercial handling practices, then the mortality of dis- carded spiny lobster from handling practices on commer- cial vessels is extreme, ranging from an average of 25% to 45% and 70% to 77%, respectively, depending on how da- ta are pooled. Handling mortality for slipper lobster also appears high (estimated at 31%) but is considerably less than that for spiny lobster; thus spiny lobsters may have a lower handling tolerance than slipper lobsters. Although there are no published estimates of handling mortality for P. i7iarginatiis and S. squa?nmosus, studies on other lobster species suggest that handling mortality in the NWHI lobster fishery is high. Lyons and Kennedy (1981), reporting on P. argiis. estimated that 12.3% of lob- sters died after 30 minutes of exposure to direct sunlight and an average 24.1% after 1-4 hours of exposures. They also found that lobsters exposed for 2-4 hours tended to die within a week following exposure, whereas those ex- posed for only 1 hour survived longer. Laboratory experi- ments by Brown and Caputi ( 1983) on small western rock lobster, Paniiliri/s cygniis, exposed to direct sunlight re- sulted in an expected time to 50%f. mortality that decreased with increasing temperature, ranging from 233 minutes at 27°C to 99 minutes at 31-35°C. Time to 50% mortality was 387 minutes for lobsters held in the shade at 26.5-32°C. Handling mortality, however, represents only a portion of the total mortality of discarded lobsters resulting from their capture, shipboard processing, and subsequent re- lease in the NWHI. Additional mortality resulting from habitat displacement, predation, and other factors associ- ated with discarding might result in total discard mortal- ity estimates that approach 100%. Qualitative evidence suggests that discarded lobsters are subject to high preda- tion from the giant trevally, Caranx ignohilis, which ag- gregate around vessels during fishing operations (Good- ing, 1985). Sorting and discarding lobsters immediately after they are placed on deck appears to reduce total discard mortal- ity. The discarded lobsters would need to be returned to the general vicinity of their capture and as close to the sea floor as possible to avoid the gauntlet of predators in the water column. Brown and Caputi (1986) reported a reduction in recapture rates of displaced undersized rock lobsters compared with nondisplaced rock lobsters and re- lated the reduction directly to predation mortality. Adoption of the retain-all fishery by the WPRFMC in 1996 significantly reduced fishery-induced handling mortality. Although sporadic discarding occurs, current discard rates are less than 1% and will have no detect- able consequence at the population level. The research does, however, provide insight into past fishery-induced handling impacts, which likely contributed to the de- cline in NWHI lobster catches. If future management again reverted to mandatory discarding practices, this research provides information to assess its impact on fishery-induced mortality. However, to fully understand the synergistic effects of catching, handling, and dis- carding practices on mortality in the NWHI lobster fish- ery, additional research to assess the impacts associat- ed with shipboard sorting and releasing (e.g. postrelease predation) is required. NOTE DiNardo et a\ Estimates of lobster mortality in the Northwestern Hawaiian Islands 133 Acknowledgments Robert Moffitt, Donald Kobayashi, Jerry Wethi'iall. and the anonymous referees provided helpful comments on the draft, for which we are very gi-ateful. We also acknowledge the crew of the Townsend Cromnell and the scientific staff who worked to ensure that good biological information is available for analyses such as this. Literature cited Brown. R. S., and N. Caputi. 1983. Factors affecting the recapture of undersized western rock lobster Pomiliru.s cygnuf: George returned by fisher- men to the sea. Fish. Res. 2:103-128. 1986. Conservation of recruitment of the western rock lob- ster (Panulirus cygnus) by improving survival and growth of undersize rock lobster captured and returned by fisher- men to the sea. Can. J. Fish. Aquat. Sci. 43:2236-2242. Everson. A. R., R. A. Skillman, and J. J. Polovina. 1992. Evaluation of rectangular and circular escape vents in the Northwestern Hawaiian Islands lobster fishery. N. Am. Fish. Manag. 12:161-171. Fogarty. M. J., and D. V. D. Borden. 1980. Effect of trap venting on gear selectivity in the inshore Rhode Island American lobster. Homarus americanus. fish- ery. Fish. Bull. 77:925-933. Gooding, R. M. 1985. Predation on released spiny lobster, Paniihriit: nuirgin- atus, during tests in the Northwestern Hawaiian Islands. Mar Fish. Rev. 47:27-35. Harris, K. 1980. Escape gaps in rock lobster pots. Fish. Ind. News Tasmania 2:40-43. Krouse, J, S. 1978. Effectiveness of escape vent shape in traps for catch- ing legal size lobsters, Homarus americanus, and harve.st- able-sized crabs. Cancer borealis and Cancer irroralus. Fish. Bull. 76:425-432. Lyons, W. G., and F. S. Kennedy. 1981. Effects of harvest techniques on sublegal spiny lob- sters and on subsequent fishing yield. Proc. Gulf Caribb. Fish. In.st. 33:290-300. Manly, B. F. J. 1994. RT: a program for randomization testing., version 1.02C. Centre for Applications of Statistics and Mathe- matics, Univ. Otago, P.O. Box 56, Dunedin, New Zealand. Uchida, R. N., J. H. Uchiyama, D. T. Tagami, and P. A. Shiota. 1980. Biology, distribution, and estimates of apparent abun- dance of the spiny lobster, Panultrus marginatus I Quoy and Gaimard) in waters of the Northwestern Hawaiian Islands. Part II: size distribution, legal and sublegal ratio, sex ratio, reproductive cycle, and morphometric characteristics. /;; Proceedings of the symposium on status of resource inves- tigations in the Northwestern Hawaiian Islands, Honolulu, Hawaii, April 24-25, 1980 (R. W. Grigg and R. T. Pfund, eds.), p. 131-142. Univ. Hawaii Sea Grant College Pro- gram MR-80-04. Honolulu. Hawaii. 134 An evaluation of pop-up satellite tags for estimating postrelease survival of blue marlin {Makaira nigricans) from a recreational fishery* John E. Graves Virginia Institute of Marine Science College of William and Mary Rt. 1208 Greate Road Gloucester Point, Virglna 23062 E-mail address gravesfaJvims edu Brian E. Luckhurst Division of Fisheries Department of Agnculture and Fistienes PO Box CR 52 Crawl CRBX, Bermuda Eric D. Prince National Marine Fistieries Service 75 Virginia Beach Dnve Miami, FL 33146 Blue marlin [Makaira nigricans) rep- resent an important commercial and recreational resource throughout trop- ical and subtropical oceanic waters. In the Atlantic Ocean, blue marlin are managed as a single, oceanwide stock. In the most recent assessment of Atlantic blue marim ( ICCAT, 2001), the Standing Committee for Research and Statistics (SCRS) of the International Commission for the Conservation of Atlantic Tunas (ICCAT) estimated the current biomass of blue marlin to be about 40% of that required for maxi- mum sustainable yield (MSY). Further- more, the assessment indicated that the current level of fishing mortality (F) was about four times higher than F,y,gY and that catch levels in recent years were more than twice the equi- librium yield, contributing to a fur- ther decline of the overexploited stock. Based on the most recent stock assess- ment, fishing-induced mortality must be reduced by about 60% to halt the decline of the stock (Goodyear, 2000). The greatest source of billfish (Istio- phoridae) mortality occurs as a result of incidental catches by longline gear deployed for tunas and swordfish (IC- CAT, 1997, 2001). These highly migra- tory species co-occur in the subtropical and tropical epipelagic environment, and all are vulnerable to the pelagic longline. Not all billfish are dead at the time of capture (haulback) on longline gear; data from observers on vessels in the Venezuelan industrial longline fishery indicate that about 49% of blue inarlin caught on pelagic longline gear are alive at the time of capture (Jack- son and Farber, 1998). To reduce billfish mortality ICCAT in 1997 required nations to reduce their landings of Atlantic blue marlin by 25% from 1996 levels. Furthermore, the IC- CAT SCRS has recommended that the Commission consider requiring the re- lease of all live billfish taken on long- line gear (ICCAT 1997, 2001). It is believed that such a management mea- sure would be more acceptable to mem- ber nations than an overall reduction in longline effort that would also re- duce catches of target species. How- ever, representatives from several na- tions have pointed out that there are not sufficient data to estimate postre- lease survival of billfish; therefore the conservation impact of a recommenda- tion requiring live released fish cannot be evaluated. In fact, low recovery rates of billfish tagged and released with conventional tags by recreational and commercial fishermen ( <2%; Jones and Prince, 1998; Ortiz et al, 1998) are con- sistent with high postrelease mortality. However, factors such as tag shedding and failure to report tag recaptures could also account for low rates of tag returns ( Bayley and Prince, 1994; Jones and Prince, 1998). Clearly, data are needed to support or refute the hypothe- sis that the release of live billfish would significantly eliminate fishing mortali- ty of blue marlin (Graves et al.'). Acoustic tracking studies designed to investigate billfish physiology and behavior have provided insights into the postrelease survival of billfish tak- en on recreational gear Specifically, ob- sei-ved and inferred mortalities during the course of the acoustic tracks indi- cate that not all released billfish sur- vive (reviewed in Pepperell and Davis, 1999). Unfortunately, it is not possible to estimate levels of postrelease mor- tality of billfish from previous acoustic tracking studies for several reasons. First, owing to the high cost of ship and personnel tiine, relatively few ani- mals have been investigated in acous- tic tracking studies. Second, because ocean conditions can deteriorate quick- ly, many of the acoustic tracks were for less than 12 hours duration, pro- viding a limited opportunity to ob- ser\'e mortality after 12 hours. Third, billfish were caught and subsequently tracked under a variety of conditions, making cross-study comparisons diffi- cult. Finally, an estimate of postrelease mortality rates resulting from acoustic studies may be biased because in soine cases only healthy fish were selected to carry acoustic transmitters. The development of pop-up satellite tag technology may present a possible means to estimate postrelease mortali- ty of billfish. Although relatively expen- sive pop-up satellite tags reduce the need to use a tracking vessel to follow * Contribution 2416 of the Virginia Insti- tute of Marine Science, College of William and Mary, Gloucester Point, VA 2.3062. ' Graves, J. E., G. Skomal. and E. D. Prince. 1995. Report of the billfish mortality workshop. Contribution rep. MIA-94/ 95-49, 7 p. Southeast Fisheries Science Center, Miami, FL. Manuscript accepted 30 July 2001. Fish, Bull. 100434-142 (2002). NOTE Graves et al An evakialion of iatellite tags for estimating postrelease srirvivai of Mnkaim nigiican^ 135 hillfish on the liigli seas. Pop-up satellite tags are capable (if recording environmental variables over predefined in- ten-als, of detaching from an animal at a designated time, lioating til the surface, and of transmitting the stored data to a satellite. Until now, these tags have been deployed pri- marily on bluefin tuna for relatively long durations (up to nine months) to determine movement patterns (Block et al.. 1998a; Lutcavage et al., 19991. Recovery of tag data has been very good in most cases, with some reported rates in excess of 90% (Block et al., 1998a; Lutcavage et al., 1999). These results suggest that the technology may be well suit- ed for shorter term studies, including the determination of postrelease sui-vival. In this paper we present the results of a preliminary study to evaluate the feasibihty of applying pop-up satellite tag technology to estimate short-term sur- vival of blue marlin. We also include a brief analysis of the movement and behavior of blue marlin that we inferred from the pop-up tagging results. Materials and methods Pop-up satellite tags The Microwave Telemetry. Inc. PTT-100 pop-up satellite tag was used in this study. The tag can withstand a pressure of 1000 psi (equivalent to a depth of about 650 meters) and is sufficiently small (38 cm by 4 cm diameter) that it would not appear to impose a major drag on a large marine tele- ost. such as blue marlin (Block et al.. 1998a). Tags were progi-amed to measure water temperature every hour and record the mean value for each two-hour period for a total of 61 cycles (122 hours). Inclinometer values were taken every two minutes and summed for the periods before tag detachment (pre-pop-up) and after tag detachment (post- pop-up). For each period the inclinometer started with an initial value of 128. If at the time of measurement (every two minutes) the tag was oriented below 30 degrees above horizontal, a value of one was subtracted from the total. If the tag was above 30 degrees above horizontal at the time of measurement, the inclinometer total was increa.sed by one, but could not exceed 255. Final values below 255 indi- cated sufficient forward propulsion such that the positively buoyant tag was depressed below 30 degrees above horizon- tal for certain periods, demonstrating forward propulsion. All nine tags were progi-amed to detach from the fish 122 hours after activation, at which time the memory within each tag would contain 61 direct temperature measure- ments and the pre-pop-up inclinometer value. The five-day attachment period of the pop-up tag was chosen, in part, as a result of a review of data from conventional tag-re- captured blue marlin in the Cooperative Tagging Center (CTC) database (E. Prince, unpubl. data). Of the 160 blue marlin tag returns in CTC that have been validated, ten individuals were recaptured within five days of release, suggesting that some blue marlin are able to survive the catching and tagging event and commence feeding again within a few days. In addition, acoustical tagging studies have shown high sui"vival rates of different marlin species in the first 1-2 days following release, demonstrating that mortality, when it occurred, generally happened within the first 48 hours of release (Pepperell and Davis, 1999). With these considerations in mind, we assumed that the five- day period of tag attachment was an adequate period for catch and release mortality to be expressed. As indicated by Goodyear (in press), the duration of this type of experi- ment should be the minimum number of days necessary to account for postrelease mortality events. Longer periods would allow for greater influence of tag shedding, tag mal- function, and natural mortality, all of which could compro- mise estimates of postrelease survival. Tag deployment Pop-up satellite tags were activated and tested at the start of each fishing day. Blue marlin were caught south- west of Bermuda in the vicinity of Challenger and Argus Banks on standard recreational gear for the blue marlin fishery in Bermuda ( 130 lb test line) by using trolled high- speed lures or skirted dead baits (in most cases with two hooks). All hooks employed in this study were "J" hooks (no. 16/0-20/0). We tagged the first nine fish available to us. Six blue marlin were caught on the vessels we were aboard. Three individuals were taken on other vessels and transferred to the tagging vessel after the fish were brought to leader (brought to the boat): one blue marlin was caught and attached to a drifting buoy until the tag- ging vessel, which was several miles away, could gain access to the fish; and two fish were directly transferred after capture from the fishing vessel to the tagging vessel by using a procedure described in Block et al. ( 1998b). Once fish were brought to leader (reeled to the side of the boat), quieted and secured, the pop-up satellite tag and a conventional (streamer) tag were deployed. Pop-up satellite tags were attached to one end of a 400-lb ( 182 kg) test monofilament leader about 18.5 cm in length, with an outside diameter of 1.8 mm. The other end of the leader was attached to a double barb nylon anchor (about 33 mm long and 10 mm wide) made of medical-grade nylon. The anchors were implanted by using a stainless steel tag ap- plicator modified to accomplish placement to a depth of about 10 cm into the dorsal musculature, about 10 cm pos- terior and 5 cm below the base of the peak of the first dor- sal fin (Fig. 1). Hook location, as well as observations on foul-hooking (tissue damage, bleeding, etc.), were noted at the time of hook removal. Analyses After detachment from the animal, the positively buoyant tags floated to the surface and began transmitting data to satellites of the Argos™' system. Position information and sections of the temperature and inclinometer data were captured with each satellite pass and transmitted to a ground station and ultimately to the investigators by the internet. Data were analyzed to determine net movement from the point of detachment to the point when the tag popped-up (usually the first tag transmittal; however, if the first satellite pass was near the horizon, the location of the second transmittal was used to obtain greater accu- 136 Fishery Bulletin 100(1) Figure 1 Blue marlin with pop-up satellite tag attached. The tag is located below the anterior portion of the dorsal fin (within outlined area). racy). Water temperature was determined from tempera- ture sensor readings by using a calibration provided by the manufacturer Depth was estimated from water tempera- ture values by using temperature and depth relationships provided by the Bermuda Biological Station for Research, which maintains an oceanographic sampling station (sta- tion S) about 13 miles (24 km) to the southeast of the island. Results and discussion Nine blue marlin. with estimated weights ranging from 150 to 425 lb (68 to 193 kg), were tagged between 25 July and 11 August 1999 (Table 1). Four specimens were below the minimum size for tagging recommended by the tag manufacturer (200 lb or 90.9 kg). Fight times ranged between 15 and 35 minutes. Seven fish were initially hooked in the jaw, and two were "foul-hooked" (i.e. outside the jaw and mouth): one in the operculum and one in the dorsal musculature. After tag placement, but before release, one fish that was originally hooked in the jaw became foul-hooked in the ventral musculature. Three of the nine fish were transferred to the tagging vessel after capture. Fish generally quieted down shortly after being brought to the side of the vessel, which maintained a head- way of 4-5 km/li during the tagging operation. Only a few minutes were required to implant the satellite and con- ventional tags, photograph the fish, estimate weight, mea- sure lower jaw fork length (most individuals), and remove the hook. Condition of the fish varied, and three individu- als required resuscitation prior to release. Eight of the nine tags became detached from their re- spective host fish after five days, floated to the surface, and transinitted to the Argos™"" satellite system. Based on the first accurate location of the tags, net displacements ranged from 40 to 134 nmi (72-248 km) with a mean lin- ear displacement of 90 nmi (167 km) for each individual (Fig. 2). These values are in the range reported for blue marlin by Block et al. ( 1992) who followed six blue marlin with acoustic transmitters for periods of one to five days. They noted individual total movements (as opposed to net displacements) of 253 km in about three days, 100 km in five days, and four animals with movements of less than 100 km over the course of the respective tracking periods. Individual marlin in our study dispersed in all directions from their point of release ( Fig. 2 ). The blue marlin tracked by Block et al. ( 1992) and Holland et al. (1990) in Hawaiian waters moved away from the point of capture in several dif- ferent directions. However, the authors noted an orientation of movements to the coastline of the Hawaiian Islands. Our r-eleases were farther offshore and an affinity to the Bermu- da coastline was not evident from the net movement data. Depending on the time of tag activation and the time of tag deployment, up to 61 direct water temperature read- ings, taken every two hours, were obtained for each blue NOTE Graves et al : An evakiation of satellite tags for estimating postrelease survival of Makaiia nigncam 137 Table 1 I'op-up s tagging. under its itellite tag deploynicnt information. "Tran.s ■Resuscitation" indicates the time spent to own power. Tag no. 24040 failed to report. er" mda move a ales hlue whether a ti.sh was moved b marlin through the water ..'tween vessels aftei ifter capture until capture to allow t could swim off Tag no. Deployment Fight time Transfer (yes/no) Hook location E;stimated weight (lb) Resuscitation (no/yes — time) date hour min. 24519 25 Jul 1999 1610 25 yes jaw 400 no 24059 28 Jul 1999 1110 35 yes jaw and ventral musculature 200 no 24522 lAug 1999 1030 20 yes jaw 175 no 24520 2 Aug 1999 1015 15 no jaw 180 no 24033 2 Aug 1999 1245 30 no jaw 425 yes-10 min. 24040 2 Aug 1999 1500 17 no jaw 200 no 24523 3 Aug 1999 1550 15 no operculum 150 yes-8 min. 24527 11 Aug 1999 1255 15 no jaw 350 no 24029 11 Aug 1999 1340 23 no dorsal mu.sculature 150 yes-3 min. marlin (Fig. 3). Temperature readings demonstrated that tagged individuals spent the majority of their time at tem- peratures above 26°C (Fig. 4). The maximum temperature range recorded for any of the eight individuals was 9°C (22-3rC, tag no. 24033). Block et al. ( 1992), using acoustic tracking, determined that the six blue marlin which they tracked spent half of their time in the upper 10 m of the water column in water temperatures 2.5-27°C, and Hol- land et al. (1990) reported that blue marlin in waters off Hawaii remained at temperatures of 26° or greater. Differences in thermal histories were evident among the individuals in our study. Blue marlin no. 24029 (Fig. 3G) spent the vast majority of time at temperatures equal to that of the surface waters (30-31°C). In contrast, individu- al no. 24527 (Fig. 3H) spent much less time in the warmer surface waters and repeatedly moved up and down in wa- ters between 23° and 31°C. Several individuals appeared to remain at or very near the surface for extended peri- ods, evident in Figure 3 as a continuous string of tempera- ture readings at or slightly above 30°C. An analysis of the data examining diurnal-nocturnal periods with tempera- ture (inferred depth) indicated a high level of variability between individuals and no clear pattern was apparent (Fig. 3). In contrast, Holland et al. ( 1990) determined that blue marlin spent a higher proportion of their time (-50%) in the upper 10 m at night than during the day (-25% ). It was possible to infer swimming depths of blue marlin by comparing water temperature values with the temper- ature-depth profiles at station "S" provided by the Bermu- da Biological Station for Research.- All blue marlin en- Although this station is situated 24 km to the southeast of the island, similar temperature-depth profiles would be expected for the general region (Johnson, R. 2000. Personal commun. Bermuda Biological Station for Research, 17 Biological Lane, Ferry Reach, St. George's GEOl Bermuda). This allowed us to use the station S profiles to infer swimming depth, realizing that, depending on when an animal was tagged and where it moved, there would be some differences for which we could not account. tered cooler waters at various times during the five-day period, with excursions to depths as great as 40 meters. Temperature records were consistent with the tagged blue marlin actively undertaking vertical movements in the upper 40 meters of the water column. However, six of the eight fish spent >75% of their time in the upper 10 m of the water column for the five-day duration of the study. If the data from all eight fish are pooled, this yields a mean value of 79.9% (SD 15.8%) of the time spent in this zone. This is a higher percentage of time spent in the upper 10 m than that obsei-ved by Block et al. ( 1992), who reported that fish spent about half of the time in this zone. How- ever, this comparison should be viewed with some caution because the Block et al. (1992) data were based on con- tinuous tracking, whereas each data point in our analysis was the average of two hourly measurements. All post-pop-up inclinometer values were 254 or 255, where 255 represented the maximum (vertical) inclinom- eter value expected for an upright, floating tag. Pre-pop- up inclinometer values ranged from 203 to 251, with three individuals at 233 and four between 247 and 251. These values indicate tags were inclined at an angle below 30 degrees above horizontal for more than 40% of the 1830 sampling times for each individual, and are consistent with sufficient forward propulsion to depress the positive- ly buoyant tag more than 60 degrees from vertical. There was no correlation between pre-pop-up inclinometer values and net displacement. The fish with the largest net dis- placement (no. 24059) had the second highest inclinometer value. This was not unexpected because the difference be- tween the lowest and highest pre-pop-up inclinometer val- ues represents a minor difference in the time the tag was depressed below 30 degrees above horizontal. Also, the re- lationship between total movement and net displacement could be quite different for different individuals. Three different lines of evidence provided by the pop- up satellite tags (net movement, water temperature, and tag inclination) each suggested that at least eight of the 138 Fishery Bulletin 100(1) Sft-W 63°W ws Jl^N 13^2405^1134 4 Nml Tag 2-10:>) 1113 1 Nm) \ Tae24033(')S,6Nm) Tag 24520 (94. R NmJ Ias2«22(8S2Nm) 'Tjg245l9(55 2 Nm) ^^/T^g 24523 (J")*) Nm) Bermuda Islands \Tjji24527(89 8Nm) 32°N 63°W Figure 2 Map showing points of release (squares) and points of recovery (end of straight lines) for eight of nine blue marlin equipped with pop-up satellite tags near Bermuda, 25 July-11 August 1999. The pop-up satellite tag number for each fish and straight line distance between point of release and geolocation where tag transmitted data to the Argos satellite (given in parentheses) are provided. nine blue marlin caught on recreational gear survived for five days following capture, tagging, and release events. The net movement data indicated a broad dispersal of the eight fish in different directions that cannot be explained by local currents. In contrast to the differing direction of movement, the net displacements of the eight fish were fairly similar. The mean displacement of 89.25 nmi over a five-day period, compares favorably with blue marlin swimming velocities of 1-2 nmi/h reported from acoustic tracking studies (Holland et al., 1990) and is consistent with the constant slow swimming of the individuals. Al- though currents could have accounted for some of the net displacement, inclinometer values indicated that all eight individuals were actively swimming. The water temperature measurements indicated that each blue marlin actively undertook dives into cooler wa- ter throughout the course of the five days. All eight in- dividuals spent the vast majority of their time in waters with temperatures of 26°C or greater, and no readings be- low 22°C were recorded. The successful data recording, tag detachment, and transmission of eight of the nme pop-up satellite tags begs the question of what happened to the one tag that failed to report. It is not possible to distinguish between the postre- lease mortality of a tagged blue marlin and the mechani- cal failure of a pop-up satellite tag. If a marlin dies and sinks in deep water, the attached pop-up satellite tag even- tually will be crushed by increasing hydrostatic pressure. NOTE Graves et a\ An evaluation of satellite tags for estimating postrelease survival of Makaim nigricans 139 c tag number 24519 0 10 20 30 40 50 60 70 80 90 100 110 12 B tag number 24059 0 10 20 30 40 50 60 70 80 90 100 110 tag number 24522 50 60 70 Time (hours) Figure 3 Temperature and inferred depth for eight of nine blue marlin equipped with pop-up satellite tags near Bermuda. 25 July-11 August 1999. See text for explanation of inferred depth and Table 1 for release information correspond- ing to each tag number Hours of darkness are shaded on the time line. The blue marlin whose tag did not report (tag no. 24040. Table 1 ) was hooked in the jaw, caught in less than 20 min- utes, did not require resuscitation, was quickly tagged, and actively swam away from the boat when released. Shark predation on released billfish has been reported (Holland et al., 1990; Pepperell and Davis, 1999); therefore mortal- ity cannot be excluded despite the apparent vigorous con- dition of the fish. Failures in component subsystems could account for the failure of reporting from a pop-up tag. A detailed analysis of the reliability of each tag component could be undertaken, but several factors external to the tag could also result in a failure of reporting. Tag man- ufacturer innovations and upgrades of the systems will allow researchers to better identify mortalities, but they will not completely solve the problem of discriminating between tag failure and fish mortality. Nonreporting tags would have significant consequences for efforts to make ocean-wide estimates of postrelease sui-vival. The ability to account for all pop-up satellite tags deployed is directly related to the accuracy of the resulting estimates of postre- lease survival (Goodyear, in press). Nonreporting satellite tags introduce uncertainty that cannot be quantified in the estimates of postrelease survival, thus compromising meaningful conclusions. Excluding nonreporting tags from the analysis decreases precision of the estimate, and in- cluding mortalities biases the survival estimate down- ward. Further, any extension of the 5-day pop-up period to allow study of possible delayed effects of tagging should involve careful consideration of the benefits and the li- abilities that longer durations might have on estimating postrelease survival (Goodyear, in press) Successful tagging and reporting of pop-up tags from four fish under 200 lb (90.9 kg) indicate that the size and design of the PTT-100 tag is tolerated by smaller blue mar- 140 Fishery Bulletin 100(1) tag number 24033 60 70 80 Time (hours) Figure 3 (continued) 100 110 120 NOTE Graves et al,: An evaluation of satellite tags for estimating postrelease survival of Makaira nigricans 141 % 15 31+ 30 29 28 27 26 25 24 23 22 Temperature ( C) Figure 4 Distribution of time at temperature for eight of nine blue marlin equipped with pop-up satellite tags near Bermuda. 25 July-11 August 1999. Histogram represents combined data for the five-day period of data transmission for each individual. lin than that recommended by the manufacturer-, at least in the short term. Thus a pop-up tag of this size might be tested on even smaller specimens or other target species to expand the study of behavior in a wider size range of species than was originally thought possible. Acknowledgments The authors would like to thank Captains Alan DeSilva, Andrew Card, Gilbert Amaral and Bobby Rego for pro- viding assistance in deploying the pop-up tags. We also express our appreciation to the anglers and to the orga- nizing committee of the Bermuda Billfish Tournament for their support. Paul Howey and Molly Lutcavage provided expert technical advice. Phil Goodyear made helpful com- ments on the manuscript. Support for this project was obtained from the Bermuda Department of Agriculture and Fisheries, the ICCAT Enhanced Research Program for Billfish, and Hewit Family Foundation. Literature cited Bayley, R. E., and E. D. Prince. 1994. A review of tag release and recapture files for Istiophoridae from the Southeast Fisheries Science Cen- ter's Cooperative Gamefish Tagging Program. Int. Comm. Cons. Atl. Tunas (ICCAT) Coll. Vol. Sci. Pap. 41:,527-548. Block, B. A., D. T Booth, and F. G. Carey. 1992. Depth and temperature of the blue marlin. Mahaira nigricans, observed by acoustic telemetry. Mar Biol. 114: 17.5-183. Block, B. A.. H. Dewar, C. Fai-well, and E. D. Prince. 1998a. A new satellite technology for tracking the move- ments of Atlantic blucfin tuna. Proc. Natl. Acad. Sci. 95:9384-9389. Block, B. A., H. Dewar, T D. Williams, E. D. Pnnce, C. Farwell, and D. Fudge. 1998b. Ai'chival tagging of Atlantic bluefin tuna iThunnus lliynniis tliynniis). Mar Tech. Soc. J. 32:37-46. Goodyear, C. P. 2000. Biomass projections for Atlantic blue marlin: potential benefits of fishing mortality reductions. Int. Comm. Cons. Atl. Tunas (ICCAT) Coll. Vol. Sci. Pap. 52:1502-1506. In press. Factors affecting robust estimates of the catch and release mortality using pop-up tag technology. In Sympo- sium on catch and release in marine recreational fisheries (A. Studholme, E. D. Prince, and J. Lucy, eds). Spec. Pub. Am. Fish. Soc. Holland. K., R. Brill, and R. K. C. Chang. 1990. Horizontal and vertical movements of Pacific blue marlin captured and released using sportfishing gear. Fish. Bull. 88:397-402. ICCAT (International Commission for the Conservation of Atlantic Tunas). 1997. Report for biennial period. 1996-97. Part 1 (1996), vol. 2, 204 p. ICCAT, Madrid, Spain. 2001. Report of the fourth ICCAT billfish workshop. Int. Comm. Cons. Atl. Tunas (ICCAT) Coll. Vol. Sci. Pap 53:1- 22. Jackson, T L., and M. I. Farber 1998. Summary of at-sea sampling of the western Atlantic Ocean, 1987-1995, by industrial longline vessels fishing out of the port of Cumana, Venezuela: ICCAT Enhanced Research Program for Billfish 1987-1995. Int. Comm. Cons. Atl. Tunas (ICCAT) Coll. Vol. Sci. Pap. 47:203-228. Jones, C. D., and E. D. Prince. 1998. The cooperative tagging center mark-recapture data- 142 Fishery Bulletin 100(1) base for Istiophoridae (1954-1995) with an analysis of the west Atlantic ICCAT billfish tagging program. Int. Comm. Cons. Atl. Tunas (ICCAT) Coll. Vol. Sci. Pap.47:31 1-322. Lutcavage, M. E., R, W. Brill, G. B. Skomal, B. Chase, and P. Howey. 1999. Results of pop-up satellite tagging of spawning size class fish in the Gulf of Maine: Do North Atlantic bluefin tuna spawn in the mid-Atlantic? Can. J. Fish. Aquat. Sci. 56:173-177. Ortiz, M., D. S. Rosenthal, A. Venizelos, M.I. Farber, and E. D. Prince. 1998. Cooperative Tagging Center Annual Newsletter: 1998. U.S. Dep. Commer . NOAATech. Memo., NMFS-SEFSC-423, 23 p. Pepperell, J. G., and T. L. O. Davis. 1999. Post-release behavior of black niarlin Makaira indica caught and released using sportfishing gear off the Great Barrier Reef (Australia). Mar. Biol. 135:369-380. 143 Reproduction of the blacknose shark (Corcharhinus acronotus) in coastal waters off northeastern Brazil Fabio H. V. Hazin Paulo G. Oliveira Matt K. Broadhurst Depaitamento de Pesca, Laboratorio de Oceanografia Pesqueira Universidade Federal Rural de Pernambuco Av Dom Manuel de Medeiros, s/n Dois Irmaos, Recife PE, Brasil, CEP: 52.171-900 E mail address (for F Hazin) fhvhazin gelogica com br The blacknose shark, Carcharhiin/s acronotus, is a relatively small car- charinid, typically inhabiting continen- tal shelf areas in the western Atlantic Ocean, from North Carolina through- out the Gulf of Mexico (Bigelow and Schroeder. 1948) and along the South American coast to Rio de Janeiro (Com- pagno, 1984). The abundance of this shark in nearshore areas throughout its distribution makes it accessible to commercial fishing, mainly from in- shore hook-and-line and gill-net fish- eries (Trent et al.. 1997; Mattes and HazinM. Aspects of the biology of C. aci-ono- tus have been reported by Springer 11938); Bigelow and Schroeder (1948); Clark and von Schmidt ( 1965); Dodrill (1977); Branstetter (1981); Schwartz ( 1984 ); Castro ( 1993 ); and Carlson et al. (1999). Schwartz (1984) provided the most comprehensive synopsis, includ- ing information on their reproduction and life cycle off North Carolina. Many of the other studies were based on rel- atively few specimens collected off the southeastern United States and cor- roborate much of Schwartz's (1984) work, including patterns in spatial and temporal abundances, size at maturi- ty, fecundity, and time of parturition. However, some inconsistencies exist with respect to the duration of the ovarian, gestation, and breeding cycles: i.e. Dodrill (1977) proposed a biennial breeding cycle with gestation taking between 10 and 11 months, Schwartz (1984) suggested a gestation of ap- proximately 9 months, and Branstet- ter ( 1981) observed two gravid females with large ovarian eggs (having con- current ovarian and gestation cycles, copulation having occurred shortly af- ter parturition). Because C. acronotus are represent- ed in catches from various inshore fish- eries and carcharhinids typically are characterized by low rates of popu- lation increase, adequate information about their reproductive cycle is re- quired to facilitate management of stocks. Given the uncertainty of knowl- edge about reproduction in C. acrono- tus and the lack of information for the southern part of their range, our aims in the present study were to provide a preliminary overview of the repro- ductive biologv' and life cycle of the blacknose shark in coastal waters off northeastern Brazil, using available fishery-dependent data. Material and methods Fishing gear used and data collected Carcharhinus acronotus (79 females and 45 males) were collected from the catches of commercial gillnetters and vessels using bottom longlines off the coast of northeastern Brazil (approx. 7°30' to 9°30'S — near Recife) between August 1994 and January 1999. The configuration of fishing gears used remained similar over this period. Gill nets were monofilament, 900 m in length, and had a stretched mesh size of 17 cm and a depth of 70 meshes. Nets were set perpendicular to the beach at depths between 5 and 10 m. Bottom longlines consisted of a multifilament mainline (6 mm in diameter) with up to 100 secondary lines, each approx. 5 m in length and constructed from 3-mm diameter monofilament attached to a wire snood ( 1 m in length). Types of hooks varied among brands, but rela- tive sizes (i.e. 9/0 ) remained similar The main baits were sardine (SardineUa hrasiliensis) and mackerel (Scomber spp. ), although some other species, including sting ray (Aetobarus nari- nari) and skipjack tuna iKatsuwonus pelamis) were occasionally used. Long- lines were set on the continental shelf at depths between 10 and 60 m, but most sets were shallower than 40 m. All fishing gears were set at dusk and hauled the following morning at dawn. All specimens were measured (total length ITLj in the natural position ) and dissected. Reproductive organs were removed and stored in a solution of lO^-'f formalin in seawater prior to be- ing transported to the laboratory. Data collected from females included weight and width of the oviducal gland and the functional right ovary, maximum ovar- ian follicle diameter (MOFD), width of the largest uterus, and, if present, the TL, sex, and number of embryos. Using the methods described by Pratt ( 1993), we examined the oviducal glands of 10 mature females for the presence of spermatozoa. Length and calcification stage of claspers, width of epididymi- des, and the presence of seminal fluid in the ampullae of the ductus deferens were recorded from males. Reproduc- tive organs were measured to the near- est 0.1 mm with vernier calipers. Inferences on stages of reproduction were made according to definitions pro- vided in previous studies on carcha- rinids (e.g. Pratt, 1979; Hazin et al., 2000). Females were categorized into six stages, mainly based on develop- ' Mattos, S. M. G. and F. H. V. Hazin. 1997. Analise dc viabilidade economica da pesca de tubaroes no litoral do estado de Per- nambuco. Boletim Tecnico-cientifico do Cepene. 5: 89-114. IBAMA (Instituto Brasiliero de Meio Ambiente e dos Recur- sos Naturals Renovaveis), Rua Samuel Hardman, s/n Tamandare - PE, Brazil. Manuscript accepted .5 July 2001| Fish. Bull. 100:143-148(2002). 144 Fishery Bulletin 100(1) ment of the oviducal gland, ovary, and uterus. Specimens were considered juvenile if they had undeveloped sexual organs, filiform uteri, and no vitellogenic activity in their ovaries. Pre- ovulatory females had relatively larger ova- ries with orange vitellogenic follicles, but no uterine eggs. Ovulating females had uterine eggs and ripe ova still in the ovary, whereas in gravid specimens ovulation was complete. Postpartum females showed similar-size ova- ries, oviducal glands, and ovarian follicles as those of gravid specimens, but had flaccid uteri that were still slightly enlarged (compared with other nongravid females) indicating re- cent parturition. Individuals that had simi- lar-size uteri as those of pre-ovulatory and ovulating individuals, but smaller ovarian fol- licles with little or no vitellogenic activity were termed "resting" females. Male maturation was evaluated according to development of the tes- tes and claspers. Individuals with relatively short, flexible claspers, and filiform ampullae of the ductus deferens were considered juve- niles. Adults were characterized by elongate and calcified claspers and relatively large epi- didymides (compared with juveniles). Statistical analyses Size-frequency distributions of males and fe- males were compared by using a two-sample Kol- mogorov-Smirnov test (P=0.05). Chi-squared goodness-of-fit tests were used to examine the hypothesis of an equal sex ratio between num- bers of juvenile and adult C. acronotus sampled and among embryos in gravid females. To help define time of parturition, analysis of variance (ANOVA) was used to investigate the hypoth- esis of an increase in TL of embryos from near- term females captured in November and December. To provide a balanced analysis, three gravid females captured on 10 December 1998 were analyzed with three individu- als captured almost one month earlier on 11 November 1998. Data were assessed for normality by using the Sha- piro and Wilk procedure (Zar, 1996), tested for heterosce- dasticity with Cochran's test, and then analyzed in the appropriate one-nested factor ANOVA (Undei-wood, 1981). Gravid females were considered a random-effects factor nested in months and the four embryos for each gravid female were the replicates. Results The Kolmogorov-Smirnov test detected significant differ- ences in size-frequency compositions between males and females; proportionally more larger-size females ( 121-131 cm TL) were captured. Both sexes showed two distinct cohorts. The first consisted of juveniles (46-65 cm) cap- tured by gill nets at depths between 5 and 10 m, and I "1 A A -D 30- ■ A 033 D 1 "- ■■' o°^^ 3 \ To *^ P 1 5- > A ^ D ° 10- o • : £ 05- ^M. ^^ > 0 0- • ^%^te 30- 25- B ^^ Juvenile n - 21 ^ ^1 Pre-ovulatory n = & ■ J^ Ovulating n ^ 4 ■ ■ I 20- Q Gravid n = 21 ■ D ,5- 1 1 Postpartum n = 8 ^ A S ,0- 05- 0 0 - • 16- — 14 - C 0° O 0 B 12- r. 9fi^ ^ 10- „^^ % B 8- 3 o O 6- £ 4- 1^°^ ■o 5 2- . „i.^ ^^A A - - # 0 0 >— 45 55 65 75 85 95 105 115 125 13! Total length (cm) Figure 1 Rt'lationshi p between total length and (A) width of oviducal gland. (B)MOFD maximum ovarian follicle diameterl. and (C) width of uterus. the second included larger specimens (i.e. 85-131 cm), and mostly adults, captured on bottom longlines in deeper water ( 10 to 60 m). The ratio of juvenile males to females was not significantly different from 1:1 (X"=0.027,P>0.011, whereas significantly fewer adult males than females were sampled (ratio of 1:2.34) (x-=14.08.P<0. 01). Female maturation and reproduction Juveniles ranged in size from 46 to 101 cm TL and had narrow oviducal glands, light ovaries with undeveloped white follicles, and thin uteri (Fig. 1, A-C, Table 1). Pre- ovulatory specimens ranged in size from 103 to 129.5 cm TL, had well-developed oviducal glands (width of between 1.9 and 3 cm), and mature ovaries with thick orange fol- hcles (1.5 to 2.8 cm in diameter) (Fig. 1, A and B, Table 1). The shortest pre-ovulatoi\v specimens (e.g. 103 and 106 cm TL), had oviducal glands that were substantially wider (i.e. >2x) than those in the longest juveniles ( 101 cm TL) (Fig. lA, Table 1); therefore sexual maturity probably was approached within this size range. This should be con- NOTE Hazin et a\. Reproduction of Caicharinus acronotus 145 Table t Characti'ristics of fi'inale C. acronotus in each matura ion stage and the total 1 -■ngth (TL) range and number ol specimens. All widths and lengths are in cm , weights in g. MOFD = maximum ovarian follicle diameter. Characteristic Juvenile Pre-ovulatory Ovulating Gravid Postpartum Resting Width of oviducal gland <1.1 1.9- -3.0 2.3-3.5 1.7- -3.0 1.5- -3.0 1.4-2.6 Weight of oviducal gland 2.2-3.0 5.7- -8.7 8.6-10.0 2.3- -5.0 3.0- -5.0 3.0-7.7 Width of uterus <0.7 1.9-4.7 2.6-3.9 6.3 -16 4.2- -5.6 1.9-3.5 MOFD 0.1-0.4 1.5- -2.8 2.8-3.0 0.2- -1.0 0.2- -1.2 0.4-1.4 Width of ovary 0.4-1.7 2..3- -6.0 4.9-5.7 1.5- -3.8 2.2- -3.5 1.7-4.0 Weight of ovary 2.7-4.6 12.2- -18.2 20.0-24.6 4.5- -16.5 12.0- -23.0 4.97-16.5 TL of specimens 46.0-101.0 103.0- -129.5 108.5-123.8 114.0- -130.0 123.0- -131.0 107.0-130.0 Number of specimens 21 8 4 21 8 17 sidered a preliminary estimate of size at sexual maturity because few individuals were caught between 95 and 105 cm TL and none between 75 and 95 cm TL. Compared with pre-ovulatory females, those in the process of ovulation had slightly wider oviducal glands and heavier ovaries with thicker follicles (2.8-3 cm in diameter) (Fig. 1, A and B, Table 1 ). Ovulation appeared to occur only when MOFD was at least 2.8 cm (Fig IB). The majority of gravid, post- partum, and resting females ranged in size from 111 to 131 cm TL and had similar-size oviducal glands (Fig. lA, Table 1 ). Gravid females had undeveloped ovaries with no vitellogenic activity (MOFD was less than 1 cm). Uteri in postpartum females were flaccid and slightly distended (4.2-5.6 cm in width), whereas uteri of resting females were comparable to pre-ovulatory and ovulating individu- als (Fig. IC, Table 1). No mating scars were observed on any of the females. There was no evidence of spermatozoa stored in the oviducal glands of 10 females examined. The temporal abundance of females according to stages of reproduction showed that juveniles were present in catches between January and November, but mainly from February to June (Fig. 2A). Pre-ovulatory and ovulating individuals were sampled between February and April and April and May, respectively (Fig. 2A). Except for one gravid female caught during August, all gravid, postpar- tum, and resting females were caught between November and January (Fig. 2A). MOFD of mature females was low- est in August and between November and January, after which it steadily increased to May (Fig. 2B). Examination of the uterine contents of the 22 gravid fe- males (Table 2) revealed that individual litter size was al- ways four with both sexes present in varying proportions, although the total pooled ratio of males to females (1:1.25) was not significantly different from 1:1 (x^=1.13, P>0.05). All embryos from individual gravid females were at simi- lar stages of development and showed small variation in TL. ANOVA detected significant differences in mean TLs of embryos among near-term females (F=20.51, P<0.01) and for the main effect of months (F=10.45, P<0.05). ANOVA showed that the mean (±SE) TL of embryos in three near- term females caught on 10 December 1998 (45.96 ±0.38 cm) was significantly longer than in three near-term females Table 2 Date of capt ure and total length (TLl of gravid C. acrono- tus and the mean TL(±E) and sex ratio of embryos. Date of TL of gravid Embryos Mean TL Males: capture specimen (±SE) "emales 17 Aug 94 118.6 29.15(1.23) 1:3 1 1 Nov 98 121 42.50(0.14) 1:3 11 Nov 98 122 45.32 (0.60) 2:2 1 1 Nov 98 123 40.75(0.25) 3:1 1 1 Nov 98 125 43.25(0.43) 3:1 11 Nov 98 129 45.12(0.43) 3:1 13 Nov 98 130 38.50 (2.06) 2:2 17 Nov 98 114 34.27 (0.19) 3:1 17 Nov 98 127.7 42.50(0.15) 3:1 21 Nov 98 115.2 41.37(0.37) 3:1 22 Nov 98 127.2 45.47 (0.62) 2:2 25 Nov 98 126 43.87(2.01) 2:2 25 Nov 98 127.5 46.07 (0.22) 1:3 6 Dec 98 117 44.75(0.14) 2:2 6 Dec 98 123 43.37(0.85) 2:2 6 Dec 98 123 48.50 (0.28) 2:2 7 Dec 98 125 46.87(0.12) 2:2 10 Dec 98 124 45.62 (0.37) 2:2 10 Dec 98 124 47.50(0.28) 3:1 10 Dec 98 122 44.75(0.25) 2:2 12 Dec 98 123 42.50(0.35) 3:1 5 Jan 95 114,5 45.37 10.24) 2:2 caught on 11 November 1998 (42.31 ±0.07 cm), suggesting that embryos continued to develop between these periods. Male maturation and reproduction Of the 45 males examined, 23 were juvenile with thin epi- didymides (Fig. 3A) and filiform ampullae of the ductus deferens without seminal fluid. Juveniles ranging from 49 146 Fishery Bulletin 100(1) to 63 cm TL had relatively short flexible claspers (Fig. 3B) and testes that were not fully differentiated from the epigonal organ. No males between 64 and 87 cm TL were caught. Three juveniles (88, 93. and 94 cm TL) had claspers that were enlarged and beginning to calcify (Fig. 3B) and two larger specimens also had thick epididymi- des (Fig. 3A). Males appear to approach sexual maturity before 104 cm TL because all specimens longer than this had completely developed sexual organs, including elon- gate and calcified claspers, thick epididymides, and cir- cumvoluted ampullae of the ductus deferens (Fig. 3). Discussion Prevalence of adult females in catches was similar to that from observations made by Schwartz (1984) for stocks off the southeastern United States and can be attributed to a reproductive migration involving relatively large num- 24 22 20 18 16- 14- 12 ^■Juvenile Ig^Pre-ovulalory ^Sovulaling ^Gravid ^Postpartum ^^ Resting 3.0- Q O 00 B Jan Feb Mat Apr May Jun Jul Aug Sep Oct Nov Dec Month Figure 2 Monthly distribution of (A) females in catches according to their stage of reproduction and (B) MOFDs (maximum ovarian follicle diameters) for mature females. hers of gravid individuals (114 to 130 cm TL) into the sampled area. Because there was no evidence of disequi- librium between sexes (i.e. there were equal numbers of male and female embryos and juveniles), large numbers of adult males were probably segi'egated. It is unlikely males were in the sampled area and not caught because females, although proportionally larger, were captured across the same size range (Fig. 1), implying that the selectivity of the gear encompassed the range of sizes of males. Evaluation of the maturation stages of males and fe- males showed delineation between juveniles and adults. All females longer than 103 cm TL had enlarged oviducal glands and developed ovaries (Fig. 1, A and B, Table 1), and males longer than 104 cm TL had elongate and cal- cified claspers and thick epididymides (Fig. 3), indicating that sexual maturity was probably approached at these lengths. Although these estimates were derived from few individuals, they are comparable to those proposed by most researchers for specimens collected off the southeast- ern United States (e.g. Springer, 1938; Clark and von Schmidt, 1965; Compagno, 1984), with the ex- ceptions of Branstetter ( 1981) who suggested 110 cm TL for both sexes and Schwartz ( 1984 ) who suggested 110 cm TL for males. Size at birth, number of embryos, sex ratio of embryos, and the time of parturition of C. acrono- tus in our study are consistent with correspond- ing data from earlier works (Springer, 1938; Clark and von Schmidt, 1964; Dodrill (1977); Brans- tetter, 1981; Compagno, 1984; Schwartz, 1984). Size at birth has been suggested to be between 45 and 50 cm TL (e.g. Branstetter, 1981; Castro, 1993) and litter sizes commonly range from 3 to 6 (Springer, 1938; Dodrill, 1977; Compagno, 1984). We observed gravid females caught in November and December (late spring to early summer) with embryos longer than 45 cm TL (Table 2). Given the significant increase in size of embryos between these months (indicating that embryos were still developing) and the capture of several neonates (46-51 cm TL) in February and March (late sum- mer to early autumn), we conclude that parturi- tion off northeastern Brazil probably occurs from December to January (mid to late summer). A sim- ilar seasonal timing has been proposed for stocks off the southeastern LInited States (e.g. during June — Schwartz, 1984) and is generally typical for the majority of carcharhinids (e.g. Castro, 1993). In contrast to the inference made by Branstet- ter (1981) but in agi'eement with obsei-vations of Schwartz (1984), we showed that vitellogenisis and gestation occur consecutively in C. acronotus. Ovaries of adult females off northeastern Brazil begin to mature (pre-ovulatory stage) in Febru- ary (late summer) with ovulation occurring two to three months later (Fig. 2, A and B). This se- quence of events is illustrated by a rapid increase in MOFD from February (1.5-2 cm in diameter) to May (3 cm in diameter) (Fig. 2B). Given the proposed summer parturition (December to Janu- NOTE Hazin et a\: Reproduction of Carchannus aaonotus 147 arv>, mating and fertilization during April and May (autumn ) would result in a gestation peri- od of approximately 8 months — slightly short- er than that suggested by Schwartz ( 1984) and Dodrill 1 1977) (9 and 10 to 11 months, respec- tively) for stocks off the southeastern United States. Further, the periods required for vi- tellogenisis and gestation indicate that repro- duction of C acronotiis off northeastern Brazil could be completed within 10 to 11 months. With the simultaneous capture of nongravid and gi-avid females off Florida, Dodrill (1977) proposed a biennial reproductive cycle for C. ac- ronotus. Although we also collected gravid and nongi-avid (i.e. postpartum and resting) females together (i.e. mostly during December and Jan- uary), the latter individuals could have given birth 3 or 4 weeks earlier If these females sub- sequently ovulated 2 months later (in April), then reproduction could conceivably be annu- al. A fast vitellogenic period, combined with clear reproductive progress, is also supportive of a 1-year cycle. An alternative hypothesis that supports biennial reproduction is that the rest- ing females represented some proportion of seg- regated nongravid females that moved with the relatively large numbers of gravid females into the parturition area. Given the lack of adult fe- males in catches between June and November (winter to late spring), it is likely, however, that they frequent other areas after copulation. Without additional fishery-indepen- dent data, it is impossible to determine these locations and whether some proportion of the female population consists of nongravid or resting individuals throughout the yean Our evidence suggests a shorter reproductive cycle for C. acronotus than that previously noted in the literature. Given the 6-month difference between times of ovulation and parturition for females off northeastern Brazil and those off the southeastern United States, our results also indicate the existence of at least two separate stocks. Not- ing temporal differences in the sizes of embryos from fe- males, Schwartz (1984) proposed two partially separated populations off the southeastern United States: one off North Carolina and the other comprising individuals from Florida and the Gulf of Mexico. Additional research is needed to determine if the population in the southwest- ern Atlantic is separate from those in the north because the existence of a unit stock off northeastern Brazil would require separate management measures according to the status of that stock. Acknowledgments This study was funded by the Comissao Interministerial para os Recursos do Mar (CIRM) through the Programa Nacional de Avalia^ao dos Recursos Vivos da Zona Economica Exclusiva ( RE VIZEE ), the Fundagao de Amparo a Ciencia e Tecnologia do Estado de Pemambuco ( FACEPE ), and the Conselho Nacional de Ensino e Pesquisa (CNPq). E 25 o S 20 ■D 1 .5 ■o TD Q- 10 0) ° 05 ■g 5 00- 16- ^ 12- ^, '1' where t = age; / . = asymptotic length; /?,,/;., = instantaneous growth rate coefficients; and f J, t., = age intercept parameters. The second, dubbed the "linear" von Bertalanffy curve (Hoese et al., 1991; Vaughan, 1996), expresses the asymp- totic length as a linear function of age: /, =(f)fi +bit){l-e -kit-t„). (2) Of course other generalizations of the von Bertalanffy cui-ve may also be appropriate, such as the Richards (1959) equation /, =La-de- ')' where 5 5^0. (3) tp=ik.,t.,-k,ti)Hk.2-k,), The double von Bertalanffy curve ac- commodates the possibility that older, larger fish might grow more slowly in proportion to their length than young- er, smaller fish. (The linear von Ber- talanffy curve has no biological inter- pretation.) In reality, one might expect the gi'owth rate in proportion to length to decrease gradually with the age of the fish rather than at some abrupt piv- otal point. Moreover, the growth pattern of juvenile red drum seems to have a : ^ + /i,e"^'' + k^e'^'' sin(2;r(f ■ (5) where A, and A., are damping coeffi- cients and t^ is a shifting parameter for the sine wave valued between 0 and 1. Substituting Equations 5 and 4 and integrating with / = 0 when ^ = ?o gives ' Condrey, R., D. W. Beclcman and C. A. Wilson. 1988. Management implications of a new growth model for red drum. Appen- dix D. In Louisiana red drum research, J. A. Shepard (ed.), 26 p. U.S. Dept. Commerce Cooperative Agi'eement NA87- WC-H-06122. Marine Fisheries Initiative IMARFIN) Program. Louisiana Depart- ment of Wildlife and Fisheries, Seafood Division, Finfish Section. Baton Rouge, Louisiana 70803-7.503. - Goodyear, C. P. 1996. Status of the red drum stocks of the Gulf of Mexico: report for 1996. Rep. MIA 95/96-47, 21 p. Miami Lahoratorv, .Southeast Fish. Sci. Cent., Natl. Mar" Fish. Serv., NOAA. 75 Virginia Beach Dr, Miami, Fl. 33149. Manuscript accepted 17 May 2001. Fish. Bull. 100:149-152 (2002). 150 Fishery Bulletin 100(1) ^1 Po fc 4;r-+(A,)' 2KCOs{2!t{t^.'t)}-' A.,sin{2nit^ -t)] '2;rcos{2;r(/, -^o)}-' A._,sin{2;r(*,, -<„)} (6) Assuming the animal will not shrink with age, i.e., dl I dt >0, implies the constraint /fo + V"^''+ V~ 'sin(2;r(?-r))>0. (7) Equation 6 may appear formidable, but typically re- quires only a minute or two more to enter into standard statistical fitting packages. It reduces to a form similar to the Gompertz equation when k., = 0 and to the von Berta- lanffy equation when ^j = ^2=0- Fitting the model to data Equations 1, 2, 3, and 6 were fitted to observations of length-at-age from red drum collected in the northern Gulf of Mexico between September 1985 and October 1998 (see Beckman et al.. 1988. or Wilson et al.'^ for further details regarding the data collection and aging procedures). The fitting was accomplished by ordinary least squares by using a Nelder-Mead simplex search^ and, as a check, proc NLIN of SAS ( 1990). The least-squares solution is equiva- lent to the maximum likelihood solution when the distri- bution of length at age is normal with constant variance, which seems to be approximately true of this particular data set (Porch, unpubl. data). Akaike's (1973) information criterion (AIC) was used to rank the gi-owth models in terms of their ability to provide statistically parsimonious explanations of the data. The formula for the AIC may be written A/C=-21og(L) + 2p, where L is the likelihood function and p is the number of parameters (see Buckland et al., 1997). In this case, -21og(L) is equal to the residual sums of squares. Wilson, C. A., D. L. Nieland and A. L. Stanley. 1993. Varia- tion of year-class strength and annual reproductive output of red drum Sciaenops ocellatus and black drum Pogonias cromis from the northern Gulf of Mexico. Final Report 1991-1992. 31 p. U.S. Dept. Commerce Cooperative Agi-eement NA90AA- H-MF724. Marine Fisheries Initiative (MARFINi Program. Coastal Fisheries Institute. Louisiana State University, Baton Rouge, La 70803-7503. Shaw, D. E., R. W. M. Wedderburn, and A. Miller 1991. A Program for function minimization using the simplex method. CSIRO, Division of Mathematics and Statistics, P.O. Box 218, Lmdfield, N.S.W. 2070. Australia. Table 1 Akaike's information criteria (AIC) quantifying the fit of the various growth models to the red drum length-at-age data. Smaller AIC values indicate statistically better fits. Model Number of parameters Negative log- likelihood AIC (-25,000) von Bertalanffy 3 16045.9 7098 Richards 4 14169.8 3348 linear von Bertalanffy 4 12883.8 776 double von Bertalanffy 5 12876.6 763 damped (Eq. 6, ^^=0) 5 12651.3 313 seasonal + damped (Eq. 6) 8 12584.9 186 The parameter estimates for the Richards equation tended to be unstable unless good initial estimates were provided. This was accomplished by conducting the esti- mation in two stages. In the first stage the exponent 5 was fixed to 1 and the other parameters were estimated, reduc- ing the Richards equation to the von Bertalanffy form. In the second stage the initial guesses were set equal to the final estimates from the first stage and then all four pa- rameters were estimated simultaneously. Results and discussion All five alternative growth models fitted the data signif- icantly better than the von Bertalanffy equation accord- ing to the AIC statistic (Table 1). The Richards equation, however, did not fit the data nearly as well as the other alternative formulations and suffered from well-known instability problems (Ratkowsky, 1983), therefore it can probably be dropped from any future consideration with respect to red drum. The double von Bertalanffy cui-ve fitted the data better than the linear von Bertalanffy curve, but the comparison is rendered moot by the perfor- mance of the new model. The five-parameter version with- out seasonal oscillations (Eq. 6 with /?.,=0) fitted the data significantly better than either The eight-parameter ver- sion with seasonal oscillations fitted the data significantly better still (see Fig. 1). The estimated seasonal component to the growth rate was fairly substantial initially, having an amplitude at age 0 of 0.301 (k.-,'> and a peak in June, but declined rap- idly with age (Fig. 2). It is possible that an even stronger seasonal signal would have been estimated if age-0 fish, which exhibit the strongest seasonal pattern (Goodyear*), had been adequately represented in the sample. Some of the parameter estimates were highly correlat- ed, as is the case in most growth studies. In particular, the correlations between the estimates for the growth rate and asymptotic length coefficients were typically above 0.8. However, the asymptotic variance-covariance matrix NOTE Porch el a\ A growth model for Sciaenops occllatus 151 ou - A •■'■'>' 40 iiiiMW mm^ 30 iW ■ t||fn?p'r- 20 10- n —> — 1 — ■ — I — 1 — 1 — 1 — 1 — 1— H 1 1 1 1 1 1 10 15 20 25 30 35 40 35 J 30 - 25 20 - 15 -- 10 - 5 B 0 12 3 4 5 Age Figure 1 Fit of the proposed seasonal growth model to red drum length at age data: (A) the fit for all ages; iBi the fit for the younger ages. g o 2 3 4 5 6 Age Figure 2 Growth rate coefficient from seasonal model as a function of age. suggests that the parameters for all of the models were estimated fairly precisely (Table 2). The new model, either with or without the seasonal component, has both practical and theoretical advantages over the four other models examined in this study. By vir- tue of its greater flexibility, it was able to fit the red drum Table 2 Parameter estimates and correspond ing coefficient h 0.412 1 h 0.0530 23 K 0.114 2 h -8.41 3 damped (Eq. 6. A-.=Oi /, 44.1 1 *0 0.0416 7 to 0.362 3 K 0.667 2 A, 0.464 1 seasonal and damped /,, 43.4 1 Ao 0.0475 5 «o 0.443 3 K 0.695 2 K 0.476 1 K 0.301 16 h 0.344 22 t,-' 0.439 3 data significantly better than the linear and double von Bertalanffy curves (its nearest competitors). Moreover, the Richards and linear von Bertalanffy curves are theoret- ically disadvantaged because their parameters have no physical interpretation. The double von Bertalanffy curve, although it has a physical interpretation, suffers because it allows only a single discontinuous change in the growth rate at one age rather than a continuous change through time. For these reasons, the new model should be more widely applicable than the others, particularly for species that change habitat preferences with age or are subject to strong seasonal environmental fluctuations. Acknowledgments We thank C. Legault, G. Scott, S. Turner, D. Vaughan, and an anonymous reviewer for their helpful comments. Literature cited Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In Second international symposium 152 Fishery Bulletin 100(1) on information theory (B. N. Petrov and F. Csaki. eds.), p. 267-281. Akademiai Kiado, Budapest. Beckman, D. W., C. A. Wilson, and A. L. Stanley. 1988. Age and growth of red drum, Sciaenops ocellatus from offshore waters of the Gulf of Mexico. Fish. Bull. 87: 17-28. Buckland, S. T. K. P. Burnham. and N. H. Augustm. 1997. Model selection: an integral part of niference. Bio- metrics 53:603-618. Hoese, H. D., D. W. Beckman, R. H Blanchet, D. Drullinger, and D. L. Nieland. 1991. A biological and fisheries profile of Louisiana red drum Sciaenops ocellatus. Fishery management plan series, number 4, part 1. Louisiana Department of Wildlife and Fisheries, Baton Rouge, LA, 93 p. Murphy, M. D., and R. G. Taylor 1990. Reproduction, growth, and mortality of red di-um, Sciae- nops ocellatus. in Florida waters. Fish. Bull. 88:.531-.542. Ratkowsky, D. A. 1983. Nonlinear regression modeling. Marcel Dekker, New York, NY, 276 p. Richards. F J. 19.59. A flexible growth function for empirical use. J. Exp. Botany 10:290-.300, SAS. 1990. SAS/STATu.sers guide, vol. 2, version 6, 4th ed. SAS Institute Inc., Gary NG, 1686 p. Vaughan, D. S. 1996. Status of the red drum stock on the Atlantic Goast: stock assessment report for 1995. U.S. Dep. Commer. NCAA Tech. Memo. NMFSF-SEFC-380. 50 p. Vaughan, D. S., and T. E. Reiser. 1990. Status of the red drum stock of the Atlantic Goast: stock assessment report for 1989. U.S. Dep. Gommer., NCAA Tech. Memo. NMFS-SEFC-263, 53 p. Fishery Bulletin 100(1) 153 Superintendent of Documents Publications Order Form *5178 I I Yli/O, please send nie the following publ ications: 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. MastarCaid 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 RO. 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. hiolog>'. 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 English 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 oirculated 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 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 1 6th ed.) for editorial style and the most current issue of 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 ( I ). 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 cfted. 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, laboratery 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; Stahnsky 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 forrnat: 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 tOibe 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 anatomical parts (line drawings) or variables (graphs) within figures; use Times Roman bold font to label the dif- ferent sections of a figure (e.g. A, B, C). Figure leg- ends should explain all symbols and abbreviations seen within the figure and should be typed in dou- ble-spaced format on a separate page at the end of the manuscript. We advise authors to peruse a recent issue of Fishery Bulletin for standard for- mats. 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 oiFishery 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 John V. Merriner Scientific Editor, Fishery Bulletin Center for Coastal Fisheries and Habitat Research, NOS 101 Pivers Island Road Beaufort, NC 28516 Once the manuscript has been accepted for pub- licatioUj 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 i. Additional copies may be purchased in lots of 100 when the author receives page proofs. U.S. Department of Commerce Volume 100 Number 2 April 2002 Fishery Bulletin ?■ ?- mi U.S. Department of Commerce Donald L Evans Secretary National Oceanic and AtmospHieric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Acting Assistant Administrator for Fisfieries The Fishery Bulletin (ISSN 0090-06561 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 Js appreciated. f' 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. John V. Merriner Editorial Assistant Sarah Shoffler . Center for Coastal Fisheries and Habitat Research, NOS 101 Pivers Island Road Beaufort, NC 28516 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. Andrew E. Dizon 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 National Marine Fisheries Service University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Sen/ice Universitat Kiel, Germany 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 Fisher^' 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 100 Number 2 April 2002 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 INOAA) or any other agency or institution. The National Marine Fisheries Service (NMFSi does not approve, recommend, or endorse any proprietary product or pro- prietary material mentioned in this pub- hcation. 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 prnpnetary 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 155-167 Brill, Richard, Molly Lutcavage, Greg Metzger, Peter Bushnell, Michael Arendt, Jon Lucy, Cheryl Watson, and David Foley Horizontal and vertical movements of juvenile bluefin tuna (Thunnus thynnus) in relation to oceanographic conditions of the western North Atlantic, determined with ultrasonic telemetry 168-180 Chase, Bradford C. Differences in diet of Atlantic bluefin tuna (Thunnus thynnus) at five seasonal feeding grounds on the New England continental shelf 181-192 Comeau, Michel, and Fernand Savoie Movement of American lobster (Homarus americanus) in the southwestern Gulf of St. Lawrence 193-199 Else, Page, Lewis Haldorson, and Kenneth Krieger Shortspine thornyhead (Sebastolobus alascanus) abundance and habitat associations in the Gulf of Alaska 200-213 Hatfield, Emma M.C., and Steven X. Cadrin Geographic and temporal patterns in size and maturity of the longfin inshore squid (Loligo pealeii) off the northeastern United States 214-227 Hesp, Sybrand A., Ian C. Potter, and Norman G. Hall Age and size composition, growth rate, reproductive biology, and habitats of the West Australian dhufish (.Glaucosoma hebraicum) and their relevance to the management of this species 228-243 Kitada, Shuichi, and Kiyoshi Tezuka Longitudinal logbook survey designs for estimating recreational fishery catch, with application to ayu (Plecoglossus altivelis) 244-257 MacFarlane, R. Bruce, and Elizabeth C. Norton Physiological ecology of juvenile chinook salmon (Oncorhynchus tshawytscha) at the southern end of their distribution, the San Francisco Estuary and Gulf of the Farallones, California Fishery Bulletin 100(2) 258-265 McFee, Wayne E., and Sally R. Hopkins-Murphy Bottlenose dolphin (.Tursiops truncatus) strandings in South Carolina, 1992-1996 266-278 Natanson, Lisa J., Joseph J. Mello, and Steven E. Campana Validated age and growth of the porbeagle shark (Lamna nasus) in the western North Atlantic Ocean 279-298 Olson, Robert J., and Felipe Galvan-Magana Food habits and consumption rates of common dolphinfish (Coryphaena hippurus) in the eastern Pacific Ocean 299-306 Powles, Perce M., and Stanley M. Warlen Recruitment season, size, and age of young American eels (Anguilla rostrata) entering an estuary near Beaufort, North Carolina 307-323 Shima, Michiyo, Anne Babcock Hollowed, and Glenn R. VanBlaricom Changes over time in the spatial distribution of walleye pollock (Theragra chalcogramma) in the Gulf of Alaska, 1984-1996 324-337 Starr, Richard M., John N. Heine, Jason M. Felton, and Gregor M. Cailliet Movements of bocaccio (Sebastes paucispinis) and greenspotted (5. chlorostictus) rockfishes in a Monterey submarine canyon: implications for the design of marine reserves 338-350 Steele Philip, Theresa M. Bert, Kristine H. Johnston, and Sandra Levett Efficiency of bycatch reduction devices in small otter trawls used in the Florida shrimp fishery 351-375 Vaughan, Douglas S., and Michael H. Prager Severe decline in abundance of the red porgy (Pagrus pagrus) population off the southeastern United States Notes 376-380 Abookire, Alisa A., John F. Piatt, and Suzann G. Speckman A nearsurface, daytime occurrence of two mesopelagic fish species (Stenobrachius leucopsarus and Leuroglossus schmidti) in a glacial fjord 381-385 Link, Jason S., and Frank P. Almeida Opportunistic feeding of longhorn sculpin (Myoxocephalus octodecemspinosus): Are scallop fishery discards an important food subsidy for scavengers on Georges Bank? 386-389 Romanov, Evgeny V., and Veniamin V. Zamorov First record of a yellowfin tuna (Thunnus albacares) from the stomach of a longnose lancetfish (Aleplsaurus ferox) 390 Subscription form 155 Abstract— Wi' omplin'od ultrasonic trans- mittLT.s lo follow (for up to 48 h) the hor- izontal and vertical movements of five juvenile (6.8-18.7 kg estimated body mass) bluefin tuna [Thunnus thynntis^ in the western North Atlantic (off the eastern shore of Virginia ). Our objective was to document the fishes' behavior and distribution in relation to ocean- ogi'aphic conditions and thus begin to address issues that currently limit pop- ulation assessments based on aerial surveys. Estimation of the trends in adult and juvenile Atlantic bluefin tuna abundance by aerial sun'eys, and other fishery-independent measures, is con- sidered a priority. Juvenile bluefin tuna spent the major- ity of their time over the continental shelf in relatively shallow water (gen- erally less then 40 m deep). Fish used the entire water column in spite of rela- tively steep vertical thermal gradients (=24°C at the surface and =12°C at 40 m depth), but spent the majority of their time (=90'"r) above 15 m and in water warmer then 20°C. Mean swimming speeds ranged from 2.8 to 3.3 knots, and total distance covered from 152 to 289 km (82-156 nmi). Because fish gen- erally remained within relatively con- fined areas, net displacement was only 7.7-52.7 km (4.1-28.4 nmi). Horizontal movements were not correlated with sea surface temperature. We propose that it is unlikely that juvenile bluefin tuna in this area can detect minor horizontal temperature gi-adients (gen- erally less then 0.5°C/km) because of the steep vertical temperature gi-adi- ents (up to =0.6°C/m) they experience during their regular vertical move- ments. In contrast, water clarity did appear to influence behavior because the fish remained in the intermediate water mass between the turbid and phytoplankton-rich plume exiting Ches- apeake Bay ( and similar coastal waters ) and the clear oligotrophic water east of the continental shelf Horizontal and vertical movements of juvenile bluefin tuna (Thunnus thynnus), in relation to oceanographic conditions of the western North Atlantic^ determined with ultrasonic telemetry Richard Brill Pelagic Fisheries Research Program Joint Institute for Marine and Atmospheric Research School of Earth and Ocean Science Technology University ol Hawaii at Manoa Honolulu, Hawaii 96822 E mail address rbnllia'honlab nmfs hawaii.edu. Molly Lutcavage Edgeiton Research Laboratory New England Aquanum Boston, Massachusetts 02110-3399 Greg Metzger Department of Biology Southampton College Long Island University Southampton, New York 11968-4198 Peter Bushnell Department of Biological Science Indiana University South Bend South Bend, Indiana 46634-7111 Michael Arendt Jon Lucy Sea Grant Program Virginia Institute ol Marine Science College of William and Mary Gloucester Point Virginia 23062-1346 Cheryl Watson Department of Biological Sciences Central Connecticut State University New Britain, Connecticut 06053-2490 David Foley NOAA CoastWatch Program, Hawaii Regional Node Pelagic Fisheries Research Program Joint Institute for Marine and Atmospheric Research School of Earth and Ocean Science Technology University of Hawaii at Manoa Honolulu, Hawaii 96822 Manuscript accepted 6 Julv 2001, Fish. Bull. 100:1,5.5-167 12002). Current estimates of spawning biomass for Atlantic bluefin tuna iThiintnis thynnus) remain controversial (Butter- worth and Punt. 1993; Restrepo et al., 1994; Restrepo, 1996), although the most conservative predicts that a pop- ulation eight times the current size would be needed to produce maximum sustainable yields (Sissenwine et al,, 1998), The current strict catch quotas are based on abundance assessments for both adult and juvenile (i.e, "school- ing") fish (age classes 1-5 years, body mass =6-60 kg). Adult abundance is derived from commercial landings data; juvenile abundance has, since 1985, been based on fishing effort and land- ings data obtained from dockside inter- cepts and telephone polling of the largely recreational fishery for juvenile bluefin tuna conducted by the National Marine Fisheries Service's Large Pelag- ics Survey (Turner et al,, 1993, 1997), The usefulness of both data sets can be compromised, however, because the relationship between catch-per-unit-of- effort (CPUE) data and real abundance is not known with certainty (Bakun et al,, 1982; Hilborn and Walters, 1992; Lauck. 1996), This problem is especially critical with highly mobile schooling fishes like tunas because of environ- mental influences on fish distribution and vulnerability to specific fishing gears, as well as the introduction of new fishing techniques (Sharp, 1978; Clark and Mangel, 1979; Brill, 1994; Bertrand and Josse, 2000). Juvenile Atlantic bluefin tuna appear in the surface waters off the east coast 156 Fishery Bulletin 100(2) 40"N United States ^^ *^ 35 30 / N J A 25 V4 A \ "^ -^^ Fish 3 & Fish 4 \5 // ?^ "^^5 ' 85 80 75 70 76 Figure 1 (A> Map of the east coast of the United States. The rectangle shows the area enlarged in panel B. (B) Movements of five juvenile bluefin tuna tThunnun thynnui^). The limit of the continental shelf is shown by the 50-, 100-, and 200-m isobath lines. The topographic features considered by local fisher- men to aggregate juvenile bluefin tuna are shown by the shaded areas. Place names are taken from local fishing charts. The rectangles show the areas enlarged in Figures 3 and 4. of the United States, from North Carohna to Rhode Island, usually during June and July (Rivas, 1978; Sakagawa, 1975; koffer, 1987; Lucy et al.," 1990; Mather et al, 1995 1. Their presence provides an opportunity for direct popu- lation assessments with aerial surveys similar to those conducted on adult Atlantic bluefin tuna (Lutcavage and Ki'aus, 1995; Lutcavage et al., 1997), southern bluefin tu- na tTluininis maccoyii).^ and other fish species (e.g. Lo et al., 1992). Assessments of juvenile bluefin tuna abundance are considered particularly crucial for effective stock man- agement because these will allow the forecasting of re- cruitment and long-term population trends iPolacheck et al., 1996; Sissenwine et al., 1998). There is, however, a need to establish the probability of detecting schools and estimating school size before aerial survey data can pro- vide robust population assessments. This need is present regardless of whether the census techniques are simple photography (Lutcavage and Kraus, 1995; Lutcavage et al., 1997) or new laser-based digital remote sensing tech- niques (Oliver et al., 1994; Lo et al., 1999). As with tra- ditional CPUE-based abundance estimates, knowledge of the effects of oceanographic conditions on depth distribu- tion, surfacing frequency, travel speeds, and residence pat- terns is critical because these conditions will affect vul- nerability to "capture," either on photographic film or as digital data. To obtain the necessary data, we undertook a study of the horizontal and vertical movements of juvenile Atlan- Cowling, A., C. Millar, and T. Polacheck. 1996. Data analysis of the aerial surveys (1991-19971 for juvenile southern bluefin tuna in the Great Australian Bight. Rep. RMWS/96/4, 87 p. Recruitment Monitoring Program, CSIRO Division of Marine Research, GPO Box 1538, Hobart 7001, Australia. tic bluefin tuna using depth sensitive ultrasonic telemetry devices. LHtrasonic telemetry is a proven technique for ac- quiring the required precise and detailed data on the be- haviors of pelagic fishes in relation to oceanographic condi- tions (e.g. Holland et al., 1990; Dagorn et al., 1999, 2000a; Lutcavage et al., 2000). Besides being useful for improv- ing stock assessments (Brill and Lutcavage, 2001), the re- sultant data can also help clarify basic ecological relation- ships and provide inferences on physiological abilities and species-specific behaviors (Carey, 1983; Brill. 1994; Brill et al., 1999). Materials and methods Fishing operations were conducted from a 16-m commer- cial fishing boat (FV Gruryipy) in the western North Atlan- tic off the eastern shore of Virginia (Fig. lA) during June and July 1998. Bluefin tuna were captured with standard recreational trolling gear. The fish were brought aboard with a plastic sling and detached from hooks. Straight line fork length was measured, and a Vemco (Halifax, Nova Scotia, Canada) ultrasonic transmitter (model V32) was attached near the second dorsal fin with nylon straps as described by Holland et al. ( 1986, 1990). The transmit- ted signal was detected with a Vemco VR-60 ultrasonic receiver connected to a directional hydrophone mounted on the end of an aluminum pipe. The pipe was clamped to the side of the vessel with a custom designed alumi- num bracket that allowed the hydrophone to be rotated to find the relative bearing to the transmitter Fish depth, encoded by the interval of the transmitters pulsed signal, was decoded by the receiver and the resultant digital data recorded by an attached laptop computer. Geographic Brill et a\: Horizontal and vertical movements of iiivenlle Thunnw, thynniK 157 Table 1 Summary of tracks Body masses were of'fiv :alcu] ? juvenile Atlantic bluofin ated from fork lengths wi tuna th the Tlumiuts thynnus) equipped with ultrasonic depth sensitive transmitters, weight-length regression equation provided by Coull el al. (1989). Kish 11(1. Fork loiigth cm Body mass kg Dates of trai (1988) k Duration of t rack h Total distance covered km (nmi) Distance between start and end points km (nmi) Mean(±SEM) swimming speed knots 1 74 6.7 17-19 Jun 47.2 217(117) 52.7(28,4) 2.8 ±0.03 2 91 12.1 23-25 Jun 47.8 267(144) 11.4(6.1) 3.0 ±0.03 3 79 8.0 2-3 Jul 30.0 152(82) 7.7(4.1) 2.7 ±0.04 4 99 15.4 6-7 Jul 31.2 192(104) 14.7(7.9) 3.3 ±0.03 5 106 18.8 10-12 Jul 47.9 289(156) 32.2(17.4) 3.2 ±0.03 positions were obtained by using a GPS satellite receiver and were recorded on a second laptop computer every minute. The tracking vessel's position was assumed to be the same as that of the fish. Sea surface temperature and bottom depth were recorded manually every 15 minutes by using a hull-mounted electronic temperature sensor and color fathometer, respectively. Depth-temperature profiles were taken approximately every four hours with a Sippican (Marion, MA) portable XBT system (model MK12). Aggregate time-at-depth and time-at-temperature dis- tributions were calculated fromlO-m and 1°C bins (respec- tively), as described by Holland et al. (1990). These data were subsequently expressed as a fraction of the total time each fish was followed, and the fractional data bins were averaged across all fish. Speed over ground (henceforth re- ferred to simply as "speed") was calculated by assuming that the fish moved in a straight line between successive geographic locations. Sea surface temperature (SST) data were recorded by the advanced very high resolution radiometer (AVHRR) carried onboard the NOAA-14 polar orbiting operational environmental satellite. High resolution picture transmis- sion (HRPT) data were obtained from the National Coast Watch Active Access System at the National Oceano- graphic Data Center and had a spatial resolution of 1.25 x 1.25 km pixels. Ocean color data were recorded by the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) car- ried onboard the Orbview-2 spacecraft (Orbimage, Inc., Dulles. VA). The level-2 global area coverage (GAC) data were obtained from the NASA Goddard Space Flight Cen- ter's Distributed Active Archive Center. These 4-km reso- lution data sets included chlorophyll-a surface concentra- tion and the diffuse attenuation coefficient at 490 nmi, m vacuo. We calculated the occupancy of waters with specific chlorophyll-a concentrations and light attenua- tions from values corresponding to and coincident with the tracks of fishes derived from satellite images. These data were subsequently expressed as a fraction of the to- tal number of observations for each fish, and the fraction- al data bins were averaged across all fish. For illustrative purposes, we also generated composite images using data from the 21-day period over which all tracking operations were conduced. Results The bottom topography in the areas where the fish were tracked is generally featureless, except for small areas where the vertical relief is approximately 2 m above the surroundings. Local fishermen have named these features (Fig. IB and subsequent figures), and the names used in this study are taken from local fishing charts. Size of fish, starting and ending dates of tracks, dura- tion of tracks, distances covered, distance between start- ing and ending points, and mean (±SEM) swimming speed offish are listed in Table 1. With the exception offish num- ber 4 (referred to simply as "fish 4"), individuals tended to follow highly irregular courses that often repeatedly covered the same areas (Fig. IB). The mean distance be- tween starting and ending points for all fish was only ll'7f (range: 4— 25'/f ) of the total distance covered (Table 1). From tracking studies of yellowfin tuna [Thitnniis alba- cares) in the Pacific, Dagorn et al. (2000a) concluded that such frequent directional changes might be characteristic of foraging behavior. The frequency of observed swimming speeds is shown in Figure 2. Although all fish reached maximum speeds of =7 knots for brief periods, over 90% of the observed speeds were less than 3.6 knots. Horizontal movements Fish 1 was captured and released at 1340 h, approximately 1.8 km (1 nmi) west of the "26 Mile Hill" (Fig. 3). It pro- ceeded on a southerly course for about 33 h, a direction that carried it over the "Hot Dog" and "Southeast Lumps." After sunset on the second day, the fish reversed its course and eventually recrossed both features. The fish was approxi- mately 5.6 km (3 nmi) south of the "Southeast Lumps," and moving south, when the track was terminated at 1300 h. Fish 2 was captured approximately 5.6 km (3 nmi) south of the "Southeast Lumps" (Fig. 3) at 1547 h, adjacent to where the track of fish 1 was completed four days earlier. It 158 Fishery Bulletin 100(2) 2 - 0 I 1 I U I I H I I I ]\ 1 I 11 L I, jW^. O- Q O- N- N- nk^T^- T/- 'b-^ t«- tx- tK- vnA^ ■n*^ ■^ v / — 1 — V y H 1 — H 1 — 1 — \— 1X:>-- 1 H — 1 — -1 1- — 1 — — h- 4 c -- 3 -- 2 in 1400 1800 2200 0200 0600 1000 1400 1800 8 12 16 20 24 28 Temperature ( 'C) Figure 6 Swimming speed (solid line, upper panel) and vertical movements (lower panel) of fish 4. The change in tempera- ture in the horizontal direction (expressed as sea surface temperature, SST) is shown by the broken line in the upper panel (Brill and Lutcavage, 2001). As in Figure 5. the change in temperature in the vertical direction (mean ±SEM) is shown to the right of the vertical movement plot. Note that changes in swimming speed are not cor- related with changes in SST, and that the steepest temperature change the fish could experience moving horizon- tally (generally less then 0.5°C/km) is several orders of magnitude less then that experienced moving vertically (=0.6°C/m). occur throughout the water column during daylight, are abundant in the areas where we tracked the fish, and dominate the diet of tunas in this area (Mason, 1976; Egg- leston and Bochenek, 1989). The nature offish 4's descents up to =160 m while off the continental shelf (Fig. 6) re- main unclear, although they too may be related to foraging (Dagorn et al., 2000a, 2000b). Their brevity is most likely due to the inability of Atlantic bluefin tuna to withstand temperatures below 10°C for long periods of time, rather than to an intolerance of low ambient oxygen conditions. Although no depth-oxygen profiles were obtained during our study, available data- suggest that juvenile tuna did not encounter ambient oxygen levels that were likely to be stressful (Bushnell and Brill, 1991, 1992). The behavior pattern we observed of short oscillatory dives near the surface is similar to that of both juvenile bluefin tuna in the eastern Pacific (Marcinek et al., 2001) and adult bluefin tuna tracked in the Gulf of Maine (Lut- cavage et al., 2000). In all cases, fish spent the majority of their time in the surface layer, although in the Gulf of Maine and eastern Pacific, the temperature of the warm- est water available was lower (=13-22°C) and more vari- able. As shown in Figure 11, when expressed as the rel- ative change in temperature with depth (i.e. in relation to the surface water temperature occurring during each track), time-at-temperature distributions of juvenile and adult Atlantic bluefin tuna become essentially identical. Moreover, the limiting effects of temperature change on vertical movements are independent of body size. Simi- larly, yellowfin tuna tracked near the main Hawaiian Is- lands and off the coast of California occupy the warmest water available, regardless of body size, even though sur- face water temperature in the two areas differs by more than 5°C (Holland et al., 1990; Block et al., 1997; Brill et al., 1999). Atlantic bluefin tuna, however, are more eury- thermal than yellowfin tuna. The latter will rarely expose themselves to more than an 8°C change in temperature, whereas the former regularly subject themselves to a tem- perature change of up to 13°C (Fig. 11). Surprisingly, the behavior of juvenile bluefin tuna observed by Marcinek et al., (2001) was more like that of yellowfin tuna in that these juvniles would not expose themselves to more then an 8°C temperature change. It still remains to be conclusively demonstrated, how- ever, whether the vertical movement patterns of tunas and other large pelagic fishes are (as suggested by Brill et al., 1993, 1999) limited by the effects of ambient temperature on cardiac function. Or whether, as suggested by Mar- cinek et al. (2001), that depth distributions "... may have more to do with the location of prey, and the physiological limitations of the prey, than physiological limitations of the bluefin Itunaj." Moreover, months-long observations 162 Fishery Bulletin 100(2) 30 20 10 0 10 Time at deptti (%) 20 10 0 10 20 30 40 Time at temperature (%) Figure 7 Vertical distribution of five juvenile bluefin tuna ex- pressed as percent time (mean ±SEMi spent at specific deptbs I Ai and at specific temperatures iBi. Shaded bars indicate nighttime and open bars indicate daytime. of juvenile bluefin tuna in the western Pacific recently ob- tained with archival (i.e. electronic data recording) tags have shown that the vertical movements of juvenile blue- fin tuna can have strong seasonal and geographic com- ponents (Kitagawa et al., 2000). In areas and at times (e.g. winter) when there was a strong thermocline, bluefin tuna remained in the uniform-temperature surface layer and demonstrated vertical movement behaviors similar to those observed during the short-term ultrasonic tele- metry studies. In the summer, when the themocline was less pronounced, the fish showed very distinct diel peri- odicity in their vertical movement patterns. They would remain at the stirface at night and make rapid vertical movements (from surface to =120 m and from =21''C to 14°C) during the day. Kitagawa et al. (2000) concluded that the differences in behavior patterns were related to foraging. It is also still an open question as to what ex- tent bluefin tuna's ability to conserve metabolic heat and maintain elevated muscle temperatures (Carey and Teal, 1966) enhances vertical mobility. Roffer (1987) was apparently the first to propose that movements and abundance of juvenile Atlantic bluefin tuna are controlled by the depth and thickness of the 18.5-20.5°C "preferred habitat" temperature layer Likewise, Inagake et al. (2001), using archival tags implanted into juvenile 37,8 37.6 - 75 8 75.6 75 4 75.2 75.0 74 8 74 6 74 4 74 2 74.0 Figure 8 Composite satellite sea surface temperature image (17 June-10 July 1998) and movements of the five juvenile bluefin tuna (Brill and Lutcavage, 2001 ). Figure reprinted with permission of Ameri- can Fisheries Society. bluefin tuna in the western Pacific, found evidence that this temperature range is indeed always "preferred." Dur- ing the periods of our observations, juvenile bluefin tuna spent the majority of their time (=809c) in water greater then 22°C. A plausible explanation is that under the condi- tions of our observation period, juvenile bluefin tuna simply occupy the warmest water available, although a relatively uniform temperature surface layer was evident only dur- ing tracks of fish 3, 4, and 5. We also did not find any con- clusive indication that juvenile bluefin tuna avoided sur- face water temperatures above 26°C. Although fish spent less than 'ZO'^i of time at these temperatures (Fig. 7), less than 20% of the recorded sea surface temperatures (i.e. the warmest water available) were above 26°C. We also found no relationship between sea surface tem- perature and horizontal movements (Figs. 6 and 8), al- though this relationship has been demonstrated for other tuna species in other areas (e.g. Laurs et al., 1977; Fiedler and Bernard, 1987; Uda, 1973). We argue that our results are due to the differences in the vertical and horizontal temperature gradients occurring along the Virginia coast. Juvenile bluefin tuna routinely traveled through the ther- mocline, moving from the relatively warm surface layer into the mid-Atlantic cold-pool water (Houghton et al., 1982; Houghton and Marra, 1983) underlying it. The fish thus experienced temperature gradients of up to =0.6°C/m (Figs. 5 and 6). In contrast, the steepest horizontal tem- perature gradient in the area where the fish were tracked was approximately three orders of magnitude smaller (=0.5°C/km). In other words, the frequent vertical move- ments of juvenile bluefin tuna probably prevent them from detecting and responding to SST gradients. Brill et al.: Horizontal and vertical movements of juvenile Thunnus thynnus 163 diffuse attenuation poefficienl. 490 nm ^^002 [comoos !e r^age A^^^^fi July 1998) ^ |B>>Bin 75.8 75 6 75 4 ,■. . ■ - 74 8 74.6 74 4 /4 ^' 74.0 Figure 9 Composite satellite images (17 June-lO July 1998) showing (A) chloro- phyll-Q concentrations (mg/m') and (B) water clarity measured as the diffuse attenuation coefficient (1/m, at an in vacuo wavelength of 490 nmi) and movements of the five juvenile bluefin tuna. Locations of juve- nile bluefin tuna schools recorded during aerial surveys conducted in 1997 are shown by filled circles. (Lutcavage, M. 1998. Aerial survey of school bluefin tuna off the Virginia Coast, July 1997. Report to the National Marine Fisheries Service, cooperative agreement NA77fm0.533. (Available from the author, Edgerton research Laboratory. New England Aquarium, Central Wharf Boston, MA 02110].) are shown by filled cir- cles. The edge of the continental shelf is indicated by the .50-, 100-. and 200-m isobath lines (Brill and Lutcavage, 2001). Figure reprinted with permission of American Fisheries Society. Carey ( 1992) was one of the first to appreciate the impor- tance of vertical thermal structuring and stated "Temper- ature gradients of 15° to 20°C are not uncommon within the depth ranges of pelagic fish. By moving a few hundred meters vertically, an animal may encounter a greater tem- perature change than it experiences seasonally or in mov- ing thousands of miles horizontally." As with bluefin tuna, the vertical movements of yellowfin tuna and swordfish also result in their experiencing vertical temperature gra- dients orders of magnitude greater than horizontal tem- perature gradients (Carey and Robison, 1981; Carey, 1990; Holland et al., 1990; Cayre and Marsac, 1993). The inabil- 164 Fishery Bulletin 100(2) «D N "^ 1 "^ C> ^ Si- C) O O C) O C) S 60 1 Log (chlorophyll-a) (mg/m) Diffuse attenuation coefficient (490 nmi/m) Figure 10 Frequency histogram (mean ±SEM) of chlorophyll-o concentrations ( mg/ni'^ i and water clarity measured as the diffuse attenuation coefficient ( 1/m, at an in vacuo wavelength of 490 nmi ) in waters along the track lines of five juvenile bluefin tuna (Brill and Lutcavage, 2001). The span of the horizontal axes show the approximate range of these variables present off the eastern shore of Virginia. ity of fish to sense shallow horizontal temperature gra- (Jients in the face of the steep vertical temperature gra- dients they routinely experience may explain, therefore, why Power and May (1991) and Podesta et al. (1993) could find no correlation between SST "fronts" and the appar- ent abundance of yellowfin tuna in the Gulf of Mexico and swordfish in the western north Atlantic. In contrast to SST, water clarity and phytoplankton abundance appear to have a strong influence on the hor- izontal movements of juvenile bluefin tuna (Figs. 9 and 10). Tunas are sight hunters, and possess the highest vi- sual acuity of any teleost (Nakamura, 1968). We suspect that juvenile bluefin tuna remain in water masses with a standing phytoplankton biomass sufficient to support con- centrations of prey, but where turbidity is low enough that visual prey detection and prey capture abilities are not impeded. Our conclusion is further supported by the lo- cations of juvenile bluefin tuna schools detected in aerial 50 40 30 20 10 ■! adult Atlantic bluefin tuna JjjWL..^ -? 60 -I o I 50 I 40 Cl I 30 ■ t 20- E P 10 0 60 50 40 30 20 10 T> f>' ■?> ^*• <^' N*' ■i^' -y v>' H' :S > 53 fe ^ ,* ?l kS> N^ 0- v> n"* <0^ C^ <^ <$- <^ C^ <^ '^ <^^^^^ ^ Temperature interval ( C) Figure 11 Frequency histograms (mean +SEM) showing time spent at specific temperatures by adult bluefin tuna tracked in the Gulf of Maine (western North Atlantic) with tem- peratures expressed as water temperature (A), and with temperatures expressed in relation to surface layer tem- perature (B), data taken from Lutcavage et al., 2000). Equivalent data for juvenile bluefin tuna are presented in panel C. Shaded bars indicate nighttime and open bars indicate daytime. surveys conducted in 1997.^ Although satellite data show- ing diffuse attenuation coefficients and chlorophyll-a con- centrations are not available for 1997, bluefin tuna schools were located in the areas where the fish carrying ultra- sonic transmitters remained (Fig. 9). Olson and Podesta ( 1987), Olson et al. ( 1994), and Humston et al. (2000) have also concluded that aggregations of highly mobile species Lutcavage, M. 1998. Aerial survey of school bluefin tuna off the Virginia Coast, July 1997. Report to the National Marine Fisheries Service (cooperative agreement NA77FM0533). (Avail- able from the author, Edgerton Research Laboratory, New Eng- land Aquarium, Central Wharf Boston. MA 02110,1 Brill et a! Horizontal and vertical movements of luvenlle Thunnus thynnus 165 at fronts result from cues other than SST, such as changes in the photic environment associated with phytoplankton distribution, changes in prey abundance, or enhanced for- age opportunities. Aerial survey techniques and population assessments of juvenile bluefin tuna Techniques for interpretation of aerial survey data with respect to population assessments are complex (e.g. Lo et al.. 1999; Newlands and Lutcavage, 2001), and a thorough discussion is beyond the scope of our present study. We can, however, use our data on juvenile bluefin tuna's verti- cal movements and distribution patterns to provide some inferences as to how often they are likely to be visible at the ocean's surface or detectable at a specific depth. Juve- nile bluefin tunas spent less than 13'7c of daylight hours at depths of 0-3 m (Fig. 7), where visual or photographic detection is possible. The depth distribution of juvenile fish was similar to that of adult bluefin tuna tracked in the Gulf of Maine (12*^ of daylight hours at depths of 0-4 m; Lutcavage et al.. 2000). Abundance estimates based solely on photographic data will, therefore, have to be corrected to account for the significant number offish that maybe be present, but that are beyond detection range. Fish detec- tion systems that use lasers (the so called "light detection and range" or "LIDAR" systems) are expected to have a depth detection zone of up to 60 m (Oliver et al., 1994). This detection zone encompasses almost the entire water column over the sections of continental shelf where juve- nile bluefin tuna are likely to be found. Moreover, if the behavior of the fish that moved into deeper water off the continental shelf is assumed typical, then juvenile bluefin tuna would be detected by LIDAR systems even in deep water The relatively small net displacement distance (i.e. distance between start and end points. Table 1) may require the development of filtering algorithms to reduce errors caused by double counting if parallel transects are flown less than =50 km apart, or if the same area is resur- veyed weekly or more often. Conversely, significant fish aggregations could be missed if parallel transects are too widely spaced. Acknowledgments This project was funded by a grant from the National Marine Fisheries Service to the Edgerton Research Labo- ratory. New England Aquarium. RWB's participation was funded through cooperative agreements NA37R.J0199 and NA67RJ0154 from the National Oceanic and Atmospheric Administration with the Joint Institute for Marine and Atmospheric Research, University of Hawaii. We grate- fully acknowledge Mark Luckenbeck and the staff and students of the Virginia Institute of Marine Science's East- ern Shore Laboratory for their gracious hospitality and extraordinary efforts to make this project a success. We also thank Jim Hannon and Sippican Inc. (Marion, MA) for use of their Mark 12 system and generous donation of XBT probes, and Mitch Roffer for access to historical bluefin tuna data. We acknowledge the DAAC at NASA's Goddard Space Flight Center and Orbimage Inc. for ocean color data and the NOAA Coast Watch Program and the National Oceanographic Data Center for SST data. We especially thank Captain Jack Stallings of the FV Grumpy (Virginia Beach, Virginia) for his unflagging enthusiasm and his significant contributions to making our project a success. Literature cited Bakun. A., J. Beyer. D. Pauly, J. G. Pope, and G. D. Sharp. 1982. Ocean sciences in relation to living resources. Can. J. Aquat. Sci. 39:1059-1070. Bertrand. A., and E. Josse. 2000. Acoustic estimation oflongline tuna abundance. ICES (Int. Council for the Exploration of the Sea) J. Mar. Sci. .57:919-926. 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 iTInin- nus albacares) at the northern extent of their range. Mar. Biol. 130:119-132. Brill, R.W. 1994. A review of temperature and oxygen tolerance stud- ies of tunas pertinent to fisheries oceanography, movement models, and stock assessments. Fish. Ocean. 3:204-216. Brill, R.. B. Block, C. Boggs. K. Bigelow, E. Freund, and D. Marcinek. 1999. Horizontal movements and depth distribution of large adult yellowfin tuna iThunnus albacares) near the Hawai- ian Islands, recorded using ultrasonic telemetry: implica- tions for the physiological ecology of pelagic fishes. Mar Biol. 133:395-408. 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 {Tetrapturusaudax) near the main Hawaiian Islands, deter- mined by ultrasonic telemetry, with simultaneous mea- surement of oceanic currents. Mar Biol. 117:567-574. Brill, R. W., and M. E. Lutcavage. 2001. Understanding environmental influences on move- ments and depth distributions of tunas and billfishes can significantly improve population assessments. In Island in the stream: oceanography and fisheries of the Charles- ton Bump (G. Sedberry ed.), p. 179-198. Am. Fish. Soc. Symp. 25. Am. Fish. Soc, Bethesda, MD. Bushneil, P.G., and R.W. Brill. 1991. Responses of swimming skipjack [Katsuwonus pela- mis) and yellowfin (Thunnus albacares) tunas exposed to acute hypoxia, and a model of their cardio-respiratory func- tion. Physiol. Zool. 64:787-811. 1992. Oxygen transport and cardiovascular responses in skipjack tuna (Katsuwonus pelamis) and yellowfin tuna iThunnus albacares) exposed to acute hypoxia. J. Comp. Physiol. B 163:131-143. Butterworth D. S., and A. E. Punt. 1993. The robustness of estimates of stock status for the western north Atlantic bluefin tuna population to violations of the assumptions underlying the associated as.sessments models. ICCAT (Int. Commission for the Conservation of Atlantic Tunas) Coll. Vol. Sci. Pap. 43(1): 192-210. 166 Fishery Bulletin 100(2) Carey, F. G. 1983. Experiments with free-swimming fish. In Oceanog- raphy— the present and future ( P. G. Brewer, ed. I, p. 57-68. Springer- Verlag, New York, NY. 1990. Further acoustic telemetry observations of swordfish. In Plannmg the future of billfishes ( R. H. Stroud, ed. ), p. 102- 122. National Coalition for Marine Conservation Inc., Savanna, GA. 1992. Through the thermochne and back again. Oceanus 35:79-85. Carey, F. G., and B. H. Robison. 1981. Daily patterns in the activities of swordfish, Xi'p/j/as gladius. observed by acoustic telemetry. Fish. Bull 79:277- 292. Carey FG., and. J. M. Teal. 1966. Heat conservation in tuna fish muscle. Proc. Natl. Acad. Sci. 56:1464-1469. Cayre P., and F Marsac. 1993. Modelling the yellowfin tuna iThunnus albacares) vertical distribution using sonic tagging results and local environmental parameters. Aquat. Living Resour. 6:1-14 Churchill, J. H., E. R. Levme, D. N. Connors, and P. C. Cornillon. 1993. Mi.xing of shelf slope and Gulf Stream water over the Continental Slope of the Middle Atlantic Bight. Deep Sea Res. 40:1063-1065. Clark, C. W., and M. Mangel. 1979. Aggregation and fishery dynamics: a theoretical study of schooling and the purse seine tuna fisheries. Fish. Bull. 77:317-329. Coull, K. A., A. S. Jermyn, A. W. Newton, G. I. Henderson, and W. B. Hall. 1989. Length/weight relationships for 88 species offish en- countered in the North Atlantic. Scottish Fish. Res. Rep. 43, 80 p. Dagorn, L., P. Bach, and E. Josse. 1999. Movement patterns of large higeye tuna [Tli tin n us obe- sus) in the open ocean, determined using ultrasonic tele- metry Mar. Biol. 1.36:361-371. Dagorn, L., E. Josse, and P. Bach. 2000a. Individual differences in horizontal movements of yellowfin tuna iThunnus albacares) in nearshore areas in French Polynesia, determined using ultrasonic telemetry. Aquat. Living Resour 13:19.3-202. Dagorn, L., F. Menczer, P. Bach, and R. J. Olson 2000b. Co-evolution of movement behaviors by tropical pelagic predatory fishes in response to the environment: a simulation model. Ecol. Model. 134:325-341. Eggleston, D .B., and E. A. Bochenek. 1989. Stomach contents and parasite infestation of school bluefin tuna Thunnus thynnus collected from the Middle Atlantic Bight, Virginia. Fish. Bull. 88:389-395. Fiedler, R C, and H. .J. Bernard. 1987. Tuna aggregations and feeding near fronts observed in satellite imagery Continental Shelf Res. 7: 871-881. Hilborn, R., and C. .J. Waters. 1992. Quantitative fisheries stock assessment: choice, dy- namics and uncertainty. Chapman and Hall. New York, NY, 570 p. Holland, K. N.. R.W. Brill, J. S. Ferguson, R. K. C. Chang, and R. Yost. 1986. A small vessel technique for tracking pelagic fish. Mar Fish. Rev 47:26-32. 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. Houghton R. W., and J. Marra. 1983. Physical/biological structure and exchange across the thermochne ShelfSlope in the New York Bight. J. Geo- phys. Res. (C Oceansi 88:4467-4480. Houghton, R. W., R. Schultz, R. C. Beardsley B. Butman, and .J. L. Chamberlin. 1982. The middle Atlantic Bight cold pool: evolution of the temperature structure during summer 1979. J. Phys. Oceanogr. 12:1019-1029. Humston, R., J. S. Ault, M. Lutcavage, and D. B. Olson. 2000. Schooling and migration of large pelagic fishes rela- tive to environmental cues. Fish. Oceanogr. 9: 136-146. Inagake, D., H. Yamada, K. Segawa, M. Okazaki, A. Nitta, and Tltoh. 2001. Migration of young bluefin tuna, Thunnus orientalis Temminck et Schlegel, through archival tagging experi- ments and its relation with oceanographic conditions in the western north Pacific. Bull. Natl. Res. Inst. Far Seas Fish. 38:.53-81. Kitagawa, T, H. Nakata, S. Kimura, T Itoh, S. Tsuji, and A. Nitta. 2000. Effect of ambient temperature on the vertical dis- tribution and movement of Pacific bluefin tuna Thunnus thynnus orientalis. Mar Ecol. Prog. Sen 206:251-260. 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. Lauck, T 1996. LIncertainty in fisheries management. In Fisheries and uncertainty: a precautionary approach to resource management (D. V. Gordon and G. R. Munro, eds.), p. 91- 105. Univ. Calgary Press, Calgary. 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. Biol. 75:.347-355. Lo, N. C. H., J. R. Hunter, and J. H. Churnside. 1999. Modeling statistical performance of an airborne LIDAR survey system for anchovy. Fish. Bull. 98:264282. Lo, N. C. H., L. D. Jacobson, and J. L. Squire. 1992. Indices of relative abundance from fish spotter data based on delta-log normal models. Can. J. Fish. Aquat. Sci. 49:251.5-2526. Lucy, J. A., E. A. Bochenek, and N. J. Chartier. 1990. Fleet characteristics and boat-owner expenditures associated with Virginia's recreational marlin-tuna fishery. In Planning the future of billfishes: research and manage- ment in the 80's and beyond: proceedings of the second international, billfish symposium, part 2 (R. H. Stroud, ed.) p. 253-262. Marine Recreational Fisheries 13. National Coalition for Marine Conservation, Savannah. GA. Lutcavage, M. E., R. W. Brill, G. B. Skomal, B. C. Chase, J. L. Goldstein, and J. Tutem. 2000. Movements and behavior of adult North Atlantic blue- fin tuna {Thunnus thynnus) in the northwest Atlantic deter- mined using ultrasonic telemetry. Mar. Biol. 137:347-358. Lutcavage M.. and S. Kraus. 1995. The feasibility of direct photographic assessment of giant bluefin tuna in New England waters. Fish. Bull. 93: 49,5-503. Lutcavage, M., S. Kraus, and W. Hoggard. 1997. Aerial assessment of giant bluefin tuna in the Bahama Banks-Straits of Florida, 1995. Fish. Bull. 95:300-310. Brill et a\: Horizontal and vertical movements of luvenile Thunmis thynnus 167 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. Marsac. F., and P. Cayre. 1998. Telemetry applied to behavior analysis of yellowfin tuna iThunnus alhoca/Vf:. Bonnaterre. 17881 movements in a network of fish aggi-egating devices. Hydrobiolgia 371/ 372:155-171. Mason. J. M. 1976. Food of small, northwestern Atlantic bluefin tuna, Thunnus thynnus (L), as ascertained through stomach con- tent analysis. M.S. thesis, Univ. Rhode Island, Kingston, RI, 27 p. Mather, F. J., J. M. Mason, and A. C. Jones. 1995. Historical document: life history and fisheries of Atlantic bluefin tuna. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-370, 165 p. Nakamura, E. L. 1968. Visual acuity of two tunas, Katsuivonus pelatnis and Euthynnus affinis. Copeia 1968:41-49. Newlands, N., and M. Lutcavage. 2001. From individuals to local population densities: move- ments of north Atlantic bluefin tuna (Thunnus thynnus) schools in the Gulf of Maine/Northwest Atlantic. In Elec- tronic tagging and tracking in marine fish (J. Sibert and J. L. Neilsen, eds.), p. 421-433. Kluwer Academic Pubis.. Dordrecht. Oliver C. W., W. A Armstrong, and J. A. Young. 1994. Development of an airborne LIDAR system to detect tunas in the Eastern tropical Pacific purse-seine fishery U.S. Dep. Commer , NOAA Tech. Memo. NMFS-SWFSC-204, 67 p. Olson. D. B., G. L. Hitchcock, Mariano, A. J.. C. J. Ashjian, G. Peng, R. VV. Nero, and G. P. Podesta. 1994. Life on the edge: marine life and fronts. Oceanogra- phy 7:52-60. Olson, D. B., and G. P. Podesta. 1987. Oceanic fronts as pathways in the sea. In Signposts in the sea (W. F. Hernkind and A. B. Thistle, eds. I, p. 1-15. Florida State Univ. Press, Tallahassee, FL. Polacheck. T. E. Pikitch. and N. Lo. 1996. Evaluation and recommendations for the use of aerial surveys in the assessment of Atlantic bluefin tuna. ICCAT Coll. Vol. Sci. Pap. 68( 1 ):61-68. Podesta, G. P., J. A. Browder, and J. J. Hoey. 1993. Exploring the association between swordfish catch rates and thermal fronts on U.S. longline grounds in the western North Atlantic. Continental Shelf Res. 13:253- 277. Power, J. H., and L. N. May Jr 1991. Satellite obsen'ed sea-surface temperatures and yellowfin tuna catch and effort in the Gulf of Mexico. Fish. Bull. 89:429-439. Reiss, C. S., and J. R. McConaugha. 1999. Cross-frontal transport and distribution of ichthyo- plankton associated with Chesapeake Bay plume dynam- ics. Continental Shelf Res. 19:151-171, Rcsprepo, V. 1996. An implementation of an age-structured production model with application to western Atlantic bluefin tuna fisheries. ICCAT Coll. Vol. Sci. Pap. 66(21:348-353. Restrepo, V. R., C. E, Porch, S. C.Turner, G. P. Scott, and A. A. Rosenberg. 1994. Combination of spawner-recruil. spawning biomass- per-recruit and yield-per-recruit computations for the esti- mation of the long term potential for west Atlantic bluefin tuna. ICCATColl.Vol. Sci. Pap. 42(1 ):214-222. Rivas, L. R. 1978. Preliminary models of annual life history cycles of the north Atlantic bluefin tuna. In The physiological ecology of tunas (G. D. Sharp and A. E. Dizon, eds.), p. 369-393. Academic Press. New York, NY. RofTer. M. A. 1987. Influence of the environment on the distribution and relative apparent abundance of juvenile Atlantic bluefin tuna along the United States east coast. Ph.D. diss. Univ. Miami, Miami. FL, 154 p. Sakagawa, G.T 1975. The purse seine fishery for bluefin tuna in the north- western Atlantic Ocean. Mar Fish. Rev. 37:1-8. 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. Sissenwine, M. P., P. M. Mace, J. E. Powers, and G. P. Scott. 1998. A commentary on western Atlantic bluefin tuna assessments. Trans. Am. Fish. Soc. 127:838-855. Turner, S. C. C. A. Brown, and H. Huang. 1997. Standardized catch rates of small bluefin tuna, Thun- nus thynnus. from the rod and reel fishery off Virginia- Rhode Island in 1980-1995. ICCAT Coll. Vol. Sci. Pap. 46(l):295-297. Turner, S. C, H, A, Forster, J. Witzig, C. Brown, M. Terceiro, D. A. Van Voorhees. G. P. Scott, J. A. Browder, D. J. Christensen, and C. E. Porch. 1993. Estimation and projection of the 1992 landing of blue- fin tuna in the rod and reel fishery for large pelagic fish off the northeast United States. ICCAT Coll. Vol. Sci. Pap. 40(11:1-16. Uda, M. 1973. Pulsatile fiuctuations of oceanic fronts in association with the tuna fishing grounds and fisheries. J. Fac. Mar. Sci. Technol.. Tokai Univ 7:245-266. Yuen. H. S. H. 1970. Behavior of skipjack tuna. Katsuwonus pelamis, as determined by tracking with ultrasonic devices. J. Fish. Res. Board Can. 27:2071-2079. Walker. M. M. 1984. Learned magnetic field discrimination in yellowfin tuna, Thunnus albacares. J. Comp. Physiol. A. 155:673- 679. 168 Abstract-The stomachs of 819 Atlan- tic bluefin tuna (Thiinnus thynnusi sampled from 1988 to 1992 were ana- lyzed to compare dietary differences among five feeding grounds on the New England continental shelf (Jef- freys Ledge, Stellwagen Bank. Cape Cod Bay, Great South Channel, and South of Marthas Vineyard) where a majority of the U.S. Atlantic commer- cial catch occurs. Spatial variation in prey was expected to be a primary influence on bluefin tuna distribution during seasonal feeding migi'ations. Sand lance iAmmodytes spp.), Atlantic herring (Clupea harengus). Atlantic mackerel (Scomber scombr-us), squid (Cephalopoda), and bluefish (Pomato- mus saltatrix) were the top prey in terms of frequency of occuiTence and percent prey weight for all areas com- bined. Prey composition was uncorre- lated between study areas, with the exception of a significant association between Stellwagen Bank and Great South Channel, where sand lance and Atlantic herring occurred most fre- quently. Mean stomach-contents bio- mass varied significantly for all study areas, except for Great South Channel and Cape Cod Bay. Jeffreys Ledge had the highest mean stomach-contents bio- mass (2.0 kg) among the four Gulf of Maine areas and Cape Cod Bay had the lowest (0.4 kg). Diet at four of the five areas was dominated by one or two small pelagic prey and several other pelagic prey made minor contri- butions. In contrast, half of the prey species found in the Cape Cod Bay diet were demersal species, including the frequent occurrence of the sessile fig sponge iSuberites ficus). Prey size selec- tion was consistent over a wide range of bluefin length. Age 2-4 sand lance and Atlantic herring and age 0-1 squid and Atlantic mackerel were common prey for all sizes of bluefin tuna. This is the first study to compare diet composition of western Atlantic bluefin tuna among discrete feeding grounds during their seasonal migi'ation to the New Eng- land continental shelf and to evaluate predator-prey size relationships. Previ- ous studies have not found a common occurrence of demersal species or a pre- dominance of Atlantic hening in the diet of bluefin tuna. Differences in diet of Atlantic bluefin tuna iThunnus thynnus) at five seasonal feeding grounds on the New England continental shelf* Bradford C. Chase Massachusetts Division of Marine Fisheries 30 Emerson Avenue Gloucester, Massachusetts 01930 E mail address brad chase gistale ma us Manuscript accepted 21 September 2001. Fish. Bull. 100:168-180 (2002). Atlantic bluefin tuna iTIuinmis thyn- nus) are widely distributed throughout the Atlantic Ocean and have attracted valuable commercial and recreational fisheries in the western North Atlantic during the latter half of the twentieth century. The western North Atlantic population is considered overfished by the International Commission for the Conservation of Atlantic Tunas ( NMFS, 1999). Bluefin tuna are the largest scombrid species, and the largest tele- ost occurring in the Gulf of Maine (Big- elow and Schroeder, 1953). Bluefin tuna migrate to coastal waters off New Eng- land during warmer months, feeding on local concentrations of prey. This migration supports a major component of the U.S. Atlantic commercial fishery for bluefin tuna; from 1978 to 1992 New England accounted for between TS'/r and 98% of annual commercial land- ings (Chase, 1992; and NMFS, 1995). Substantial annual variation has been seen in the harvest locations on the New England continental shelf (Chase, 1992). Spatial variation in prey popu- lations is suspected to be the primary influence on these annual aggregations of bluefin tuna. Adaptations in the cir- culatory system of the bluefin tuna allow these fish to retain metabolic heat, facilitating the regulation of body temperature (Carey and Teal. 1969) and assisting in the efficient transfer of energy from consumed prey to fast growth rates and large body size. These adaptive features also allow bluefin tuna to make extensive migrations into cold-temperate waters in search of prey. Diet information is necessary to improve the understanding of sea- sonal movements of bluefin tuna and predator-prey relationships in the New England continental shelf region, and as a baseline for bioenergetic analyses. Information on the feeding habits of this economically valuable species and apex predator in the western North Atlantic Ocean is limited, and nearly absent for the seasonal feeding grounds where most U.S. Atlantic commercial catches occur. Previous food habit studies have shown that western North Atlantic bluefin tuna are opportunistic feeders on a wide variety of finfish, cephalo- pods, and crustaceans (Crane, 19.36; Krumholz. 1959; Dragovich. 1970; Ma- son, 1976; Holliday, 1978; Eggleston and Bochenek, 1990; Matthews et al.M. Pinkas (1971) reported similar results for Pacific bluefin tuna iThunnus thyn- nus orientalis). These bluefin tuna food habit studies either reported on small numbers of samples or were qualitative studies on samples distributed over a broad geographic range. Previous studies have not evaluated size relationships between bluefin tuna and their prey and the amount of food consumed. The present study tar- geted regions of the New England gi- ant tuna (>140 kg) fishery to quan- titatively analyze stomach contents of bluefin tuna among discrete sea- sonal feeding grounds and to investi- gate predator-prey size relationships for this species in the Gulf of Maine. Contribution 3 of the Massachusetts Divi- sion of Marine Fisheries, Gloucester. MA 01930. Matthews, F. D., D. M. Damkaer, L. W. Knapp, and B. B. Collette. 1977. Food of western North Atlantic tunas (Thun- nus) and lancetfish lAlepisaurus). Nat. Oceanic Atmos. Admm. Tech. Rep. NMFS SSRF-706, Washington. D.C., 19 p. Chase: Differencesin diet of Thunnus thynnus at seasonal feeding grounds off New England 169 42 N Materials and methods Study area Five fishery areas (Jeffreys Ledge, Stelhva- gen Bank, Cape Cod Bay, Great South Chan- nel, and South of Martha's Vineyard IFig. 11) were selected for their traditional associa- tion with the bluefin fishery and because geo- graphic and bathymetric differences among areas could result in distinct prey communi- ties. The northernmost area, Jeffreys Ledge, is a major bathymetric feature in the Gulf of Maine and is important for commercial fisheries from northern Massachusetts, New Hampshire, and southern Maine. Stellwagen Bank is a distinct bathymetric ridge located on the eastern boundary of Massachusetts Bay (DOC^) that provides close access for ports from Gloucester to Cape Cod. Cape Cod Bay is a large, relatively shallow Gulf of Maine bay that is semi-enclosed on three sides by the land mass of Cape Cod and the south shore of Massachusetts. The two southern areas. Great South Channel and the area south of Martha's Vineyard, are larger regions with less distinct bathymetric features and are separated by Nantucket Shoals. The Great South Channel covers a wide nearshore region running east of Cha- tham and Nantucket and is bordered by the slopes of Nantucket Shoals on the west. The area south of Martha's Vineyard is located on the continental shelf off southern New England and is distinguished from the other areas by its warmer water, predominance of smaller bluefin tuna (<50 kg), and the sea- sonal occurrence of other large pelagic fish, such as marlins and tropical tunas. Commercial catch records and trawl sur- vey indices of abundance indicate that At- lantic herring (Clupea harengus) and Atlan- tic mackerel (Scomber scombrus) are the principal pelagic fish species for all these study areas (Clark and Brown, 1976; and NOAA, 1998). Atlantic herring occur in lower relative abundance in the region south of New England than in northern regions. Jeffreys Ledge and Great South Channel are primary spawning locations for Atlantic her- ring. Catch records and trawl survey indices indicate that the following prey species have been abundant in the study areas: silver hake (Merluccius bilineahs), butterfish (Peprilus triacanthus), northern shortfin squid (lllex illece- brosus), and longfin inshore squid (Loligo pealeii). 70°W 2 DOC (Department of Commerce). 1991. Stellwagen Bank Na- tional Marine Sanctuary, Draft environmental impact statement/ management plan. U.S. Dep. Commerce (DOC», Nat. Oceanic Atmos. Adm., Sanctuaries and Reserves Division, Washington, D.C., 238 p. Figure 1 Map of the five bluefin tuna feeding grounds used as study areas ( * ) July- October 1988-92. Depths are given in fathoms. Sample collection and analysis Bluefin stomach samples were collected primarily from commercial landings at ports in Massachusetts during the 1988-92 seasons (July-October). Sportfishing tour- naments were a secondary source of stomach samples. Handgear landings (rod and reel, handline, and harpoon) were primarily collected in Gloucester and Cape Cod. Purse-seine landings were also a large source of samples, and were collected in Gloucester, New Bedford, and Cape Cod. A majority of stomach samples was collected during 1989 and 1990. A reduced number of samples after 1990 was influenced by the increasing practice of gutting blue- fin at sea to sustain a quality product for the sashimi export market. Most samples were collected at the docks and process- ing locations of commercial tuna buyers. The vessel cap- tain or tuna buyer was interviewed for location of catch. 170 Fishery Bulletin 100(2) The cui-ved fork length (CFL) and weight of the catch were recorded from the National Marine Fisheries Ser- vice (NMFS) tuna logbook. Stomachs were removed by cutting the esophagus above the pylorus and were stored on ice until analysis later that day or were frozen for anal- ysis at a later date. Stomach samples were typically re- moved from bluefin tuna the day of capture, except those from purse-seine landings, where the catch was often sub- merged in ice for one or two days prior to landing. Stomach contents were identified to the lowest possible taxon and weights and counts were made of individual prey species. Wet weights of prey were measured with Homs tubular scales (±5 g) after contents were rinsed through a standard testing sieve (2.00-mm mesh). Prey counts were made only when all individual prey items could be identified and counted for a given stomach sam- ple. Because bluefin tuna consume relatively large prey items and swallow prey whole, species identification was possible for nearly all contents. Skeletal remains were compared with the skeletons of known specimens to assist with identification. Fish contents that could not be identi- fied were categorized as "unidentified fish." Otoliths, squid beaks, and skeletal traces less than .5 g were rounded up to a minimum weight of 5 g. Stomach-contents data were analyzed by frequency of occurrence, percent composition by number, and percent composition by weight for each prey item (Hyslop 1980; Bowen 1986). Frequency of occurrence can indicate prey composition and availability, and number and weight per- centages can represent the quantity a prey item contrib- utes to a diet. "Stomach-contents biomass" refers to all prey in stomach contents, and "prey weight" refers to the weight of individual prey species. Stomach samples were assigned a status of "empty," "chum," "chum and prey," or "prey only." Chum refers to cut pieces of bait fish that fishermen use to attract bluefin. Both empty and chum stomach samples were eliminated from further data analysis. For "chum and prey" stomachs, chum weight was eliminated from the analysis, and prey weight was included because chum and natural prey were easily distinguished in bluefin tuna stomachs. Wlien the status of contents, or the veracity of catch location could not be re- solved, the samples were eliminated from the analysis. Statistical analysis Stomach-contents biomass data were analyzed to deter- mine differences among the five fishery areas. Prey weight data were tested for normality (Shapiro and Wilk, 1965) and equality of variance (BMDP. 1990). Stomach-contents biomass data were transformed to natural logarithms and tested for area differences by using the Brown-Forsythe test for unequal variances and the Welch test for painvise comparisons of areas (BMDP, 1990). The species composition of stomach contents were test- ed among areas with the Spearman rank correlation test, under the null hypothesis of no association between spe- cies composition and feeding area. The twelve most com- mon prey species were ranked according to frequency of occurrence for each location. The Spearman rank correla- tion test was applied by pairwise ranking for all locations, excluding missing cases. Bluefin tuna size was compared with prey length and stomach-contents biomass data by using the Pearson prod- uct moment correlation coefficient (Sokal and Rohlf, 1981) to test for significant associations between prey and pred- ator length, and to correlate bluefin tuna weight and stom- ach-contents biomass for all samples from the Gulf of Maine. The relationship between bluefin tuna size and food consumed was also evaluated by comparing the ratio of stomach-contents biomass and tuna weight (% kg wet weight of prey biomass/kg wet weight of tuna) to tuna length (cui-ved fork length) (Young et al., 1997). The area south of Martha's Vineyard area was excluded from tests on size relationships because of the limited sample num- bers of juvenile bluefin tuna collected there. Results A total of 819 bluefin tuna stomachs were analyzed during 1988-92 (Table 1) and 568 contained prey; empty stom- achs (206) and "chum only" samples (45) were eliminated from further analysis. Approximately equal quantities of the samples with prey came from purse-seine landings (273) and from rod and reel and handline landings (264). The fishing method used for the hook-and-line landings was recorded for 242 samples and was evenly divided between chumming (where chum was used as bait) (123) and trolling (119). Size composition of sampled bluefin tuna was similar for the four Gulf of Maine study areas; a large majority of fish were large, mature adults, esti- mated to be age 10 and older (Mather and Schuck, 1960). In contrast, nearly all fish sampled from the area south of Martha's Vineyard were small juveniles, ages 2-6. Tuna sampled from Cape Cod Bay had the largest average size <251 cm CFL, and 273 kg). " Prey composition All areas combined Stomach contents comprised at least 21 species of teleosts, two species of elasmobranchs, and at least nine species of invertebrates (Table 2). Stomach- contents biomass in terms of taxonomic composition was dominated by Osteichthyes (Fig. 2A). Of the invertebrates, only squid ("squid" refers to two species, Loligo pealei and lllex illecebrosus) were a consistent component of prey composition. Although squid accounted for only about 2'7c of the stomach-contents biomass. it was the second most common prey, occurring in a third of all stomach samples. The only other common invertebrate was the fig sponge iStiberites ficus) found in Cape Cod Bay. Despite the large diversity of prey items, few species made major contri- butions to overall prey composition. Sand lance (Amino- dytes ssp). squid, Atlantic herring, Atlantic mackerel, and bluefish {Pomatomus saltatrix) exceeded all other prey in terms of frequency of occurrence and accounted as a group for 88% of total stomach-contents biomass (Fig. 2B). In the Gulf of Maine areas, sand lance, and Atlantic herring were the major prey in the diet of bluefin tuna during Chase: Differencesin diet of Thunnus thynnus at seasonal feeding grounds off New England 171 Table 1 Summary of numtx'r of l)lui fin tuna stomach samples col ected at five study areas on thf New England continental shelf. July- October 1988-92. Also included are lengths (cui'ved fork length ICFLl, cm) and weights kg of sampled tuna. "Other" column refers to twelve samples with prey collected at nearshore fishing areas in the Gulf of Maine that were outside the five study areas Sample Jeffreys Stellwagen Cape Cod Great South South of category Ledge Bank Bay Channel Marths's Vineyard Other Total By year 1988 13 30 16 0 0 0 59 1989 13 33 28 88 13 1 176 1990 57 15 26 61 33 2 194 1991 34 8 13 34 2 8 99 1992 6 7 26 0 0 1 40 By condition of stomach Number with prey 123 93 109 183 48 12 568 Number with chum 19 17 6 0 1 2 45 Number empty 5 1 158 27 8 7 206 Total 147 111 273 210 57 21 819 Mean length of tuna 221 240 251 221 124 221 227 SD 32 35 19 38 30 37 44 Mean weight of tuna 186 243 273 196 36 205 215 SD 78 94 58 90 38 83 97 the study period. Twenty-one bluefin tuna stomach sam- ples were collected outside of the five study areas; mostly at two inshore locations north of Gloucester and south of Stellwagen Bank. Jeffreys Ledge Atlantic herring were the dominant prey in the 123 stomach samples from Jeffreys Ledge (Table 3). The frequency of occurrence for Atlantic herring ( 74*7^ ) was the second highest and the percentage of stomach-contents biomass iST^'i ) was the highest for any individual prey item among areas. Squid were in nearly half of all samples, but comprised less than 2% of stomach-contents biomass. Atlantic mackerel were the third most common prey for this region {32"r occurrence). All other prey species occurred inci- dentally. Menhaden iBrevoortia tyrannus) and pollock (Pol- lachius virens) were unique to samples from Jeffreys Ledge. Mean stomach-contents biomass (-2.0 kg) was the highest among study areas and few empty stomachs were found. Stellwagen Bank As at Jeffreys Ledge, a single species dominated the stomach-contentss from Stellwagen Bank. Sand lance were found in nearly 80% of 93 stomachs and accounted for nearly 70'^^ of the stomach-contents biomass (Table 3). Atlantic herring, squid, spiny dogfish (Squalits acanthias), Atlantic mackerel, and bluefish tuna were sec- ondary prey items, with frequencies of occurrence ranging from 9% to 14%. All other prey species found in stomach contents from Stellwagen Bank occurred incidentally, and there were no species from this area that were unique to the overall studv area. dicate a dominant pelagic prey but did include the common occuiTence of demersal prey. Six prey species were only found in Cape Cod Bay. Squid occurred most frequently but accounted for only 2% of stomach-contents biomass biomass (Table 3). The fig sponge was the top prey in terms of per- centage by weight (27%). The diet of bluefin tuna caught in Cape Cod Bay displayed the most diversity among study areas. A total of 16 prey species were identified, of which eight were demersal species. Three species of flounder were identified, representing 9% of the stomach-contents biomass. The occurrence of bluefish tuna as prey in Cape Cod Bay (25%) was the highest among study areas. The amount of food in Cape Cod Bay stomach samples was the lowest for the four Gulf of Maine locations; 60% of the stomachs collected from this area were empty. Great South Channel A large number of stomach sam- ples with prey ( 183 ) were collected from the Great South Channel during 1989-91. The most abundant prey was sand lance, occurring in 62% of the stomach samples and accounting for 28% of the stomach-contents biomass. Atlantic herring was also important, with a frequency of occurrence of 27% and stomach-contents biomass of 48%. As with Stellwagen Bank and Jeffreys Ledge, squid was an important secondary prey item, and bluefish and Atlantic mackerel were secondary prey of lesser impor- tance. Four species were unique to Great South Channel samples: shrimp i Panda! us spp.), finger sponge (Haliclona oculata), silverstripe halfbeak [Hyporhaniphus unifascia- tus), and Atlantic cod (Gadus morhua). Cape Cod Bay Unlike diet composition for the other study areas, diet composition for Cape Cod Bay did not in- South of Martha's Vineyard Only 48 stomachs with prey were analyzed from the area south of Martha's Vineyard. 172 Fishery Bulletin 100(2) Table 2 Stomach contents of bluefin tuna caught off New England during 1988-92. Prey species are combined for all five study areas, including 12 samples from outside of the study areas. Percent frequency of occurrence C^'t O) and percent weight C^^'r W I data were determined from the 568 stomach samples that contained prey. Prey species Sand lance Atlantic herring Atlantic mackerel Bluefish Butterfish Silver hake Windowpane Hake Winter flounder Atlantic menhaden Sea horse Atlantic cod Fourspot flounder American plaice Wrymouth Pollock Filefish Halflieak Longhorn sculpin Unidentified fish Spiny dogfish Skate Skate egg case Squid Octopus Shrimp Lobster Argonaut Crab Salp Fig sponge Finger sponge Frequency of occurrence Total weight (gl Ammodytes (spp. ) Cliipca harengus Scomber scui/ibrus Pom a torn ii s saltatrix Pi-prihis tnacanthiis Mcrlucciua bilineans Scoptha/miiK aqiiosus Urophycia (spp.) P.vuclopleuroncctes americanus Brevoortia ty ran mis Hippocampus erect us Gadus morhua Paralichthys oblongus Hippoglussoidcs platessoides Ciyptacanthodes masculatiis Pullachius virens Monocanlhus luspidus Hyporhamphus unifasciatus Myoxocephalus octodecimspinosus Teleostei Squahis acanthias Raja (spp. I Raja ( spp. 1 Cephalopoda Cephalopoda Pandalus (spp.) Homarus americanus Argonauta argo Cancer (spp.) Salpidae Suberites ficus Haliclona oculata Stomachs with chum and prey Stomachs with chum only Empty stomachs Total stomachs with prey Total stomachs sampled 194 167 108 55 21 16 12 11 8 7 7 6 2 2 2 1 1 1 1 34 13 13 3 186 1 5 2 2 2 2 30 1 95 45 206 568 819 YcO % W 128,240 34.2 22.6 299,550 29.4 .52.8 18,930 19.0 3.3 40,830 9.7 7.2 1685 3.7 0.3 1490 2.8 0.3 1890 2.1 0.3 2500 1.9 0.4 1820 1.4 0.3 5500 1.2 1.0 40 1.2 <0.05 22,840 1.1 4.0 370 0.4 0.1 310 0.4 0.1 1300 0.4 0.2 940 0.2 0.2 30 0.2 <0.05 120 0.2 <0.05 280 0.2 <0.05 605 6.0 0.1 10,490 2.3 1.9 4670 2.3 0.8 20 0.5 <0.05 10,835 32.8 1.9 30 0.2 <0.05 25 0.9 <0.05 230 0.4 <0.05 25 0.4 <0.05 15 0,4 <0.05 60 0.4 <0.05 11,510 5.3 2.0 50 0.2 <0.05 107,430 (chum) 49,040 (chum) 567,230 723,700 Giant bluefin were scarce in this area during the study period and numerous samples could not be used because the bluefin tuna were caught in association with trawler fleet discards. This area is distinguished from the others by a predominance of juvenile bluefin tuna . Four prey species were unique to this area: lined sea horse (Hip- pocampus erectus), argonaut {Argonauta argo), planehead filefish (Monocauthus hispiduxl, and octopus (Cephalop- oda ). The filefish and seahorse were associated with bluefin tuna foraging at sargassum weed communities. Squid and Atlantic mackerel were the two most important prey for this area. The frequency of occurrence and percentage of prey weight for mackerel and butterfish were the highest among study areas. Combined stomach-contents biomass for squid, mackerel, and butterfish represented nearly 809;^ of the stomach contents for this area. Chase: Differencesin diet of Thunnus thynnus at seasonal feeding grounds off New England 173 Comparison of study areas The top 12 prey items, overall, were lanked i'or each area by frequency of occur- rence, and area differences were tested with Spearman rank correhition. Of the ten painvise comparisons, only Stelhvagen Bank and Great South Cliannel showed a significant association in the ranking of prey items (;=0.98. P<0.02). attributable to a high rank of sand lance and a similar ranking of squid. Atlantic herring. and Atlantic mackerel for both areas. Stomach-contents biomass Comparison of study areas Large differences in stom- ach-contents biomass were found: Jeffreys Ledge aver- aged nearly 2 kg, followed by approximately 1 kg for Stellwagen Bank and Great South Channel, and less than 0.5 kg for the remaining areas. Stomach-con- tents biomass data from the five areas were positively skewed and heteroscedastic. The natural logarithm- transformed data for Stellwagen Bank, Cape Cod Bay, and Great South Channel were normal (Wilk-Shapiro test, P>0.05). Transformed biomass data for the other two areas still differed significantly from normality. Transformation of biomass data reduced the inequal- ity of variances, but significant differences (Levenes test, P<0.05) remained, which precluded use of analy- sis of variance. The Brown-Forsythe test for unequal variances showed a significant effect of area on stom- ach-contents biomass (P<0.0001). Painvise compari- sons of the equality of prey weight means were made with the Welch test (Bonferroni corrected significance level of P=0.005). All area paii-wise comparisons of stomach-contents biomass were significantly different except that between Cape Cod Bay and Great South Channel(P=0.816). The similarity in the amount of food found at Cape Cod Bay and Great South Channel was also indicated by the geometric mean of stomach-contents biomass (Table 4). The arithmetic mean of stomach-contents biomass was higher at the Great South Channel than Cape Cod Bay, but this value was biased by a few samples with large amounts of prey. The stomach- contents biomass range for Cape Cod Bay did not exceed 3.0 kg, in contrast to the wider range for the Great South Channel up to 16.0 kg, including 13 samples over 3.0 kg. The use geometric means reduced the bias of skew- ness and indicated that samples from Jeffreys Ledge con- tained the most prey and that the amounts declined mov- ing southward. Effect of tuna size Increased stomach-contents biomass with increasing body size (due to increasing gape and stomach size) was not clearly demonstrated from these data. Correlation of stomach-contents biomass to bluefin weight was not significant for Gulf of Maine samples. A scatterplot of data for the Gulf of Maine samples revealed that a large majority of the samples contained small amounts of food, regardless of body size (Fig. 3). A size- related trend was observed in that only very large bluefin tuna (>250 kg) contained more that 6 kg of food. Size Osteichttiyes 93% Elasmobranchii 3% Cephalopoda 2% Ponfera 2% B other invertebrates sand lance 2% 23% ^^ If—.. other fish squid 2% Atl mackerel 3% Atlantic herring 53% Figure 2 Percent prey weight composition in stomach contents of bluefin tuna caught off New England during 1988-92 ln=568). The taxo- nomic composition (A) includes 0.05'^ Crustacea and O.Ol^r Uro- chordata. The comparison by major prey type (B) comprises the five most common prey and all remaining fish and invertebrates. effects were also compared by using the ratio of stomach- contents biomass and tuna weight C* kg/kg, wet weight) to tuna length. The percentage of food to body mass declined with increasing bluefin tuna length (Fig. 4). Food obsei-ved in 120-149 cm bluefin tuna averaged over 1** of their body weight, and declined to approximately 0.5% for bluefin tuna over 230 cm. The high ratios primarily resulted from large meals of Atlantic herring or sand lance. Cape Cod Bay ratios were consistently the lowest among the four areas, ranging from 0. 1 to 0.2%. Characteristics of prey species Prey size Prey size was evaluated for 190 stomach sam- ples that contained measurable prey from the four Gulf of Maine areas. A total of 1866 prey items were measured, of which 95% were either sand lance, Atlantic herring, squid, or Atlantic mackerel (Table 5). A significant positive cor- 174 Fishery Bulletin 100(2) Table 3 Stomach contents of bluefin tuna caught off New England durii g 1988-1992, listed by the five study areas Percent frequency of | occurrence (% 0) and percent by tveightC/c W) were calculated for each prey species. South of Prey species Jeffreys Ledge Stellwagen Bank Cape Cod Bay Great South Channel Martha's Vineyard 'i 0 ^■, W '-r O ^,\\ KiO ^i W 't 0 r^w ^'f 0 C^ W Sand lance 3.3 1.8 79.6 69.3 0 0 62.3 28.3 0 0 Atlantic herring 74.0 87.2 14.0 6.0 8.3 3.1 27.3 48.4 2.1 2.5 Atlantic mackerel 3L7 2.0 10.8 2.6 18.3 12.7 8.2 0.6 33.3 56.2 Blucfish 7.3 3.5 8.6 17.5 24.8 14.7 4.9 5.7 0 0 Butterfish 2.4 <0.05 11 <0.05 5.5 1.5 1.6 0.2 16.7 10.4 Silver hake 3.3 0.2 0 0 0 0 2.2 0.3 10.4 2.9 Window-pane 0 0 0 0 10.1 4.1 0 0 0 0 Hake 4.1 0.7 1.1 0.2 0,9 0.2 0 0 8.3 9.9 Winter flounder 0 0 0 0 6.4 3.9 0 0 0 0 Atlantic menhaden 5.7 2.3 0 0 0 0 0 0 0 0 Sea horse 0 0 0 0 0 0 0 0 14.6 0.7 Atlantic cod 0 0 0 0 0 0 3.3 13.5 0 0 Fourspot flounder 0 0 0 0 1.8 0.9 0 0 0 0 American plaice 0.8 0.1 1.1 <0.05 0 0 0 0 0 0 Wrymouth 0 0 0 0 1.8 0.9 0 0 0 0 Pollock 0.8 0.4 0 0 0 0 0 0 0 0 Filefish 0 0 0 0 0 0 0 0 2.1 0.5 Halfteak 0 0 0 0 0 0 0.5 0.1 0 0 Longhorn sculpni 0 0 0 0 0 0 0 0 0 0 Unidentified fish 4.9 <0.05 3.2 <0.05 2.8 0.1 8.7 0.1 12.5 2.9 Spiny dogfish 0 0 10.8 3.7 2.8 16.6 0 0 0 0 Skate 0 0 0 0 9.2 9.3 0.5 0.3 0 0 Skate egg case 0 0 1.1 <0.05 1.8 <0.05 0 0 0 0 Squid 48.8 L8 14.0 0.5 35.8 1.8 22.4 2.5 60.4 12.9 Octopus 0 0 0 0 0 0 0 0 2.1 0.5 Shrimp 0 0 0 0 0 0 2.7 <0.05 0 0 Lobster 0 0 0 0 1.8 0.5 0 0 0 0 Argonaut 0 0 0 0 0 0 0 0 4.2 0.4 Crab 0 0 0 0 0.9 <0.05 0.5 <0.05 0 0 Salp 0 0 0 0 1.8 0.1 0 0 0 0 Fig sponge 0 0 0 0 27.5 27.2 0 0 0 0 Fmger sponge 0 0 n n 0 0 0.5 <0.05 0 0 relation (P<0.001) between prey and bluefin tuna lengths was found for all prey-size data (Fig. 5). Despite this cor- relation, there appeared to be httle association between predator and prey length for most species. The positive correlation was influenced by 29 larger prey items (>40 cm) all consumed by bluefin tuna larger than 230 cm. The larger prey were spiny dogfish, skate, bluefish, or Atlantic cod. Size data on the four most common prey species pro- vided evidence of the consistency of prey size across a wide range of bluefin lengths in the Gulf of Maine. A signifi- cant positive size relationship was found for sand lance and Atlantic mackerel (both P<0.001), although both may have contained biases. The relationship for sand lance was influenced by smaller bluefin tuna that ate smaller sand lance (/-test, P<0.001) in Great South Channel than at Stellwagen Bank. All mackerel were YOY or age-1, except for two large mackerel consumed by larger bluefin tuna (>200 cm). The predator-prey size relationships for Atlantic herring (P=0.36) and squid (P=0.16) were not sig- nificant. A large majority (939f ) of Atlantic herring prey were 18-27 cm in length, which corresponds to age-2 to age-4 cohorts for the western Gulf of Maine (Penttila et Chase: Differencesin diet of Thunnus thynnus at seasonal feeding grounds off New England 175 Table 4 Summary statistics on stomacli-contonts liioniass 1^,' w( Eiifjland, 1988-92. The geometric mean, confidence into ral logarithms. ■t weight 1 from bluefin tuna caught at five seasonal feeding areas ofTNew n-als (CI), and coefficient of variation ((-V) are backtransformed from natu- Statistic All areas JefTreys Ledge Stellwagen Bank Cape Cod Bay Great South Channel South of Martha's Vineyard Number of stomach samples with prey 568 123 93 109 183 48 Minimum stomach biomass 5 5 5 5 5 5 Maximum stomach biomass 16100 5800 6240 2900 16100 1200 Arithmetic mean 999 1957 1012 389 925 124 Geometric mean 214 8.32 438 133 126 37 Lower 95^f CI 180 608 324 96 91 23 Upper 95% CI 255 1140 593 184 174 60 CV 39 26 24 35 46 44 16 0 1 m 14 0 ■ ^£. cn 12 0 - m H o 100 ■ .Q C 80 - CD c- o CJ 60 a en F 40 ■ n fl 2 0 - 00- al., 1989). and the youngest were age 2. There were no young-of-the-year (YOY) sand lance. Most sand lance from Great South Channel samples were age 2. in contrast to age 3 or age 4 at Stellwagen Bank (Weston et al, 1979). Most squid were YOY or age 1. Numbers of prey species Few prey species were found in large numbers in a given stom- ach sample. Only sand lance, squid, Atlantic herring, and Atlantic mackerel had sample counts higher than 20 individual fish (Table 6). Data on prey numbers are limited because prey counts were made for only 208 samples. Many samples with large numbers of well-digested sand lance were difficult to count. From this subsample, the mean number of sand lance per stomach was 159, much higher than the next highest mean of 19 for Atlantic herring. Squid and mackerel commonly occurred as prey, although typically only a few individuals were found per stomach. Weight of prey species Sand lance and Atlantic herring (combined) accounted for 75'7i of the total stomach-contents biomass for all areas combined. The next highest species was bluefish at T^i. Despite a high frequency of occurrence, squid accounted a low percentage of overall biomass (2%) and a mean stomach prey weight of only 58 g. The highest mean prey weight was 1794 g for Atlantic herring. Only three other prey items averaged over 500 g in stomach-con- tent weight: spiny dogfish (807 g), bluefish (742 g), and sand lance (661 g). With few exceptions, stomachs that were full or near full, contained only sand lance or Atlantic herring. Only five stomachs contained over 10 kg of prey contents: three with Atlantic herring, and one each with sand lance and Atlantic cod ( 16.0 kg, the largest meal obsei-ved). In summary, five prey items occurred in frequency and mass to be considered important dietary components of n=520, r=0 048. P=0,276 40 cm ) by large bluefin tuna (>230 cm). Prey this large were not common in stom- achs and were probably limited by mouth and esophagus gape. Otherwise, the sizes of prey were consistent across the range of bluefin tuna sampled in the Gulf of Maine. The findings on prey size and numbers (per stomach) provide evidence of three selective foraging strategies used by bluefin tuna in the Gulf of Maine. Feeding on indi- vidual, fast-swimming pelagic prey was evident from con- sumed bluefish and Atlantic mackerel that are abundant pelagic species in the Gulf of Maine, but which occurred much less frequently in stomach contents and with few in- dividuals per stomach. Ram feeding (swimming through a dense school prey with mouth open) of small prey may have resulted in high average number of sand lance found in stomachs and contributed to the similar prey sizes in most sizes of bluefin tuna. Bluefin tuna in Cape Cod Bay displayed a different foraging behavior, selecting larger, individual demersal prey. The variation in prey compo- sition and different feeding strategies is consistent with previous descriptions of bluefin tuna as an opportunistic predator However, the dominance of sand lance and At- lantic herring in the Gulf of Maine diet suggests a depen- dence on these species as an optimal energy source. Trophic influences on bluefin tuna distribution Changes in biomass and spatial availability of forage pop- ulations may affect the distribution of bluefin tuna on the New England continental shelf Major changes in the prey community of the Gulf of Maine have occurred in recent decades. After Atlantic mackerel and Atlantic her- ring stocks off New England were severely overharvested in the 1960s and 1970s (NOAA, 1998), sand lance popula- tions increased, presumably as a result of decreased pre- dation and competition for food (Sherman et al., 1981). Atlantic herring and Atlantic mackerel stocks off New England increased steadily during the 1980s and 1990s ( NEFSC, 1998). By the mid-1990s, the U.S. Atlantic coastal spawning stock biomass for these species increased to the highest levels on record (NEFSC, 1998). Commercial bluefin tuna catches have increased in areas where Atlantic herring abundance has increased (western Gulf of Maine and Great South Channel) and diminished at traditional areas south of the Gulf of Maine (Chase, 1992). The northward shift in bluefin tuna distribution on the New England continental shelf may be influenced by improved foraging opportunities on Atlantic herring in the Gulf of Maine. I suspect that the timing of bluefin tuna migi-ations to the New England continental shelf are as- sociated with seasonal spawning and feeding aggregations of Atlantic herring. Sand lance populations appear to be an important influence on bluefin tuna feeding migrations. Chase: Diffeiencesin diet of Thunnus thyninis at seasonal feeding grounds off New England 179 but they occur on a limited spatial scale in relation to At- lantic herring in the (nilf of Maine, and changes in tlieir population abundance are not well documented. Atlantic herring and sand lance are also important in the diet and distribution of marine mammals in the Gulf of Maine (Payne et al., 1990; Weinrich et al., 1997; Gannon ct al., 1998). Changes in bluefin tuna stock composition and the long- term impact of small-mesh trawling gear on commercially important prey items (squid, silver hake, and butterfish) in southern New England waters and Mid-Atlantic areas where bluefin tuna catches have diminished are two poten- tially confounding factors in this discussion. In the man- agement of Atlantic bluefin tuna, care should be taken to recognize that the fluctuations in the regional abundance of this species can be influenced by more than changes in stock structure. Changes in major prey populations, pro- duced either by environmental features or by fishery prac- tices, can have a profound effect on regional aggregations of bluefin tuna. Investigations should be conducted on in- troducing forage-base information into the interpretation of catch-per-unit-of-effort indices of abundance for Atlan- tic bluefin tuna populations. There is also a need for future research to improve our knowledge on the bioenergetics of this warm-bodied tuna and its associated relationships with prey species. Acknowledgments This study was conducted by the Massachusetts Division of Marine Fisheries (DMF), Commonwealth of Massachu- setts, and funded by the Federal Aid in Sportfish Res- toration Program. I would like to thank the DMF and National Marine Fisheries Service staff who assisted with field sampling and the technical review of this project. I am especially thankful to Steve Cadrin for his review of the manuscript and assistance with the statistical analy- ses. I am very grateful to the many fishermen and tuna buyers who contributed stomach samples and catch infor- mation. Special thanks are due to the following for the contribution of large numbers of samples: Bill Raymond, Mark Godfrey, Rodman Sykes, and the crew of FV Debra Lynn, Great Circle Fisheries, Ralboray, Crocker and Sons, Canal Marine, and Atlantic Coast Fisheries. Literature cited Amundsen, P. A., and A. Klemetsen. 1986. Within-sample variabilities in stomach contents weight of fish- implications for field studies of consump- tion rates. In Contemporary studies on fish feeding (C. A. Simenstad and G. M. Cailliet. eds.), p. .307-314. Dr. W. Junk Publishers. Boston. MA. Bigelow, H. B., and W. C. Schroeder. 19,53. Fishes of the Gulf of Maine. U.S. Fish and Wildl. Serv. Bull. 74, vol. 53, 577 p. BMDP (Biomedical Dynamic Programs). 1990. BMDP statistical software manual, vol. I. UnivCal. Press, Berkeley, CA, 617 p. Bowen, S. H. 1986. Quantitative description of the diet. In Fisheries tech- niques (L. A. Nielson, and I). I,. Johnson, eds.), p. 325-336. Am. Fish. Soc, Bethesda, MD. Butler, M. J., and J. M. Mason Jr 1978. Behavioral studies on impounded bluefin tuna. Int. Comm. Conserv. Atl. Tunas Ooll. Vol. Sci. Pap. 7(2);379-381. Carey, F G., and J. M. Teal. 1969. Regulation of body temperature by the bluefin tuna. Comp. Biochem. Physiol. 28:20.5-213. Chase, B. C. 1992. A profile of changes in the Massachusetts bluefin tuna fishery: 1928-1990. M.S. thesis, Univ. Rhode Island, Kingston, RI, 266 p. Clark, S. H., and B. E. Brown. 1976. Changes in biomass of finfishes and squids from the Gulf of Maine to Cape Hatteras, 1963-1974, as determined from research vessel survey data. Fish. Bull. 75:1-21. Crane, J. 1936. Notes on the biology and ecology of giant tuna Thun- nus thynnus, L., observed at Portland, Maine. Zoologica 212:207-212. Dragovich, A. 1970. The food of bluefin tuna tThunnus thynnus) m the western North Atlantic Ocean. Trans. Am. Fish. Soc. 99: 726-731. Eggleston, D. B., and E. Bochenek. 1990. Stomach contents and parasite infestation of school bluefin tuna Thunnus thynnus collected from the Middle Atlantic Bight, Virginia. Fi.sh. Bull. 88:389-395. Gannon, D. P., J. E. Craddock, and A. J. Read. 1998. Autumn food habits of harbor porpoises, Phocoena phocoena, in the Gulf of Maine. Fish. Bull. 96:428-437. Hodgson, J. R., S. R. Carpenter, and A. P. Gripentrop. 1989. Effect of sampling frequency on intersample variance and food consumption estimates of nonpiscivorous large- mouth bass. Trans. Am. Fish. Soc. 118:11-19. Holhday. M. 1978. Food of Atlantic bluefin tuna, Thunnus thynnus (L.), from the coastal waters of North Carolina to Massachusetts. M.S. thesis. Long Island Univ., Long Island, NY, 31 p. Hyslop, E. J. 1980. Stomach contents analysis — a review of methods and their application. J. Fish. Biol. 17:411-429. Ki'umholz, L. A. 1959. Stomach contents and organ weights of some bluefin tuna, Thunnus thynnus (Linnaeus), near Bimini, Bahamas. Zoologica 44:127-131. Mason, J. M. 1976. Food of small, northwestern Atlantic bluefin tuna, Thunnus thynnus (L.) as ascertained through stomach con- tent analysis. M.S. thesis, Univ. of Rhode Island, Kings- ton, RI, 31 p. Mather, F. J, III, and H. A. Schuck. 1960. Growth of bluefin tuna of the western north Atlantic. Fish. Bull. Fish Wildl. Serv. 179:39-52. NEFSC (Northeast Fisheries Science Center). 1998. A report of the 27"' Northeast regional stock assess- ment workshop. Stock Assess. Rev. Comm. (SARC ), consen- sus summary of assessments. Lab. Ref Doc 98-15, 350 p. [Available from Northeast Fisheries Science Center ( NEFSC), Woods Hole, Woods Hole, MA.] NMFS (National Marine Fisheries Service). 1995. Final environmental impact statement for a regulatory amendment for the western Atlantic bluefin tuna fishery. 180 Fishery Bulletin 100(2) National Marine Fisheries Service i NMFS ), Silver Springs, MD, 142 p. 1999. Final fishery management plan for Atlantic tunas, swordfish. and sharks. In Chapter 3: Rebuilding and main- tainmg HMS fisheries, p. 1-.321. (Available from NMFS, Silver Springs, MD.l NOAA (National Oceanic and Atmospheric Administration). 1998. Status of fishery resources off the northeastern United States for 1998. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NE-115, 149 p. Payne, P. M., D. N. Wiley, S. B. Young, S. Pittman, P. J, Clapham, and J. W. Jossi. 1990. Recent fluctuations in the abundance of baleen whales in the southern Gulf of Maine in relation to changes in selected prey Fish. Bull. 88:687-696. Penttila, J. A., G. A. Nelson, and J. M. Burnett III. 1989. Guidelines for estimating lengths at age for 18 northwest Atlantic finfish and shellfish species. U.S. Dep. Commen, NOAA Toch. Memo., NMFS-F/NEC-66, 39 p. Pinkas, L. 1971. Bluefin tuna food habits. //; Food habits of albacore. bluefin tuna, and bonito in California waters (L. Pinkas, M. S. Ohphant, and I. K Iverson. eds. i. p. 47-63. Cal. Fish Game Fish. Bull. 1.52:1-105. Shapiro, S. S., and M. B. Wilk. 1965. An analysis of variance test for normality (complete samples). Biometrika 52: 591-611. Sherman. K. C, C. Jones, L. Sullivan, W. Smith, P. Berrein, and L. Ejsymont. 1981 . Congi'uent shifts in sand eel abundance in western and eastern north Atlantic ecosystems. Nature 291:486-489. Smagula, C. M., and I. R. Adelman. 1982. Day-to-day variation in food consumption by large- mouth bass. Trans. Am. Fish. Soc. 111:543-548. Sokal, R. R., and F J. Rohlf 1981. Biometry, 2nd ed. W. H. Freeman and Co., New York, NY, 859 p. Weinrich, M., M. Martin, R. Griffiths, J. Bove, and M. Schilling. 1997. A shift in distribution of humpback whales, Megap- tera novaeangliae. in response to prey in the southern Gulf of Maine. Fish. Bull. 95:826-836. Weston, D. T, K. J. Abernethy, L. E. Meller, and B. A. Rogers. 1979. Some aspects of biology of the American sand lance, Ainnuidytes amcncanua. Trans. Am. Fish. Soc. 108:328- 331. Young, J. W., T. D. Lamb, D. L. Russel, W. Bradford, and A. W. Whitelaw. 1997. Feeding ecology and interannual variations in diet of southern bluefin tuna, Thunnus maccoyii, in relation to coastal and oceanic waters off eastern Tasmania, Australia. Environ. Biol. Fi.sh. .50:275-291. 181 Abstract-A total of 42,445 American lobsters iWomaru.s americanus) wore tagged in thirty-one sites throughout the southwestern Gulf of St. Lawrence betw^een 1980 and 1997. Results from the recapture of 8503 tagged lobsters showed small distances traveled be- tween the release and the recapture position for animals ranging in size from 51 to 152 mm carapace length. The average distance traveled ranged from 2 km in parts of Bale des Chal- eurs and western Cape Breton to 19 km in central Northumberland Strait. Lob- sters moved generally along the shore (939J of the dispersion was in areas between the shore and the 20-m bathy- metric contour). As a result, lobsters traveled longer distances in sites char- acterized by a gradually sloping bottom where the distance between the shore and the 20-m contour line was exten- sive in contrast to areas characterized by rapidly changing depths and by a relatively small amount of habitat shal- lower than 20 m. In the majority of sites (14 of 19) there was no significant difference between males and females in the average distance they traveled. In four of the five sites females moved farther than males. In general, the average distance traveled by berried females was shorter than that traveled by males or nonberried females. No relationship was obsen.'ed between the distance traveled and the size of the animal. There was no strong evidence of a relationship between the average distance traveled and the number of days at liberty. In general, lobsters in the southwestern Gulf of St. Lawrence traveled short distances and dispersion was restricted to the nearshore habi- tat. Further, the distance traveled was not correlated to size, sex, or years at large. These findings show that there is little interaction between American lobsters from different fishing areas at the benthic level and that American lobster movements should have mini- mal consequences for management of the species in the southwestern Gulf of St. Lawrence. Movement of American lobster (Homorus americanus) in the southwestern Gulf of St. Lawrence Michel Comeau Fernand Savoie Department of Fisheries and Oceans 343 University Ave. Moncton, New Brunswick, Canada EIC 9B6 E-mail address (lor M Comeau) comeaumm'dfo mpo gcca Manuscript accepted 14 September 2001. Fish. Bull. 100:181-192 (2002). The American lobster (Homarus ameri- canus Milne-Edwards, 1837) fishery in the southwestern Gulf of St. Lawrence (GSL) and for the entire Canadian Mari- time Provinces has become the most eco- nomically important fishery of coastal communities. Consequently, there is increasing interest by fishermen and the fishing industry to better under- stand the biology of the species and factors that may play a role in the fluctuations of landings, including pos- sible lobster movements between lob- ster fishing areas (LFAs). Fishermen are particularly concerned by lobster movements because the minimal legal size increased in different LFAs in the southwestern GSL throughout the 1980s and 1990s, and they want to know whether lobsters returned at sea in a given area could be recaptured elsewhere. Several tagging projects have been conducted in the past to study lobster movements in the GSL (Table 1). These tagging projects, initiated in the 1930s by Templeman ( 1935), showed that the average distance traveled by lobsters in the GSL was generally less than 15 km and that very few animals traveled up to 70 km (for review see Stasko, 1980; Lawton and Lavalli, 1995). Other tag- ging studies conducted in inshore wa- ters outside the GSL in Nova Scotia (Wilder, 1974; Campbell, 1982, 1989; Campbell and Stasko, 1985; Miller et al., 1989; Tremblay et al, 1998), Bay of Fundy (Campbell, 1986; Campbell andStasko, 1986), Maine (Cooper, 1970; Cooper et al., 1975; Krouse, 1981), New Hampshire (Watson et al, 19991. Mas- sachusetts (Karnofsky et al., 1989) and Rhode Island (Fogarty et al., 1980) have also shown that lobster move- ments were generally similar (4 to 18 km) to those from the GSL. However, long-distance movements of more than 90 km for up to 20% of the animals have also been observed for lobsters tagged inshore (Dow, 1974; Fogarty et al., 1980; Campbell and Stasko, 1985, 1986; Campbell, 1989; Robichaud and Law- ton, 1997); the farthest distance trav- eled reported was 798 km (Campbell and Stasko, 1986). These long distances traveled are more similar to those re- ported for offshore lobsters tagged on the continental shelf and over the off- shore deep canyons (Saila and Flowers, 1968; Cooper and Uzmann, 1971; Uz- mann et al, 1977; Fogarty et al., 1980; Campbell et al., 1984; Campbell and Stasko, 1985). Movements of more than 70 km have itever been reported for lob- sters tagged in the southwestern GSL. Since 1980, forty-six tagging studies have been conducted throughout the southwestern GSL, mostly in areas where information on lobster movements has been unavailable. These tagging studies have covered fishing grounds characterized by a flat bottom and hav- ing a relatively smooth transition from shore to 30 m and a narrow habitat close to shore where changes in depths occur over a relatively short distance. The purpose of our study was to in- vestigate the benthic movement of lob- sters tagged in different locations with- in the southwestern GSL by comparing the distance traveled, number of days at liberty, and size and sex of lobsters. 182 Fishery Bulletin 100(2) Materials and methods The recapture position was documented for 8503 lobsters, ranging in size from 51 to 152 mm carapace length (CL), from a total of 42,445 tagged and released during forty-six Table 1 Locations and authors of lob ster tagging studies on move- | ments of American lobster iHoma rus americanus) con- ducted m the Gulf of St. Lawrence. Location Authors and date Various locations in Gulf of St. Lawrence Wilder (1974) West coast of Newfoundland Templeman (1940) Magdalen Islands (Quebec) Templeman ( 1935 ) Bergeron (1967) Munro and Therriauh(1983) Gaspe Peninsula (Quebec) Corrivault(1948) Malpeque ( Prince Edward Is land) Templeman ( 1935) Western Prince Edward Island Wilder (1963) Northumberland Strait Templeman ( 1935 ) Wilder (1963) tagging studies in thirty-one sites throughout the south- western GSL (Fig. 1) between 1980 and 1997 (Table 2). The animals were captured by traps and tagged between July and November after the commercial fishing seasons (May and June in LFAs 23, 24, 26A, and 26B, and from mid-August to mid-October in LFA 25). The CL, measured to the nearest mm, sex of each animal, and presence or absence of eggs under the female abdomen were recorded. Prior to 1989, all animals were tagged with orange sphy- rion anchor tags, and blue streamer tags were used after 1992. Between 1989 and 1992 inclusively, both type of tags were used. A description of the tags and the tagging tech- nique are presented in Moriyasu et al. ( 1995) and Comeau et al.(1998). The positions of recaptured lobsters came from fisher- men. Because all tagging projects were undertaken after each commercial fishing season, no lobsters were recap- tured during the same year of their tagging. To increase fishermen's participation in reporting tagged animals, a major awareness campaign was conducted. Prior to 1993, representatives of the Department of Fisheries and Oceans (DFO) were present at each wharf to measure and collect information on tagged lobsters that were recap- tured. Beginning in 1993, letters were sent to all lobster fishermen in regions where tagging projects were conduct- ed; in these letters the tagging project was described, in- Gulf of St. Lawrence Figure 1 Locations (*) of American lobster {Hnniaruf; amencanus) tagging studies conducted in the southwestern Gulf of St. Lawrence between 1980 and 1997. The names of each site are presented in Table 2. The lobster fishing areas (LFAs) are indicated on the map. Comeau and Savoie: Movement of Homaivs amencanus in the southwestern Gulf of St Lawrence 183 structions given for returninfi lobster tag information, and the cooperation of the fishermen was sought (Comeau et al., 1998). Similar information was posted at wharves. For each tagged lobster, fishermen were asked to record date of capture, tag number, position of capture, and depth. They were then asked to freeze the animal with the tag still attached and contact an information collection center, where they could leave their names, addresses, and tele- phone numbers. A DFO representative collected tagged lobsters for measurement and reimbursed fishermen ac- cording to the market value of the recovered lobsters. No reward was issued. Fishermen also had the option to bring the lobsters and the information to a fisherman-represen- tative. If a tagged lobster under the legal size or a tagged berried female was captured, fishermen were asked to re- cord the above information and release the lobster to the water with the tag still attached. The distance traveled by each recaptured animal was calculated as the linear distance between release and re- capture positions. The Kruskal-Wallis and Mann-Whitney (t/-test) tests were used to compare the average distance traveled for males, females, and berried females for ani- Table 2 Average distances traveled by American lobsters tagged in the southwestern Gulf of St. Lawrence between 1980 and 1997. The number of each site corresponds to its geographical position on the map in Figure 1. n = number of recaptured lobster with tags indicating release location for which distance traveled to recapture site could be calculated. Average carapace Carapace Average distance length ±SD length range traveled ±SD Tagging site Year of tagging n (mml (mml (km) 1 Belledune 1980 604 70.5 ±4.9 50-115 7.4 +6.5 2 Pointe-Verte 1996 26 78.2 ±6.4 71-90 6.1 +4.0 3 Stonehaven 1994-97 618 73.2 ±5.4 53-111 2.4 ±3.0 4 Anse-Bleue 1994 290 73.5 ±4.8 54-91 3.5 ±6.3 5 Caraquet 1993- 97 1416 74.2 ±9.2 55-133 8.0 ±8.8 6 Petit-Shippagan 1994 134 76.2 ±10.4 54-116 8.7 ±8.8 7 Miscou 1995 58 69.9 ±8.2 66-117 4.5 ±6.5 8 Le Goulet 1996 101 73.8 ±6.2 59-98 5.3 ±3.5 9 Val Comeau 1985. 1996-97 519 76.3 ±9.1 57-117 7.0 ±6.5 10 Neguac 1996 74 78.0 ±8.7 65-105 14.2 ±8.2 11 Escuminac 1997 17 73.9 ±5.2 66-88 9.6 ±9.9 12 Seacow Pond 1996 62 71.1 ±6.0 61-94 5.3 ±7.8 13 Alberton 1996 62 74.0 ±6.9 65-97 5.7 +5.9 14 Malpeque 1989 492 68.1 ±7.2 51-100 10.1 ±7.0 15 Tracadie 1984 27 68.1 ±12.4 55-111 10.4 ±10.1 16 North Lake 1996 78 72.8 ±5.0 61-89 2.5 ±3.4 17 Pointe-Sapin 1996 22 72.5 ±5.0 65-83 6.7 ±4.1 18 Kouchibouguac 1997 53 67.7 ±4.3 60-74 8.0 ±7.3 19 Cap-Pele 1997 41 72.2 ±6.0 60-84 12.2 ±11.9 20 Skinner's Pond 1996 13 71.4+5.8 63-131 5.4 ±6.7 21 Miminegash 1996 37 74.1 ±8.5 64-100 8.6 ±10.8 22 Howard's Cove 1995 79 66.9 ±7.1 54-84 15.3 ±13.2 23 Egmont Bay 1982 15 66.1 ±7.0 58-77 19.4 ±16.0 24 Souris 1996 213 75.4 ±7.1 59-109 4.9 ±6.8 25 Fortune 1996 208 72.7 ±3.9 66-84 3.3 ±3.6 26 Beach Point 1982 243 88.0 ±12.9 61-120 7.4 ±8.9 27 Lismore 1997 51 78.2 ±11.4 58-103 7.8 ±12.4 28 Ballantynes Cove 1986 226 74.0 ±11.2 55-130 9.3 ±13.0 29 Port Hood 1988 339 67.4+11.6 51-150 4.5 ±2.9 30 Margaree 1984, 1988, 1992 1332 70.7 ±8.5 52-152 3.2 ±5.1 31 Pleasant Bay 1988, 1992 1053 69.8 ±8.1 54-130 2.3 ±2.1 184 Fishery Bulletin 100(2) mals recovered during the first recapture period following their tagging. Correlation (/•) was used to determine the relationship between the distance traveled and the size of the animal. To determine the relation between the average distance traveled and days at liberty, sites with recaptures over multiple years were used. The relationship between the average distance traveled and the extent of shallow waters was also established. The extent of shallow waters was quantified as the distance from shore to the closest 30-m bathymetric contour for each tagging site. Results A total of 7565 tagged lobsters were returned during their first recovery period, with size and geogi'aphical position at recapture. Only sites with fifty recaptures or more were con- sidered. There was no evidence of a relationship between the size of the animal and distance traveled in the southwestern GSL because the correlation coefficient ir) ranged from -0.19 Table 3 Correlation coefficient (;l for the relationship of the dis- tance traveled and the carapace length (CL) of American lobsters tagged in the southwestern Gulf of St. Lawrence. The number of each site corresponds to its geogi'aphical position on the map in Figure 1. n = number of recaptured lobster with tags indicating release location for which dis- tance traveled to recapture site could be calculated. Tagging site /) CL range (mm) r 1 Belledune 536 50-115 -0.12 3 Stonehaven 580 53-111 0.04 4 Anse-Bleue 232 54-91 0.02 5 Caraquet 1288 55-133 0.08 6 Petit-Shippagan 117 54-116 0.10 8 Le Goulet 90 59-98 -0.03 9 Val Comeau 500 57-117 -0.05 10 Neguac 72 65-105 0.22 12 Seacow Pond 61 61-94 0.08 13 Alberton 61 65-97 -0.08 14 Malpeque 251 51-100 0.11 16 North Lake 64 61-89 0.18 18 Kouchibouguac 51 60-74 -0,06 22 Howards Cove 73 54-84 -0.19 24 Souris 202 59-109 0.14 25 Fortune 207 66-84 0.04 26 Beach Point 243 61-120 -0.02 27 Lismore 51 58-103 0.13 28 Ballantynes Cove 188 55-130 0.23 29 Port Hood 339 51-150 0.20 30 Margaree 1319 52-152 -0.05 31 Pleasant Bay 1040 54-130 0.04 to 0.23 (Table 3). In tliis study small lobsters (<70 mm CL) traveled as far as the large animals (>90 mm CL) (Fig. 2). There was no significant difference in the average dis- tance traveled between males and females in eleven out of nineteen sites (Table 4). Females traveled significantly farther than males (Table 4) at three sites located in the upper part of Baie des Chaleurs (Fig. 1, sites 1, 3, and 4) and one site on the northeastern tip of Prince Edward Is- land (Fig. 1, site 25). The only site where males traveled significantly farther than females was in Val Comeau (Ta- ble 4, Fig. 1, site 9). No significant differences were ob- served in the average distance traveled by berried fe- males compared with males or nonberried females in four out of nine sites where data were available for berried females (Table 4). The average distance traveled by ber- ried females was significantly shorter than that by both males and females in three sites (Table 4, sites 9, 16, and 25) and significantly farther only in Port Hood (Table 4, site 29). In Souris (Table 4, site 24), no significant differ- ence was obsei-ved for average distance traveled between males and berried females, but they were both significant- ly shorter than the average distance traveled for nonber- ried females. Only seven out of thirty-one sites had a sufficient num- ber of recaptures over multiple years to allow a compari- son between distance traveled and days at liberty. A sub- stantial decrease in the percentage of tags recaptured from the first to the second and third recovery periods was observed (Table 5), reflecting high exploitation rates by the fishery. There is no strong evidence of a positive relationship between the average distance traveled and days at liberty. No significant difference (P>0.05) was ob- sei-ved in the average distance traveled over time in Stone- haven (site 3), Anse-Bleue (site 4), Caraquet (site 5) and North Lake (site 16) (Table 5). The average distance trav- eled decreased significantly (P=0.0002) in Belledune (site 1), whereas it increased significantly in Souris (site 24) (P=0.0119) and Margaree (site 30) (^=0.004) over multi- ple years recovery (Table 5). Even in these areas where distances were significantly different, the differences were not large (2.4, 2.3, and 5.3 km). Also, the longest distance traveled obsei^ved for the second or third (or both) recovery periods was equal or less than the one obsei"ved for the first recovery period for all 7 sites (Table 5). Lobster movements in the southwestern GSL seemed to be restricted to short distances along the coast near shore in areas where the lobster habitat is restricted to a few kilometers from the shore and longer distances over a broader gradually sloping bottom (Fig. 3). In general, 93"^^ of lobster dispersions were limited to the 20-m ba- thymetry contour. The shorter average distances traveled (2.4-4.9 km. Table 2) were observed in part of Baie des Chaleurs (Figs 1 and 4, sites 3 and 4), the northeastern tip of Prince Edward Island (Fig. 1, sites 16. 24, and 25) and Cape Breton (Figs 1 and 5, sites 29, 30, and 31). Relative- ly short average distances (5.3-8.6 km. Table 2) were ob- served around northeastern New Brunswick (Figs 1 and 6, sites 1, 2, 5, 6, 8, and 9), the northwestern tip of Prince Ed- ward Island (Fig. 1, sites 12, 13, 20, and 21), eastern New Brunswick (Fig. 1, sites 17 and 18) and the eastern end of Comeau and Savoie: Movement of Homaius amencanus in the southwestern Gulf of St. Lawrence 185 60 • A 50-- ♦Vt n=1288 ^^ ^ ♦ ♦ mean distance = 8.0 km ♦«»<♦♦ . SD = 8.8 40- ♦ ^ * r=0.08 30 - 20 ■ 10 ■ 55 75 95 115 135 155 60 ■ B ? 50 ■ ♦ ♦ n = 73 « mean distance = 15.3 Distance traveled 3 o o o o SD = 13.2 ♦ • ♦ 50 60 70 80 90 60 - 50 - ♦ ♦^ n=1319 mean distance = 3.2 40 - 30 - SD = 5 1 • ♦ r=-0.05 • 20 ■ 10 - 0 4 5 65 85 105 125 145 165 Carapace length (mm) Figure 2 Relationship between the distance traveled and the size of the animal for American lobsters recaptured in (A) Caraquet, (B) Howard's Cove, and iC) Margaree. the Northumberland Strait (Fig. 1, sites 26 and 27). The longest average distances traveled (9.3-19.4 km, Table 21 were observed in the immediate vicinity of Miramichi Bay (Figs 1 and 6, sites 10 and ll),Malpeque Bay andTracadie Bay in northern Prince Edward Island (Fig 1, sites 14, and 15), St. Georges Bay (Fig. 1, site 28) and central Northum- berland Strait (Figs 1 and 7, sites 19, 22, and 23). Lobster movements in the Northumberland Strait were related to whether the site was located toward the center or at either extremity (either end) of the Strait. Lobsters at sites close to the external boundaries (Fig. 1, sites 17, 18, 20, 21, 26, and 27) traveled on average shorter dis- tances (5.4-8.6 km) than those located in the center of the Strait (Fig. 1, 12.2-19.4 km, sites 19, 22, and 23). Although tags were recovered in the western portion of Northum- berland Strait (LFA 25) at a different time (a different fishing season) compared with tag recoveries at the other LFAs, movements seemed to be related to the extent of shallow waters (<20 m) rather than the time of the recov- ery period. Discussion Lobster movements in the southwestern GSL are related to the local bottom topography and are depth-dependent, i.e. lobsters traveled on average longer distances in areas where the shallow waters (<20 m) extended farther from shore. We observed that on the narrow coastal shelf of western Cape Breton and in some areas in Baie des Chal- eurs, lobsters traveled on average less than 5 km compared with distances ranging from 9.3 to 19.4 km in the grad- ually sloping bottom of the Northumberland Strait and some shallow bays. Similarly, Templeman (1935) reported 186 Fishery Bulletin 100(2) Table 4 Comparison of the average distance traveled between the average distance traveled by male, female. and beiTied female American lobsters. The number of each site corresponds to its geographical position on the map in Figure 1. ;i = number of recaptured lobster with tags indicating release location for which distance traveled to recapture site could be calculated . Average distance [^-Test or Biological traveled ±SD Ki-uskal-Wallis' Tagging site category ;( (kml P 1 Belledune male female 360 176 7.0 ±5.9 9.0 ±8.3 0.0003 3 Stonehaven male female 316 264 2.1 ±2.8 2.7 ±3.3 0.0248 4 Anse-Bleue male female 127 105 2.1 ±3.3 3.8 ±5.2 0.0056 5 Caraquet male female berried female 746 487 55 8.1 ±9.1 7.7 ±8.6 5.5 ±6.6 0.0937 8 Le Goulet male female 45 45 5.3 ±2.7 5.2 ±2.7 0.4385 9 Val Comeau male female 279 218 7.7 ±7.1 6.3 ±5.6 0.0260^' berried female 5 1.3 ±0.7 0.0018 12 Seacow Pond male female 20 41 5.0 ±7.7 5.5 ±7.9 0.5800 13 Alberton male female 33 29 6.0 ±6.7 5.4 ±5.0 0.8849 14 Malpeque male female berried female 114 95 42 11.0 ±7.6 11.2 ±7.4 10.5 ±9.9 0.7310 16 North Lake male female 26 33 2.1 ±1.4 3.5 ±4.9 0.4700-' berried female 5 0.8 ±0.6 0.0243 18 Kouchibouguac male female 25 26 8.9 ±7.8 8.0 ±7.0 0.5847 22 Howard's Cove male female 43 31 17.3 ±12.3 12.8 ±14.0 0.1333 24 Souris male 131 3.6 ±5.3 0.5826' female 66 7.1 ±9.0 0.0140^' berried female 5 1.6 ±0.7 0.0355 25 Fortune male female 133 62 3.2 ±3.6 4.0 ±3.7 0.015.5- berried female 12 0.5 ±0.2 0.0001 26 Beach Pomt male female 178 65 6.4 ±6.2 10.1 ±13.4 0.4729 28 Ballantynes Cove male female 56 132 8.5 ±10.0 8.0 ±9.7 0.6193 29 Port Hood male female 139 1.58 4.4 ±3.1 4.3 ±2.6 0.7616- berried female 42 5.6 ±3.2 0.0294 30 Margaree male female berried female 449 553 317 3.1 ±4.9 3.5 ±5.0 2.8 ±5.0 0.1478 31 Pleasant Bay male female berried female 322 565 1.53 2.1 ±1.8 2.3 ±2.2 2.5 ±2.4 0.2638 ' The ['-test and the Kru *kal-Wallis test were used to compare t vo and three groups. respectively. ■■' [/-test between male.s and fe nales. ' U-test between males and be rried females. Comeau and Savoie: Movement of Homarus amencanus in the southwestern Gulf of St Lawrence 187 Table 5 The average and the longest distances trav eled bv American lobsters for sites with t igs returned over multiple years. The time between tagging and recapture of the animal (number ofd ays at hberty), and the number of observations («) are indicated. The number of each s te corresponds to its geogi aph ical position on the map in Figure 1. Number Average t/-test or Longest of days distance traveled Kruskal-Wallis' distance traveled Tagging site at hberty n (km) P (km) 1 Belledune 295-378 678-747 1048-1102 497 72 35 7.6 ±6.1 6.7 ±7.5 5.2+3.9 0.0002 44 37 17 3 Stonehaven 231-291 598-648 580 38 2.4 ±3.0 2.4 ±4.1 0.5827 28 17 4 Anse-Bleue 217-270 577-628 942-972 235 48 7 2.9 ±4.3 6.2+11.9 5.0 +7.3 0.2753 32 56 17 5 Caraquet 213-360 584-725 965-1057 1302 97 17 7.9+8.4 7.7 ±11.8 11.8 ±14.5 0.6416 51 49 50 16 North Lake 232-280 605-641 65 13 2.7 ±3.7 1.7 ±1.5 0.2075 24 6 24 Souris 230-285 584-625 202 11 4.7 ±6.9 7.0 ±4.2 0.0119 31 13 30 Margaree 250-308 622-671 147 13 3.3 ±3.7 8.6 ±7.6 0.004 23 21 ' Tlie C'-te. from the 1994 tagging project conducted in Stonehaven, New Brunswick. The release sites are indicated by a star symbol. Recently, a trawl survey conducted at a depth of 40 m in the Caraquet area (Bale des Chaleurs) over a 7-month pe- riod produced lobsters in mid-May, late-October, and No- vember, but not between June and early-October ( Comeau, personal obs.). Further, the recapture positions during the fishing season showed that lobsters tagged during these trawl surveys were recaptured along the coast at depths less than 20 m from Stonehaven to Miscou (Comeau, per- sonal obs.). This finding suggests that there is an inshore- offshore movement on the south shore of Bale des Chal- eurs similar to the one observed by Corrivault (1948) on the north shore of that bay. Unfortunately, our tagging projects were not designed to study this type of movement and did not allow us to speculate more on inshore-offshore movements. The lack of long-range movements across the south- western GSL could be explained by the presence of an extensive cold (<1.5°C) intermediate layer (CIL). In the southwestern GSL, the CIL is a large volume of water sandwiched between the coastal water and the deep wa- ter located in the Laurentian channel. The top of the lay- er ranges from 20 to 40 m depth from June to October and rises to the surface from January to April (Gilbert and Pettigrew, 1997). As it was hypothesized by Stasko (1980), there seems to be no advantage in long-distance movement to deeper water (>40 m) for lobsters in the GSL because it is cold (<1.5°C) in both summer and winter (CIL). Although lobsters can tolerate temperatures rang- ing from -1.5° to 30°C, at temperatures below 0°C they are in a state of hibernation, and below 5°C molt induc- tion is blocked (Waddy et al., 1995). It is clear that lob- sters can "tolerate" cold temperature but to be active they need warmer waters. Lobster movements of more than 40 km were rare in the southwestern GSL and movements exceeding 70 km through deep, colder waters (>40 m) were not observed. In contrast, lobsters from the Bay of Fundy, from coastal waters of southwestern Nova Scotia, and from coastal waters off New England can take advan- tage of warmer temperatures in deep waters in the Gulf of Maine and on the continental shelf during the winter (Campbell and Stasko, 1986). Campbell and Stasko ( 1985) and Campbell (1989) showed that lobsters tagged in the inshore waters of southwestern Nova Scotia traveled up to 240 km to the edge of the continental shelf off Georges Bank to depths below 200 m, seemingly without crossing a wide area of cold water. Similarly, lobsters tagged in the Bay of Fundy were also recaptured in deep waters at the edge of the continental shelf across the Gulf of Maine and along the coastal waters of the United States (Campbell and Stasko, 1986), at a distance of more than 780 km. These types of movements were not observed in the south- western GSL. Lobster movements between mainland New Brunswick, Prince Edward Island, or Cape Breton, and the Magdalen Islands, for example, have not been report- ed. The Magdalen Islands are an archipelago with a sub- stantial lobster fishery located in the middle of the south- western GSL, surrounded by >60 m depths at about 80 to 90 km from Prince Edward Island, and Cape Breton. Lob- ComeaLi and Savoie: Movement of Homarus amencanus in the southwestern Gulf of St- Lawrence 189 ly / -'}^ / ^ '/ w /yy 35. V / 46°53' - m. . , /- . ■ ' w Gulf of St. Lawrence ' ,■ / ' ' •'t^s^ / ' / ; JF Pleasant Bay /-■^ /'■// / - ---^, \ / ■■/ / / i Cape Breton 46°4r — /.://' / / . ; // / / 7 Nova Scotia \^ & /■//■■' / J 0 2-5 5 Kilometers 60 mm CL) in the southwestern GSL were not sex- or size-dependent, ex- cept for berried females. Similar to our findings, the re- sults of most recent studies do not indicate significant differences between the distance traveled in relation to size or sex for lobsters tagged in coastal waters (Fogarty et al, 1980; Ki'ouse, 1981; Campbell. 1982; Tremblay et al.. 1998). In contrast. Templeman (1935) and Bergeron ( 1967) suggested that lobster movements were sex-depen- dent, but neither author reported whether the differences were supported in a statistical or biological sense. Camp- bell and Stasko ( 1985. 1986) and Campbell ( 1989) indicat- ed that lobster movements were size-dependent because they observed that large mature animals (>95 mm CL) on average traveled significantly farther than small im- mature ones. They explained that mature animals would move more extensively to reach the warmest seasonal temperature to maximize their degree-days (the accumu- lative sum of daily mean temperatures recorded above 0°C) needed for somatic and gonadic development. This was not the case in our study. We observed, however, that on average berried females in the southwestern GSL traveled shorter distances than males and nonberried fe- males. Saila and Flowers (1968) indicated that the dis- tance traveled by berried females was related to their physiological state. In a tagging experiment, they cap- tured berried females on the continental shelf tagged, and released them in the inshore waters off the coast of Rhode Island at about 220 km from their captured posi- tion. When females were carrying eggs, they traveled on- ly short distances within the inshore waters. Once they shed their eggs, however, these females returned to the continental shelf where they were originally captured. Berried females tagged in inshore waters in the Magda- len Islands (Munro and Therriault. 1983). Jeddore Har- bour and Clam Bay. Nova Scotia (Jan'is. 1989). and New Hampshire (Watson et al., 1999) also traveled short dis- tances. In the Cape Cod area, berried females tagged in the inshore waters were reported to have traveled an av- erage distance of 30 km, mostly parallel to the coast (Mor- rissey, 1971; Estrella and Mornssey, 1997). Off Grand Manan Island, Campbell (1986, 1990) reported relatively small inshore-offshore migration, less than 15 km, for 75"^ of the berried females and attributed this migration to an effort to maximize egg development by exposure to warmer water. In general, it seems that the condition of carrying eggs could influence the extent of movements in the southwestern GSL, not the size or sex of the animal. In terms of fishery management, there is relatively lit- tle interaction between lobsters at different LFAs at the benthic level because lobster traveled on average small Comeau and Savoie: Movement of Homaivs amencanus in the southwestern Gulf of St, Lawrence 191 distances. The main concern of fishermen in the south- western GSL was the minimal legal size (MLS) disparity between LFA 24 (MLS of 63.5 mm CD located on the north side of Prince Edward Island and the other LFAs (MLSs from 65.1 to 70.0 mm CL). More precisely, they were interested in lobster movements in relation to time. i.e. if lobsters released in a given area in one season would be recaptured in the same area in future seasons. From our findings, lobsters do not move farther if they are at large for a longer period. Distances traveled by lobsters were not time-dependent for lobsters at large between 200 (with at least one winter season) and 1102 days. Results of our tagging studies were consistent with results from ear- lier studies carried out in the southwestern GSL and dem- onstrated that lobsters in their benthic stages have little long distance interaction. Hence, lobster movements in the southwestern GSL should have minimal consequences in terms of lobster management. Acknowledgments The authors wish to thank all fishermen from the south- western Gulf of St. Lawrence who returned lobster tags. We also want to thank Bruno Comeau, Daniel Ferron, Marc Lanteigne, Wade Landsburg, Manon Mallet. Pierre Mallet. Donald R. Maynard. Gilles Paulin and Guy Robichaud for their technical assistance in the field and during tag col- lection. We especially thank J. Mark Hanson. Marc Lan- teigne. and David Robichaud for critically reviewing the manuscript and three anonymous reviews for thoughtful suggestions that improved the quality of this manuscript. Literature cited Bergeron. T. 1967. Contribution a la biologie du homard {Homarus ameri- canus M. Edw. idesIles-de-la-Madeleine. NaturalisteCan. 94:169-207. Campbell, A. 1982. Movements of tagged lobsters released off Port Mait- land. Nova Scotia, 1944-80. Can. Tech. Rep. Fish. Aquat. Sci. 1136, 41 p. 1986. Migratory movements of ovigerous lobsters. Homa- rus americanus, tagged off grand Manan, eastern Canada. Can. J. Fish. Aquat. Sci. 43:2197-2205. 1989. Dispersal of American lobsters, Homarus americanus. tagged off southern Nova Scotia. Can. J. Fish. Aquat. Sci. 46:1842-1844. 1990. Aggregations of berried lobsters tHomarus america- nus) in shallow waters off Grand Manan, eastern Canada. Can. J. Fush. Aquat. Sci. 47:520-523. Campbell. A., D. E. Graham. H. J. MacNichol. and A. M. Williamson. 1984. Movements of tagged lobsters released on the conti- nental shelf from Georges bank and Baccaro bank, 1971-73. Can. Tech. Rep. Fish. Aquat. Sci. 1288, 16 p. Campbell. A., and A. B. Stasko. 1985. Movements of tagged American lobsters. Homarus americanus. off southwestern Nova Scotia. Can. J. Fish. Aquat. Sci. 42:229-238. 1986. Movements of lobsters [Homarus americanus) tagged in the Bay of Fundy, Canada. Mar. Biol.. 92:393-404. Comeau, M., W. Landsburg, M. Lanteigne, M. Mallet, P. Mallet, G. Robichaud, and F. Savoie. 1998. hohster {Homarus american us ) tagging project in Cara- quet (1993)— tag return from 1994 to 1997. Can. Tech. Rep. Fish. Aquat. Sci. 2216, 35 p. Cooper. R. A. 1970. Retention of marks and their effects on growth, behav- ior and migrations of the American lobster. Homarus amer- icanus. Trans. Am. Fish. Soc. 99:109-417. Cooper, R. A., and J. R. LTzmann. 1971. Migrations and growth of deep-sea lobster, Homarus americanus. Science 171:288-290. Cooper. R. A.. R. A. Clifford, and C. D. Newell. 1975. Seasonal abundance of the American lobster, Homa- rus americanus, in the Boothbay region of Maine. Trans. Am. Fish. Soc. 104:669-674. Corrivault, G. W. 1948. Contribution a I'etude de la biologie du homard {Homarus americanus) des eaux de la province de Quebec. Ph.D. diss., Universite Laval. Quebec. 283 p. Dow. R. L. 1974. American lobsters tagged by Maine commercial fish- ermen. 1957-59. Fish. Bull. 72:622-623. Ennis. G. P. 1984. Small-scale seasonal movements of the American lob- ster.Homarus americanus. Trans. Am. Fish. Soc. 113:336- 338. Estrella. B. T. and T D. Mornssey. 1997. Seasonal movement of offshore American lobster. Homarus americanus, tagged along the eastern shore of Cape Cad, Massachusetts. Fish. Bull. 95:466-476. Fogarty, M. J., D. V. D. Borden, and H. J. Russell. 1980. Movements of tagged American lobster, Homarus americanus, off Rhode Island. Fish. Bull. 78:771-780. Gilbert. D.. and B. Pettigrew. 1997. Interannual variability ( 1948-19941 of the CIL core temperature in the Gulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 54 (suppl. l):57-67. Jarvis. C. 1989. Movement patterns of late-stage ovigerous female lobsters {Homarus americanus Milne-Edwards I at Jeddore, Nova Scotia. M.S. thesis, Dalhousie Univ., Halifax. Nova Scotia, Canada, 148 p. Karnofsky, E. B.. J. Atema, and R. H. Elgin. 1989. Natural dynamics of population structure and habitat use of the lobster, Homarus americanus, in a shallow cove. Biol. Bull. 176:247-256. Krouse, J. S. 1981. Movement, growth, and mortality of American lob- sters, Homarus americanus, tagged along the coast of Maine. U.S. Dep. Commer, NOAA,Tech. Rep. NMFS SSRF- 747, 12 p. Lawton. P. and K. L. Lavalli. 1995. Postlarval. juvenile, adolescent, and adult ecology. In Biology of lobster. Homarus americanus (J. R. Factor, ed.). p. 47-81. Academic Press. New York. NY. 528 p. Miller, R. J., R. E. Duggan. D. G. Robinson, and Z. Zeng. 1989. Growth and movement o{ Homarus americanus on the outer coast of Nova Scotia. Can Tech. Rep. Fish. Aquat. Sci. 1716. 17 p. Moriyasu M., W. Landsburg, and G. Y Conan. 1995. Sphvrion tag shedding and tag induced mortality of the American lobster, Homarus americanus H. Milne Edwards, 1837 ( Decapoda, Nepbropidae 1. Cru.staceana 68: 184- 192. 192 Fishery Bulletin 100(2) Morrissey, T. D. 1971. Movements of tagged American lobsters. Homarus americanus, liberated ofTCape Cod, Massachusetts. Trans Am. Fish. Soc. 100:117-120. Munro, J., and J.-C. Themault. 1983. Migi'ations saisonnieres du homard (Homarus ameri- canus) entre la cote et les lagunes des Iles-de-la-Madeleine. Can. J. Fish. Aquat. Sci. 40:905-918. Robichaud, D. A., and P. Lawton. 1997. Seasonal movement and dispersal of American lob- sters. Homarus americanus, released in the upper Bay of Fundy Can Tech. Rep. Fish. Aquat. Sci. 2153, 21 p. Saila, S. B., and J. M. Flowers. 1968. Movements and behaviour of berried female lobsters displaced from offshore areas to Narragansett Bay, Rhode Island. J. Cons. Int. Explor Mer 31:342-351. Stasko, A. B. 1980. Tagging and lobster movements. Can. Tech. Rep. Fish. Aquat. Sci. 932:141-150. Templeman, W. 1935. Lobster tagging in the Gulf of St. Lawrence. J.Biol. Board Can. 1:269-278. 1936. Local differences in the life history of lobster on the coast of the maritime provinces of Canada. J. Biol. Board Can. 2:41-88. 1940. Lobster tagging on the west coast of Newfoundland 1938. Nfld. Dep. Nat. Resource Fish. Res. Bull. 8, 16 p. Tremblay, M. J., M. D. Eagles, and G. A. P. Black. 1998. Movements of the lobster, Homarus americanus. off northeastern Cape Breton Island, with notes on lobster catchability Can. Tech. Rep. Fish. Aquat. Sci. 2220, 32 p. Uzmann, J. R.. R. A. Cooper, and K. J. Pecci. 1977. Migration and dispersion of tagged American lob- sters, Homarus americanus. on the southern New England continental shelf U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-705, 92 p. Waddy S. L., D. E. Aiken, and D. P V. De Kleijn. 1995. Control of growth and reproduction. In Biology of lobster, Homarus americanus (J. R. Factor, ed. ), p. 217-266. Academic Press, New York, NY, 528 p. Watson III, W. H.. A. Vetrovs, and W. H. Howell. 1999. Lobster movements in an estuary Mar. Biol. 134:6.5— 75. Wilder, D. G. 1963. Movement, gi'owth and survival of marked and tagged lobsters liberated in Egmont Bay, Prince Edward Island. J. Fish. Res. Board Can. 20: 305-318. 1974. Inshore and offshore lobster stocks. Fish. Res. Board Can. MS Rep. 1293, 14 p. 193 Abstract— Short.spino thoniyhead (St-- Ixi.stolohii.-i ulaacanus) abundance was es- timated from 107 video transects at 27 stations recorded from a research sub- mersible in 1991 off southeast Alaska at depths rantrinf; from 165 to 355 m. Num- bers of invertebrates in seven major taxa were estimated, as was substrate type. Thornyhead abundance ranged from 0 to 7.5/100 m-, with a mean of 1.22/100 m'-, and was positively correlated with depth and amount of hard substrate. Invertebrate abundances were not sig- nificantly correlated with numbers of thornyheads. Shortspine thornyhead abundance estimates from this study were several times higher than esti- mates produced by bottom trawl sur- veys off southeast Alaska in 1990 and 1993, the two years of survey that encompassed the submersible tran- sects: however, the trend of increasing abundance with depth was similar in the trawl surveys and in the submers- ible transects, suggesting that trawl surveys systematically underestimate abundance of shortspine thornyheads. Shortspine thornyhead iSebastolobus alascanus) abundance and habitat associations in the Gulf of Alaska Page Else Lewis Haldorson School of Fisheries and Ocean Sciences University of Alaska 11120 Glacier Highway Juneau, Alaska 99801 E-mail address (for L Haldorson, coniaci author) lewhaldorsoniwuafedu Kenneth Krieger Auke Bay Laboratory National Manne Fishenes Service Juneau, Alaska 99802 Manuscript accepted 24 August 2001. Fish. Bull 100:193-199 (2002). Fish distributions are affected by physi- cal conditions such as depth, tempera- ture and substrate type and by biotic variables such as prey distribution, pred- ator presence, and habitat features (e.g. kelp forests). Patterns of habitat use are important factors in resource assess- ments, and stratification based on habi- tat characteristics are common features of survey design. For example, trawl sur- veys in the Gulf of Alaska are stratified by depth and general bottom type (flats, gullies, shelf break, slope) (Stark and Clausen, 1995). Assessment offish popu- lations based on traditional fishing gear, such as trawls or longlines, provide rel- atively economical surveys with broad geographic coverage; however, they pro- vide limited detailed information on spe- cies associations or habitat preferences (Matlock etal., 1991). Shortspine thornyhead (Sebastolobus alascanus) is a member of the family Scorpaenidae, which includes the rock- fishes {Sebastes — over 60 species in the northeast Pacific) and three species of thornyheads {Sebastolobus). Shortspine thornyhead range from Baja California, Mexico, to the Bering Sea and are found at depths to 1500 m (Moser, 1974). Off southeast Alaska, most fish sampled from 200-310 m depths were between 15 and 30 years old and had lengths of 25-35 cm (Miller, 1985). The maxi- mum age observed by Miller ( 1985 ) was 62 years, and age at 509!^ maturity was 12 years for both sexes. These life-histo- ry features are similar to those of rock- fishes, and such long-lived fishes are difficult to manage because they are easily overexploited and recover slowly from overfishing (Adams, 1980). Short- spine thornyhead have been commer- cially valuable in the Gulf of Alaska, where catch (including discards) ranged from 1298 to 2020 metric tons from 1991 tol996 (lanelli and ItoM. Bycatch of thornyhead has the potential of forc- ing closure of other high-value fisher- ies (lanelli and Ito^). Estimates of their abundance and size distribution are currently based on bottom trawl and longline surveys (lanelli and Ito^); how- ever, direct observations from submers- ibles can provide an alternate method for assessing abundance ( Krieger, 1993 ). Assessments of thornyheads and rock- fishes often result in population esti- mates with high variances; resulting in considerable uncertainty for assigning ' lanelli.J. N.andD. H. Ito. 1998. Status of Gulf of Alaska thornyheads iSebastolo- biis sp. ) in 1998. In Stock assessment and fishery evaluation report for the ground- fish resources of the Gulf of Alaska, p. 371-402. [Available from North Pacific Fishery Management Council, 605 West 4'*' St., Anchorage, Alaska 99501.] - lanelli,.:. N., and D.H. Ito. 1994. Thorny- heads. In Stock assessment and fishery evaluation report for the gi-oundfish re- sources of the Gulf of Alaska for 1995. (Available from North Pacific Fishery Man- agement Council. 605 4"' St., Anchorage, Alaska 99501.1 194 Fishery Bulletin 100(2) 1 40'W 55 30'N Figure 1 Locations of stations ( • ) where transects were run off southeast Alaska. The offshore line marks the 200-m isobath near the edge of the continental shelf harvest levels. Submersible obsei-vations with line-transect methods have been used in attempts to improve estimates of rockfish abundance and to understand rockfish habitat associations (Richards, 1986; Pearcy et al., 1989; Krieger, 1992; Stein et al.. 1992; O'Connell and Carlile, 1993). Our goal was to assess abundance and habitat use by shortspine thornyhead in the eastern Gulf of Alaska. based on data from existing video records taken during submersible transects. We estimated abundance of fish and invertebrates and quantified substrate type. We ex- plored the relationships between thornyhead abundance and both physical and biotic environmental variables and compared abundance estimates from submersible tran- sects with those from trawl survevs in the same area. Materials and methods Sources of data were video tapes of 107 bottom transects recorded at 27 stations during submarine dives in June 1991 on the outside coast of northern southeast Alaska from Cape Ommaney to Yakatat (Fig.l). All transects were conducted with the Delta submersible. This battery-pow- ered two-man submersible is 4.7 m long, dives to 365 m, and travels 2-6 kni/h. It is equipped with ten 150 W exter- nal halogen lights, internal and external video cameras. a 35-mm external camera, magnetic compass, directional gyro compass, and underwater telephone and transponder that allowed the submersible to be tracked from the sur- face support vessel. The surface vessel recorded LORAN fixes at the beginning and end of each transect. A pilot and observer formed the crew of the submersible: the pilot attempted to maintain the submersible within 0.5 m above the bottom at 3-4 km/h while the observer made obsei-va- tions through a starboard porthole. Video recordings were made from a downward project- ing external Hi-8 color video camera tilted obliquely for- ward on the starboard side and included a digital read- out of depth, temperature, and height above bottom. Data were collected either in strip transects by using the entire length of each transect, or in quadrats by freezing individ- ual frames from the video. The width of the strip transect was calculated from the height above bottom and field of view of the camera. The field of view formed a trape- zoid on the seafloor beginning almost directly beneath the camera and projecting forward; its maximum width iW) was estimated by calibrations performed on a subsequent cruise that had the same camera and camera configura- tion (Zhou and Shirley, 1997). When the submarine was resting on the bottom, the camera height iH) was 0.93 m and the width of the field of view (i.e. longest side of trap- ezoidal field) was 1.78 m. Transect width i \V) was estimat- ed by W= {1.78/0.93) H. llZhou and Shirley 1997)]. The height above bottom (H) was recorded at one minute intervals during each transect and mean height was used to estimate width ( W) of that transect. All dives were made during daylight between 0600 and 1900 h. At each station a series (usually 4) of parallel transects was run. and spac- ing between transects was about 200 m. Transect lengths Else et a\ Abundance of Sebaslolobus alasconus in the Gulf of Alaska 195 were calculated from the position fixes taken from the sur- face vessel. The entire length of each transect was viewed and all thornyhcads in the field of view were counted. Inverte- brates were also counted in seven categories: starfish, seap- ens, sea urchins, anemones, corals, sponges, and sea cu- cumbers. Depth and temperature were recorded once every minute during each transect and averaged over the com- plete transect. Substrate type was estimated by scoring vid- eo quadrats at one-minute intei-vals during the transect. A mylar grid of about twenty-five 50-mm squares was placed on the screen over the freeze-framed image, and substrate within the squares was scored in three categories of soft (mud, sand, and gravel), cobble, and rock-boulder, the lat- ter two of which were then combined into a single catego- ry, hard-bottom. The proportions of each category (soft and hard) for each transect were estimated as the mean from the one-minute quadrats. Abundance (number/100 m-) of shortspine thornyheads and invertebrate categories was es- timated for each transect by dividing the number counted by the area estimate ( W x Ti-ansect length x 100). To evaluate variables that may have affected thorny- head densities, we assembled a correlation matrix and a partial correlation matrix of three physical variables (depth, substrate, temperature) and transformed (log(.v-i-l)) biotic variables (shortspine thornyheads and seven inver- tebrate categories). Based on those matrices, we selected depth and substrate for further analyses. We used nonparametric procedures (Kruskal-Wallis and Mann-WTiitney) to determine if thornyhead densities var- ied among sampling stations and to evaluate the effects of substrate and depth. Substrate and depth were coded into nominal categories for use as independent variables in those analyses. We used Scheffe's (Zar, 1984) post-hoc test to evaluate between-categorv differences when Krus- kal-Wallis results were significant. An ANOVA factorial model was used to explore the joint effects depth and substrate on shortspine thorny- head abundance (transformed by logtx-i-l)). In addition, we used stepwise linear regression with thornyhead abun- dance as the dependent variable to evaluate the relative importance of all possible independent variables. 0.2 0.3 0.4 0.5 0,6 0.7 0.8 0.9 Distance (km) 50 B 40 c 30 0) o I 20 10- 0 100 150 200 250 300 350 400 450 Depth (m) 0.4 0.6 O.i Hard substrate Figure 2 Distribution of transects by (A) distance (km). (B) average depth, and (C) the proportion of bottom categorized as hard. Results Transect lengths ranged from 320 to 800 m and had a mean of 580 m (Fig. 2A). The mean transect depth ranged from 165 to 355 meters, and the highest numbers of tran- sects were at 200-250 m (Fig. 2B). Most transects (85'7f ) had substrate that was completely soft (silt, sand, and gravel) or hard (cobble and rock-boulder) (Fig. 2C). Abun- dance of shortspine thornyheads per transect ranged from 0 to 13.6/100 m'-; and the mean abundance at the 27 sta- tions ranged from 0 to 7.5/100 m- i Table 1). Difference in abundance among stations was very highly significant (P <0.0001, Kruskal-Wallis). A correlation matrix of all variables indicated that depth, substrate, and sponge abundance were related to variation in thornyhead abundance; however, the partial correlation matrix indicated that among those three vari- ables, substrate type and depth were most strongly re- lated to shortspine thornyhead abundance (Table 2). The high correlation of sponge and thornyhead abundances was apparently spurious because of the relationship be- tween sponge abundance and substrate type (Table 2). For further analyses of the relationship between thorny- head abundance, depth, and substrate, we coded depth into three nominal categories, <200 m, 200-300 m, and >300m, chosen to correspond to the depth intervals used in National Marine Fisheries Service (NMFS) triennial trawl surveys (Stark and Clausen. 1995). We also coded substrate into two nominal categories, soft bottom (>90'7( sand, mud, and grav- el) and hard bottom (>40% cobble and rock-boulder). Abun- dance increased with depth (Table 3), and differences that were highly significant occurred among three depth catego- ries (P<0.001, Kruskal-Wallis). Significant differences exist- 196 Fishery Bulletin 100(2) Table 1 Mean means numbers, with standard deviation, c of depth, bottom temperature, and f shortspine thornyheads ^Seha^tvlobus alascanus) per 100 proportion of the bottom categorized as hard substrate. m- at transect stations, with Station n No./lOO m- SD Mean depth (ml Temp. Proportion of bottom as hard substrate Lat. Long. 55 4 0.4 0.002 210.1 4.00 0.27 56°37'11" 135°57'50" 56 4 0.5 0.002 209.0 4.33 0.00 56°32'19" 135°54'39" 57 4 2,7 0.007 319.9 3.83 1.00 56°06'06" 135°39'72" 59 4 0.2 0.002 280.1 4.51 0.00 55°46'03" 134°77'97" 60 4 0.0 0 227.9 4.83 0.00 55°61'28" 134°55'86" 61 4 0.0 0 202.9 4.90 0.00 55°71'53" 134''76'89" 62 4 0.0 0 165.5 5.42 1.00 55°86'00" 134°84'00" 63 4 0.9 0.004 355.9 4.04 0.04 55°98'36" 134°93'97" 64 4 3.8 0.011 220.3 4.86 0.78 56°80'97" 135°80'97" 65 4 2.5 0.007 237.9 4.73 0.54 56°69'22" 135°87'69" 67 3 2.6 0.012 247.9 3.95 0.11 57°24'72" 1.36°29'31" 68 4 3.6 0.02 246.2 4.50 0.75 57°21'67" 136°25'03" 69 4 7.5 0.045 242.7 4.53 0.72 57°55'33" 1.36°55'22" 70 4 0.7 0.01 217.7 4.84 0.20 57°95'94" 137°21'61" 71 4 0.1 0.002 177.4 4.95 1.00 58°02'00" 137°10'00" 72 4 0.1 0.001 289.9 4.02 0.00 58°10'92" 136°98'39" 73 4 1.2 0.007 210.8 4.79 1.00 59''03'75" 14n3'75" 75 4 0.1 0.001 208.4 4.25 0.86 59°24'50" 14r62'14" 77 4 2.9 0.015 299.8 3.76 0.71 59°31'44" 142°11'42" 78 4 0.0 0 205.8 4.25 0.77 59°22'22" 141°70'86" 79 4 0.0 0 183.9 4.67 0.51 58°00'00" 140°50'00" 80 4 0.4 0.001 251.4 4.43 0.00 58''66'81" 139°37'33" 81 4 0.3 0.002 208.4 5.04 0.00 58°65'61" 139°56'08" 82 4 0.3 0.001 262.8 4.59 0.00 58°53'81" 139°6r.56" 83 4 0.0 0 218.9 4.81 0.00 58°35'00" 139°32'89" 84 4 0.0 0 172.4 5.32 0.76 58°22'00" 139°00'00" 85 4 2.4 0.016 215.2 5.26 0,74 .58°10'17" 138°65'3.3" ed between the shallowest depth category and both deeper depth groups (Table 3, Sheffe test). Substrate type also af- fected abundance (Table 3); transects with hard substrate had a significantly higher density of thornyheads than tran- sects on soft bottom (P=0.016, Mann- Whitney). There was a significant interaction (P=0.01) of depth and substrate in the ANOVA factorial analysis (Table 4) due to the sharp in- crease in thornyhead abundance in depths >200 m on the hard substrate than on the soft substrate (Fig. 3). Stepwise multiple regi'ession with all variables resulted in a model that incorporated two variables: substrate type and depth, with substrate type entered first. The final two- variable regression model was A = 0.017 S + 0.00016 D - 0.033, (/•■-=0.2081 where A = thornyhead abundance; S = proportion of bottom that is hard substrate; and D = depth (m). Thornyhead abundance increased with depth and amount of hard substrate, although there was an indication that abundance may have reached maximum levels at depths of200-300m(Fig. 3). Discussion Film (still or motion) and videotape recordings of tran- sects have been used to assess abundance of aquatic or- ganisms (Auster et al., 1989; Butler et al.'l. Potential biases in such data include systematic underestimation Butler, J. L., W. W. Wakefield, P. B. Adams, B. H. Robison, and C. H. Baxter 1991. Application of line transect methods to sui'vcying demersal communities with ROVs and manned sub- mersibles. Proceedings of the IEEE Oceans '91 Conference, p. 689-696. histitute of Electrical and Electronic Engineers, Pis- cataway, NJ. Else et a\ Abundance of Sebastolobus alascanus in the Gulf of Alaska 197 Table 2 Correlation (lower left diagon substrate type, temperature. al) and partial correlation matrices (upper right diagonal) for shortspine thornyhcad (SSTHi, depth, ind invertebrates. Significant correlations are indicated in bold. SSTH Depth Substrate Temperature Sea star Urchin Sea pen Anemone Coral Sponge Cucumber SSTH 0.35 0.37 0.08 -0.13 0.24 0.04 -0.13 0.04 0.27 -0.12 Depth 0.23 -0.34 -0.56 0.02 0.03 0.00 0.09 -0.16 -0.08 0.11 Substrate 0.32 -0.31 0.00 -0.12 -0.16 -0.01 0.39 -0.15 0.23 0.25 Temp. -0.10 -0.62 0.17 -0.02 -0.07 0.24 0.04 -0.05 -0.15 -0.22 Seastar -0.18 0.05 -0.16 -0.04 0.23 0.18 0.24 0.04 -0.06 0.01 Urchin 0.11 0.18 -0.21 -0.12 0.35 0.19 0.04 0.08 -0.13 -0.04 Sea pen -0.01 -0.09 0.04 0.13 0.36 0.33 -0.06 0.35 0.10 0.43 Anemone 000 -0.07 0.38 0.03 0.22 0.04 0.15 0.07 0,07 0.09 Coral -0.05 -0.09 -0.03 0.03 0.30 0.27 0.61 0.16 0.05 0.27 Sponge 0.37 -0.08 0.41 -0.02 -0.18 -0.16 -0.02 0.14 -0.04 -0.17 Cucumber -0.06 0.08 0.13 -0.15 0.27 0.20 0.59 0.24 0.53 -0.09 Table 3 Mean abundance (number/100 ni-) of shortspine thorny- heads in three depth categories ( Kruskal-Wallis, P<0.0001). and in two substrate categories (Mann-Whitney, P=0.016), with sample sizes and standard errors. Results of nonpara- metric analyses are included. Abundance u SE Depth 100-200 0.1 18 0.1 200-300 1.4 80 0.2 >300 1.9 9 0.4 Substrate soft 0.6 57 0.1 hard 1.9 50 0.4 due to gear avoidance or overestimation due to attraction of fish to the submersible. The behavior of shortspine thornyheads obsei^ed in the videotapes from our study indicated that they were not affected by the submersible. They typically remained relatively motionless unless the submersible came very close (almost touching the fish). It is also unlikely that they had moved away, or toward, the submersible, before coming into the camera field. On all dives there was an observer and another video camera recording a broader area, and there was no indication that shortspine thornyheads were responding to the sub- marine. The passive behavior of shortspine thornyheads in response to submersibles has been noted previously (Krieger, 1992). In transect assessments, the probability of detecting or- ganisms typically is not equal over the entire field of view. Detection is affected by such factors as lighting, orienta- tion of the organism and its reflectivity, sea floor relief. Table 4 Factorial analysis of effect of depth and substrate (propor- tion of bottom categorized as hard substrate) on shortspine thornyhead abundance (number/100 m^, log(.v-)-l) trans- formed). Variable df Sum of squares F-value P Depth 2 1.01 10.28 <0.0001 Substrate 1 0.35 7.10 0.009 Depth-substrate interaction 2 0.41 4.17 0.018 Residual 101 4.96 suspended particles, and size of the organism. Butler et al.-^ examined the detection functions for three types of fish (flatfish, hagfish, and thornyhead) from data collected off California. For thornyheads, the probability of detec- tion was relatively constant to a distance of about 180 cm. This distance is larger than the width of the transects in our study, indicating that our use of a constant detection function (probability of 1.0) was appropriate. We found that shortspine thornyheads preferred habi- tat with hard substrate. Submersible transects have been used to identify habitat used by rockfishes and thorny- heads in the northeast Pacific Ocean (Richards, 1986; Pearcy et al., 1989; Stein et al., 1992; O'Connell and Car- lile, 1993; Ki'ieger and Ito, 1999). Off Oregon, thornyheads were included in an assemblage of fishes associated with mud bottom (Stein et al., 1992), and Pearcy et al. (1989) observed that in deep water they occurred on mud bot- tom, but in shallower water were found over both rock and mud. We have no explanation for the differences in hab- itat association between Oregon and southeast Alaska. 198 Fishery Bulletin 100(2) Our obsei-vations are based on a larger number of tran- sects than those surveyed off Oregon, but the consistency of results from the two Oregon studies suggests that the difference is not due to a low sample size there. Our ob- servation that abundance of shortspine thornyheads in- creases with depth is consistent with results from trawl surveys off Alaska, Oregon, and California (Martin and Clausen, 1995, Stark and Clausen, 1995, Jacobson and Vetter, 1996). Off Oregon, their abundance was highest in the 200-400 m depth zone, and decreased sharply between 3- 2.5- DSoft R E 2- BHard £1 o o 1 15- .Q E i 1- 0.5- 0- 48 '■ 1 7ni3 ■ "' <200 20&-300 >300 Depth (m) Figure 3 Mean atiundance of shortspiiiL' thornyhead un soft and hard bottom substrates in three depth intervals. Sample size (number of transects 1 and standard errors are indicated m the number and vertical line over each bar 2.5 2 - E 01990 Trawl Q1991 Sub 01 993 Trawl o ? 1-5 . o 1 - 0 5 - 18 ^ <200 200-300 Depth (m) >300 Figure 4 Mean abundance of shortspine thornyhead in three depth inter- vals from submersible transects in 1991 and from the 1990 and 1993 trawl surveys off southeast Alaska. Sample size (number of transects) and standard errors of the submersible estimates arc indicated in the number and vertical line with each bar 400 m and 1400 m (Jacobson and Vetter, 1996). Off Califor- nia, the highest abundance of shortspine thornyheads oc- curred at 400-600 m, probably because of the warmer wa- ters off California (Jacobson and Vetter, 1996). Thus, our sampling depths covered the most important depth zones for this species in the waters off southeast Alaska, which are colder than those off Oregon. Bottom trawl sui-veys of fishes in the Gulf of Alaska are conducted triennially, and occurred one year before ( 1990) and two years after ( 1993 ) our submersible sui-vey ( Martin and Clausen, 1995; Stark and Clausen, 1995). For many species, trawl sui^vey results may be biased because some species may be herded by the trawl doors into the path of the net, resulting in overestimates of abundance when the "area-swept" method is applied (Ki'ieger, 1992). Other fish in the water column above the bottom may swim over the net and be underestimated by the survey (Balsiger et al., 1985). Escape routes under the foot rope and through the larger meshes in the trawl wings are also possibilities. The NMFS trawl survey does not cover exceptionally nig- ged rocky habitats that would destroy equipment; conse- quently, species that select high-relief habitats may be un- dei'estimated by the survey, whereas species that select low-relief soft-bottom habitats may be overestimated. Sub- mersible obsei-vations provide a means to quantify the bi- ases inherent in bottom trawl sui-veys. The depth catego- ries we used to analyze the submersible data matched the depth strata used in the NMFS triennial trawl sui-veys. Mean abundance of shortspine thornyheads in submersible sui^veys were several times higher that those in the 1990 and 1993 trawl surveys; however, the ratios of abundance in the three depth zones were very similar ( Fig. 4 ). We suggest that trawl sui-veys underestimate the abundance of short- spine thornyheads in the Gulf of Alaska but may be good indicators of relative abundance, patterns of dis- tribution, and stock trends. Acknowledgments We thank Dan Ito for his encouragement and support for this project. This research was supported by a grant (43ABNF401902) from the U.S. Dept. of Com- merce, National Marine Fisheries Ser\'ice, Auke Bay Laboratory, Juneau, Alaska. Literature cited Adams, P. B. 1980. Life history patterns in marine fishes and their consequences for fisheries management. Fish. Bull. 78:1-12 Auster, P. J., L. L. Stewart, and H. Spunk. 1989. Scientific imaging with ROVs: tools and tech- niques. Mar Technol. Soc. J. 23:16-20. Balsiger J. W.. D. H. Ito. D. K. Kimura. D. A. Somerton. and J. M Terry. 1985. Biological and economic assessment of Pacific ocean perch (Sebastea alutus ) in waters of Alaska. U.S. Dep. Commer, NOAA Tech. Memo. NMFS F/NWC-72, Else et al : Abundance of Sebastolobus alascanus in the Gulf of Alaska 199 NMFS Alaska Fish. Sci. Cent., Seattle, WA. 210 p. Jacobson, L. D.. and R. D. Vetter. 1996. Bathymctric demog^raphy and niche separation of thornyhead rockfish: Sebastolobus alascanus and Sebastol- obus altivelis. Can. J. Fish. Aquat. Sci. 53:600-609. Krieger, K. J. 1992. Shortraker rockfish, Sebastes horealis. observed from a manned submersible. Mar Fish. Rev. .54(4):34-36. 1993. Distribution and abundance of rockfish determined from a submersible and by bottom trawHng. Fish. Bull. 91 : 87-96. Kreiger, K. J., and D. H. Ito. 1999. Distribution and abundance of shortraker rockfish, Se- bastes borealis, and rougheye rockfish, S. aleutianus, deter- mined from a manned submersible. Fish. Bull. 97:264-272. Martin, M. H., and D. M. Clausen. 1995. Data report: 1993 Gulf of Alaska bottom trawl sui-vcy. U.S. Dep. Commor. NOAATech. Memo. NMFS-AFSC-59. 217 p. Matlock, G. C. W. R. Nelson. R. S. Jones, A. W. Green, T. J. Cody E Gutherz, J. Doerzbacher 1991. Comparison of two techniques for estimating tilefish, yellowedge grouper and other deepwater fish populations. Fish. Bull. 89:91-99. Miller, P. 1985. Life history of the shortspine thornyhead. Sebasto- lobus alascanus. at Cape Omnaney, S.E. Alaska. M.S. thesis. Univ. Alaska, Juneau, AK, 75 p. Moser, H. G. 1974. Development and distribution of larvae and juveniles of Sebastolobus (Pisces; Family Scorpaonidac). Fish. Bull. 72:865-884. O'Connell, V. M.. and D. W. Carlile. 1993. Habitat-specific density of adult yelloweye rockfish Sebastes ruberrimus in the eastern Gulf of Alaska. Fish. Bull. 91:304-309. Pearcy W. G.. D. L. Stein, M. A. Hixon, E. K. Pikitch, W. H. Barss, and R. M. Starr 1989. Submersible observations of deep-reef fishes of Heceta Bank, Oregon. Fish. Bull. 87:955-965. 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. Fishes 17: 13-21. Stark, J. W., and D. M Clausen. 1995. Data report: 1990 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-AFSC-49, 221 p. Stein, D. L., B. N. Tissot, M. A. Hi.xon, and W. Barss. 1992. Fish-habitat associations on a deep reef at the edge of the Oregon continental shelf Fi.sh. Bull. 90:540-551. Zar, J. H. 1984. Biostatistical analyses. Prentice-Hall. Englewood Cliffs, NJ. 718p. Zhou, S., and T. C. Shirley. 1997. Distribution of red king crabs and Tanner crabs in the summer by habitat and depth in an Alaskan fjord. Invest. Mar Valparaiso 25:59-67. 200 Abstract— Analysis of 32 years of stan- dardized survey catches ( 1967-98 ) indi- cated differential distribution patterns for the longfin inshore squid iLoligo pealeii) over the northwest Atlantic U.S. continental shelf, by geographic region, depth, season, and time of day. Catches were greatest in the Mid- Atlantic Bight, where there were sig- nificantly greater catches in deep water during winter and spring, and in shallow water during autumn. Body size generally increased with depth in all seasons. Large catches of juve- niles in shallow waters off southern New England during autumn resulted from inshore spawning observed during late spring and summer; large propor- tions of juveniles in the Mid-Atlantic Bight during spring suggest that sub- stantial winter spawning also occurs. Few mature squid were caught in sur- vey samples in any season; the major- ity of these mature squid were cap- tured south of Cape Hatteras during spring. Spawning occurs inshore from late spring to summer and the data suggest that winter spawning occurs primarily south of Cape Hatteras. Geographic and temporal patterns in size and maturity of the longfin inshore squid iLoligo pealeii) off the northeastern United States Emma M.C. Hatfield Steven X. Cadrin Northeast Fisheries Science Center National Manne Fishenes Service, NOAA 166 Water Street Woods Hole, Massachusetts 02543 Present address (for E.M.C, Hatfield) FRS Marine Laboratory Victoria Road Aberdeen ABll 9DB Scotland, United Kingdom E-mail address (for E M C Hatfield) e hatfield a marlab ac uk Manuscript accepted 17 mav 2001 Fish. Bull. 200-213 (2002). ' The longfin inshore squid, Loligo pea- leii. is distributed in the northwest Atlantic from Canada to the Carib- bean (Cohen, 19761. Within its range of commercial exploitation (from southern Georges Bank to Cape Hatteras) the population is considered to be a unit stock ( NEFC ' ), although heterogeneous subpopulations may exist (Garthwaite etal., 1989). North of Cape Hatteras, L. pealeii migrate seasonally. The migration has been described as a movement offshore during late autumn (so that the species can overwinter in warmer waters along the edge of the continental shelf ) and a return movement inshore during the spring and early summer (Summers, 1969; Serchuk and Rathjen, 1974; Tib- betts, 1977). Murawski (1993) defined L. pealeii as a member of a migratory, warm-water group of species, centered primarily in mid-Atlantic waters (par- ticularly in the spring), that make in- shore and northward migrations in the spring and offshore and southward mi- grations in late autumn. Geographic patterns in Northeast Fisheries Science Center (NEFSC ) sur- vey catches, from the Gulf of Maine to Cape Hatteras, show that L. pealeii are distributed over the entire conti- nental shelf (from inshore to offshore) in the autumn, are concentrated at the edge of the continental shelf and at the southern end of the survey area during winter and spring, and are con- centrated inshore in summer (Sum- mers, 1967; 1969; Serchuk and Rathjen, 1974; Vovk, 1978; Lange, 1980; Whita- ker, 1980; Lange and Waring, 1992). Analyses of sui-vey catches indicate that depth, time of day, and tempera- ture all influence cross-shelf distri- bution patterns (Summers, 1969; Ser- chuk and Rathjen, 1974; Lange and Warmg, 1992; Murawski, 1993; Brod- ziak and Hendrickson, 1999). Diel cor- rection factors have been applied to sui-vey indices in various studies to ad- just nighttime bottom trawl catches to daytime equivalents (daytime catches are higher when squid are concentrat- ed close to the bottom) (Lange and Sis- senwinel983; Lange and Sissenwine-). Research by Lange and Waring (1992) and Brodziak and Hendrickson (1999) demonstrated that the diel differences were size specific and that further con- sideration of these differences in cor- rection factors was warranted. Until recently, L. pealeii was thought to have a life span of up to three years, and the stock was assessed accord- ingly (Sissenwine and Tibbetts, 1977; Lange, 1981; Lange and Sissenwine, 1 NEFC (Northeast Fisheries Center I. 1986. Report of the second NEFC stock assess- ment workshop. NEFC Lab. Ref Doc. 88- 02, 114 p. [Available from NEFSC, 166 Water Street, Woods Hole. MA 02,543.1 - Lange, A. M. T, and M. P. Sissenwine. 1977. Lo/(go pea/ei stock status. North- east Fisheries Science Center Lab. Ref Doc. 77-28, 9 p. [Available from NEFSC, 166 Water Street, Woods Hole, MA 002543.] Hatfield and Cadrin: Geographic and temporal patterns in size and maturity of Loligo pcaleii 201 44"N 42" 40° 38° 1983; Lange''; Lange et al.''). Recent advances in the use of statoliths for age determination of squid (see reviews in Rodhouse and Hatfield, 1990; Jereb et al., 1991; Jackson, 1994) have enabled now esti- mates of life span to be derived for L. pcaleii ( Macy, 1995; Brodziak and Macy, 1996; Macy"*), which in- dicate that the life span of L. pealeii can be less than nine months. Back-calculations of hatching date from age data revealed that there is more than just a spring-summer spawning component of the population (Brodziak and Macy, 1996; Ma- cy''), with a small proportion of squid hatching dur- ing winter This winter spawning is presumed to occur offshore (Brodziak and Macy, 1996), in the vicinity of the submarine canyons along the edge of the northeastern U.S. continental shelf, from Hudson Canyon up to Georges Bank (Fig. 1). The possibility of winter spawning was raised initially by Summers ( 1969), based on length-frequency da- ta, but squid were not presumed to spawn until their second year because their growth was as- sumed to be too slow to allow spawning during their first summer Our study reports on two studies: 1) an analy- sis of survey data from spring and autumn NEFSC surveys from 1967 to 1998, and from winter NEF- SC surveys from 1992 to 1998, to describe gi-oss distribution patterns of L. pealeii over the north- west Atlantic continental shelf from Cape Hat- teras to the Gulf of Maine; 2) some results of a field study initiated in 1997 to investigate geographic and seasonal patterns of growth and maturity to determine if the winter spawning component off the northeastern United States can be defined by time and area. Materials and methods Survey analysis Length-frequency data for L. pealeii were analyzed from NEFSC bottom-trawl surveys conducted in the autumn (generally from mid-September to late October) from 1967 to 1997; in the spring (generally from March to early April) from 1968 to 1998; and in the winter (generally in Febru- ary) from 1992 to 1998. Data collection and processing and archiving methods are described by Azarovitz (1981). In 66'W 36 /y Hudson Canyon r/ MAB MAB - Mi(d-Atlantic Bight NE - New Englatitj GOM - Gulf of Maifie /Cape Hatteras Ja i—i , I , L_ J , I I I . L 3 Lange, A. M. T. 1984. An assessment of the long-finned squid resource ofFthe northeastern United States. Northeast Fisher- ies Science Center (NEFSC) Lab. Ref Doc. 84-.37. 24 p. [Avail- able from NEFSC, 166 Water St. Woods Hole, MA 02543.] '' Lange, A. M. T, M. P. Sissenwine, and E. D. Anderson. 1984. Yield analysis of long-finned squid, Loligo pealei (LeSueur). Northwest Atlantic Fisheries Organization (NAFO) SCR Doc. 84/1X/97, 29 p. ^ Macy W. K. 1995. Recruitment of long-finned squid in New- England (USA) waters. ICES CM 1995/K:35, 18 p. [Available from W. K. Macy, Graduate School of Oceanography, Univ. Rhode Island, South Ferry Road, Narragansett, RI 02882]. Figure 1 Map of the survey areas for longfin inshore squid off the northeastern coast of the United States 1 1967-98). the autumn and spring sui-veys the same trawl-sampling gear (Yankee-36 trawl) has been used since 1967, except during 1973-81, when a Yankee-41 (high rising) trawl was substituted in the spring surveys. In the winter sui-veys the trawl gear was larger and the Gulf of Maine was not sampled. The NEFSC sui-vey area was divided into two geograph- ic regions (the region north of Hudson Canyon to the Gulf of Maine [designated New England, NE] and the region south of Hudson Canyon to Cape Hatteras (designated Mid-Atlantic Bight [MAB]) and into four bottom depth zones (27-55 m, 56-110 m, 111-185 m, and 186-366 m [Fig. 1] ). The 1-26 m depth zone was not sampled in NEF- SC offshore surveys, but "inshore" strata were added to the survey in 1972. For the spring and winter surveys, the combined effects of annual abundance (numbers of squid per standardized trawl haul), survey stratum, and time of day (night, 20:00-03:59; dawn and dusk, 04:00-07:59, 16:00-19:59; and day, 08:00-15:59), as described by Brodziak and Hen- drickson (1999) for the autumn survey, were analyzed to determine adjustment factors for diel differences in log- transformed survey catches of prerecruit squid (<80 mm dorsal mantle length [ML], the minimum size in com- mercial catches) and recruits (>80 mm ML). The derived factors were then used to adjust all survey catches to 202 Fishery Bulletin 100(2) their daytime equivalent. The size groups (<80 mm ML, >80 mm ML) were chosen to allow comparisons with re- sults from previous studies (e.g. Lange, 1980: 1981; Lange and Sissenwine, 1983: Brodziak and Hendrickson. 1999: NEFC': Lange and Sissenwine-: Lange'^). Alternative anal- yses were performed for different size groups (<50 mm ML, >50 mm ML and <100 mm ML, >100 mm ML) to assess the sensitivity of the results to the choice of size groups. The combined effects of geographic region (NE and MAE), depth zone, and year on sui-vey catches were tested by using generalized linear models (GLM) to derive main effects and coefficients for each sui-v^ey. Paii^wise compari- sons were tested by using a Mest with Bonferroni adjust- ments (Sokal and Rohlf, 1995) to compare specific regions, seasons, and depth zones. All tests were analyzed at the 5'y( significance level. Differences between seasons and re- gions were tested between autumn and spring sui'veys for the years 1968 to 1997, between autumn and winter for 1992 to 1997, and between spring and winter surveys from 1992 to 1998. Proportion of catches <50 mm ML were ana- lyzed to evaluate the relative distribution of juvenile L. pealeii. Biological analysis Subsamples of 50-100 individuals were obtained from five different survey time series: NEFSC autumn (September- October 1997), winter (February 1998), and spring (March 1998), inshore Massachusetts (Howe^) (October 1997), and Connecticut (Johnson") (Long Island Sound, May 1998). The samples were analyzed from each of five depth zones (1-26 m, 27-55 m, 56-100 m, 111-185 m, 186-366 m), within each of three geographic regions (Gulf of Maine [GOM): Georges Bank-Southern New England, north of Hudson Canyon |SNE|: and Mid-Atlantic Bight, see above [MABl). A fourth region, south of Cape Hatteras (SOH) was added later. Each sample comprised a nonrandom selection of lengths to represent the size range present in a tow. In total, 2156 individuals were subsampled from 53 sun'ey tows. Sexes were determined and specimens were measured to enable the morphometric maturity analyses of Macy (1982): each individual squid was also weighed on a top-loading balance to 0.1 g. The morphometric method uses a suite of length measurements for female and male squid to determine maturity stage (measured on a scale of 1 to 4, where 1 is immature and 4 is fully mature). Oppor- tunistic commercial samples from early winter (December 1998 and January 1999) were also analyzed (118 individu- als) to bridge the temporal gap in sui-vey coverage. Data on dorsal mantle length (ML, mm) and total body mass (BM, g) for each maturity stage were used to esti- mate proportions for each maturity stage across the length ^ Howe, A. B. 1989. State of Massachusetts inshore bottom trawl survey. Atlantic States Marine Fisheries Conimi-~sion (ASMFC) Spec. Rep. 17:33-38. [Available from ASMFC, 1444 Eye Street, N.W., sixth floor, Washington. DC 2000.5.] " Johnson, M. 1994. State of Connecticut marine finfish trawl survey. Atlantic States Marine Fisheries Commission (ASMFC ) Spec Rep. 35:24-26. [Available from ASMFC, 1444 Eve Street, N.W., sixth floor, Washington, DC 20005.1 and weight range. These proportions were used to deter- mine the sizes at which squid of both sexes changed from one maturity stage to the next. Maturity-at-length data were weighted by diurnally ad- justed catch-at-length data for each depth zone and region to provide population-weighted maturity patterns, assuming that sui"vey length distributions accurately represent rel- ative proportions of population components. Catch-weight- ed data were analyzed to derive 1 ) the patterns of matu- rity for each sex at different times of the year: 2 ) estimates of proportions of each maturity stage sampled by survey: 3 ) mean length for each maturity stage of each sui-vey: 4 ) mean length for each region of each survey: and 5) mean length for each depth stratum of each survey. A small pro- portion (8.4^^ ) of survey catches <50 mm ML were not sub- sampled: these were assigned to the juvenile stage. Catches at larger sizes, which were not subsampled (6.8% of sui-vey catches), were removed from the analysis because sex or maturity stage could not be assigned with any degree of cer- tainty Individual squid, or size classes of squid, were not weighed during NEFSC surveys: therefore the maturity da- ta from the biological analysis could only be catch-weighted by length because length was measured on a random sub- sample of squid caught at each station in NEFSC surveys. Results Survey analysis Patterns of diurnal distribution were different among sea- sons surveyed (Table 1). In winter surveys, from 1992 to 1998, prerecruit (i.e. <80 mm ML) catch was lower at night and during dawn and dusk than during daylight hours (65'/J and 81% of daytime catch, respectively). However, for recruits (i.e. >80 mm ML), catch was higher both at night and at dawn and dusk than during the day ( 131% and 115% of daytime catch respectively) in winter sui-veys. In autumn and spring sui'veys, from 1968 to 1998. both prerecruits and recruits showed a lower catch at night and during dawn and dusk than by day: recruits showed a lesser diurnal variation than prerecruits. Results from analyses with dif- ferent size groups (<50 mm ML, >50 mm ML and <100 mm ML, >100 mm ML) were very similar, suggesting that the interaction of size and time of day is gradual. Catch rates varied significantly by season (Table 2). During winter and spring, survey catches were greater in the MAB by a factor of approximately four (Fig. 2). How- ever, there was no significant difference in sui'vey catches between geographic regions in autumn (Fig. 2). Pairwise comparisons showed that mean number-per-tow was sig- nificantly greater in autumn than in spring within both geographic regions and was greater in autumn than in winter in the NE. There were no significant differences be- tween autumn and winter means in the MAB nor between spring and winter means in either the MAB or the NE. Catch by depth, pooled over the MAB and NE, varied by season (Table 3). Pairwise comparisons of each depth for each season showed that winter and spring survey catch- es were lowest in the shallowest stratum (27-55 mi, in- Hatfield and Cadrin Geographic and temporal patterns in size and matunty of Loligo pealeii 203 Table 1 Ki'lativo catch rates for small i<80 mm dorsal mantle length (ML) 1 and large (>80 mm ML) L bottom-trawl surveys, 1967-98, by time of day (in relation to catch rates during daytime). ollfil l)C(ll< 11 in three seasonal NEFSC Winter Spring Autumn Time of day <80 mm >80 mm <80 mm >80 mm <80 mm >80 mm Night 0.65 1.30 0.51 0.72 Dawn and dusk 0.81 1.14 0.79 0.92 Day 1.00 1.00 1.00 1.00 0.09 0.46 1.00 0.34 0.83 1.00 Table 2 Results of generalized linear model (GLM) of !■ of freedom; SS=sum of squares; F=F-statistic ui-vey mean numbers-per-tow by P=probability ), for Loligo pealeii vear, depth zone, and geographic off the northeast LInited States region basec (df=degrees on NEFSC bottom-trawl sui-vey data 1967 -98. Season and effect df Type III SS Mean square F P Winter year 6 42.51 7.09 2.71 0.0133 depth 3 339.62 113.21 43.36 0.0001 region 1 233.73 233.73 89.53 0.0001 Spring vear 30 409.47 13.64 4.44 0.0001 depth 3 1696.53 565.51 183.79 0.0001 region 1 983.51 983.51 319.64 0.0001 Autumn year 30 557.85 18.59 5.17 0.0001 depth 3 1134.09 378.03 105.20 0.0001 region 1 9.12 9.12 2.54 0.1113 Table 3 Diurnally adjusted, mean numbers-per-tow of Loligo pea- leii from the three annual NEFSC bottom-trawl surveys, 1967-98, by season. Depth »ne (m) 27-55 56-110 111-185 186-366 Winter 42.9 103.1 215.5 30.3 Spring 42.6 90.7 342.5 91.9 Autumn 853.3 352.8 377.2 66.5 creased to peak values in deeper strata ( 111-185 m, great- er than 10 times the catches in the shallowest strata), and were low in the deepest stratum (>185 m). Converse- ly, autumn survey catches were highest in the shallowest stratum (27-55 m) and lowest in the deepest stratum (>185 m). These patterns were also generally significant within both geographic regions (Table 4). Table 4 Diurnally adjus tec , mean numbers-per tow of Loligo pea- \ leii from the th ree annual NEFSC bott om-trawl sui-veys. 1967-98, by area (NE=north of Hudson Canyon to the 1 Gulf of Maine; MAB=south of Hudson Canyon to Cape Hatteras. - Depth zone ( m I 27-55 56-110 111-185 186-366 Winter NE 6.0 38.9 160.4 28.7 Winter MAB 83.5 396.2 510.7 36.0 Spring NE 2.3 24.7 259.5 81.4 Spring MAB 86.8 392.3 787.3 129.7 Autumn NE 644.2 365.9 392.8 72.4 Autumn MAB 1082.5 293.1 293.7 45.5 In the pairwise comparisons of the winter and spring surveys, catches were significantly greater at 111-185 m than in other depth zones. In the 27-55 m depth zone, au- 204 Fishery Bulletin 100(2) tumn catches were considerably higher than in winter or spring. For all other depth zones and sui-vey comparisons, the differences were not significant. In the autumn survey, the proportion of small squid (<50 mm ML) was highest at 27-55 m depths (over 50% of the sampled squid in that depth zone in the NE and almost 75*^7^ of the squid in that depth zone in the MAB, Table 5). Proportions of small squid at greater depths were consid- erably lower. These patterns show higher relative recruit- ment into the population in the shallow waters of the con- tinental shelf in the autumn. Similarly in the MAB during winter and spring, small squid form a higher proportion of squid sampled in the two shallowest depth zones than at greater depths. A higher percentage of small squid was present in spring than in winter, with over GC/f of squid sampled in the MAB from '27-55 m being <50 mm ML. However, the highest propor- tion of small squid in the NE during winter and spring was at intermediate depths. Biological analysis The raw data of numbers sampled for each sex, length, and maturity stage are given in Table 6. For all seasons combined, the ML at 50'7( maturity during 1997-98 was approximately 200 mm ML for females and males (Table 7, Hatfield and Cadrin Geographic and temporal patterns in size and matunty of Loligo pealeii 205 Table 5 The percentage o( Loligo pealeii <50 mm ML in each depth zone, for each region and survey, and for the number these data were available (NE=north of Hudson Canyon to the Gulf of Maine; MAB=south of Hudson Canyon to of years for which Cape Hatteras). Depth zone (m) Winter Spring Aut umn Winter Spring Autumn NE (No. of years) NE (No. of years) NE (No. of years) MAB (No. of years I MAB (No. of years ) MAB (No. of years) 27-55 3 6 17 20 51 31 33 7 64 30 73 31 .56-110 11 7 40 31 19 31 27 7 48 31 27 31 111-185 27 7 24 31 31 31 14 7 22 31 9 31 186-366 7 3 2 31 14 30 10 3 14 31 2 31 Table 6 Numbers oi Loligo pealeii measured, for each sex and maturity stage, rity was based on a four-stage scale for sexual maturity for each sex length. from samples taken for biological analy where 1 was immature and 4 was fully sis in 1997- mature. ML 99. Matu- = mantle ML ( mm ) Females Males Stage 1 Stage 2 Stage 3 Stage 4 Total 2 Stage 1 Stage 2 Stage 3 Stage 4 Total S 30 1 1 40 5 5 2 2 50 31 1 32 18 18 60 59 3 1 63 30 5 1 36 70 73 17 1 91 32 22 2 1 57 80 57 30 1 1 89 27 37 12 1 77 90 51 45 5 101 10 45 16 4 75 100 27 51 2 3 83 2 50 17 2 71 110 10 62 5 6 83 2 35 30 8 75 120 8 86 14 4 112 1 32 40 4 77 130 3 79 6 4 92 24 37 11 72 140 3 71 10 8 92 14 39 10 63 1.50 2 38 16 14 70 14 43 10 67 160 25 6 7 38 14 28 8 50 170 13 2 7 22 12 31 7 50 180 7 2 3 12 6 18 7 31 190 9 1 2 12 3 19 9 31 200 1 5 6 5 9 14 210 1 1 1 3 2 8 11 220 1 3 4 4 7 12 230 5 6 240 1 5 7 250 1 3 5 260 1 1 270 1 1 3 3 280 1 1 290 1 1 Total 330 539 74 69 1012 124 318 345 126 913 206 Fishery Bulletin 100(2) 00 1 A 80- 60- 40- 20- 0- 200 300 nn-i E 80- A V ^ 60- / / / 40- male / / I — A 20- IL-^ — 3+4 —2+3+4 100 100 200 Mantle length (mm) 300 100 200 Body mass (g) Figure 3 Percent mature ofLoligopealeii for dorsal mantle length (ML mm) (Al and wet body mass IBM gi iB). The percentage at each maturity stage is shown for female ML (C) and BM (D) and for male ML lE) and BM iFi. Table 7 Size at maturity for Loligo pealcii in mantle length iML, mml and body mass IBM, g), by sex. i — denotes a missing value). Female Male Proportion mature ML BM ML BM 25% 166 111 50% 207 — 75% 2.38 — 184 113 196 146 241 184 Fig. 3, A and B). In terms of body mass, the size at 50% maturity for males was approximately 1.50 g, but, accord- ing to our samples, female maturity did not seem to be as closely associated with body mass (e.g. even squid in the heaviest size class were less than 50% mature). In females, maturity stage 2 was reached at a relatively small size (Fig. 3, C and D). To reach stage 3 requires a considerable increase in length or body mass, whereas fe- males in stage 4 are neither much longer, nor heavier than stage-3 females. Thus the transition from stage 3 to stage 4 (full maturity) takes place over a lesser period of somatic growth (and therefore possibly a shorter time period) than the transition from stage 2 to stage 3. In Macy's ( 1982) maturity stage notation, stage-3 females have no mature Hatfield and Cadrin Geographic and temporal patterns in size and maturity of Loligo pealeii 207 Table 8 DiLU'iially adjusti'd. t iTspond to unscxcd j (SOH: south of Cape atch-vveighted uvcnilcs and a Hatteras). mean numbers-per four-stage scale for tow o{ Li ill i^d H'xual maturi pealeii sampled in each maturity stage. Maturity stages cor- ty for each sex where 1 was immature and 4 was fully mature Juvenile Female Me le Totals Stage 1 Stage 2 Stage 3 Stage 4 Stage 1 Stage 2 Stage 3 Stage 4 Autumn 240.0 94.0 16.0 0.0 2.0 64.0 49.0 4.0 3.0 472.0 Winter 5.0 13.9 10.9 0.4 0.2 4.8 6.1 6.2 1.1 48.6 Spring plus SOH 26.3 12.7 10.0 2.3 1.4 5.4 5.0 5.0 2.4 70.5 Long Island Sound 436.0 0.0 52.0 378.0 176.0 161.0 0.0 196.0 144.0 1543.0 Commercial 0.0 5.0 155.0 0.0 0.0 8.0 198.0 331.0 15.0 712.0 Totals 707.3 125.6 243.9 380.7 179.6 243.2 258.1 542.2 165.5 2846.1 Table 9 1 Maturity patterns in Loligo pealeii (percentage of sample at each sex and stage derived from diurnally adjusted, catch-weighted mean numbers-per-tow. Maturity stages correspond to unsexed juveniles and a four stage scale for sexual maturity for each sex where 1 was immature and 4 was fully mature (SOH: south of Cape Hatteras). % juvenile Female Male % stage 1 % stage 2 % stage 3 % stage 4 % stage 1 % stage 2 7f stage 3 7f. stage 4 Autumn 50.7 20.0 3.4 0 0.5 13.6 10.3 0.9 0.6 Winter 10.3 28.7 22.4 0.7 0.4 9.8 12.5 12.7 2.2 Spring plus SOH 37.3 18.0 14.2 3.2 1.9 7.7 7.0 7.1 3.5 Long Island Sound 28.3 0 3.4 24.5 11.4 10.4 0 12.7 9.3 Commercial 0 0.7 21.8 0 0 1.1 27.8 46.5 2.1 oocytes (therefore they would not be considered to be close to maturity) and stage-4 females are fully mature. In comparison, the transition of male squid (Fig. 3, E-F) from stage 2, to stage 3, to stage 4, seems more evenly spaced, with a more gradual development seen over the course of the maturation process. If anything, there seems to be a greater transition from stage 3 to stage 4, than from stage 2 to stage 3. The difference between stage-2 and stage-3 males is measured only as elongation of the testis in conjunction with a reduction in the ratio of mantle cir- cumference to mantle length. In stage-4 males "elongate mature spermatophores are visible both in the Needham's sac and the penis." In subjective terms, stage 2 represents definitely immature, stage 3, maturing, and stage 4, fully mature males. Thus the rate at which full maturity is ap- proached is very different between the sexes. The raw data (numbers sampled in each size class, sex, and maturity stage ) were then catch-weighted to be repre- sentative of the squid sampled in the different surveys and from commercial data. Catch-weighted proportions of fe- male and male L. pealeii at each maturity stage are shown in Tables 8 and 9. In the autumn and spring surveys, and in the May Long Island Sound (LIS) samples, juvenile squid were the most abundant stage within the sampled popula- tion. In winter surveys the juveniles were one of the least abundant stages. Mature squid were never abundant in the NEFSC survey subsamples, nor in the inshore autumn (October) Massachusetts survey (combined with NEFSC data for autumn surveys). Most mature squid were seen in the LIS samples. In NEFSC surveys, the majority of both sexes were immature and stages 1 and 2. In LIS samples more squid were either stage 3 or 4 than immature. No ma- ture females were observed in the commercial samples, al- though a small proportion of males were mature. The catch-weighted mean ML for each maturity stage, by season, is given in Table 10. There was little difference between the size of juvenile squid between seasons. For both sexes, squid at stages 1, 2, and 3 were all longest in the autumn samples and shortest in the spring and LIS samples. Conversely, mature squid of both sexes (stage 4) were considerably larger in the spring than in autumn. In the LIS samples, mature feinale squid were the same size as in spring survey samples; mature male squid, on the other hand, were smaller than in the spring survey but larger than the autumn sui-vey samples. Commercial sam- ples showed a larger size for each maturity stage sampled. 208 Fishery Bulletin 100(2) Table 10 Mean dorsal mantle length (ML, mm) o{ Lotigo pealeii for each sex and maturity stage (con-esponding to four stage scale for sexual maturity for each sex where 1 was immature and 4 was fully mature), from d weighted mean numbers-per-tow data (SOH: south of Cape Hatteras). unsexed juveniles and a lurnally adjusted, catch- Juvenile Female Male Stage 1 Stage 2 Stage 3 Stage 4 Stage 1 Stage 2 Stage 3 Stage 4 Autumn 34 77 133 115 70 105 161 94 Winter 41 71 122 166 164 54 93 131 169 Spring plus SOH 37 58 90 122 139 61 83 114 143 Long Island Sound 47 90 79 138 70 83 109 Commercial 83 160 79 150 169 180 Table 11 Mean dorsal mantle length (ML mm) and perce ntage of each sex foi each sample of Loligo pcalcii by depth stratum per survey. Derived from diurnally adjusted. catch-wei ghted tow data. Depth zone Depth zone Depth zone Depth zont Depth zone <27m "'r 27-55 m % 56-110 m I'/ 111-185 m % 186-366 m % Autumn juvenile ML 32 73 29 T2 47 58 48 5 0 Autumn female ML 67 13 87 13 95 25 92 21 123 68 Autumn male ML 67 14 105 15 111 17 80 74 118 27 Winter juvenile ML 42 10 39 4 39 5 46 5 Winter female ML 100 68 84 63 97 47 111 47 Winter male ML 105 22 93 33 97 48 131 48 Spring juvenile ML 35 42 43 3 37 54 40 3 43 1 Spring female ML 60 32 113 61 78 26 100 47 123 48 Spring male ML 59 25 129 36 90 18 104 49 136 51 Long Island Sound juvenile ML 47 28 Long Island Sound female ML 97 39 Long Island Sound male ML 86 33 Commercial female ML 158 23 Commercial male ML 161 78 The distribution of the catch-weighted mean ML, by sex, by depth zone, for each season separately is shown in Ta- ble 11. For juvenile L. pealeii, in autumn, at 1-55 m depth, there was little difference in size at depth. In the 56-110 m depth zone, juveniles in autumn were considerably larg- er than at shallower depths. In winter, there was evidence for a slight increase in the size of juveniles with increasing depth. In spring survey samples, juveniles were similar in size across the depth range sampled. Female and male L. pealeii were generally smaller in the shallowest depth zone (1-26 m, only sampled in au- tumn and spring sui-veys), and much larger at depths greater than 185 m, for each survey. There was no clear pattern for intermediate depths. In autumn surveys, squid were generally smaller at 27-55 m depth than in deeper water. In winter and spring, however, squid at this depth were longer than at 56-185 m. The LIS samples showed larger mean sizes for each group at <27 m depth than in autumn and spring samples. In the autumn survey, squid of all maturity stages (ex- cept juveniles) were generally largest in the south (MAB) and smallest in the north (SNE and GOM, Table 12). Some of this distribution may have been an artifact of the sam- pling design because no squid were sampled in the MAB region in the 1-26 m depth zone, whereas this zone was sampled in the SNE, and was the only zone for which data were available for the GOM region. In the winter survey, the general pattern was the re- verse of that seen in the autumn. In this survey squid were generally smaller in the south (MAB) than in the Hatfield and Cadrin: Geographic and temporal patterns in size and maturity of Loligo pealeii 209 Table 12 Mean dorsal mantl e length (ML. mml by sex and maturitv stage (corresponding to unsexed juveniles and a four-stage scale for sexual maturitv for each sex, where 1 was immature and 4 was fully mai\\re)oi Loligo peateii by region per survey. Derived from | diurnallv adjusted. catch-weighted mean n umbers-per-tow data. MAB = Mid-Atlantic Bight; SNE = Georges Bank-Southern New | England, north of Hudson Canyon; GOM = Gulf of Mexico; SOH = South of Cape Hatteras. Juvenile Female Ma le Stage 1 Stage 2 Stage 3 Stage 4 Stage 1 Stage 2 Stage 3 Stage 4 Autumn MAB •23 98 132 162 69 120 172 170 Autumn SNE 38 71 130 85 69 102 147 96 Autumn GOM 33 68 65 69 71 Winter MAB 42 67 116 186 180 59 94 128 151 Winter SNE 50 80 127 148 159 52 94 134 171 Spring MAB 37 62 93 127 154 66 82 110 150 Spring SNE 41 51 95 132 151 54 88 125 183 Spring SOH 56 70 11 137 54 81 98 162 Long Island SNE 47 90 79 138 70 83 109 Commercial SNE 83 160 79 150 169 180 north (SNE only, no GOM samples were taken in the win- ter survey). The exception in winter were the few stage-3 and stage-4 females sampled. In the spring survey, the same pattern as in the winter survey was observed; squid in the south were smaller than in the north. In the spring survey, samples available from south of Cape Hatteras (the limit of the MAB samples) followed the same trend because the observed mean sizes were smaller than the IVLAB samples. The exception to this were maturity-stage- 1 squid. Discussion Survey analysis The high proportion of small squid (<50 mm ML) in the winter and spring surveys corroborated the occurrence of an early winter hatching event, documented from age data determined by squid statolith analysis (Brodziak and Macy, 1996; Macy^). Mean numbers-per-tow of juvenile squid in the MAB were considerably higher than in the NE in all seasons surveyed. Recruitment of squid into the population was highest in autumn, but juvenile squid were distributed more widely over the continental shelf in the spring. Per- haps the MAB component of the L. pealeii stock was larger because it is more stable — a result of the higher propor- tion of squid recruited into the area each winter, spring, and autumn. Murawski (1993) inferred a centering of the population in the MAB subject to the issue that portions of the stock are outside the area of the NEFSC surveys. Our data sug- gested that the area south of Cape Hatteras may play an important role in reproductive dynamics and recruitment to the population, suggesting that a considerable portion of the stock is south of the surveyed area, particularly dur- ing winter and spring. South of Cape Hatteras a second loliginid species, L. plei. is abundant (Roper et al., 1984). In our study, all Loligo specimens were examined carefully to ensure that only L. pealeii were measured and included in the biological analyses. Diel differences in catches of L. pealeii have been ob- sei-ved in a number of studies (Summers, 1969; Serchuk and Rathjen, 1974; Roper and Young, 1975; Sissenwine and Bowman, 1978; Lange and Sissenwine, 1983; Lange and Waring, 1992), where catches were consistently high- er in daytime than at night. To account for diel effects on minimum swept-area estimates of L. pealeii biomass and stock size, nighttime catches were adjusted to daytime equivalents by using the diel correction factors of Lange and Sissenwine (1983). However, these correction factors were not size specific. Brodziak and Hendrickson (1999) applied size-specific diel correction factors to squid from the autumn sui-vey (1967-94), splitting the data into pre- recruits (<80 mm ML) and recruits (>80 mm ML). In the autumn surveys the nighttime catch of prerecruits was only 8.79c of the daytime catch. The nighttime catch of recruits was 34'7( of the daytime catch. These differences were attributed to the different feeding behavior of juve- nile and adult squid, in that juvenile squid might need to undertake more vertical migrations at night to meet their higher metabolic requirements. In our present study, pre- recruit nighttime catch differed to a lesser degree from the daytime catch in the winter and spring (65% in the win- ter surveys, and 51% in the spring surveys) than in the autumn. In the winter surveys, nighttime catches of re- cruits exceeded daytime catches by 31%. In spring, night- time tows showed a lower catch of recruits, 72% of day- time values, than in winter. Patterns of diel differences reported in another study by Lange and Waring (1992) for spring catches were similar to those in our study. Their 210 Fisher/ Bulletin 100(2) reported autumn catches (Lange and Waring, 1992) were higher at night than in our study, but still lower than day- time catches. The behavior of squid at both prerecruit and recruit sizes therefore appears to be different in the winter and spring than ui the autumn. The prerecruit nighttime catch in winter and spring was half, or more than half of day- time catches, as opposed to 9% in autumn. For recruits, there was an even gi-eater difference among seasons. In winter, almost 1.5 times as many squid were caught at night, than by day. In spring the nighttime catch was T2''k of the daytime catch, twice the proportion of the autumn catch. Vovk (1978; 1985) and Maurer and Bowman ( 1985) have documented large changes in the diet of squid in dif- ferent seasons, relating these changes in feeding activity and dietary preference to changes in the size composition of the squid population, movements of squid in search of food concentrations, seasonal abundance of prey, and envi- ronmental conditions that affect both prey and predator Vovk (1985) noted that in autumn L. pealeii are daytime predators and do not feed extensively at night when they occur at shallow depths. In autumn, squid are more abun- dant near the seafloor by day (Brodziak and Hendrickson, 1999). Vovk (1985) also noted that feeding activity was generally low from December to April and related this to possible prey abundance. Perhaps the different diel behav- ior patterns of L. pealeii in winter and spring, when they appear to be more available to capture by bottom-trawls, are related to a lower prey abundance and a requirement for more time to be spent searching for prey. Some of these differences might be temperature related as well. In the autumn, with strong vertical stratification of the water column, there may be some physiological benefit for squid to move off the bottom at night into warmer waters. In winter and spring when there is no vertical stratification, no advantage is conferred by a strong diel migration as seen in the autumn. Comparisons in performance and catchability of the two trawl nets used in the spring and autumn sui-\'eys (Yankee 36 and Yankee 41) have been conducted (Sissenwine and Bowman, 1978), but any differences in catchability were confounded by diel and vessel differences. In our study, we corrected for diel differences, and vessel differences were not found to be significant (NEFSC/'*). Squid catches are more abundant in the autumn sur- veys than in winter and spring (Serchuk and Rathjen. 1974; Lange and Waring, 1992; Lange and Sissenwine-). This difference may be related to the recruitment of large numbers of small squid, present m shallow water, in au- tumn. Temperature, however, is a major factor limiting distribution. In winter and spring much of the continental shelf water is below the preferred temperature minimum for the species (ca. 8°C) (Summers, 1969; Serchuk and Rathjen, 1974; Murawski, 1993); therefore squid are ap- parently less abundant than in the autumn surveys when *• NEFSC (Northeast Fisheries Science Center). 1996. Report of the '21st Northeast Regional Stock Assessment Workshop (SAW 21). Center Reference Document 90-05, 200 p. jAvail- able from NEFSC, 166 Water St., Woods Hole, MA 02543.1 temperature is not a limiting factor The greatest differ- ences between autumn and spring or winter sui-veys are most apparent in the NE, whei-e a large portion of the stock may be outside the sui-veyed area owing to tem- perature limits (e.g. off the shelf or south of Cape Hat- teras). In the MAB there are few differences in catches be- tween spring and autumn at depths deeper than 27-55 m. Spring numbers are higher than autumn numbers from 111 to 185 m. The patterns between autumn and winter catches are similar to each other at that depth. Catches are always lower in winter than in autumn, but the differ- ences arc only significant from 27-110 m depth. In winter and spring sui-veys, catches were highest from 1 1 1 to 185 m, both in the NE and the MAB. Catches were higher, in both surveys, in the MAB than in the NE. These patterns also were obsei-ved for L. pealeii in survey anal- yses from 1967 to 1971 (Summers, 1969; Serchuk and Rathjen, 1974), from 1970 to 1977 (Lange, 1980), and from 1975 to 1986 (Lange and Waring, 1992). This finding may imply that geographic distribution is relatively stable for L. pealeii. during February and March, at least within the areas sui-veyed. In autumn sui-veys. Serchuk and Rathjen (1974). Lange (1980). and Lange and Waring (1992) also reported highest catches in the MAB. However, Serchuk and Rathjen ( 1974) showed a relatively higher abundance of squid taken from 56-110 m depth than that obsei-ved in our study, where mean numbers per tow in the depth zone 27-55 m were greater than three times higher than those in the zones 56-110 m and 111-185 m. The difference be- tween the 27-55 m zone and deeper strata was less notice- able in the NE, although mean numbers-per-tow were al- most twice as high in the 27-55 m zone. The differences may result from diurnal adjustments, or from the inclusion of more years of data in the sui-vey database. Lange ( 1980) found mean numbers-per-tow in the NE autumn to be high- est from 111 to 185 m depths. Biological analysis We found that squid mature at greater lengths than pre- viously reported for L. pealeii (NEFSC"*). Figure 3 (C-F) shows that using stage 3 or greater to indicate maturity may be an adequate proxy for females. For example, the size at which 50*^^ of females are mature is 198 mm with stages 3 and 4, and 207 mm with only stage 4 (Fig. 3C). Such a proxy may be valuable for samples with few obser- vations of stage-4 females (e.g. the body mass at 50% maturity is 120 g with stages 3 and 4 [Fig. 3D1 ). For males, however, the size difference between stage 3 and stage 4 is considerable, and substantial somatic growth is required to develop from stage 3 to stage 4. Com- bining the two maturity stages in males is therefore un- supported biologically, and the combined data would un- derestimate size at 50'^r maturity. That mature squid are largest in winter and smallest in autumn samples (and intermediate in size in the early INEFSCl and late |LIS| spring) has been noted previous- ly (Summers, 1971; Lange, 1980; Macy, 1980). Prior to the availability of age data for the species, the size dif- ferences at maturitv were ascribed to different year class- Hatfield and Cadrin Geographic and temporal patterns in size and maturity of Loligo pealeii 211 es (Summers, 1971; Lange, 1980; Macy 1980). The obser- vation that immature squid are larger in autumn than in winter and spring was documented by Lange (1980) but not interpreted. Some of this variabiHty might be ex- plained as a function of temperature. If L. pealeii have a life-span of 9-12 months (Brodziak and Macy, 1996), then females that are mature in September-October sam- ples would have hatched between November and January. Females that are mature in March would have hatched around May or June. Brodziak and Macy (1996) showed that squid hatched between November and April had a lower growth rate than those that hatched between May and October. Recent laboratory studies on small L. pealeii have indicated that squid grow significantly faster at high- er temperatures (Hatfield et al., 2001), in accord with results described for other cephalopod species (Octopus bimaculoideg — Forsythe and Hanlon. 1988; L. forbesi — Forsythe and Hanlon. 1989; Sepia officinalis — Forsythe et al., 1994). These laboratory studies also found that the effect of temperature on growth is most pronounced dur- ing the early life cycle of these cephalopods, nominally the first three months. If temperatures experienced by L. pealeii hatching in May and June are warmer than tem- peratures experienced by squid hatching from November to January, then the gi'owth potential will be lower for winter-hatching squid (seen as mature squid in autumn sui^veysi, resulting in the obser\'ed lower size at maturity in autumn versus winter and spring. The same phenom- enon would explain the size differences of immature squid among seasons. The large immature squid caught in the autumn survey are probably the same squid that become the large, mature squid in the winter, spring, and LIS sam- ples. The small immature squid in the spring are probably those squid which become the small mature squid seen in autumn survey samples. If the winter-hatching squid are from southern spawning events, then the temperature difference between winter-spawned and summer-spawned squid may not be very large. However, age data for L. pealeii show that growth rates are generally slower for winter hatched squid. Also, growth studies on L. forbesi have shown that a temperature difference of just 1"C can change growth rates of squid by 2% body weight/day and produce a threefold difference in weight at 90 days after hatching (Forsythe and Hanlon, 1989). The high numbers of juvenile squid in the autumn sur- vey were from protracted spawning in inshore waters some 4-5 months previously (documented since Verrill [1882] first reported inshore spawning). The high propor- tion in spring therefore reflects a period of spawning, pos- sibly also some 4-5 months earlier, around September or October of the previous year. Juvenile squid in winter and spring survey samples denote the presence of a hatching component other than the main inshore autumn compo- nent. The high proportion obser\'ed in LIS (May) samples probably reflects an extended winter-spring spawning pe- riod because squid of the size found in Long Island Sound in May could not have been the result of that seasons in- shore spawning. The inshore spawning season does not usually begin until late April and incubation time may require up to 4 weeks at the temperatures at that time of year (27 days at 12°C, McMahon and Summers, 1971). Brodziak and Macy ( 1996) showed a pattern of year-round hatching, which is consistent with the patterns suggested by our data. Summers (1969), Serchuk and Rathjen ( 1974), and Brod- ziak and Hendrickson (1999) all reported an increase in size of L. pealeii with increasing depth. We found the smallest squid in the shallowest water and the largest squid in the deepest water, confirming that nearshore wa- ters of the continental shelf are a preferred habitat for ju- venile L. pealeii during the autumn (as described by Brod- ziak and Hendrickson [1999]). The pattern of ontogenetic descent exhibited by other loliginid species (L. vulgaris — Worms, 1983; L. gahi—Hatfiek\ et al., 1990; L. vulgaris reynaudii — Augustyn et al., 1992) is consistent in L. pea- leii, but less marked at intermediate depths. In winter and spring, mean length is generally higher in the NE and lower in the MAB. In autumn, mean length is generally lower in the NE and higher in the MAB. Lange (1980) showed a similar pattern for immature females from the autumn survey. Males, however, showed the op- posite pattern. Langes (1980) winter and spring survey data showed the same pattern as our study for immature females and all males, except the fully mature squid. Commercial samples from early winter (December and January) contained no mature female squid that might produce the winter hatching component evident from age data and aggi-egated sui-vey length-frequency distribu- tions. Egg masses are only found consistently in one small offshore area (off Chesapeake Bay) by commercial fisher- men in the early winter (see Fig. 4). However, the com- mercial samples were from the southern edge of Georges Bank, and the survey data suggest that the winter recruit- ment originates from the southern part of the MAB. The scarcity of mature squid in NEFSC sui-vey samples sug- gests that sampling did not occur consistently in the right areas or seasons to identify major spawning peaks. Whita- ker 1 1978) documented that about 40% of males were fully mature in January and February in the region south of Cape Hatteras, off the coast of South Carolina. There was a large proportion of mature squid in samples from March and April, with 74% of females and 56% of males fully ma- ture. There is evidence for both spawning and hatching in the SNE from March to April; as in 1999, egg masses were caught incidentally from northeast of Hudson Can- yon and up towards the southern flank of Georges Bank, at depths of about 200 m, from mid-March to late April (StommelP). In the 1998 Massachusetts spring survey, in mid-May, a high abundance of small squid, about 30 mm ML, were caught south of Marthas Vineyard, at depths of <27 m (senior author, personal obs.). These observations suggest that spawning is probably protracted, from early to late winter, and early winter spawning is more dom- inant in the southern end of the U.S. continental shelf Thus, the available fishery-dependent samples may not be indicative of the population. The information on age and maturity in Brodziak and Macy (1996) was derived from ^ Stommell, M. 1999. Personal, comniun. ¥W Nobska, Woods Hole, MA 02543. 212 Fishery Bulletin 100(2) data collected from the winter fishery, most of which occurs north of the MAB region. A more structured design is required to address some of these issues. The entire size and maturity stage range needs to be sampled across the geographic range of each survey and these data should be augmented with opportunistic sampling outside the area or time frame in which the sui-veys are carried out. In summary, results from these two studies complement each other to reveal patterns of re- productive dynamics for L. pealeii that have been suggested previously in other studies but that have never thoroughly been investigated. The high frequency of small squid present in spring sui-vey catches indicates that winter spawning is indeed an important component of reproduction for the population, and biological analyses sug- gest that this winter spawning occurs primarily south of Cape Hatteras, rather than in the vicin- ity of the offshore submarine canyons along the edge of the northeastern U.S. continental shelf from Hudson Canyon up to Georges Bank. Acknowledgments 76" 44"N 42" 40" 36= i\. 74" TTTJ- 72° -J- 70° 68° 66 W ( V '^1 V \ / f May-Jul f Jun-Aug A-' Hudson Canyon ■s. Mar-Apr . May-Jun // 38"k..>-y We would like to extend our thanks to John Gal- braith of NEFSC. Ai-nie Howe of the Massachu- setts Division of Marine Fisheries, Dave Simpson of the State of Connecticut Marine Division, for coordinated field sampling from sui-veys; Glenn Goodwin and Gier Monsen of Seafreeze and Joe Mantineo of Ruggierio Seafoods for commercial samples. Lars Axelson, Glenn Goodwin, and Jim Ruble shared their obsei^vations on spawning grounds of Loligo pealeii with us. Chad Keith and Lynette Suslowicz provided technical help in the cutting room. Jon Brodziak provided the code for. and assistance with, the diel correction analyses. We would like to thank John Boreman. Steve Murawski, and Fred Serchuk for their thoughtful and instructive reviews of this manuscript. EMCH would like especially to acknowledge Steve Murawski's guidance and help throughout the course of her Research Associate- ship with NMFS. We also thank other scientific personnel on NEFSC, MA, and CT bottom-trawl surveys and last, but most definitely not least, the thousands of squid that have sacrificed their lives for the advancement of sciencel Fund- ing for EMCH was provided though the National Research Council Research Associateship program. Literature cited Augustyn. C. J.. M. R. Lipinski, and W. H. H. Sauer. 1992. Can the Loligo squid fishery be managed effectively? A synthesis of research on Loligo vulgaris reynaudii. In Benguela trophic functioning (A. I. L. Payne, K. H. Brink, K. H. Mann, and R. Hilborn. eds.i. p. 90.3-918. S. Afr. J. Mar. Sd. 12. f V >. i*, Dec-Jan -May Atlantic Ocean Figure 4 Persistent spawning areas and seasons for Loligo pealeii. as indicated from incidental catches of eggs in commercial squid trawls. Azarovitz, T. R. 1981. A brief historical review of the Woods Hole Laboratory trawl survey time series. Can. Spec. Publ. Fish. Aquat. Sci. 58:62-67. Brodziak, J. K. T, and L. C. Hendrickson. 1999. An analysis of environmental effects on survey catches of squids. Loligo pealeii and Ille.x illecehrotiua in the north- west Atlantic. Fish. Bull. 97:9-24. Brodziak. J. K. T. and W. K. Macy III. 1996. Growth of the long-finned squid. Loligo pealei. in the northwest Atlantic. Fi.-^h. Bull. 94:212-236, Cohen, A. C. 1976. The systematics and distribution of Loligo i Cephalop- oda, Myopsida) in the western North Atlantic, with descrip- tions of two new species. Malacologia 15:299-67. Forsythe, J. W., and R. T. Hanlon. 1988. Effect of temperature on laboratory growth, repro- duction and life span of Octopun himaculoides. Mar. Biol. 98:369-379. 1989. Growth of the Eastern Atlantic squid. Loligo forbesi Steenstrup (Mollusca: Cephalopoda). Aquaculture Fish. Manag. 20:1-4. Forsythe. J. W., R. H. DeRusha. and R. T. Hanlon. 1994. Growth, reproduction and life span of Sepia officina- lis (Cephalopoda: Mollusca) cultured through seven consec- utive generations. J. Zool. 233:175-192. Hatfield and Cadnn: Geographic and temporal patterns in size and maturity of Loligo pealeii 213 Garthwaite, R. L., C. J. Berg, and J. Harrigan. 1989. Population genetics of the common squid Loligo pealei LeSueur. 1821. from Cape Cod to Cape Hatteras. Biol. Bull. 177:287-294. Hatfield, E. M. C, R. T. Hanlon, J. W. Forsythe, and E. P. M. Grist. 2001. Laboratory testing of a growth hypothesis for juve- nile squid Loligo pealeii (Cephalopoda: Loliginidae). Can. J. Fish. Aquat. Sci. 58:845-857. Hatfield, E. M. C, P G. Rodhouse, and J. Porebski. 1990. Demography and distribution of the Patagonian squid {Loligo gahi. d'Orbigny) during the austral winter J. Cons. Int. Explor. Mer 46:306-312. Jackson, G. D. 1994. Application and future potential of statoHth incre- ment analysis in squids and sepioids. Can. J. Fish. Aquat. Sci. 51:2612-2625. Jereb, P. S., S. Ragonese. and S. von Boletzky. 1991. Squid age determination using statoliths: proceedings of the international workshop held in the Istituto di Tec- nologia della Pesca e del Pescato (ITPP-CNR), Mazara del Vallo, Italy 9-14 October 1989. N.T.R.-I.T.RP (Note Tec- niche e Reprints dellTnstituto di Tecnologia della Pesca e del Pescato) special publication 1, 128 p. Lange,A. M.T. 1980. The population dynamics of the squids {Loligo pealei and Illex illecebrosus) from the northwest Atlantic. M.S. thesis, Univ. Washington, Seattle, Wa, 178 p. 1981. Yield-per-recruit analyses for squid, Loligo pealei and Illex illecebrosus. fi-om the Northwest Atlantic. J. Shell- fish Res. 1:197-207. Lange, A. M. T, and M. P. Sissenwine. 1983. Squid resources of the northwest Atlantic. In Ad- vances in assessment of world cephalopod resources. FAO Fish. Tech. Pap. 231:21-54. Lange, A. M., and G. T Waring. 1992. Fishery interactions between long-finned squid (Lo//go pealei) and butterfish tPeprilus triacanthiis) off the north- east USA. J. Northwest Atl. Fish. Sci. 12:49-62. Macy W. K. 1980. The ecology of the common squid, Loligo pealei LeSueur, 1821, in Rhode Island waters. Ph.D. diss., Univ. Rhode Island, Narragansett, RI, 249 p. 1982. Development and application of an objective method for classifying long-finned squid, Loligo pealei. into sexual maturity stages. Fish. Bull. 80:449-459. 1995. The application of digital image processing to the aging of long-finned squid, Loligo pealei, using the stato- lith. In Recent developments in fish otolith research (D. H. Secor, J. M. Dean, and S. E. Campana, eds.), p. 283-302. Univ. South Carolina Press, Columbia, SC. Maurer, R.O., and R. E. Bowman. 1985. Food consumption of squids t Illex illecebrosus and Loligo pealei) off the northeastern United States. NAFO (Northwest Atlantic Fisheries Organization! Sci. Coun. Studies 9:117-124. McMahon. J., and W. C. Summers. 1971. Temperature effects on the developmental rate of squid (Loligo pealei) embryos. Biol. Bull. 141:561-567. Murawski, S. A. 1993. Climate change and marine fish distributions: fore- casting from historical analogy. Trans. Am. Fish. Soc. 122: 647-658. Rodhouse, P.G., and E.M.C, Hatfield. 1990. Age determination in squid using statolith growth increments. Fish. Res. 8:323-334. Roper, C. F. E., M. J. Sweeney, and C. K. Nauen. 1984. FAO species catalogue. Vol. 3. Cephalopods of the world. An annotated and illustrated catalogue of species to interest to fisheries. FAO Fish. Synop. ( 125) Vol. 3, 277 p. Roper, C. F E., and R. E. Young. 1975. Vertical distribution of pelagic cephalopods. Smith- sonian Contributions to Zoology 209, 51 p. Serchuk, F M., and W. F Rathjen. 1974. Aspects of the distribution and abundance of long- finned squid, Loligo pealei. between Cape Hatteras and Georges Bank. Mar Fish. Rev. 36:10-17. Sissenwine, M. P., and E. W. Bowman. 1978. An analysis of some factors affecting the catchability of fish by bottom trawls. ICNAF (Int. Comm. for the Northwest Atlantic Fisheries) Res. Bull. 13:81-87. Sissenwine, M. P., and A. M. Tibbetts. 1977. Simulating the effect of fishing on squid i Loligo and Illex) populations off the northeastern United States. ICNAF (Int. Comm. for the Northwest Atlantic Fisheries) Sel. Pap. 2:71-74. Sokal, R. R. and F J. Rohlf 1995. Biometry, the principles and practice of statistics in biological research. W. H. Freeman and Company, New York., NY, 887 p. Summers, W.C. 1967. Winter distribution of Loligo pealei determined by exploratory trawling. Biol. Bull. 133:489. 1969. Winter population of Loligo pealei in the mid-Atlantic bight. Biol. Bull. 137:202-216. 1971. Age and gi-owth of Loligo pealei. a population study of the common Atlantic coast squid. Biol. Bull. 141:189-201. Tibbetts, A. M. 1977. Squid fisheries (Loligo pealei and Illex illecebrosus) off the northeastern coast of the United States of America, 1963-1974. ICNAF (Int. Comm. for the Northwest Atlan- tic Fisheries) Sel. Paps. 2:85-109. Verrill, A. E. 1882. Report on the cephalopods of the northeastern coast of America. U.S. Comm. Fish and Fish., part 7, Rep. Comm. 1879:211-450. Vovk, A. N. 1978. Peculiarities of the seasonal distribution of the North American squid Loligo pealei (Lesueur 1821). Malac. Rev. 11:130. 1985. Feeding spectrum of Longfin squid (Loligo pealei) in the northwest Atlantic and its position in the ecosystem. NAFO (Northwest Atlantic Fisheries Organization) Sci. Coun. Studies 8:33-38. Whitaker, J. D. 1978. A contribution to the biology of Loligo pealei and Loligo plei (Cephalopoda, Myopsidal off the southeastern coast of the United States. M.S. thesis. College of Charles- ton, Charleston, SC, 165 p. 1980. Squid catches resulting from trawl surveys off the Southeastern United States. Mar Fish. Rev. 42:39-43. Worms, J. 1983. Loligo vulgaris. /;i Cephalopod life cycles, vol. I, spe- cies accounts. Academic Press, London, 475 p. 214 Abstract— Samples of the commer- cially and recreationally important West Australian dhufish iGlaucosoma Iwbra- icum) were obtained from the lower west coast of Australia by a variety of methods. Fish <300 mm TL were caught over flat, hard substrata and low-lying limestone reefs, whereas larger fish were caught over larger limestone and coral reef formations. Maximum total lengths, weights, and ages were 981 mm, 15.3 kg, and 39 years, respectively, for females and 1120 mm, 23.2 kg, and 41 years, respectively, for males. The von Bertalanffy growth curves for females and males were significantly different. The values for L^, k, and t„ in the von Bertalanffy growth equations were 929 mm, 0. Ill/year, and -0.141 years, respectively, for females, and 1025 mm, 0.111/year, and -0.052 years, respectively, for males. Preliminary estimates of total mortality indicated that G. hebraicum is now subjected to a level of fishing pressure that must be of concern to fishery man- agers. Glaucosoma hebraicum, which spawns between November and April and predominantly between December and March, breeds at a wide range of depths and is a multiple spawner The /.^(I's for females and males at first maturity, i.e. 301 and 320 mm, respec- tively, were attained by about the end of the third year of life and are well below the minimum legal length (MLL) of 500 mm. Because females and males did not reach the MLL until the end of their seventh and sixth years of life, respectively, they would have had. on average, the opportunity of spawning during four and three spawning sea- sons, respectively, before they reached the MLL. However, because G. hebra- icum caught in water depths >40 m typically die upon release, a MLL is of limited use for consei-\'ing this spe- cies. Alternative approaches, such as restricting fishing activity in highly fished areas, reducing daily bag limits for recreational fishermen, introducing quotas or revising specific details of cer- tain commercial hand-line licences (or doing both) are more likely to provide effective conservation measures. Age and size composition, growth rate, reproductive biology, and habitats of the West Australian dhufish (Glaucosoma hebraicum) and their relevance to the management of this species Sybrand A. Hesp Ian C. Potter Norman G. Hall Centre for Fish and Fisheries Research School ol Biological Sciences and Biotechnology Murdoch University South Street Murdoch, Western Australia 6150, Australia Email address (for I C Potter, contact author) i-potlenSpossum murdocti edu au Manuscript accepted 3 October 2001. Fish. Bull. 100:214-227 (2002). The West Australian dhufish (Glauco- soma hebraicum ), also known as the Westrahan jewfish (McKay, 1997), is one of the most commercially valuable and recreationally sought after finfish in Western Australia ( Sudmeyer et al. ' ). This species is confined to southwest- ern Australia, where its distribution ranges southwards from Shark Bay at 26°00'S, 113°00'E down the west coast and then eastwards along the south coast to the Recherche Aj-chipelago at 34°10'S, 122n5'E(HutchinsandThomp- son. 1995). Glaucosoma hebraicum is one of four members of the monogene- ric family Glaucosomatidae, which also includes the pearl perch (Glaucosoma scapulare) that is fished commercially and recreationally in eastern Australia (McKay. 1997). Despite the high quality of the flesh of Glaucosoma species and the commercial and recreational importance of G. hebra- icum in particular, there are no refer- eed papers on the biology of any species in this genus. Furthermore, the results of undergi-aduate studies on the biology of G. hebraicum, which were collated by Sudmeyer et al.,' included estimates of age that were based on the number of growth zones in whole otoliths, an ap- proach that would almost certainly have underestimated the age of many older fish (see "Results" section). Numerous commercial and recre- ational fishermen report that they now fish much farther offshore in order to obtain catches of dhufish comparable with those they had previously been able to obtain from nearer the coast. This consistent circumstantial evidence strongly suggests that the abundance of G hebraicum in more inshore waters has declined in recent years and that this is particularly the case in areas near the city of Perth where this spe- cies has been targeted by recreational fishing crews. The indications that the abundance of dhufish in nearshore wa- ters was declining led the Australian Fisheries Research and Development Corporation to fund the current study, with a view to producing biological da- ta for managing G hebraicum . During the present study, we first determined whether the otoliths of G hebraicum had to be sectioned to de- tect all of their opaque zones and we then validated that these opaque zones are formed annually and could thus be used to age this species. As accurate estimates of the age of a fish are de- pendent on a reliable birth date, the trends exhibited by reproductive vari- ables were used to estimate the dura- tion of the spawning period and, in par- ticular, when spawning activity peaked. The age of each fish was then deter- mined and the resulting length-at-age Sudmeyer, J. E.. D. A. Hancock, and R. C. J. Lenanton. 1992. Synopsis of Westralian jewfish iGlaucosoma hebraicum^ (Richard- son, 1845 1 (Pisces: Glaucosomatidae). Fish- eries Research Report 96, Western Austra- lian Marine Research Laboratory, Fisheries Western Australia, PO Box 20, North Beach 6020, Western Australia. Hesp et a\ Age and size composition, growth rate, reproductive biology, and habitats of Glaucosoma hebraicum 215 data employed to determine the age compositions and fiirowtli rales ot female and male G. hebraicum. The repro- ductive variables were also used to help ascertain where (1. hcbr'ciiciim spawns and whether this species is a mul- tiple spawner sciisu deVlaming (1983), i.e. whether indi- vidual females release eggs on more than one occasion in a spawning season. The lengths of both sexes at first ma- turity were calculated to determine whether they lay be- low the minimum legal length (MLL) of 500 mm and were thus appropriate for helping to conserve this species. The ages at which females and males mature were also deter- mined in order to elucidate whether fish might spawn in one or more spawning seasons before they attain the MLL. Attempts were also made to ascertain the types of habitat occupied by G. hebraicum at different stages during its life cycle and to obtain preliminary mortality estimates which could be used as an indicator of whether this species is being lightly or heavily fished. Finally, the data collected during this study were used to discuss ways in which the fishery for G. hebraicum might be managed most appro- priately in the future. Materials and methods Glaucosoma hebraicum, that were less than the MLL of 500 mm total length (TL), were collected between May 1996 and June 1999 by commercial trawls, hand-lines, and a recreational spear diver under a research collection permit issued by Fisheries Western Australia — the gov- ernment agency responsible for managing the fishery for this species. Filleted carcasses of G. hebraicum >500 mm TL, together with their gonads, were obtained monthly between May 1996 and April 1998 from commercial fish- processing plants and weigh-ins at local recreational fish- ing club competitions. These fish had been caught by commercial or recreational rod and hand-lines along the lower west coast of Australia between Mandurah (32°32'S) and the Houtman Abroholos <28°35'S), i.e. within that part of the distribution of G. hebjxiicum where this spe- cies is considered to be most abundant and is most heavily fished. The total length of each fish was measured to the near- est 1 mm and the weight of each fish <500 mm was weighed to the nearest 1 g. The weights of 334 females and 442 males >500 mm TL were weighed to the nearest 10 g prior to filleting. The relationship between total length (L) in mm and total wet weight ( W) in g of each sex was Females logW = logO.0000417 + 2.859 logL (/(=486, r'^=0.995) Males logW = logO.0000322 -f- 2.898 logL (w=572, r2=0.995). These relationships were then used to estimate the weights of the female and male fish that had been filleted but not weighed. Note that all of the logarithm values recorded in this paper are natural logarithms. On several occasions, a video camera, attached by cable to a television monitor and video recorder, was lowered over the substrata during commerical hand-line fishing for dhufish. Video footage of the substrate over which dhufish were caught was later examined to determine the types of habitat occupied by this species. Age determination The two sagittal otoliths of each fish were removed, cleaned, dried, and then stored in paper envelopes. All sagittal oto- liths were sectioned, except for those which, when placed in methyl salicylate and examined microscopically under reflected light against a black background, could clearly be seen to possess either no opaque zones or only a single opaque zone. However, because the opaque zones in the whole otoliths of large fish were so numerous and closely spaced that they were often difficult to distinguish from one another and because previous estimates of the age of dhufish were based on counts of opaque zones in whole otoliths (Sudmeyer et al.'), the number of opaque zones visible in 100 otoliths, obtained from a wide size range of fish, were compared prior to and after sectioning to ascer- tain whether sectioning increased one's ability to detect the opaque zones. For sectioning, the otoliths were mounted in clear ep- oxy resin and cut into 500 pm sections with a low-speed diamond saw (Buehler). The sections were cleaned and mounted on slides with DePX mounting medium and ex- amined under reflected light with a dissecting microscope attached to a video camera (Panasonic WV-CD20). The im- age was analyzed by using the computer imaging package Optimas 5 (Optimas, 1995). The number of opaque zones in each otolith was always counted twice and on different days and without knowledge of either the date of capture or the size of the fish from which the otolith came, and also, in those cases where the two counts differed, on a third occasion. Although the number of times that a third count did not agree with either of the two previous counts was negligible for otoliths with less than 15 opaque zones, such disagreement increased to ca. 10% for otoliths with 15-25 opaque zones and ca. 30% for those with more than 25 opaque zones. When a third count was necessary and was not the same as either of the two previous counts, fur- ther counts were made until successive counts did not dif- fer by more than two opaque zones. On such occasions, the final count was recorded. An independent reader counted the number of opaque zones on 110 otoliths from a wide size range offish. Eighty four percent of the counts of the number of opaque zones made by this independent reader were the same as those of the senior author for 50 sectioned otoliths that had been judged by the senior author to have up to 10 such zones and, in those cases where there were discrepancies, the dif- ferences were never more than one opaque zone. Eighty percent of the counts made by the independent reader of the number of opaque zones on 50 sectioned otoliths re- corded as possessing between 11 and 25 such zones by the senior author were the same or differed by only one from those of the senior author and, where there were discrepan- cies, these rarely exceeded three opaque zones. In the case of ten otoliths with >25 opaque zones, the maximum dis- 216 Fishery Bulletin 100(2) crepancy between the counts recorded by the independent reader and the senior author was five. After consultation, it was agreed that, in many of the cases where the counts differed by one opaque zone, the independent reader had failed to discern the outermost opaque zone at the periph- ery of the otohths. Moreover, the extent of any discrepancy between the counts of the independent reader and the se- nior autiior declined if the independent reader continued to recount the number of opaque zones on the otoliths. Validation that the opaque zones in the otoliths of G. he- braiciim are formed annually was carried out by analyzing the trends exhibited throughout the year by the marginal increments on whole otoliths, when only one opaque zone was present, and on sectioned otoliths when two or more opaque zones were present. For this purpose, the marginal increment on each otolith, i.e. the distance between the out- er edge of the single or outermost opaque zone and the edge of the otolith, was expressed either as a proportion of the distance between the primordium and the outer edge of the opaque zone, when only one opaque zone was pres- ent, or as a proportion of the distance between the outer edges of the two outermost opaque zones, when two or more opaque zones were present. Each of the above requisite dis- tances was measured perpendicular to the opaque zone(s) and without knowledge of the date of capture of the fish and was recorded to the nearest 0.01 mm by using Optimas 5. The values for the marginal increments were separated into groups according to the number of opaque zones on the otoliths, i.e. 1, 2-5, 6-8, 9-11 etc., after which the values for each of those groups in each corresponding month of the year between May 1996 and April 1998 were pooled. Von Bertalanffy growth equations The time when spawning peaked was estimated from the trends exhibited throughout the year by gonadosomatic indices, gonadal maturity stages, and pattern of oocyte development. Tliis time was considered to coiTespond to the birth date of G. hehraicum and could thus be used, in combi- nation with the number of opaque zones on the otolith and the time when the annulus becomes delineated on the oto- lith, to determine the age of individual fish on their date of capture. Because the sex offish <150 mm could not be deter- mined, the lengths-at-age of these small fish were randomly allocated in equal numbers to the data sets for female and male fish used for constructing the gi-owth curves. Assumptions are made concerning the distribution of er- rors when fitting von Bertalanffy growth cui-ves to length- at-age data. Kimura (1980) discussed the implications of the following three assumptions, namely that 1) the indi- vidual lengths-at-age have a constant variance, 2) the mean lengths-at-age have a constant variance and 3) the vari- ance of the lengths-at-age is dependent on age. The assump- tion most frequently adopted in growth studies is that the individual lengths-at-age have a constant variance. As dis- cussed by Kimura (1980), different assumptions regarding the error variance require modifications to the objective function to ensure that the parameters are estimated ac- curately and that any comparisons between gi-owth cui-ves, that are based on the likelihood ratio, are appropriate. A von Bertalanffy growth equation was fitted to the lengths-at-age of female and male fish with the traditional assumptions that the lengths-at-age are normally distrib- uted around the values predicted from the growth equa- tion and that the variance of this distribution is constant for each sex over all ages. However, visual examination of the residuals for each curve suggested that it was not ap- propriate to make the latter assumption. Further study showed that the variance of the residuals is approximately proportional to the age of the fish, as above in assump- tion 3 of Kimura ( 1980). Thus, the von Bertalanffy growth equation was fitted to the length-at-age data for each sex by using the assumption that the residuals were normally distributed, where the variance of this distribution was proportional to age but dependent on sex. That is, L, =L.{l-exp[-/?(^, -^„)]} and Li = L, 4- f,, where, for each sex, L^ = the observed length at age; L = the estimated length-at-age; t^ = the age; and fj= the error associated with the 7th fish. For the growth curves for females and males, L .^ is the mean asymptotic length predicted by the equation, k is the growth coefficient, and t^^ is the hypothetical age at which fish would have zero length if growth had followed that predicted by the equation. The errors are assumed to be normally distributed, such that f~NiO, cj^), where c,. is the constant of proportionality between the variance of the residuals and age for fish of sex .s. The growth equa- tions were fitted to the observed length-at-age data for both sexes by maximizing the log-likelihood of the data. The log-likelihood for the combination of male and female fish. A, may be written as A: \m-^ ^f' where A^ = the log-likelihood associated with females or males and may be calculated as A, = -^log(2ff)-- > log(cJ,) > \ — '- ^— k and n^ = the number offish of that sex in the length-at- age data. The maximum likelihood estimate off,, for each sex is given by The SOLVER routine in Microsoft EXCEL (Microsoft Corp., 2000) was used to estimate the parameters that Hesp et a\ . Age and size composition, growth rate, reproductive biology, and liabitats of Glaucosoma hebraiaim 217 would maximize the log-likelihood function, by fitting the equations to the combined set of length-at-age data for both females and males. The growth curves for the fish of each sex were com- pared following the Hkelihood ratio method described by Ivimura (1980) and Cerrato (1990). The test of the like- lihood ratio, A. that was applied was to reject the null hypothesis il (that there was no difference between the cui-v'es) at the « level of significance when fM+ff -F", ih,*U where A = . 2 \-"f '2 ' Fil ) and where n - /; ,, + nf, f = n - 3; and q = the number of linear constraints of the form 0J^^ = Op, where 6 is one of the parameters M and F = males and females, respectively. This test was developed by Gallant (1975), as described by Cerrato (1990). The gi'owth curves were fitted under all possible para- meter sets, and the best of both the 4- and 5-parameter, models, i.e. those that maximized the log-likelihood, were selected. The resulting 3-, 4- and 5-parameter models were compared with the 6-parameter model by using the above test to determine which of these three models, w. was of minimum complexity and not significantly different from the 6-parameter model i2. The model selected on the basis of these tests was the simplest model that, in the statisti- cal sense, provided the best description of the data. Reproductive biology The gonads of each fish that could be sexed macroscopi- cally were removed and weighed to the nearest 0.01 g. Each gonad was allocated to a maturity stage, based on the scheme of Laevastu (1965), but which, in the case of females, also took into account the histological character- istics of the ovaries (see "Results" section). The percentage contributions made by the different go- nadal stages in sequential 50-mm length intervals were calculated for both female and male G. hebraicum. The lengths at which 50'7f of female and male G. hebraicum reach sexual maturity (L50' were determined by fitting the logistic curve to the percentage of female and male fish which, during the spawning period, possessed gonads at stages III to VIII (see "Results" section for rationale for us- ing these six stages for this purpose). The logistic curve was fitted by employing a nonlinear technique (Saila et al., 1988) and by using a routine statistical method pro- vided in SPSS (SPSS Inc., 1988). The logistic equation is Pf^ = 1/(1 -H e'"*'''-'), where P^ is the proportion offish with mature gonads at the mid-point of the length class, L, and a and b are constants. The L-,, for each sex was derived from the equation Lr,,, = y . The ages at which 50% of fe- males and males reached maturity, i.e. the Ar,,,, were esti- mated, as follows, from the inverse von Bertalanffy growth equations for the two sexes (see Stergiou, 1999): ■^0 ~ 'i ■o-(i|log h. Gonadosomatic indices (GSIs) of females and males >Lr^Q at first maturity were determined from the equation W1/W2 X 100, where W\ = wet weight of the gonad; and W2 = wet weight of the whole fish. Mortality PreliminaiT analysis of catch curves demonstrated that the mortality estimates derived for commercially and rec- reationally caught fish that were greater than the MLL of 500 mm and eight years old and thus fully recruited (see "Results" section) were similar. Thus, the data from the commercial and recreational samples were pooled for esti- mating mortality. An estimate of the instantaneous coef- ficient of natural mortality, M, was determined from the von Bertalanffy gi'owth coefficient, k. with the regression equation developed by Ralston (1987), i.e. M = 0.0189 -i- 2.06/;. The instantaneous coefficient of total mortality, Z, was determined by maximizing the likelihood, when fitting the estimated age composition resulting from that mortal- ity to the observed age composition data for those dhufish that were gi'eater than the MLL of 500 mm and eight years old. In order to assess whether the observed age composi- tion data reflected decreasing levels of total mortality in earlier years, the catch cui-ve analysis was repeated with different initial ages, ranging from 10 to 30 years. Values of Z were also estimated by using the observed maximum age '^jiov* fo'' *'^^ sampled dhufish, employing both the regres- sion equation reported for fish by Hoenig ( 1983), i.e. log(Z)= 1.46- 1.01 log(C„,,,,^), and the equation for the expected value of the maximum age in a sample of size n, i.e. £(C) = 1 V- 1 Ir'^' where /, = the age at which fish become fully recruited to the fishery (Johnson and Kotz, 1970, p. 216, as reported by Hoenig, 1983). Results Habitats of Glaucosoma hebraicum Glaucosoma hebraicum <150 mm TL and <14 months old were caught regularly by trawlers offshore in water 218 Fishery Bulletin 100(2) depths of 27 to 33 m. The depth sounder indicated that these small G. hebratcum were most consistently caught over hard substrate that lay adjacent to reefs — a con- clusion later confirmed by video footage. Although a considerable amount of effort and a variety of techniques were employed in attempts to catch fish with lengths of 150-300 mm, only a small number offish of this size were collected. However, a few G. hebraiciim of this length class were caught by an experienced spearfisher while diving over low-lying reefs with rock ledges <30 cm high. Large numbers of dhufish >300 mm in length were obtained from rod and hand- line fishermen who were fishing in waters that were shown by video camera and com- mercial echo sounders to be located over limestone and coral reef formations and, in particular, where the "drop-offs"( reef edges) were two or more metres in height. Comparisons between number of opaque zones visible in whole and sectioned otoliths Sectioned otoliths Numbef ol otolil Number of opaque zones §5 hs 10 3 7 10 10 10 10 10 10 10 12 10 10 10 10 10 9 10 6 4 2 2 2 2 111 0 2 3 4 5 6 7 8 9 10 11 12 13141516 17 18 1920 2122 23 24 26 28 31 20 30 ao do 50 60 40 40 50 10 - 1 - • • • • • • • • • 10 20 40 30 20 10 40 50 2 ~ • • • 20 • • 10 • • 50 10 33 • 25 40 3 ~ • • 20 • • 30 30 • 33 • • - 4 " • • • • ?S 40 -5 ' 33 • • 20 50 6 ~ • • • 50 50 50 - 7 ~ • • • 100 -8 50 • 100 -9 " • 50 • 50 10 - • • u - 12 — 100 • The number of opaque zones observed in each sectioned otolith, in which up to six such zones could be seen, was the same as those visible on the same otolith prior to sectioning (Fig. 1). However, this fre- quently did not apply when a greater number of opaque zones were present. Fur- thermore, where such discrepancies occurred, the differ- ences between the number of opaque zones detected prior to and after sectioning rose as the number of opaque zones increased. In all cases where there were discrepan- cies, the number of opaque zones detected after section- ing was greater than prior to sectioning. Underestimates of the number of growth zones with whole otoliths, based on comparisons with those detected in sectioned otoliths, rose from one in whole otoliths with seven to nine opaque zones to between one and seven in those with 10-21 opaque zones (Fig. 1). In otoliths with a large number of opaque zones, the differences sometimes exceeded eight and for one such otolith was as high as twelve. These com- parisons demonstrated that, for validation that opaque zones are formed annually and that these zones can thus be used for aging G. hehraicum. experiments should be conducted on sectioned otoliths. Validation that opaque zones are formed annually The mean monthly marginal mcrements on sectioned oto- liths with 2 to 16 or more opaque zones rose from a low level in January to a maximum in September, before declining precipitously to a minimum in October and then rising slightly in December (Fig. 2). They thus reached high levels in early spring, before declining markedly in mid-spring, as the outermost opaque zone became delin- Figure 1 Comparisons between the number of opaque zones obsei-ved on the otoliths oi Glaucosoma hebraicum prior to and after the sectioning of those otoliths. The numbers above enclosed circles represent the percentage number of underestimates of the number of opaque zones when using whole rather than sectioned otoliths. eated through the formation of a new translucent zone, and then increased progressively in the ensuing inonths as the translucent region increased in width. Although fish possessing otoliths with one opaque zone were not caught in all months, the trends exhibited by the mean monthly marginal increments for those months when such fish were caught were consistent with those exhibited by otoliths with a larger number of opaque zones. Because the mean monthly marginal increment rose and declined only once during the year, irrespective of the number of opaque zones in the otolith, a single opaque zone IS laid down in the otoliths of G. hebraicum each year. The number of opaque zones in sectioned otoliths can thus be used, in conjunction with the birth date of G. hebrai- cum and the month when the opaque zone(s) become de- lineated, to age this species. Growth of Glaucosoma hebraicum Because the trends exhibited by the GSIs and stages in gonadal maturation and oocyte development demonstrated that the spawning of G. hebraicum peaked from late Janu- ary through early February, this species was assigned a birth date of 1 Februai'v. Age O-i- G. hebraicum were first caught by trawling over hard substrate in April and May, when their lengths ranged from 57 to 81 mm (Fig. 3). How- ever, substantial numbers of the O-i- age class were not Hesp et al : Age and size composition, growth rate, reproductive biology, and habitats of Glaucosoma hebraicum 219 0.4 r 02 25 55 08 0.4 2-5 opaque zones g 7 28 16 '' 20 14 6-8 opaque zones 04 0 9-11 opaque zones 0.8 0.4 12-15 opaque zones g .j. 0 >- > 16 opaque zones 0.4 JFMAMJJASOND Month Figure 2 Mean monthly marginal increments ±1SE for sag- ittal otoliths of Glaucosoma hebraicum. Sample size is given for each month. In this Figure and Figure 5. the closed rectangles on the horizontal axis refer to summer and winter months and the open rectangles to autumn and spring months. 10 - 5 0 15 10 15 10 5 0 ■5 10 5 D 15 10 5 0 15 10 5 0 '- ' 5 - ■0 - ■0 15 10 ■0 5 0 ■5 10 [H] r"~^ April n=3 r\/1ay n=8 June n=13 July n=3 August n=11 October n=51 November n=63 December Q n=60 January n=27 February n=27 Marcti n=45 Total length (mm) Figure 3 Length-fi-equency distributions for Glaucosoma hebraicum caught by trawling along the lower west coast of Australia by using data for corresponding months in the period May 1996 to June 1999. "denotes mean lengths of 0-t- and early 1+ fish. His- tograms in gray refer to 0+ fish, and those in black and white refer to 1+ and 2-1- fish, respectively. caught until October, presumably reflecting the time typi- cally required for the 0-f age class to be recruited into these areas from those in which spawning occurs. The mean length of the O-i- age class had reached 95 mm by October, in which month the first opaque zone became delineated on the otoliths, and 108 mm by January, when fish were approaching the end of their first year of life. The mean length of the corresponding cohort, now early l-i-, was 127 mm in March, after which month the number of l-i- fish caught in trawl samples declined markedly (Fig. 3). The best of the 4- and 5-parameter growth curves were selected as the models with the largest log-hkelihood at that level of model complexity. The best 4-parameter model was the cui-ve that assumed different asymptotic lengths for the sexes, and the best 5-parameter model was that which assumed that the growth coefficients were equal. Comparisons between the curves demonstrated that the model that assumed common growth coefficients for fe- males and males was not significantly different {P>0.05) from the more complex model, which assumed that all 220 Fishery Bulletin 100(2) parameters of the gi-owth curves differed between the sexes. For this cui^ve, the parameters L., k, and t„ and their 959; confidence hmits were estimated to be 929 (908 to 949) mm, 0.111 (0.107 to 0.116)/year, and -0.141 (-0.183 to -0.100) years, respectively, for females, and to be 1025 (1003 to 1048) mm, 0.111 (0.107 to 0.116)/year. and -0.052 (-0.088 to -0.0161 years, respectively, for males. The growth curves for females and males were significant- ly different (P<0.001), with the asymptotic lengths hav- ing the most influence on the difference between the sex- es. The estimated constant of proportionality between the variance of the residuals and age were 363 for females and 320 for males. The von Bertalanffy growth cui-ves demonstrated that females gi'ow slightly slower than males. Thus, at ages 2 to 5. females had reached lengths of 196, 273, 342, and 404 mm, compared with 209, 294, 371, and 440 mm for males. By the time G. hebraicum had attained 10, 15, and 20 years, the females had reached ca. 628, 756, and 830 mm, respectively, and the males had reached ca. 689, 832, and 914 mm, respectively (Fig. 4). The maximum ages re- corded for females and males were 39 and 41 years, re- spectively, and the maximum total lengths of females and males were 981 mm (=ca. 15.3 kg) and 1120 mm (=ca. 23.2 kg), respectively. The ages at which female and male G. he- braicum reach the minimum legal length for capture (500 mm TLi were 7.0 and 6.0 years, respectively. Trends exhibited by reproductive variables The macroscopic characteristics of the different stages in gonadal development, and of the cytological characteris- tics of the ovaries at different stages based on an examina- tion of histological sections, are given in Table 1. Because stages I (virgin) and II (immature) in the de- velopment of both the ovaries and testes of G. hebraicum were difficult to separate macroscopically, data for these two stages were pooled in the case of both sexes. Further- more, it is also important to recognize that spawning stage (VI) ovaries are distinguished from prespawning stage (V) ovaries almost exclusively on the basis of their posses- sion of hydrated oocytes or postovulatory follicles (or both) when histological sections were employed to examine the ovary at a finer scale. However, because G. hebraicum is a multiple spawner, i.e. produces eggs in batches at in- tervals, any "prespawning" stage ovary may already have produced some hydrated oocytes, but been at an interme- diate phase in which the next batch of yolk granule oocytes had not yet become hydrated. The prevalence of females with prespawning ovaries that had already spawned on one or more occasions would be expected to increase dur- ing the spawning period. Likewise, the main difference be- tween prespawning and spawning testes, i.e. the ability of the testes to produce milt when subjected to physical pressure, may often represent different phases in the cy- clical changes undergone in the testis during the spawn- ing period. For the above reasons, the data on stage-V and stage-Vl ovaries and testes were pooled for describing the change in compositions of the gonadal maturity stages of each sex during the year. 1250 1000 750 ■0^^ 500 ^k'-' '■ Females 250 - /" o (UIUJ) 1 / c 1 1 H 1250 1000 >-^^— ^ ' ■ ■'■'S^^^^^' 750 m 500 ap.' Males jJF n=799 250 f 0 J 0 10 20 30 40 50 Age (years) Figure 4 Von Bertalanffy growth curves fitted to length-at-age data for females and males of Glaucosoma hebraicum caught on the lower west coast of Australia. ;; = number offish used to construct growth curves. Between May 1996 and April 1998, the mean monthly GSIs for females of G. hebraicum that were greater than the L-ii of 301 mm at first maturity were always low in winter (June to August) and early spring (September), i.e. <1.0 but then rose sharply to reach a peak of ca. 2.8 in mid-summer (January), before declining markedly during early to mid-autumn ( March and April i ( Fig. 5 ). The trends displayed by the mean monthly GSIs for males of G. he- braicum, that were greater than the L-^of 320 mm at first maturity, paralleled those just described for females. Because the trends exhibited by the mean monthly GSIs for females and males were the same during both 12 month periods, the percentage contributions of the different go- nadal stages of the females and males of G. hebraicum, that were longer than the L^,,, were pooled for each of the cor- responding calendar months. The gonads of female G. he- braicum in July were at stages I-II, i.e. virgin or immature (Fig. 6). Fish with ovaries at stage III (developing) were first found in August, albeit only a few fish, and those at stage IV Hesp et al : Age and size composition, growth rate, reproductive biology, and habitats of Glaucosoma hebmicum 221 Table 1 Characteristics of the macroscopic stages in the development ofthe gonads of Glai/cuKDiua lichmiciini and, in the case of the ova- ries, the histological characteristics of each corresponding ovarian stage. Terminology for oocyte stages follows Khoo ( 1979). Stage Macroscopic appearance Histological characteristics l-ll (virgin and immature) III (developing) rV (maturing) V (prespawning) VI (spawning) VII (spent) VIII (.recovering spent) Gonads very small. Ovaries transparent and oocytes not visible. Testes strandlike and gray-white. Gonads slightly larger than at stage I or II. Ovaries pinkish, blood capillaries visible in ovary walls. Testes white and more lobular. Gonads markedly larger. Ovaries reddish-orange, capillaries more conspicuous and some yolk granule oocytes visible through ovary wall. Milt is not extruded when pressure is applied to testes. Ovaries orange and occupy most of space in body cavity. Extensive capillaries in ovary walls. Milt appears when testes placed under firm pressure. Same as for stage V, but with hydrated oocytes visible through ovarian wall and only slight pressure required to release milt from testes. Gonads smaller than at stages V or VI. Ovaries flaccid. Some yolk granule oocytes still visible through ovary wall. Testes pinkish-red. Gonads greatly reduced in size and dark red. Testes strandlike. Ovigerous lamellae highly organized. Oogonia and chromatin nucleolar oocytes and, in more advanced ovaries, early perinucleolar oocytes are present. These oocyte stages are present in all subsequent ovarian stages. Early and late perinucleolar oocytes and yolk vesicle oocytes present. Yolk vesicle and yolk granule oocytes abundant. Yolk granule oocytes abundant and in tight groups. Hydrated oocytes or postovulatory follicles (or both) present. Remnant yolk granule oocytes present, typically under- going atresia. Lamellae not organized as in early stages of develop- ment and contain extensive scar tissue. Any remain- ing yolk gi'anule oocytes are atretic. (maturing) and stages V and VI (prespawning and spawn- ing) were first recorded in September and October, respec- tively. Stage-V and stage-VI ovaries collectively became the most prevalent group in females in November and formed the most dominant group by far in December to March. The samples in February and March contained a few fe- male fish with stage I-II ovaries, but none with ovaries at either stage III or FV (Fig. 6). These trends provide over- whelming circumstantial evidence that any female whose ovaries have developed to at least stage III by November will progress through to maturity during the following months of the spawning period. Thus, the L^,, for females at first maturity was calculated by using the percentage of ovaries with stages III and FV, as well as those with stages V-VIII. Although females with stage-VII (spent) and stage- VIII (recovering spent) ovaries were found between Janu- ary and May, the majority of ovaries were at stages I-II in the latter month and all were at stages I-II in June. The trends exhibited by the pattern of gonadal development in males were essentially the same as those just described for females and thus the L-q of males was likewise calculated with the percentage of testes at stages III- VIII (Fig. 7). The following account of the trends exhibited by the oo- cyte composition of ovaries is based on an histological ex- amination of the ovaries of large fish well above the L^g at first maturity. The oocytes in ovaries in July and Au- gust were almost exclusively at the chromatin nucleolar stage. Ovaries with yolk vesicles first appeared in Septem- ber, and those with yolk granules were first found in Oc- tober. Yolk granule oocytes became increasingly prevalent in ovaries in November and dominated the complement of their larger oocytes between December and March. Some of the residual yolk granule oocytes in April and all of those in May were undergoing atresia. No yolk vesicle or yolk granule oocj^tes were found in June. Hydrated oocytes were first found in ovaries in November and were present in many ovaries between December and March and in a few ovaries in April, but were found neither in May nor in the immediately ensuing months. Small numbers of post- ovulatory follicles were present in about a third of the ovaries of large females caught between December and March. The oocyte diameters of individual large G. hebra- iciim caught in each month of the spawning period pro- duced a series of modes (data not shown). 222 Fishery Bulletin 100(2) Length and age at maturity The sex oiGlaucosoina hebraicurn was not able to be deter- mined by macroscopic examination of the gonads until it had reached ca. 150 mm in length. During the main part of the spawning period, i.e. December to March, the gonads of all female and male G. hebraiciun <250 mm were at the earliest stages of development, i.e. I-II (Fig. 7). Gonads at stages III-VIII were first found in the 250-299 mm length class of females and in the 300-349 mm length class of males. The presence of such gonads demonstrated that the fish were maturing or that spawning was occurring or had been completed (see earlier). The prevalence of ova- ries at stages III-VIII increased to ca. SO'/r in the 300-349 mm length class and to 1009i^ in all females >450 mm. The gonads of all males >450 mm were at stages III-VIII (Fig. 7). The L^^'s for the lengths of female and male G. hebraicum at first maturity, derived from the logistic cui-ve 4 r Females o ^ 03 C5 02 0.1 Males fitted to the percentage contributions of fish with gonads at stages III-VIII in sequential 50-mm length classes, were 301 and 320 mm, respectively (Fig. 7). Individual G. hebraicum could first be sexed macroscopi- cally during their second year of life. Although relatively few two-, three- and four-year-old fish were caught, the trends exhibited by the proportion of gonads at stages III- VIII in both sexes during the spawning period were con- sistent. One female and no males at two years of age pos- sessed gonads at stage III or gi-eater (Fig. 8). However, 50'X of three-year-old female and male fish, and all five-year-old females and all six-year-old males possessed such gonads and were thus regarded as mature. The Ar^^s for the age at first maturity of females and males were 3.4 and 3.3 years, respectively. Mortality Using the regi'ession equation developed by Ralston (1987), in combination with the esti- mated value for the von Bertalanffy growth coefficient, k = 0. Ill/year, we estimated the instantaneous coefficient of natural mortality, M, to be 0.25/year The catch curve analysis of the combined age composition data, for the 620 dhufish older than 9 years and longer than the MLL (Fig. 9), produced an estimate of the instantaneous coefficient of total mortality, Z, of 0.21/year (95'^f confidence intei-val: 0.19 to 0.23/year). The estimate of Z remained at about this level as the initial age was increased to 15 years and then declined to ca. 0.15/year at 24 to 27 years (Fig. 9). It subsequently became less precise as the initial age increased. An esti- mate of Z of 0. 10/year was obtained when the obsei-ved maximum age of 41 years was substi- tuted into Hoenig's ( 1983) regression equation. However, when the sample size of 620 fish was taken into account, with the expression for the expected maximum age (Hoenig 1983, Appen- dix A), Z was estimated to be 0.22/year. M J J A S O N D J I I 1996 I FMAMJJASONDJFMA 1997 I 1998 ' Month Figure 5 Mean monthly gonadosomatic indices ±1SE for females and males of Glaucosoina hebraicum caught in offshore waters between May 1996 and April 1998. Data in this Figure and Figure 6 are restricted to females and males 2Lg„ at first maturity. Numbers offish used to calcu- late each mean GSI arc shown. Discussion Ontogenetic changes in habitat of Glaucosoma hebraicum Extensive sampling for G. hebraicum during the present study, allied with data obtained with an echo sounder and video footage, dem- onstrate that this species changes habitat as it increases in size. Thus, G. hebraicum <150 mm was found to live in areas near reefs where the substrate is firm and sponges often occur (Bergquist and Skinner, 1982). The reduction in the numbers of l-t- dhufish caught by trawl- ing in this type of habitat in late autumn, when their lengths were about 130 mm, probably reflects a movement by the members of this Hesp et al.: Age and size composition, growth rate, reproductive biology, and habitats of Glaucosoma hebmicum 223 cohort, as they increase in size, from a hab- itat that could be trawled to one where reefs occur and where it was not possible to trawl. This conclusion is supported by the fact that the few dhufish of 150-300 mm that were caught were collected from low- lying reefs, i.e. reefs that contained rock ledges up to 30 cm in height. In contrast, G. hebraiciim >300 mm typically occupy areas where there are substantial lime- stone and coral reef formations and their large size would make them less suscepti- ble to predation in a habitat where large predatory species, such as the Samson fish iScriola hippos) and the pink snapper {Pagrus auratus ) are found (Hesp, personal obs.). Aging The trends exhibited by the marginal increments on sectioned otoliths of G. heb?-aiciim show that an opaque zone is formed annually in the otoliths of this spe- cies. However, comparisons between the number of opaque zones on individual oto- liths prior to and after sectioning demon- strate that, after this species has reached six years in age, one or more of these zones often become visible only after the otolith has been sectioned. This demon- strates that earlier estimates of the age of older G. hebraiciim, which were based on the number of opaque zones visible in whole otoliths (Sudemeyer et al.M, were almost certainly often too low. An inability to detect all of the opaque zones in the whole otoliths of older fish is largely attributable to the fact that, as the otolith increases in width, it becomes in- creasingly difficult to distinguish between the zones at the periphery of the otolith. This problem parallels the situation re- corded for several other medium-size to large teleosts, such as Pacific hake (Mer- luccius productus) (Beamish. 1979), starry flounder iPlaty- ichthys stellatiis) (Campana, 1984) and blue-spotted flat- head (Platycephalus speculator) (Hyndes et al., 1992). Our results demonstrate that, although most G. hebraiciim are less than 25 years old, some females and males live for longer than 30 years and very occasionally for up to about 40 years. Other species that are typically caught by commercial and recreational rod and hand-line fishermen when fish- ing for dhufish include pink snapper iPagrus auratus), Sampson fish (Seriola hipposK silver trevally iPseudo- caranxdentexK breaksea cod iEpincphelides armatus). and occasionally also King George whiting (Sillaginndes punc- tata). The maximum total lengths and weights recorded for these five species are 1300 mm and 19.5 kg for pink ,oo.r-, Females Males 0 "=32 [!□ July n=45 100 r |_ m n=83 .; [—1 August otl- n=120 100 r ^ J September Q 1 r -'1 , . n=93 ^DDa™ n=105 100 r ^ 0 LLJc3=_ October n=85 :=□□= n=83 100 r November & □ nizjQ n=72 ;_nnn_ n=100 S 100 r 3 ~r December cr m ° =. nDl- n=25 □ n n=20 en 100 r B -, January c y 0 J„„ n=99 = Da„ n=90 ^ 100 j- February 0 n=78 Dn„ n=61 100 r = Li _o March 0 n=63 := an^ n=65 100 r April 0 D aunU n=83 .□ —Llczi n=90 100 |- J □ _□ May 0 L U C=Z1 _ IZ=1 □ n=105 100 f— June n=46 1 ■-; n=55 0 '-' L Lil 1=1 HI 11 IV V-VI VII VIII HI III IV V-VI VII VIII Gonadal stage Figure 6 Percentage frequency histogi' ams for gonadal maturity stages of females and males of Glaucosoma hebraiciim. Data have been pooled for corresponding | months in the period May 1996 to April 1998. Sample sizes (n ) are given for each month. snapper, 1753 mm and 53.6 kg for Samson fish, 938 mm and 10.0 kg for silver trevally, 550 mm and 2.9 kg for breaksea cod, and 690 mm and 4.8 kg for King George whiting, compared with 1219 mm and 25.8 kg for dhufish (Hutchins and Thompson 1995). Although reliable data have been obtained for the age and growth of a number of commercial and recreational fish species that live in nearshore coastal or estuarine waters in southwestern Australia (e.g. Chubb et al., 1981; Hyndes et al., 1992, 1996; Hyndes and Potter, 1996, 1997; Lauren- son et al, 1994; Fairclough et al., 2000; Sarre and Potter, 2000), comparable data for those species that are found in and around reefs in deeper waters in southwestern Aus- tralia are restricted to those recorded for G. hebraicum in this paper and for the King George whiting Sillagitiodes 224 Fishen/ Bulletin 100(2) 100 75 50 25 Females 4 3 5 12 16 15 15 3236242228 33 ?1 ^ 3 5 CT 0 S Males I 100 Q. 75 50 W 3 1 9 10 18 8 17 25 34 2522 24 22 2610 12 2 0 25 |y Ijir lllllllllllll 200 400 600 800 Total length (mm) 1000 1200 Figure 7 Percentage frequency of occurrence of gonads at stages I-II (D) and stages III-VIII (Dt in each sequential 50-mm class of female and male Glaucosoma hehrai- cum caught between December and March. The logistic curve has been fitted to the data for fish with gonads at stages III-VIII. The sample size is given for each length class. punctata by Hyndes et al. (1998). Using data from the low- er west coast of Australia, Hyndes et al. (1998) estimated the von Bertalanffy growth parameters, L, k, and f^, for King George whiting to be 538 mm TL, 0.47/year, and 0.13 years, respectively, for females, and 500 mm TL, 0.53/year and 0.16 years, respectively, for males. Although there are no published studies on the growth of pink snapper and sil- ver trevally in Western Australia, the growth of these two species has been investigated in New Zealand by Francis et al. ( 1992) and by James ( 1984) respectively. The param- eters L , k, and t,^ were estimated to be 720 mm FL (fork length), 0.106/year and -0.75 years, respectively, for pink snapper, and i-anged from 436 to 448 mm FL, from 0.27 to 0.43/year and from -1.6 to -0.6 years, respectively, for silver trevally Although the growth coefficient, k. for dhufish, i.e. 0. Ill/year, was similar to that for pink snapper, it was ap- preciably less that that for both Iving George whiting and Females 4 6 4 1010 7 9 12 7 58 75 50 25 =1 ^ CT (D m Males c o 36 10 96997 48 Q. 75 50 - 25 0 1 234 56789 ^10 Age (years) Figure 8 Percentage frequency of occurrence of gonads at stages I-II (D) and stages III-VIII (D) in each sequential age class of female and male Glaucosoma hebraicum caught between December and March. The sample size is given for each age class. silver trevally. Dhufish had an asymptotic length ca. 35% greater than that of pink snapper and approximately twice those of King George whiting and silver trevally. The lengths of females at maturity have been reported as ca. 350 mm FL for silver trevally (James, 1984) and 413 mm TL for King George whiting (Hyndes et al, 1998) and have been estimated as 237 mm FL for pink snapper (calculated from Crossland, 1977). Although maturity is first achieved by the females of snapper and dhufish at 30% of their respective asymptotic lengths, it is at- tained by the females of King George whiting and silver trevally at 75-80'7f of their asymptotic lengths. This find- ing implies that the last two species have a higher repro- ductive load sensu Gushing (1981). Thus, although these four species reach maturity and begin to occupy promi- Hesp et al : Age and size composition, growth rate, reproductive biology, and habitats of Claucosoma hebra/cum 225 nent reefs at similar lengths, the growth of silver trevally and King George whiting slows after mat- uration, whereas that of snapper and dhufish continues appreciably after they have reached maturity. Spawning location, period, and mode 125 r 100 •= 75 50 25 Glaucosoma hebraicum with gonads at stages VI (spawning) and VII (spent) were caught in waters rang- ing from 10 to 150 m in depth and at distances of 3 to 50 km from the shore and between latitudes 28°55' and 32°45'S. Thus, the spawning of dhufish is not apparently restricted to any particular water depth or region along the coast. However, because G. hebraicum greater than 300-320 mm in length (their size at first maturity) were almost invari- ably caught only around limestone or coral reef formations, this species apparently spawns in the vicinity of reefs. Because hydrated eggs and postovulatory follicles were found in at least some of the ovaries of large females in each month between November and April, it is evident that G. hebraicum spawn between the end of spring and middle of autumn. Although some fish commenced spawn- ing in November, the mean GSI of female fish in that month was still well below its maximum. This indication that only a small amount of spawning occurs in Novem- ber is consistent with the fact that many of the ovaries of large fish were still at stages III and IV. Although most of the ovaries of large females caught in May 1997 con- tained some vitellogenic oocytes, these oocytes were usu- ally undergoing atresia and the ovaries of other large fish in that month were either spent or resting. Furthermore, none of the ovaries of large G. hebraicum caught in May contained hydrated oocytes. This finding provides strong evidence that the spawning period does not extend into May. There is also strong evidence that spawning peaks in January and February. For example, by January, the ova- ries and testes of most large fish were at stages V or VI, i.e. prespawning or spawning, and, for the first time, some were spent or even recovering spent (stages VII and VIII). The maintenance of the GSIs of females at their maxima in both January and February is attributable to the fact that, because G. hebraicum is a multiple spawner, new batches of hydrated oocytes were continually being devel- oped in the ovary during these two months. However, the GSIs of females and males both declined precipitously in March, which demonstrates that, in the case of ovaries, the release of eggs during spawning was not being com- pensated for by a comparable production of new batches of mature eggs. As spawning activity peaked in January and February, it was appropriate to use 1 February as the birth date of G. hebraicum when assigning an age to each fish. 0.4 03 2 0.2 :i - 0 1 tB J 0 n-m-;^^ n=^ 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 Age (years) Figure 9 Age composition of Glaucosoiiia hebraicum. that were caught at lengths > the mini- mum legal length of .500 mm, and maximum likelihood estimates (±95% confidence limits) for the instantaneous coefficients of total mortality/year, Z, determined from subsets of these data selected by using different initial ages. The fact that, during the spawning season, mature ova- ries of G. hebraicum often contained yolk vesicle, yolk gran- ule, and hydrated oocytes and, in some cases, also post- ovulatory follicles, implies that this species is a multiple spawner sensu deVlaming (1983), i.e. individual females release eggs on more than one occasion in a spawning sea- son. The oocytes of individual female G. hebraicum during the spawning period ranged widely in size and, in many cases, their diameters formed relatively discrete modes in oocyte diameter-frequency distributions. The ovaries of G. hebraicum thus contain hatches of oocytes that are pre- sumably released at different times. Multiple-batch spawn- ing over a protracted period enables a greater total number of eggs to be produced and released during a spawning pe- riod and results in eggs becoming discharged at different times (McEvoy and McEvoy, 1992), which would increase the overall chance of recruitment success. Implications of the biology of Glaucosoma hebraicum for fisheries management The age composition data for dhufish older than 9 years and larger than the MLL reflect an average level of the instan- taneous coefficient of total mortality. Z. of 0.21/vear This value is consistent with the estimate obtained from the observed maximum age, taking into account the sample size of 620 fish (Hoenig. 1983). The much lower value obtained for Z, with Hoenigs ( 1983) regression equation for fish, i.e. 0.10/year, does not take into account sample size. The estimate of the instantaneous coefficient of natural mortality, M, of 0.25/year, that was calculated with Ralston's (1987) equation, exceeds the average value of 0.2L/year for the instantaneous coefficient of total mortality, which was estimated from the catch curve. Examination of the residu- als from the regression line fitted by Ralston (Fig. 8.1 in 226 Fishery Bulletin 100(2) Ralston, 1987) suggests that the precision of this estimate of the instantaneous coefficient of natural mortality is like- ly to be relatively low. It was therefore concluded that the value for M derived from Ralston's equation represented an overestimate. A more detailed examination of the catch cun'e data suggested, but was unable to demonstrate con- clusively, that the level of total mortality experienced by the older fish when they were young was less than that which is now being experienced by the population. Indeed, if the decline in the estimated value for Z, displayed in Fig. 9, was extrapolated to an age of 40 years, the total mortal- ity exhibited by the oldest age classes (when, as young fish, they first became fully vulnerable to the fishery) would be ca. 0.1/year. Such a value, which might be only slightly greater than the natural mortality, matches the estimate of Z calculated from the obsei-ved maximum age with Hoe- nig's ( 1983) regi-ession equation. However, such agi'eement may be fortuitous because the latter estimate should rep- resent the total mortality experienced by the fish within the sample, i.e. from age 9, rather than just the mortality of the older fish. Nevertheless, if the level of natural mor- tality, M, is ca. 0. l/year and the average level of instanta- neous coefficient of total mortality, Z, from age 9 years is ca. 0.21/year, the current level of fishing mortality, F. would exceed 0.1 l/year. In the nearshore waters along the lower west coast of Australia where this species is most heavily fished, the abundance of G. hebraicum has declined to a level that is of concern to fishermen. Numerous anecdotal reports indicate that commercial and dedicated recreational fish- ermen, such as those who provided the samples for this study, now tend to move further offshore in order to obtain catches of G. hebraicum comparable with those they used to obtain in waters closer to the coast. However, many rec- reational fishermen still continue to fish for G. hebraicum (and other species) in the traditional areas where dhufish were fished in the past. The expansion of the fishery for dhufish to include waters farther offshore, allied with the increasing use of global positioning systems (GPSs) to im- prove fishing efficiency, is increasing the level of exploita- tion of the stock as a whole. The fact that there are indications that the fishing-in- duced mortality of dhufish may now exceed natural mor- tality and that ongoing expansion in the extent to which fishermen are moving offshore (and also, in the case of rec- reational fishermen, in a northwards and southwards di- rection from the main metropolitan region of Perth) will further increase fishing pressure, is of concern to the man- agers responsible for the fishery for G. hebraicum. How- ever, because our sampling regime was not designed spe- cifically at determining the levels of fishing mortality to which G. hebraicum is being subjected, there is clearly a need to undertake a study in which the main aim is to achieve this objective. If such research were to confirm our preliminary findings that fishing mortality is reaching an unacceptable level, there will be an urgent need to use the biological data produced during the current study to refine the management plans designed to consei-ve this species. Female and male G. hebraicum first roach sexual matu- rity at the end of their third year of life when they are just over 300 mm in length and they reach 500 mm, the MLL for capture, when they are about 7 and 6 years old, respec- tively. Thus, on average, the female and male dhufish that live until they reach the MLL will have had the opportu- nity to have spawned for four and three years, respectively, before they can legally be retained following capture. On- going research at the state fisheries laboratory in Western Australia has indicated that ca. 50'^'i of fish caught in wa- ters of 20-30 m depth die on being released back into the water and that this percentage increases to ca. 95'r for fish brought to the surface from depths greater than 40 m (Moran-). Thus, the use of a MLL is likely to be of only lim- ited value for consei-ving this species as fishing effort con- tinues to increase. It is therefore important to introduce measures that will conserve G. hebraicum by maintaining the catches of this species at a level consistent with the re- quirements for ecological sustainability. Examples of such management controls might include closing areas to com- mercial and recreational fishing ( particularly those around reefs that are especially heavily fished) introducing quo- tas for commercial fish catches, making adjustments to the number of commercial licenses, further restricting the bag limit for recreational fishermen, and limiting the number of recreational fishermen that can fish in a given area. Furthermore, because the Fremantle Maritime Centre has successfully cultured G. hebraicum (Cleary et al.-^), there is also now the potential for restocking this species in areas in which it has become severely depleted. Acknowledgments Gratitude is expressed to those recreational and commer- cial fishermen and fish processors and many friends and colleagues for their help in the collection of fish and to colleagues at the Centre for Fish and Fisheries Research at Murdoch University for their help and advice. Grati- tude is also expressed to Gavin Sarre for providing inde- pendent counts of the number of translucent zones on otoliths and to two anonymous referees for their construc- tive comments on the original text. Funding was provided by the Fisheries Research and Development Corporation (FRDC). Literature cited Beamish, R. J. 1979. Differences in the age of Pacific hake iMcrluccius pro- ductus) using whole and sections of otoliths, J. Fish. Res. Board Can. 36:141-151. - Moran, M. 2001. Personal commun. Western Australian Ma- rine Research Laboratory, Fisheries Western Australia. PO Box 20, North Beach 6020, Western Australia. ^ Cleary, J. J., G. I. Jenkins, and G. Partridge. 1999. Prelimi- nary manual for the hatchery production of WA dhufish tGlauco- soina hebraicum ). Interim report to FRDC ( Fisheries Research and Development Corp.), 30 June 1999. (Project 96/308), 36 p. Fremantle Maritime Centre. 1 Fleet St.. Fremantle. Western Australia. 6160. Hesp et a\ Age and size composition, growth rate, reproductive biology, and habitats of Glaucosoma hebraicum 227 Bcrgquist, P. R., and I. G. Skinner. 1982. Sponges (phylum Porifera). ChapU'r 3 lit Marino invertebrates of southern Austraha, part I (S. A. Shepherd and I. M. Thomas, eds. ), p. 38-72. D. J. Woolmnn. Onvern- ment Printer, South Australia. Campana, S. E. 1984. Comparison of age determination methods for the starry flounder. Trans. Am. Fish. Soc. 1 13:365-369. Cerrato, R. M. 1990. Interpretable statistical tests for gi-owth comparisons using parameters in the von Bertalanffy equation. Can. J. Fish. Aquat. Sci. 47:1416-26. 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 mullet, Miigil ccphahis L., and yellow-eye mullet, Aldri- chetta forsteri (Valenciennes), in the Swan-Avon river sys- tem. Western Australia. Aust. J. Mar. Freshwater Res. 32; 605-628. Crossland, J. 1977. Seasonal reproductive cycle of snapper Chrysophrys aiirariis (Forster) in the Hauraki Gulf N.Z. J. Mar. Fresh- water Res. 11:37-60. Gushing, D. H. 1981. Fisheries biology: a study in population dynamics, 2"'' ed. Univ. Wisconsin Press, Madison, WI, 295 p. deVHaming, V. L. 1983. Oocyte development patterns and hormonal involve- ments among teleosts. In Control processes in fish physi- ology (J. C. Rankin, T. J. Pitcher, and R. T. Duggan, eds.), p. 176-199. Croom Helm, Beckenham. Fairclough, D. F, W. F Dimmich, and I. C. Potter 2000. Length and age compositions and growth rates of the Australian herring Arripis georgiana in different regions. Mar Freshwater Res. 51:631-640. Francis, R. I. C. C, L. J. Paul, and K. P Mulligan. 1992. Ageing of adult snapper (Pagrus aurattis) from otolith annual ring counts: validation by tagging and oxytetracy- cline injection. Aust. J. Mar. Freshwater Res. 43:1069-89. Gallant, A. R. 1975. The power of the likelihood ratio test of local nonlin- ear models. J. Am. Stat. 29:73-81. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mortality rates. Fish. Bull. 82:898-903. Hutchins, B. H., and M. Thompson. 1995. The marine and estuarine fishes of south-western Aus- tralia: a field guide for anglers and divers. Western Aus- tralian Museum, Perth, 103 p. Hyndes, G. A., N. R. Loneragan, and I. C. Potter 1992. Influence of .sectioning otoliths on marginal incre- ment trends and age and growth estimates for the flathead Platycephalus speculator. Fish. Bull. 90:276-284. Hyndes, G. A., M. E. Platell, I. C. Potter and R. C. J. Lenanton. 1998. Age composition, growth, reproductive biology, and recruitment of King George •whiting.Sillagmodes punctata . in coastal waters of southwestern Australia. Fish. Bull. 96: 258-270. Hyndes, G. A., and I. C. Potter 1996. Comparisons between the age structures, growth and reproductive biology of two co-occurring sillaginids, Sillagu rohusta and S. bassenNis, in temperate coastal waters of Australia. J. Fish Biol. 49:14-32. 1997. Age, growth and reproduction ofSillago schomburgkii in south-western Australian, nearshore waters and com- parisons of life history styles of a suite oi Sillago species. Environ. Biol. Fish. 49:435-447. Hyndes, G. A., I. C. Potter and S. A. Hesp. 1996. Relationships between the movements, growth, age structures and reproductive biology of the teleosts Sillago burrus and S. vittata in temperate marine waters. Mar Biol. 126:549-558. James, G. D. 1984. Trevally, Caranx georgianus Cuvier: age determina- tion, population biology and the fishery. Fish. Res. Bull. NZ. 25, 51p. Johnson, N. L. and S. Kotz 1970. Continuous univariate distributions-1. John Wiley, New York. NY. 300 p. Khoo. K. H. 1979. The histochemistry and endocrine control of vitello- genesis in goldfish ovaries. Can. J. Zool. 57:617-626. Kimura. D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Laevastu, T. 1965. Manual of methods in fisheries biology. FAO, Rome, 51 p. Laurenson, L. J. B., I. C. Potter and N. G. Hall. 1994. Comparisons between generalised growth curves for two estuarine populations of the eel tailed catfish Cnido- glanis macrocephalus. Fish. Bull. 92:880-889. 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, (suppl. B) 41:125-136. McKay, R. J. 1997. Pearl perches of the world. FAO species catalogue, vol. 17. FAO, Rome, 24 p. Microsoft Corporation. 2000. Excel. Microsoft Corporation, Seattle, WA. Optimas Corporation. 1995. Optimas 5, user's quide and technical reference, vol.1, 7'h ed. Bothell, WA, 524 p. Ralston, S. 1987. Mortality rates of snappers and groupers. //? Tropical snappers and groupers. Biology and fisheries management (J. J. Polovina and S. Ralston, eds.), p. 375-404. Westview Press, Boulder, CO. Saila, S. B., C. W Recksieck, and M. H. Prager 1988. Basic fisheries science programs. Elsevier, New York, NY, 230 p. Sarre, G. A., and I. C. Potter 2000. Variation in age compositions and growth rates of Acanthopagrus hulcheri iSparidael among estuaries: some possible contributing factors. Fish. Bull. 98:785-799. SPSS Inc. 1988. SPSS-X™ users guide. SPSS Inc., Chicago, IL, 806 p. Stergiou, K. 1999. Intraspecific variation in size- and age-at-maturity for red bandfish, Cepola macrnphthalma. Environ. Biol. Fish. 54:151-160. 228 Abstract— Longitudinal surveys of ang- lers or boat owners are widely used in recreational fishery management to estimate total catch over a fishing season. Survey designs with repeated measures of the same random sample over time are effective if the goal is to show statistically significant differ- ences among point estimates for succes- sive time intervals. However, estimators for total catch over the season that are based on longitudinal sampling will be less precise than stratified estimators based on successive independent sam- ples. Conventional stratified variance estimators would be negatively biased if applied to such data because the samples for different time strata are not independent. We formulated new general estimators for catch rate, total catch, and respective variances that sum across time strata but also account for correlation stratum samples. A case study of the Japanese recreational fishery for avu tPlecoglossus altivelis) showed that the conventional stratified variance estimate of total catch was about 10^:i of the variance estimated by our new method. Combining the catch data for each angler or boat owners throughout the season reduced the vari- ance of the total catch estimate by about 75%. For successive independent surveys based on random independent samples, catch, and variance estima- tors derived from combined data would be the same as conventional stratified estimators when sample allocation is proportional to strata size. We are the first to report annual catch estimates for ayu in a Japanese river by formu- lating modified estimators for day-per- mit anglers. Longitudinal logbook survey designs for estimating recreational fishery catch, with application to ayu iPlecoglossus altivelis) Shuichi Kitada Tokyo University of Fisheries 4-5-7 Konan, Minato Tokyo 108-8477, Japan Email address kiladaia'tokyo-u-fish.ac ip Kiyoshi Tezuka Nakagawa Branch Tochigi Prefectural Fishenes Expenmental Station Ogawa, Nasu Tochigi 324-0501, Japan Manuscript accepted 14 September 2001 Fish. Bull. 100:228-243 (2002). Angler surveys are widely used in fish- ery management to estimate recreation- al catch, and there is an extensive body of literature on this subject (see Guthrie et al., 1991). Pollock et al. (1994) pub- hshed a manual on angler sui-vey meth- ods and their applications in fishery management. The first purpose of our study is to make two very important points for the designers of recreational fishery surveys: 1 ) longitudinal surveys taking repeated measures on the same random sample of anglers or boats over time are better than successive indepen- dent surveys if the goal is to determine significant trends in catch and fishing effort, and 2) stratified surveys, or suc- cessive independent surveys based on random independent samples of ang- lers or boats, are better than longitu- dinal surveys if the goal is to obtain pre- cise estimates of annual total catch and fishing effort. If longitudinal fisheries data sets are used to estimate annual catches, then correlations between monthly sample observations must be taken into account when evaluating the precision of catch estimates. This problem is not addressed in the litera- ture, and the variance estimation pro- cedures for this situation are unclear. The second purpose of our study is to estimate the annual catch of ayu iPleco- glossus altivelis) in a river because no estimates have been reported in Japan. In our study, we formulated a new method for accurate variance estima- tion with longitudinal fishery data, ex- emplified by a case study of the rec- reational fishery for ayu in Nakagawa River in Tochigi Prefecture, Japan (Fig. 1). Annual catch estimates based on sums of monthly estimates were com- pared with those based on combined data for each angler or boat through- out the fishing season. We demonstrate how use of a design with repeated mea- sures facilitates determination of sig- nificant seasonal trends in catches, and show the usefulness of combined (non- stratified) data analysis. We also esti- mate the total annual catch of the ayu fishery by formulating modified esti- mators for day-permit anglers (anglers who are granted permits to take fish for one day). Materials and methods Case study of ayu We used longitudinal data collected for the Japanese ayu fishery to compare estimators of effort and catch and their associated variance estimators. Ayu is the most popular target species of rec- reational anglers in rivers in Japan. In the Nakagawa River (Fig. 1), the upstream run of wild juvenile ayu from the coast begins in late March to early April, and is completed by early July Ayu mature and spawn from September to November, and then die after spawn- ing. Cooperatives release both hatch- ery-produced and wild juveniles caught Kiiada and Tezuka: Survey designs for estimating recreational fishery catch 229 Figure 1 Location of the angler survey. Bold lines in the lower figure indicate the area of the Nakagawa River where the sui"vey was conducted. in Biwako Lake, Shiga Prefecture (Fig. 1) from early April to the end of May. Thus, recreational anglers catch both naturally recruited wild juveniles and transplanted wild juveniles from Biwako Lake and hatchery-produced ayu released by the cooperatives. Both hatchery and wild stocks consist of a single year class that recruits in the spring. The river fishing season for ayu begins on June 1st and closes at the end of October To estimate the annual ayu catch by recreational anglers in the Nakagawa River, we conducted a longitudinal log book survey in 1993. Sampling procedure There are four cooperatives that set fishing rights on the Nakagawa River in Tochigi Prefecture. Fishing permits for ayu are sold at the cooperatives and fishing tackle shops, and these permits are valid over the entire Nak- agawa River in the prefecture. The cooperatives record the total number of season- and day-permits sold, and a complete list of season-permit anglers is available. An a priori sample size of 120 anglers (an expected sampling fraction of about 0.5% of the total number of season-per- mit anglers) was allocated to the four cooperatives in pro- portion to the number of season-permits sold (Table 1). Anglers who possess a permit (season or day! can fish for ayu over the whole Nakagawa area, regardless of where the permit was purchased. Hence, we treated the samples as if they were drawn from the population by simple random sampling, even though they were drawn by strati- fied random sampling of cooperatives. We asked the cooperatives to select samples randomly, but the samples were drawn arbitrarily. The selection, however, was not a purposive sampling; therefore we treat- ed them as random samples. The sampled season-permit anglers were asked to record catch data throughout the fishing season, including each fishing date, the number of ayu caught, and the fishing site, on a printed form, which was returned after the fishing season was over. To estimate the total catch in weight, we also surveyed the body weights of ayu in recreational catches by month. The 230 Fishery Bulletin 100(2) The number of tions along the permits Nakaga sold waP Table 1 in 1993. sample size, and the number of logbooks i iver. ;; = number of anglers sampled. etu •ned by th c four fishermen's cooperative associa- Fishermen's cooperative associations Number of season permits sold Number of day permits n Number of logbooks returned Hokubu Nanbu Chuo Motegi Total 11,314 6.391 1911 2346 21,962 6520 1946 231 369 9066 70 30 10 10 120 64 21 10 9 104 primary sampling unit in a population of season-permit anglers or boat owners (i.e. party-boat owner.s or per.son- al-boat owners) was an angler or a boat owner, and the secondary sampling unit was a fishing day. We selected anglers or boat owners by simple random sampling with- out replacement from a list of anglers or boat owners, and asked a sample of anglers or boat owners to record catch data on all fishing days throughout the survey. Because all the secondary sampling units were surveyed, we regard- ed this as a single-stage cluster sampling procedure (Co- chran, 1977). Estimation of total catch by month V{R,J- N N-n Mr (2) This variance estimator is obtained by dividing Equation 6.9 in Cochran ( 1977, p. 155 ) by the total number of fishing days in the /i'th month M^,. In Equation 2, M,. is unknown; hence we approximated the variance estimate by using the estimator of M/, as follows (Thompson, 1992, p. 621: ViR,^ = —± N-u NMpiUi - 1) t" ^iC,,-Mj,f, (3) Estimation for season-permit anglers or boat owners The principal notations for estimation of the total catch for season-permit anglers or party-boat owners are as follows (Cochran, 1977; Pollock et al., 1994): N = total number of sampling units (season-permit angler or party-boat or personal-boat owner; known number); n = sample size drawn from the population A'^; M). = total number of fishing days in the population in /(•th month (to be estimated); M,;, = total number of fishing days in kth month of selected /th sample; M^_ = mean number of fishing days per sampling unit in kih month (to be estimated); C^f. = number offish caught by ;th sample in /;th month; R/^ = catch rate of /?th month (to be estimated); Cj,"' = total catch by season-permit anglers or by party- boat owners or personal-boat owners in A'th month (to be estimated). The catch rate is the number offish caught per sampling unit each day. For logbook surveys, the ratio of the mean is the preferable estimator of catch rate (Jones et al.,1995; Pollock etal., 1997). The ratio and the variance for the catch rate is estimated by R, X—' " I,.l^-'- (1) where M^, = NMj. The variance of M^. is estimated by ViM/. ) = N'ViM,. ). where M,, is the mean number of fish- ing days per sampling unit in /;th month. The estimator and the variance are as follows (Cochran, 1977, p. 249, fromEq. 9A.2): V(M,) = Nnin-1)'^ ''■ '• (4) (5) The total number offish caught in A'th month is estimat- ed by CI"' = Mi,Ri, = NM,M,, T.J'- TJ- N :7V- Ic.. (6) This results in an unbiased estimator. Wlien the total number of fishing days M,. is unknown, the ratio estima- tor coincides with the unbiased estimator. The variance is evaluated by (Cochran. 1977, p. 249, from Eq. 9A.2): Kitada and Tezuka Survey designs for estimating recreational fishery catch 231 iHn-l) ^ where Q = V C,^, / n. W'jl'" = Q'"wj(. In general, season-permit and day-permit (7) anglers fish in the same location; therefore we assumed the same mean body weight for both kinds of anglers. The samples for estimating C/. and iv f, are independent; therefore the variances are estimated by using Goodman's (1960) formula: Estimation for day-permit anglers We estimated the catch of day-permit anglers separately. The notations for day- permit anglers catch estimation are as follows: D = total number ofday-permits sold through the season (known); Dj^ = total number of day-permits sold in k th month (known); dj. = total number of day-permit anglers who returned logbooks in kth month; Rl''' = catch rate of A'th month for day-permit anglers; Cjf' = total catch by day-permit anglers in ^th month (to be estimated). The estimator of total number of fish caught by day-per- mit anglers in A'th month is v(w,j = v(w;;') + v(wi" (14) where ViWl"') = W^V(C'^-") + cf V(JZ7,, ) + V(C'^")V{UJ^ ) and V(Wl'") = ZZvfV(Ci'" ) + C','"''V{W, ) + V(C['' ' )V{Ui, ). The mean body weight offish in Ath month is estimated from a sui-vey of individual body weights (u\^, of /,, fish caught on the fishing grounds). The estimator for Ath month and its variance estimator are and ViUJ^ ' = Z !li ' "' * ' '"^^ '" / (/* (4 - !>)■ Cl'"=DX''\ (8) where the catch rate for day-permit anglers, estimated by the sample mean, is The total fishing days in Ath month is estimated as the sum of the fishing days estimates of the season-permit an- glers and the day-permit anglers by K'" ^ d,: (9) The variance estimator of Cj/' is (Cochran, 1977, p. 26, from Eq. 2.20): nCl'"}= D-iViRi'"} ^^^'A-c^^)|-(C,, - R,'^)\ d,id, (10) Total monthly catch of season- and day-permit anglers The total catch by season- and day-permit anglers in ^th month is Q, which is estimated by C,=Cl'' + C['". (11) Mj.f. =M^+Di^ (15) and the variance is estimated by V(M.,.i.) = N'-ViMi^) be- cause Di is known. Estimation of annual catch Method 1 (based on monthly estimates) The annual catch is estimated by summing monthly catch estimates over the entire fishing season (A' months). The point estimator c=£Q=£cr+|;ci'"=c-+c"^'. (16) These two total estimates are obtained from independent samples (season- and day-permit anglers); therefore the variance is estimated by adding the variances: V(C.) = V(C,") + V{CY (12) When the same sample of anglers reports catches through- out the season, the sampling is not independent in each month, and monthly catch estimates are auto-correlated. Taking this correlation into account, the variance estima- tor is The total catches in weight in kth month are estimated by using the mean body weight of the species in ^th month ( w f^) estimated from the survey of individual body weights by W,=W-' + <'=C,i^, (13) where VV^~ and W^'^'are the total catches in weight for sea- son- and day-permit anglers, given by W"^" = C^'W^ and V(C) = V(C'") + V(C'") = XV(Cr') + 2;^c7v(Cr,Ci^') K K (17) +^v(c,'")+2^cov(c;'",c;^'). 232 Fishery Bulletin 100(2) The covariance between two total estimates of season-per- mit anglers is estimated by (see Appendix 1, from Cochran, 1977, p. 25) CovCC;' N(N-n ^C-, = i^^^lZii!^,C„-C,)(C,,-C,). (18) nin-1) ^ The fourth term of Equation 17 is equal to 0 if a different sample of day-permit anglers is drawn in each month. The total catch in weight is estimated by K K K w = ^Wi = ^w;;' + '^Wi';" = w"-' + w"^' (i9) and the variance estimator is similar to Equation 17 but has a slightly different covariance which is Cov(w;-^',w;'') - w^m^.Cov{c;f\ci- (20) This covariance estimator was derived by the delta method (Seber, 1982, p. 7, see Appendix 2). which coincides with a covariance when Wj and w,. are constant. The mean annual catch rate is estimated for season-per- mit anglers or boats by The estimator of covariance between C"" and M"' is simi- lar to Equation 18 (see Appendix 1): C^{C"",M"")= ^'^~"'y (C ,-C)(M ,-M). (25) where C, and M, are the number of fish caught and the number of fishing days of (th season-permit angler or boat owner throughout the season, respectively, and ^" = l.lf-"^"- and M y" M,l n. The mean number of fishing days per sampling unit (season-permit angler or boat) is estimated by — M" M = N and the variance estimator is V(M) = -^ViM'"}, R C" M'" (21) where M'" = ^M^ = iV^M* Here M /, is given by Equation 4. The total effort is esti- mated by M'^'+ D. The approximate variance of ¥(/?'"") is given by the delta method (Seber, 1982, p. 7; Appendix 3), that is ViR"") M' V(C"") + c \2 M'- ViM'") M' (Cov(C'",M'-') (22) where V(C' ' ) is given by Equation 17, and the variance of the total number of fishing days is estimated by V(M"') = iV-'K^V(M;,)-^2^Cov(M^,M;..) . (23 Here ViM^.) is given by Equation 5 and the covariance of two sample means is estimated by (Cochran, 1977, p. 25) Cov(M,. M.. ) = ^ " T^M,, - M.)(M,,. - My ). (24 where M'"'and V(M'"') are already derived from Equation 21 and Equation 23, respectively. The catch rate of day-permit anglers over the season is estimated by R r'' D and the variance is V{k•") = A^V^C■-"\, D- where C'"'' and V(C"'') are given in Equation 8 and Equa- tion 10, respectively. Method 2 (based on total annual catch of each angler or boat owner) Another procedure for estimating annual catch is to use annual data rather than monthly catch for individual anglers or boats. The advantages of this proce- dure are that covariances between months do not have to be considered and estimators are much less complicated than those obtained using method 1. Equations derived for monthly estimation can be used without modification for this procedure. Modified estimators for day-permit anglers We could not conduct a sui-vey of day-permit anglers, so we substituted i?^ for R^" in Equation 8. In addition, the total number of day permits sold in /;th month ^D^) was Kilada and Tezuka: Survey designs for estimating recreational fishery catch 233 unknown. Hence, we slightly modified the procedure for estimating D,. by Dpi., where p^. is the proportion of day permits sold in ^th month to the annual total number of day permits sold (D). Day permits issued by the cooperatives are sold mainly in fishing tackle shops. We selected four tackle shops and surveyed the total number of day permits sold at the se- lected jth fishing tackle shop (D^), and the total number of day permits sold at the selected /th fishing tackle shop in /.■th month (D^i.). The proportion of day permits sold in ^th month was estimated by Pk = " n i;.^. where h is the number of fishing tackle shops selected from a total of// shops. The evaluation of the variance of P/. was similar to Equation 3: Vlftl = "'""'tI'"..-".'.''- D'hih- Some day permits, however, were sold at the fishing sites, and the above variance estimator was not appropriate for this situation. Assuming S'^'^j/), was selected by simple random sampling from D, we evaluated the variance by (Cochran, 1977, p. 52, Eq. 3.8) ^E>.-i)' The modified estimator for the total number of fish caught by day-permit anglers in /;th month is k DpA (26) Here p^ and Rf, are independent because these are esti- mated from different survey data. The variance of revised d''' was estimated by using Goodman's (1960) method; ViCi"'): : d-'[r'^V(p,) + pIv(R, ) + V(ft mR, )}. The total fishing days in /jth month was estimated by Equation 15 but in this case /)^ was unknown. The modi- fied estimator was Mn = M, + Dp, and the variance was slightly revised as Vmn) = N''Vm,) + E)'V(p,). The annual catch was estimated by Equation 16, sub- stituting Equation 8 by Equation 26. In this case, we estimated C[ and C,- ' from the same sample of season- permit anglers. Hence, the fourth term of the covariance in Equation 17 must be considered. The approximate covari- ance is estimated by (see Appendix 4) c'o^(c;;",c;/') = D'~p,,p,,c^v{R,^,Rf^.), ai) where the covariance between /?^, and /?;,. is Cov(R,,Rk)- N' M,M,, ^^Cov(C;-'',M^.) M,Ml M'kM^. Cov( Mi,C;?' + ± Z Cov(M^,M^.) Here Cov(Ci'",Ci?') and Cov(M,..Mi,.] are already given by Equation 18 and Equation 24. The other covariance com- ponents are c^v(&;\M,. ) = ^-'\ f (C,, - ^, )(M„. - M,.) n~{n - 1) ■^ C'^(M„C;?') = ^~'\ j^(C„. - 4 "M,, - M„ ). n (n -1)~ The annual catch in weight was estimated by Equation 19, substituting Equation 8 with Equation 26. The covariance in the fourth term of V(W) in Equation 17 was estimated with Equation 20 by c^( W;!'", vv;.'" ) = (t,(tj..(5rv(c;'",c;?' ) where Cov(C/ ,C^' ) is given in Equation 27. Results In 1993, 21,962 season permits and 9066 day permits were sold (Table 1). The total number of day permits sold at the four fishing tackle shops was 4776, and the number (pro- portion, p^) sold was 2732 (0.572) in June, 1,189 (0.249) in July, 716 (0.150) in August, 124 (0.026) in September, and 15 (0.003) in October. We received 104 logbooks from the 120 season-permit anglers sampled, a return rate of 86.7%. In addition, two anglers voluntarily submitted logbooks, but we did not in- clude these unsolicited returns because they were not ran- domly selected. The modes of the catch rates by the sam- pled anglers were from five to ten fish per month (Fig. 2). The histograms show a large variation in the catch rate among season-permit anglers. The peak fishing season was from June to July. In September, the number of anglers decreased, and the fishing season ended in October The 234 Fishery Bulletin 100(2) 1- 10 20 30 40 50 60- July 3n 1- Mil □_ 10 20 30 40 50 60- October n n 10 20 30 40 50 60- August Ha ^ r-l _ 4 2 0 10 20 30 40 50 60- Total nnn 10 20 30 40 50 60- 10 20 30 40 50 60- Catch (number of fish) Figure 2 Distributions of the catch rate of ayu by month for 104 sampled season-permit anglers in the Nakagawa River. modes of the number of fishing days per season-permit an- gler were five for all months. The variation in the niunber of fishing days among anglers was also large (Fig. 3). Monthly plots of the total number of fish caught versus the number of fishing days showed linear relationships (Fig. 4). The variation in Figure 4 indicates differences in the skill of the anglers. The monthly number of anglers decreased over the fishing season. Figure 5 shows the monthly changes in the total number of fishing days, the total number of fish caught, and the catch rate for the 104 sampled anglers. The decline in number offish caught was largely due to the decrease in fishing days. The change in catch rate indicated a decline in the abundance of the stock. The mean body weight of ayu was greatest in June (Fig. 6) and was affected by a method of fishing for ayu called "Tomo-zuri" angling, which takes advantage of the attack- ing behavior of ayu when another fish enters its territory. Anglers attach a "call" fish (a live ayu) above a treble hook that snares the territorial wild fish, as it attacks the "call" fish. Because larger individuals establish territories ear- lier than smaller ayu, fish caught in June were predomi- nantly the larger individuals. Reflecting the monthly trend in the number of fishing days, 89'7,_ of the total annual catches of season-permit an- glers and 98'>; of those of day-permit anglers were taken from June to August (Table 2). The catch by day-permit anglers was substantially smaller than anticipated, esti- mated at about 29^ of season-permit anglers' catch in both numbers and total weight. CVs ranged from 7% to 12% in June and July for all parameters; however, they were higher in August and September, ranging from 10% to 209; . In October, CVs exceeded 43% for total catch in num- ber and weight. The decreasing precision of the monthly catch rate estimates was caused by the decrease in anglers («,,) (Figs. 4 and 5). _ The CVs of annual estimates of M and Mj, by method 1 were about T'r. but that of R'"' was about 20':i (Table 2i because we evaluated the covariance terms for the number of catches and fishing days between months; those were Kitada and Tezuka: Survey designs for estimating recreational fishery catch 235 w 8- o 4- n E 8- June September £2. 5 10 15 20 25 July 8- 4- II 5 10 15 20 25 October 5 10 15 20 25 5 10 15 20 25 4n August 2- Total n^r. 5 10 15 20 25 10 30 50 70 90 Number of fistiing days Figure 3 Distributions of the ayu fistiing days per angler by niontii for 104 sampled season-permit anglers in the Nakagawa River Cov{C'if\Cl-[) in V(C'")and Cov(M;,,,M,. ) in Equation 22. The CVs of C and W were also evaluated at about 21% and were strongly affected by the covariances between months in Equation 17. The variance of the total number estimate V[C) was 1.2604 x 10'-. and variance by neglecting the covariance term in Equation 17 was 1.2230 x 10". The CV of C without the covariance term was 6.53%. If we ne- glect the covariance, the variance is substantially under- estimated. The variance was 10.31 times larger when the covariance term was included. We obtained similar point estimates of anijual catch by method 2 iTotal'"'^'^. 2 ;„ -pable 2 ). The CVs of M , Mj, C, and W for day-permit anglers were about 7%, but that of i?'" was reduced from 19.7% to 6.6% by not considering the co- variance terms. The CVs of C and W dropped about 10% from 21% without the covariance. Similar point estimates and smaller variance estimates were obtained. The vari- ance estimate of the annual catch obtained by method 1 with covariances ( 1.2604 xlO'-'l was 4.11 times larger than that by method 2 13.0667 xlO"). The relationship between the sample size and the pre- cision of the annual catch estimate for season-permit an- glers was examined. We calculated the values ViC) for var- ious values of n by using Equation 7. To obtain precision over the season for CVs of C''"( = Vviti/r) below 10%, a sam- ple size of 120 or more is required (Table 3). A high positive correlation in catches between adjacent months was detected (Table 4). We mapped anglers (ob- jects) and fishing days (categories) into a two-dimension- al graph by correspondence analysis (Hayashi, 1950; Ben- zecri, 1992) using the function "pqS.prcomp" in S version 4 (Chambers and Hastie, 1992). Correspondence analysis showed the relations between rows and columns in a fre- quency table gi'aphically as points in a common low-di- mensional space (Clausen, 1998). Both objects (rows) and categories (columns) of variables are represented as points in such a way that an object is relatively close to its catego- ry and relatively far from other categories (Leeuw and van Rijckevorsel 1988). For example, the 72nd angler fished 10 days in June, five days in July, seven days in August, three days in September, one day in October, and this angler was mapped closed to June, reflecting the month of his high- est fishing effort (Fig. 7). The results suggest several fish- ing patterns with high catch seasons in June-July, July- 236 Fishery Bulletin 100(2) 6-1 June n=95 0 September n=52 4- o ° 0 1 0 , . 0 5 10 15 20 25 0 5 10 15 20 25 o 6-| o 6-1 July o n=100 October n=16 CD O 4- 4- SI c/) O OJ E -=> c 2- 0- C „ oo oO 0 o° o O °° 8 2" 0 5 10 15 20 25 0 5 10 15 20 25 6- tir' Total Q n=102 4- 0 12- 0 2- 0- 8- ooigo o ° 4. D 5 10 15 20 25 0 o 0 O o J^^°o 0 D 10 30 50 1 1 70 90 Number of fishing days Figure 4 Relat ionship between the number of fishing days and the number of fi; h cau ?ht by 104 sampled season-permit anglers in the Nakagawa River August, August-September, and September-October, re- sulting high correlations between adjoining months, and large covariances between distant months (Table 4). The prime advantage of a longitudinal study is its ef- fectiveness for studying change, and a repeated measures analysis of variance can be applied to a complete data set with a constant correlation (Diggle et al., 1994). However, our data set was incomplete because the number of anglers who fish in each month changed (see n in Fig. 4) and had a different correlation structure among month (Table 4). We tested the differences between successive monthly catch estimates of season-permit anglers by using a parametric bootstrapping method. In the central limit theorem, the sample distribution of a monthly total catch estimate can be regarded as a normal with the mean C^'-^'and the variance ViCj."). Based on the two point estimates, vari- ance estimates and the correlation coefficient between suc- cessive two months, we generated 10,000 bivariate normal random variables (Gentle, 1998). The means and 95% con- fidence intervals of the differences between two monthly total catch estimates were -226,561 1-650,404-203,080) for June and July, 870,720 1455,091-1,277,402] for July and August, 470,594 1161,013-783,1681 for August and September, and 537,488 1290,905-782, 727| for September and October. Significance levels were corrected for multi- ple testing by using the Bonferroni ajustment factor (So- kal and Rohlf, 1995). The confidence interval for June and July straddled 0, showing no significant difference. On the other hand, three other confidence intervals did not in- clude 0; therefore the monthly differences were statisti- cally significant (Fig. 8). Discussion Bias and source of variation The estimate of the total annual catch of ayu by the rec- reational fishery was the first obtained in Japan and was much larger than expected. The total number of day-per- mits sold was 9066, and was quite small (1.9%) compared with the estimated total number of anglers (477,520). Kitadd and Tezuka Survey designs for estimating recreational fishery catch 237 Although the difference in the catch rate between day- and season-permit anglers was unknown, the influence of this bias on the total catch estimates would be minor In order to check the bias, however, one could conduct a logbook survey of day-permit anglers. Sixteen anglers of the total sample ( 13'/r I in our study did not return logbooks and therefore may have caused a bias in our estimates; however no attempt was made to evaluate the difference between nonrespondents and respondents. The angler sample was drawn arbitrarily by the cooperatives but was not a random sample in the strictest sense. If cooperative anglers tended to be selected, this could have been a source of bias. The source of variation in total catch is the variation in the catch of the sampling unit, including differences in fishing days, skill of the anglers, and the number of an- glers that a party boat could accommodate. A stratified sampling scheme based on categories of anglers or boats is effective for this situation. The weakest point in the use of logbook surveys, perhaps, is that the catch data are report- ed by those who catch the fish and by boat owners with monetary interests. To what extent the anglers might have exaggerated or under- reported their catch is not known. Party boat owners may record lower than actual catches to reduce taxes. To examine this possible source of bias, on- site sui-veys should be conducted. For the ayu fishery in the Nakagawa River, an access point survey may be prac- tical (Pollock et al., 1994). When comparatively complete lists of boat owners and anglers are available, logbook sur- veys based on these lists, combined with on-site surveys, are appropriate. Longitudinal and stratifled survey designs Longitudinal surveys taking repeated measures on the same random sample over time are better than successive independent surveys if the designer's goal is to show sta- tistically significant differences in the estimates between time intervals. Monthly estimates showing seasonal trends 238 Fishery Bulletin 100(2) 30- 20- 10- 30- 20- _g 10- E 30- 20- 10- June n=47 Mean=56 Ma 20 40 60 80 100 July n=74 IVIean=46 n n n n II n 20 40 60 80 100 August n=68 Mean=52 n nn 30-1 20- 10- September n=63 Mean=45 n n 80-1 60- 40- 20- 0 20 40 60 80 100 Total n=252 IVlean=49 _IZL M JTL 20 40 60 80 100 20 40 60 80 100 Body weight (g) Figure 6 Distributions of body weight by month foi' ayu caught in the Nakagawa River- can be obtained by the equations derived in our study. In such repeated measure designs, the most precise esti- mates of annual catch are obtained by method 2. On the other hand, stratified surveys, or successive independent sui-veys between time intervals based on random samples of anglers or boats, are better than longitudinal surveys if the designer's goal is to obtain precise estimates of total effort, or catch (or total effort and catch), over the entire season. Stratification by month would improve the preci- sion of annual estimates even more if estimates varied greatly across months. Stratified sampling allows inde- pendent monthly estimates, and monthly estimates can be summed to produce precise estimates over time. In the absence of correlations between monthly sample obser- vations, the estimated variance of annual estimates can be obtained simply by adding the estimated variances of the monthly estimates. The estimated variances of annual estimates stratified by month would be considerably less than those of annual estimates based on repeated monthly observations of a one-time annual sample. If method 2 is used to analyze data obtained by such in- dependent surveys, how would the precision of the estima- tor compare with the precision of a stratified estimator? For simplicity, we consider a population that is divided into two subpopulations of A^,, N., units, respectively. The stratified estimator of the population total and the respec- tive variance are V{Y} = N^^V{y,) + N'^V{y,), where Vj and y-, are the sample means for sample sizes of «, and /!2. On the other hand, those obtained by method 2 are - TV, + N.-, _ r, = — ' =- 1 " i.v, + n.,y.:, ), ^,^., J..(7V..N,,]-y, j,MiV, + 7V,,-^,-^, «] +n., n, + n.. Subtracting i' from Y. we have Kitada and Tezuka: Survey designs for estimating recreational fisliery catch 239 Table 2 Estima ed pai ameters and coefficients of variation lin pare nthescs). Total"""' ' = total estimated by summing iionthly estimates ( method 1 ). Total""""' ^ = total estimated by combining data throughout the season (method 2) fli- ' = catch rate for season-permit anglers M = mean number of fishing days per season-permit anglers. Mj. = total number of fishing days (s eason-i-day-permit | anglers ). C"^' = total catch in lumber for season-permit angl ers. C''' = total catch in number for day-permit angl srs. C = total catch in number (season+day-perm it anglers). W" = total catch in weight for season permit anglers. W-' = total catch in weight for day- | permit inglen . W = total catch in weight (season+day-permit anglers). Parameter June July August September October Total""-"' ' Total""^"' '^ fl'"' 11 12 12.37 10.29 9.29 5.35 11.04 11.04 (0.087) (0.075) (0.115) (0.187) (1.495) (0.197) (0.066) M 6.92 7.05 4.62 2.80 0.28 21.66 21.65 (0.073) (0.075) (0.101) (0.153) (0.291) (0.073) (0.073) M^ 157,231 157,047 102,722 61,687 6,152 484,839 484.628 (0.071) (0.0741 (0.100) (0.152) (0.290) (0.072) (0.071) QS) 1,690,440 1.914,495 1,042,772 570,800 32,732 5,251,241 5.251,241 (0.116) (0.115) (0.154) (0.179) (0.437) (0.214) (0.105) Qd) 57,658 27,915 13,982 2,186 152 101,894 100,108 (0.087) (0.077) (0.118) (0.197) (1.528) (0.056) (0.066) c 1,748,099 1.942,410 1,0.56,755 572,987 32,884 5,353,135 5,351,349 (0.112) (0.114) (0.152) (0.178) (0.435) (0.210) 10.103) w""(n 94.64 88.30 53.26 25.85 1.48 263.. 53 253.90 (0.125) (0.122) (0.159) (0.184) (0.439) (0.213) (0.107) W"''(n 3.23 1.29 0.71 0.10 0.01 5.34 4.84 (0.099) (0.086) (0.124) (0.202) (1.530) (0.065) (0.069) WU) 97.86 89.59 53,98 25.95 1.49 268.86 258.74 (0.122) (0.120) (0.157) (0.183) (0.437) (0.209) (0.105) Table 3 Coeffic ent of variations of the total catch es timate C'-" (method 2) for various sample sizes (number of anglers). ;; CV n CV n CV 10 0.3401 110 0.1025 230 0.0709 20 0.2405 120 0.0982 250 0.0680 30 0.1963 130 0.0943 300 0.0621 40 0.1700 140 0.0909 400 0.0.538 50 0.1521 1.50 0.0878 500 0.0481 60 0.1388 160 0.0850 600 0.04.39 70 0.1285 170 0.0824 700 0.0406 80 0.1202 180 0.0802 800 0.0380 90 0.1134 190 0.0780 900 0.0358 100 0.1075 200 0.0760 1000 0.0340 Y-Y.. n.>Ni-niN2 (>'i-y2'- showing that the two methods provide different estimates with the extent of the difference, depending upon the Table 4 Estimated variance-covariance matrix ( xlO"') for the monthly estimates of catch (number) by season-permit 1 anglers in number by Equation 17 (lower diagonal) and the correlation coefficient r(C,(., C,.) (upper diagonal, m bold font). The diagonal component refers to V'(Ct and the lower half refers to Cov(Cj",Cj'' ). Month Jun Jul Aug Sep Oct Jun 3.837 0.67 0.43 0.30 -0.03 Jul 6.011 4.862 0.65 0.49 0.08 Aug 9.422 4.434 2.578 0.65 0.28 Sep 10.681 6.258 1.456 1.038 0.44 Oct 10.725 6.336 1.522 0.069 0.020 sizes of the strata, the sample sizes, and the estimates of the sample means. According to Cochran (1977), method 2 works well enough if the sample allocation is propor- tional because a simple random sample distributes itself approximately proportionally among strata. With propor- tional allocation N^/n^ = NJn.^; therefore the difference of the two estimators is 0. In our case study, the annual 240 Fishery Bulletin 100(2) 100 97 O - d 20 99 ,7 12 2|4 10 ,^95 ^1 ^1, ond axis 0.0 1 ^J430 2Au^^g '' ^393 4 1 80 Juty ags ^' 77 93 31 32 34J, 90 Sept. ^.sse. ^^^^ -P'j^ ,6 71 f»6 35 57 ^^^02 3B2 June72 ^ ^^6 « in o 9 47 51 ^3 " 18 ^' 55 o 5 " 70 Oct. 1 1 1 1 1 1 1 -008 -006 -004 -002 00 0 02 0.04 First axis Figure 7 Plots of the determined quantities for sample (anglers) and category (months) from the correspondence analysis for the fishing days of 104 sam- pled anglers. The numbers in the figure refer to the number of anglers that were sampled catch estimates for the season-permit anglers C"' were the same value for metho(i 1 and method 2. The population size N and the sample size /; were in proportion constant throughout the season; therefore NJn^ = NIn was same for all strata. The ratio of the variances is V(Y^ ) {N,+ N., fU'fViy, ) + nlV(y, )| ' which also shows that the precision of both methods depends on the sizes of the strata, the sample sizes, and the variance of the sample means. If the allocation is proportional, the variance ratio becomes 1 and the two variances coincide. The objective of stratified surveys is to obtain precise total effort or catch estimates (or both) over the entire season. A proportional sample allocation is recommended, which allows a simple calculation with method 2 and improves precision of estimates at the same time. Acknowledgements We extend our gratitude to the anonymous referees and to John V. Merriner, Katherine Myers, and Sharyn Matriotti for their critical readings that greatly improved the man- uscript. We also thank John M. Hoenig, John B. Pearce, Yashushi Taga, and Geoff Gordon for their comments on earlier versions of the manuscript. A portion of this study was funded by the Fisheries Agency of Japan (Japan Sea- Farming Association ). Kilada and Tezuka: Survey designs for estimating recreational fishery catcfi 241 >. -1,000,000 it ^„„ Jul vs. Aug 500 ^ 300 0 500,000 100 _jl 500 300 100 0 200,000 600,000 1,000.000 Sep vs. Oct 0 500,000 1,500,000 0 200,000 600,000 1,000,000 Difference between two total estimates Figure 8 Bootstrap distributions of diffences between successive monthly ayu catch estimates. Literature cited Benzecri, J, P. 1992. Correspondence analysis handbook. Dekker. New York, r^, 665 p. Chambers J. M., and, T. J. Hastie (eds.) 1992. Statistical models in S. Chapman and Hall, New York, NY, 608 p. Clausen, S.-E. 1998. Applied correspondence analysis: an introduction. SAGE Publications, Thousand Oaks, CA, 69 p. Cochran, W. G. 1977. Sampling techniques, third ed. John Wiley and Sons, New York, NY, 41.3 p. Diggle, P. J, K-Y. Liang, and S. L. Zeger. 1994. Analysis of longitudinal data. Oxford Univ. Press, Oxford, 253 p. Gentle, J. E. 1998. Random number generation and monte carlo meth- ods. Springer, New York, NY, 247 p. Goodman, L. A. 1960. On the exact variance of products. J. Am. Stat. Assoc. .55:708-713. Guthrie, D.. J. M. Hoenig, M. Holliday, C. M. Jones, M. J. Mills, S. A. Moberiy, K. H. Pollock, and D. R. Talhelm (eds). 1991. Creel and angler sui-veys in fisheries management. Am. Fish. Soc. Symp. 12, Betliesda, MD. 528 p. Hayashi, C. 1950. On the prediction of phenomena from mathematical statistic point of view. Annals Inst. Stat. Math. 3:69-98. 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. Leeuw, J. de, and J. L. A. van Rijckevorsel. 1988. Beyond homogeneity analysis. In Component and corre- spondence analysis (J. L. A. van Rijckevorsel and J. de Leeuw, eds.), p. 56-57. John Wiley and Sons, New York, NY. Pollock, K. H., J. M. Hoenig, C. M. Jones, D. S. Robson, and C, J. Greene. 1997. Catch rate estimation for roving and access point sur- veys. N.Am. J. Fish. Manag. 17:11-19. 242 Fishery Bulletin 100(2) Pollock, K. H., C. M. Jones, and T. L. Brown. 1994. Angler survey methods and their applications in fish- eries management. Am. Fish. Soc. Special Publication 2.5, Bethesda, MD, 371 p. Sober, G. A. F. 1982. The estimation of animal abundance and related parameters, second ed. Griffin, London, 6.54 p. Sokal, R. R.,andF.J. Rohlf 1995. Biometry, third ed. Freeman and Company, New York, NY, 887 p. Thompson. S. K. 1992. Sampling. John Wiley and Sons, New York, NY', .343 p. Appendix 1 : The covariance of two estimators from sample means First we consider the covariance of two total estimates. Let X, and Yj be simple random samples (i=l n) from a population of size N with mean //^ and ;/^, and A' and Y be two sample means. Cochran (1977. p .2.5) derived the covariance of two- sample mean, that is Cov( X. y ) = ^^^!— ^ - Cov( X, y ) N n N-n N tAY^X -^ )(Y -u ). This is estimated by Cov(x, y ) = ^^-^ - Cov( X, y N n > (A, -X){Y, -Y Nnin- W, = C,tv, + w, (C,-C,) + C,iw,-IU,). From Taylor's series (mentioned above), the approximate covariance is obtained. Cov( w' " , w;.' ' ) = £( w; ■" - w; -^ ' ) ( w'. ' -w:?') -£[^,(Cl'^'-Cr') + C-'([Z7,. A^ Here Q''|' and iZij , are independent, and both Wk and u'a; are estimated from different samples. Therefore Cov (QVilij. ) = Cov(w,,Cj:^ = Covi w,,w,. ) = 0, then we get the covariance as only the first term. Appendix 3: Approximate variance of ^ Taylor's series of R with respect to C and M is obtained by" ^ = — -F— (C-C)--^(M-M). MM M- Then the approximate variance is obtained by V(i?) = £;(^-i?)-"- J-V(C) + -£-V'(M) M' M' -'iS-CoviCM). The covariance between two population total estimators is defined by Cov(X,y ) = CmtNX.NY) = EiNX - N^, HNY - N/u^. ) ., -^ NiN-n) — = Af-Cov(X,y)= cov(X.y). n This is estimated by Appendix 4: Covariance between C'lf' and C, By expanding C,.'' and C",''' we get CI'' ' = Dp,R, + DR, ( p, - p, ) + Dp, ( i?, - «, ). Q'^' = Dp,R, + DR,Ap,. -p,) + Dp.AR,. - R, (d) k c7viX,Y) = ^^^^^^^^±iX,-XnY,-Y). For the monthly total catches, we get c^(cr',c-' ) = ^'^";'' X ic,, - ^, )(c„. - 1 nin - 1) ■'— ' Appendix 2: Approximate covariance between Wl"and Wl" Taylor's series of W^ with respect to the random variables is obtained by then the approximate covariance is given by Cov(c;,'",c;'" ) = £(c;.'" - Dp,,/?,, kc;;" - Dp,yR,, = D- R,M,Covi p, , p, I + /?,p, Cov( p, .R,) +p,M,XoviR,.p,, ) + p,p, Cov(i?,.i?,. I If the first three covariance components are ecjual to 0 because of independent sampling, then we have Cov(c;;",c;;" i = £»-p,,p,,Cov( /?,.,/?, ). Here we can write R/. and Rf. as the ratio of two random variables from Equation 1 by Kitada and Tezuka Siirvey designs for estimating recreational fishery catch 243 R,= C(s) By using a method similar to that given in Appendix 3, we get Cov(R,.,R.,)- N' 1 M,M,. -cov(c;-'",c;f') M,A/^. -CoviCl",M,.) J^'^ Cov(M„C;r') Cov{M,,,Mi,) Cov(Q'",Af^,.) and Cov(M,,,Cj.?'). These are the covariances between a total estimate and a sample mean. By a method similar to that in Appendix 1, we have Cov(i-,y ) = EiNX - NiJ^)(Y - fi^.) = NCov(X,y ) = Cov(X.y). The covariance is estimated by Cov(i'T)= ^ " y{X~X)(Y-Y). n(n - 1) f^ For our case, the two covariances are as follows; n (« - 1) ■"' Here Cov(C'i'",C'i'')and CovlM^M,. I are already given by Equation 18 and Equation 24. Hence we can estimate Cov(M, , CI? ' ) = !y " X ' C,*- - Q- ) (^,* - M, nin- ll •^ 244 Abstract— Juvenile chinook salmon, Oncorbyncluis tshawytscha. from natal streams in California's Central Valley demonstrated little estuarine depen- dency but grew rapidly once in coastal waters. We collected juvenile chinook salmon at locations spanning the San Francisco Estuary from the western side of the freshwater delta — at the con- fluence of the Sacramento and San Joa- quin Rivers — to the estuary exit at the Golden Gate and in the coastal waters of the Gulf of the Farallones. Juveniles spent about 40 d migrating through the estuary at an estimated rate of 1.6 km/d or faster during their migration season (May and June 1997) toward the ocean. Mean growth in length (0.18 mm/d) and weight (0.02 g/d> was insignificant in young chinook salmon while in the estuary, but estimated daily growth of 0.6 mm/d and 0.5 g/d in the ocean was rapid (PsO.OOll. Condition (A' factor) declined in the estuary, but improved markedly in ocean fish. Total body pro- tein, total lipid, triacylglycerols (TAG), polar lipids, cholesterol, and nonesteri- fied fatty acids concentrations did not change in juveniles in the estuary, but total lipid and TAG were depleted in ocean juveniles. As young chinook migrated from freshwater to the ocean, their prey changed progressively in importance from invertebrates to fish larvae. Once in coastal waters, juve- nile salmon appear to employ a strat- egy of rapid growth at the expense of energy reserves to increase survival potential. In 1997, environmental con- ditions did not impede development: freshwater discharge was above aver- age and water temperatures were only slightly elevated, within the species' tolerance. Data suggest that chinook salmon from California's Central Valley have evolved a strong ecological pro- pensity for a ocean-type life history. But unlike populations in the Pacific Northwest, they show little estuarine dependency and proceed to the ocean to benefit from the upwelling-driven, bio- logically productive coastal waters. Physiological ecology of juvenile chinook salmon iOncorhynchus tshawytscha) at the southern end of their distribution, the San Francisco Estuary and Gulf of the Farallones, California* R. Bruce MacFarlane Elizabeth C. Norton Santa Cruz Laboratory Southwest Fisheries Science Center National Marine Fisheries Ser^/lce, NOAA 110 Shaffer Road Santa Cruz, California 95060 E-niail address (for R B MacFarlane) Bruce MacFarlane a noaa gov Manuscript accepted 23 August 2001. Fish. Bull. 100:244-2.57 (2002). Estuaries are considered important in the development of juvenile salmon. In the Pacific Northwest, estuaries have been shown to provide nursery and rearing conditions for juveniles emigrat- ing from streams of birth to the ocean (Reimers, 1973; Healey, 1982; Levy and Northcote, 1982; Myers and Horton. 1982; Simenstad et al, 1982; McCabe et al, 1986). The San Francisco Estuary is the largest estuary on the West Coast and is a segment in the migration corri- dor for chinook salmon (Onc-orhyru'luis tshawytscha ) from natal streams in the watersheds of the Sacramento and San Joaquin Rivers, known as California's Central "Valley The Central Valley is unique by having four runs of chinook salmon which constitute a significant socioeconomic resource. Ocean harvest south of Pt. Arena (estimated as 85-95% from Central Valley stocks) and spawn- ing escapement range between 0.5 and 1.3 X 10'' chinook salmon per year ( 1970-98) and represent about $60 mil- lion (U.S.) in personal income annually (PFMCM. Beyond the direct value of Central Valley chinook salmon, their demography and welfare significantly affect the financial and societal aspects of water rights decisions. Chinook salmon populations migrat- ing through the San Francisco Estuary are at the southern limit of the species' geographical range and are subject to the impacts of a highly urbanized, in- dustrialized, and agricultural freshwa- ter and estuarine system (Nichols et al., 1986).A11 chinook salmon runs originat- ing in the Central Valley are in jeopar- dy. Before 1900, spawning runs were es- timated at 2 X 10« adults (Fisher, 1994), but in 1998 only an estimated 0.25 X 10'' returned of which about 30% were of hatchery origin (PFMCM. The Sacramento River winter-run chinook was the first Pacific salmonid species listed under the U.S. Endangered Spe- cies Act of 1973 (ESA). Originally cat- egorized as threatened in 1989, its sta- tus was changed to endangered in 1994. Chinook salmon of the Central Valley spring run, once forming the dominant chinook race in California (Clark, 1929). were listed as threatened in 1999. Even the fall run, by far the dominant run to- day ( 92% of all Central Valley spawn- ers, 1990-98 IPFMC'l), has uncertain status and is an ESA candidate. Hatch- ery production supports the natural fall run, and the other runs to a much lesser degree. Annually, about 35 million chi- nook salmon are produced by state and federal hatcheries in the Central Val- ley; the fall run comprises 95% (Mills et ai, 1997). California and federal water develop- ment projects, such as dams and water diversions, have clearly played a role in the decline of Central Valley salmon (Movie, 1994). but other factors may al- * Contribution 114 of Santa Cruz Labora- tory National Marine Fisheries Service, NOAA. Santa Cruz, CA 95060. ' PFMC (Pacific Fishery Management Coun- cil). 1999. Review of 1998 ocean salmon fisheries. Pacific Fisheries Management Council, Portland, OR, 65 p. PFMC, 2130 SW Fifth Ave., Suite 224, Portland, OR 97201. MdcFdrlane and Norton: Physiological ecology of Oncorhynchus tshawytscho 245 38'13'N Delta Ns% 12r45'W Figure 1 The northern portion of the San Francisco Estuary and nearshore Gulf of the Farallones. Juvenile salmon col- lection locations are denoted as km (e.g. km 26) from the estuary exit at the Golden Gate. Juvenile salmon were collected throughout the nearshore area in the Gulf of the Farallones, not at specific locations. so be involved. Migration through the highly impacted San Francisco Estuary and early residence in a marine en- vironment at the southern margin of the species' distribu- tion may impair physiological development that could lead to direct mortality or, indirectly, to reduced survival poten- tial during the oceanic phase. Although some data exist on abundance, growth, and feeding for chinook salmon migrating through estuaries in southeastern Alaska and British Columbia ( Healey, 1980b, 1982; Levy and Northcote, 1982; Landingham et al., 1998); Washington and Oregon (Reimers, 1973; Myers and Hor- ton, 1982; Simenstad, et al. 1982; McCabe et al.,1986; Fisher and Pearcy, 1996); and the Klamath River estuary in northern California (Wallace and Collins, 1997), the on- ly data available for Central Valley emigrants are those of Kjelson et al. (1982). In that paper, life history descrip- tions were presented for fall-run juveniles, but the empha- sis was on fry (<70 mm fork length) in the freshwater delta at the head of the estuary. Almost nothing is known of ju- venile chinook biology in the larger saline portions of the San Francisco Estuary. Knowledge of juvenile chinook salmon biology during their first year in the marine environment is even more limited, and nonexistent for the area south of the Cali- fornia-Oregon border. Healey (1980a) presented distribu- tion, growth, and feeding information on first ocean-year chinook salmon in the Strait of Georgia, British Colum- bia. Similar data have been presented for juvenile chi- nook from the Columbia River drainage off the Oregon and Washington coasts (Miller et al., 1983; Brodeur and Pearcy, 1990; Fisher and Pearcy, 1995). The purpose of this study was to describe juvenile chi- nook salmon physiological development during their em- igration through the San Francisco Estuary and early residence in the coastal waters of central California. Resi- dence time, age, growth, condition, lipid classes and pro- tein concentrations, and feeding data are presented to characterize the significance of habitat utilization at the southern limit of the species' distribution. The information presented here can serve as a basis for comparison with other year classes from the Central Valley and with popu- lations from more northerly estuaries, as well as for as- sessments of the influences of natural and anthropogenic perturbations on salmon habitat. Methods Study area Juvenile salmon leaving California's Central Valley pass through the San Francisco Estuary, a series of embayments, before entering the ocean in the Gulf of the Farallones (Fig. 1 ). The delta, a freshwater network of channels and leveed 246 Fishery Bulletin 100(2) islands at the confluence of the Sacramento and San Joa- quin Rivers, forms the eastern boundary. Measurable salin- ity (>1 psu) occurs on the western side of the delta in Suisun Bay. Water flows through Suisun Bay into San Pablo Bay, then into the Central Bay of San Francisco Bay before exit- ing the estuary at the Golden Gate. The estuaiy has a sur- face area of about 1100 km- and is fringed on the northern shore with marshes, which have shrunk by more than 90%, to 125 km-, since 1850 (Conomos, 1979). Most of the estuary shoreline is urban, suburban, and industrial development, however. The Gulf of the Farallones, a relatively broad expanse of the continental shelf extending from Pt. Reyes (38°00'N, 123°01'W) to Pillar Pt. (37°30'N, 122°30'W) to the Farallon Islands (Southeast Farallon Island— 37°42'N, 123°00'W) on the edge of the continental shelf ranges to 90 m but is mostly 20 to 50 m deep. This hydrodynami- cally complex area is influenced by the cool southerly flow- ing California Current: freshwater discharge from the San Francisco Estuary; and seasonally strong coastal upwelling (spring and summer) and a northerly flowing countercur- rent, the Davidson Cuirent (winter). Field sampling Juvenile chinook salmon in this study were collected from the fall run, as determined by the daily length criteria used to discriminate juveniles among the four runs (John- son et al., 1992). Because salmon from the four runs are phenotypically indistinguishable, to target fall-run juve- niles we used daily length criteria and collected fish during the period when fall-run chinook salmon dominate the migration toward the ocean. We collected juvenile chinook salmon at four locations spanning the San Francisco Estuary (Fig. 1): at km 68, the west side of the delta at the confluence of the Sacramento and San Joaquin Rivers; km 46, the exit from Suisun Bay; km 26 in San Pablo Bay; and several sites within a 1-km radius of km 3 for greater coverage of the estuary exit at the Golden Gate and to avoid ship traffic. Two multiday surveys of the estuary were completed starting at km 68 and proceeding through successive downstream locations to the Golden Gate. The first sampling date was 30 April 1997 at km 68; the last was 15 July 1997 at km 3. Collections were made in the estuary with a midwater trawl towed at 2-3 knots for 15-30 min. The trawl was made of nylon mesh with a 10-m headrope and footrope. 10-m height at the mouth, and 20-m length. Mesh size was 1.6 cm at the headrope and decreased to 0.4 cm before the codend. The codend was fitted with a 1.27-cm knotless mesh liner. The net was kept open by aluminum spread- ers and depressors. For most tows, the spreaders were vis- ible at the surface, confirming that the net fished the up- per layer of the water column. Juvenile chinook salmon were also obtained from the coastal ocean in the Gulf of the Farallones. Stations where salmon were caught were within 15 km of the Golden Gate in waters 18 to 36 m deep. A high-speed midwater rope trawl was towed at 3-4 knots for 15-30 min at each site. The net had a 53-m headrope and footrope with a 1.27-cm mesh codend liner (Dotson and Griffith, 1996). Tempera- ture-depth recorders attached to the footrope and head- rope indicated that the net fished 5-10 m below the sur- face with a vertical opening of 13 m. Fish were removed from the net as soon as practicable and placed in labeled plastic zip-top bags. Juveniles cap- tured within the estuary were kept under ice: those from the ocean were stored at -80°C, until returned to the laboratory We collected hydrologic data during all field trips. With- in the estuary, a Hydrolab H20' Multiprobe connected by cable to a Surveyor 3- data logger (Hydrolab Corp., Aus- tin, TX) was used to record vertical profiles of tempera- ture, salinity, dissolved oxygen, pH, and turbidity. In the Gulf of the Farallones, a "SeaCat SBE 19-03" (Sea-Bird Electronics, Inc., Bellevue, WA) conductivity-temperature- depth sonde (CTD) recorded profiles of temperature and salinity in a grid of stations spaced at 2' latitude and lon- gitude intervals encompassing the salmon fishing sites. Laboratory analyses In the laboratory, we examined, measured, and dissected fish, usually within 24 h of capture. Fork length and total body weight were recorded. The peritoneal cavity was opened by incision along the ventral side from vent to opercular isthmus. The stomach and its contents were removed and stored in 10% buffered formalin for subse- quent analysis of prey Sagittal otoliths were removed, cleaned of membranes, rinsed in deionized water, and stored for subsequent aging. The remaining portion of the fish was placed in a Whirl" bag. purged with N,^, and stored at -80°C for lipid and protein analyses. Heads of juvenile salmon containing coded-wire tags, evident by a missing adipose fin, were removed and sent to the U.S. Fish and Wildlife Service, Stockton, CA, to obtain data on the location and time of release of each fish. Concentrations of lipid classes and total protein were determined in 15 fish randomly chosen from each location. Heads, fins, and stomachs were removed to limit analyses to body constituents. Frozen bodies were minced with a knife, then homogenized in a blender for about 30 sec to a uniform paste. We extracted lipids from a Ig- to 3g-ali- quot by using the method of Bligh and Dyer (1959). Total lipid was quantified by thin-layer chromatography with flame-ionization detection with an latroscan TH-10 Mark V" (latron Laboratories. Inc., Tokyo, Japan) according to procedures published previously (MacFarlane et al., 1990, 1993). Total lipid was separated into steryl or wax ester, triacylglycerols, nonesterified fatty acids, cholesterol, and polar lipid classes and quantified according to methods in MacFarlane and Norton ( 1999). We estimated total protein concentration by the Lowry method, using bovine serum albumin as a standard (Lowry et al., 1951). All lipid and protein values are expressed as wet weight (mg/g). Fish ages were estimated from otolith analysis (Broth- ers, 1987 1. Sagittae were embedded in epoxy, then ground and polished on the distal surface to a mid-sagittal sec- tion. Otolith concentric bands were counted under oil im- mersion with transmitted polarized light illumination in- to a video microscopy system at a monitor magnification MacFarlane and Norton Physiological ecology of Oncorhynchus tshawytscha 247 of 1500x. Counts representing daily increments of gi-owth were made from the otolith margin (dorsal edge) to a posterior primordium. Each otolith was counted at least two times by the same reader Data are presented as age, in days, between the hatching check and the otolith margin. Stomach contents were identified and quantified ac- cording to methods of Hobson and Chess ( 1976). Prey were identified to the lowest possible taxa, enumer- ated, and their relative volume and size distribution recorded. The derived parameter, index of relative im- portance (IRI), was used to determine the importance of specific taxa to juvenile salmon at each sampling location. The IRI was computed by the equation IRI = (N + V)FO. where A^, V, and FO the percent number, volume, and frequency of occurrence, respectively, of a taxon in the stomach contents (Pinkas et al.,1971). Size, age, condition, lipids, and protein data were ana- lyzed for variability among locations by means of the general linear model of analysis of variance. Differ- ences among specific locations were determined with Tukey's studentized range test, set at a = 0.05, which controls for MEEK (maximum experimentwise error rate) under complete or partial null hypotheses. All statistical procedures were performed with SAS soft- ware (SAS Institute, Inc., 1994). Results We collected and evaluated 310 subyearling chinook salmon from the 1996-97 year class. The catch con- sisted of naturally spawned and hatchery-produced fish of unknown proportions because less than 37r were marked (by adipose fin clips and coded wire tags) and there were no morphological features to distinguish hatchery from naturally produced salmon. Mean fork length (FL) was 89 mm and ranged from 68 to 113 mm. Total body weights ranged from 3.59 to 14.62 g, with a mean of 7.58 g. A power function was fitted to the relationship between FL and weight (Fig. 2A). Ages were determined for 156 juve- niles and ranged from 112 to 209 days after hatching. The relationship of FL or weight to age was not as well fitted as that for weight on FL (Fig. 2, B and C); however, length and weight were positively correlated to age (P<0.0001 ). Each sampling location within the estuary was visited more than once on successive surveys through the 30 April-15 July period of juvenile emigration. There were no statistically significant trends in size, age, lipid, and protein variables by sampling date at any location: there- fore we combined data for each location from both surveys through the estuary While in the estuary, juvenile chinook salmon grew lit- tle in length or weight; but in coastal waters, they grew 16 ^ 12 I 8 5 4 0 120 ? i- 100 X 80 o ^ 60 A y = 0.0003 x2 2'7 r2 = 0.8523 .•••i!» l!.|:|:l'5' IS : • 0 60 70 80 90 100 Fork length (mm) B y = 60.72 + 0.162X r2 = 0.2390 110 120 12 0 100 120 140 160 Age (d) c y = 2.368 + 0 029x r2 =0.1654 . .• ". 180 200 220 5 4 •tjt:'.''. 100 120 140 160 Age (d) 180 200 220 Figure 2 Fork lengths, weights, and ages of all juvenile chinook salmon collected from the San Francisco Estuary and Gulf of the Faral- lones: (A) fork length-weight relationship; (Bl age-fork length relationship; (Cl age-weight relationship. rapidly (Fig. 3). Juveniles entering the estuary at km 68 had a mean (±SE) FL of 82.8 ±0.7 mm and weight of 6.36 ±0.21 g. At the exit of the estuary (km 3), mean FL and weight of cohorts were 89.5 ±1.1 mm and 7.23 ±0.30 g, representing mean gains of about 7 mm and 0.9 g. Size changes within the estuary were not statistically signif- icant (P>0.05). Year-class cohorts in the coastal waters of the Gulf of the Farallones were significantly longer and heavier than those from the estuary (FL, P<0.0001; weight, P<0.001). Ocean-caught juveniles were 8.2 mm longer and weighed 6.5 g more than fish collected near the estuary exit. Juvenile salmon spent about 40 d migrating along the 65-km length of the estuary, according to otolith increment counts (Fig. 3C). Mean age at the entry to the estuary was 136 ±2 d; at the exit it was 176 ±3 d. Juveniles aged from the Gulf of the Farallones were 156 ±5 days old, indicating a shorter freshwater or estuarine residence than that for those captured from the Golden Gate area. Daily growth rates in the estuary can be calculated from differences in mean size over the 40-d estimated time of 248 Fishery Bulletin 100(2) transit. According to this technique, mean daily growth was about 0.18 mrn/d and 0.02 g/d. Because we aged only a subset of fish from each sam- pling location, mean age of all fish obtained at each site could be estimated with the FL-on-age regression equa- tion for age (Fig. 2B). Results of that computation revealed close agreement with measured ages at all locations ex- cept the Gulf of the Farallones (Fig. 3C), where calculated age was 189 ±8 d. Both methods of age determination in- dicated that juveniles caught in coastal waters were from the same year class. The average migration rate through the estuary was es- timated at 1.6 km/d on the basis of the difference between mean ages of juvenile salmon sampled at the estuary en- try and exit (i.e. 65 km/[176 d at km 3 -136 d at km 68 ]). Data from coded-wire-tagged fish caught within the estu- ary and the Gulf of the Farallones revealed a wide range of migration rates (Table 1). For those captured within the estuary (17 of 24), the mean migration rate was 4.0 ±0.9 km/d. Most tagged-recaptured salmon had rates <2.6 km/d 105 100 95 90 85 80 0- 20 B 47 15 y 10 145 28 26 i 64 y /\ 5 • 70 60 50 40 30 20 km 10 GF Figure 3 Mean l±SE) fork lengths i A), weights (B), and ages (Oof juve- nile chinook salmon collected from locations in the San Fran- cisco Estuary (km 68, km 46. km 26, km .3) and the Gulf of the Farallones (GF). Open circles and dashed line in (C) represent calculated ages of all fish at each location from regi'ession of fork length on age from Figure 2(B). Numbers near means are sample sizes. but were from two releases within the estuary, thus rep- resenting migration rates in the lower estuary only. Juve- niles released farther upstream in the rivers and caught in the lower estuary had somewhat faster migration rates (.r=9.1 ±2.5 km/d, 72=4). Not only did juvenile chinook salmon gi'ow slowly while in the estuary, their condition declined as they proceeded to the Golden Gate (Fig. 4). There was a significant de- crease in Fulton's condition factor (K) between fish enter- ing the estuary and those departing (P<0.001). Once the fish were in coastal waters, however, their condition im- proved markedly Body constituents and energy reserves varied little while migrating fish were in the estuary, but total lipid was depleted in fish from the Gulf of the Farallones (Fig. 5A). Total body protein concentrations were approximate- ly 150 mg/g, wet weight, in fish from all locations, and did not vary significantly. Total lipid also did not vary in fish within the estuary but was significantly lower in salmon caught in the ocean (P<0.0005). Mean lipid concentration for fish in the estuary was about 30 mg/g, or IS'^f of dry weight, and decreased to 17.7 ±1.6 mg/g in the ocean. The decline in lipids in fish from the Gulf of the Faral- lones was attributed to decreased concentrations of tri- acylglycerols (TAG), the dominant lipid class (Fig. 5B). TAG levels increased from 14.8 ±2.5 mg/g in fish en- tering the estuary to about 18 mg/g in estuarine salm- on and were depleted to 4.3 ±1.4 mg/g in coastal fish. The concentration of polar lipids, composed primarily of phospholipids, remained unchanged through the estu- ary and in the ocean. Other lipid classes — cholesterol and nonesterified fatty acids — were found at much low- er concentrations and did not vary significantly during the emigration. Steryl or wax esters were absent or at very low levels in most individuals and showed no dif- ferences related to location. We examined feeding and prey selectivity in juvenile chinook salmon migrating through the estuary and in coastal waters. A lesser proportion of fish leaving the rivers contained food items compared with those with- in the estuary. Fifty percent of juvenile salmon had stomach contents at Chipps Island (km 68, 21 of 42 sampled). In contrast, more than 80'^f from Carquinez Strait (km 46, 8 of 10 samples) and San Pablo Bay (km 26, 20 of 23 samples) had prey in their stomachs. A low- er percentage KlWc) from the Central Bay (km 3, 9 of 13) had fed recently, but in the Gulf of the Farallones, 82'^^f (23 of 28) of the juveniles contained food. The relative importance of prey items varied as juve- nile salmon migrated through the estuary and entered the ocean (Table 2). Stomachs of juveniles leaving the rivers contained primarily gammaridean amphipods Corophnim sp., and lesser amounts of crab megalopae, dipteran insects, other malacostracan crustaceans, and other insects. After the salmon passed through Suisun Bay (km 46), their stomachs contained mostly the he- mipteran Hesperocorixa sp., the calanoid copepod Eu- calanus califoniiciis, the mysid Acanthoryiysis sp., fish larvae, and other insects. Cumaceans were clearly dom- inant in juvenile salmon in San Pablo Bay (km 26), but MacFarlane and Norton: Physiological ecology of Oncorhynchus tshawytscha 249 Table 1 Minimum distance traveled by coded- wire tagged juvenik chmook salmon betw eon m arking and recapture. Release Capture Km Days Km/d Date Location km Date Location km' traveled free traveled 1 Apr 1997 Battle Creek 470 12 May 1997 Gulf of Faralloncs -19 489 41 11.93 15 Apr 1997 West Sacramento 157 9 May 1997 Gulf of Farallones -19 176 24 7.33 Hi Apr 1997 Battle Creek 470 11 May 1997 Gulf of Farallones -17 487 25 19.48 24 Apr 1997 Gridley Boat Ramp 244 29 May 1997 Pt. San Pablo 17 227 35 6.49 24 Apr 1997 San Pablo Bay 37 10 May 1997 Gulf of Farallones -22 59 16 3.69 24 Apr 1997 Gridley Boat Ramp 244 11 May 1997 Gulf of Farallones -20 264 17 15.53 28 Apr 1997 Mossdale 150 12 May 1997 Gulf of Farallones -20 170 14 12.14 29 Apr 1997 Dos Reis Park 146 27 May 1997 Carquinez Strait 46 100 28 3.57 29 Apr 1997 Dos Reis Park 146 11 May 1997 Gulf of Farallones -20 166 12 13.83 14 May 1997 Hatfield State Park 221 29 May 1997 Pt. Pinole 26 195 15 13.00 19 May 1997 Benicia 43 29 May 1997 Pt. Pinole 26 17 10 1.70 19 May 1997 Benicia 43 29 May 1997 Pt. Pinole 26 17 10 1.70 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2.60 19 May 1997 Benicia 43 29 May 1997 Pt. San Pablo 17 26 10 2,60 30 May 1997 Woodbridge Dam 149 Jun 10 1997 Golden Gate 0 149 11 13.55 'Negative km data represent distance captured seaward of the Golden Gate- insects were still important. In the Central Bay, the final embayment before the estuary exit, fish larvae were the dominant forage item. Ampelisca abdita, a gammaridean amphipod, and cumaceans were important as well. Fish larvae continued to be the most important prey of juvenile chinook salmon in the coastal waters of the Gulf of the Farallones, but euphausiids and decapod early life stages were also consumed in significant numbers. Discussion Chinook salmon from California's Central Valley streams have a largely ocean-type life history although perhaps a few spring and late-fall run juveniles overwinter in fresh- water and migrate to the ocean as yearlings (Fisher, 1994; Myers et al., 1998). Fry (<70 mm FL) are abundant in the freshwater delta from February to April. They leave as smolts (>70 mm FL) and enter the San Francisco Estuary, primarily in May and June. Kjelson et al. (19821 reported that few juveniles were present in the delta or estuary Figure 4 Changes in mean (±SE) Fulton's condition factor (K") of juve- nile chinook salmon from locations in the San Francisco Estuary (km 68, km 46. km 26, km 3) and the Gulf of the Farallones (GFl. Numbers above means are sample sizes. 250 Fishery Bulletin 100(2) after June. Our findings concur: the smallest salmon we caught was 68 mm FL, and we caught no fish after 27 June. Juveniles were approximately 4.5 months old (after hatching) when they entered the San Francisco Estuary. They spent about 40 days there, migrating at 1.6 km/d, based on mean age differences of fish enter- ing the estuary and fish leaving at the Golden Gate. This is probably a minimum estimate of migration rate, because data from tagged fish caught within the estu- ary showed rates of 1.70 to 13.55 km/d, representing residence times of 38 to 5 days. A previous mark-recap- ture study also found higher migration rates ( 10 to 18 km/d), although the data were for passage through the upstream delta (Kjelson et al.. 1982). Studies in the Pacific Northwest suggest that juve- niles of ocean-type chinook salmon make extensive use of estuaries, spending as much as 6 to 9 months in them feeding and gi'owing (Myers and Horton, 1982; Si- menstad et al., 1982: Healey, 1991). In the Columbia River estuary, subyearling chinook salmon were present throughout the year, although most abundant in May to September (McCabe et al., 1986). Fall-run juveniles entered the Sixes River estuary in the spring and were abundant from June through August (Reimers, 1973). Juvenile chinook salmon resided in Washington estuar- ies for 6 to more than 29 weeks, but some individuals were present for up to 189 d (Simenstad et al., 1982). Even in northern California's Klamath River estuary, juvenile fall-run chinook salmon remained from June to September (Wallace and Collins, 1997). The relatively short period of abundance in the San Francisco Estuary and emigration rates presented here suggest that juve- nile chinook from the Central Valley may derive less benefit from estuarine residence than do more northerly populations. Juvenile chinook salmon grew little while in the San Francisco Estuary. Though growth was not statistically significant, on average they increased in size by 7 mm FL and 0.9 g, representing daily growth of 0.18 mm/d and 0.02 g/d. These estimates indicate slower growth than has been reported for juvenile chinook in most estuaries to the north. In five estuaries on Vancouver Island and the Eraser River, juveniles grew 0.21 to 0.62 mm/d (Levy and Northcote, 1982; Healey, 1991). Growth measurements from population sampling and from following marked fish produced similar results in the Sixes River estuary, where daily growth ranged from 0.07 mm/d in summer to 0.9 mm/d in spring (Reimers, 1973). Reimers speculated that the very slow gi-owth during summer was due to food limi- tation caused by high salmon abundance. Growth can also be inferred by the change in size offish collected at the same locations from the beginning to the end of the emigi'ation period. We found no significant in- crease in size at given locations within the estuary durhig the May^une emigration season. Kjelson et al. ( 1982) al- so found no change in smolt size in the delta from April to June. In contrast, the mean size of juvenile chinook salm- on in the Columbia River estuary increased from March to December, suggesting substantial growth or immigration of larger fish during the season. 25 20 B £ ,5h 10 < TAG ^ II \ Choi \ 6 z m o 70 60 50 40 30 20 km 10 GF Figure 5 (A) Protein and total lipid concentrations, and (B) lipid class concentrations (triacylglycerols [TAG], polar lipids [PL], cho- lesterol IChol], and nonesterified fatty acids |NEFA|) in juve- nile chinook salmon from locations in the San Francisco Estuary (km 68, km 46, km 26, km 3) and the Gulf of the Far- allones (GFi. Concentrations are mg/g wet weight. Although little gi'owth occurred during estuarine resi- dence, growth was rapid in the coastal waters of the Gulf of the Farallones. Juvenile ocean cohorts were 10% longer in mean FL and 90% heavier than those from the estu- ary, although size varied more in ocean residents than in chinook salmon within the estuary. Faster growth rates of juvenile chinook salmon in coastal waters than in estu- aries have been reported from British Columbia (Healey, 1980a), and Oregon and Washington (Miller et al., 1983; Fisher and Pearcy, 1995), with rates of 1 mm/d and more. Daily growth rates of juveniles in the Gulf of the Faral- lones were difficult to determine because the mean age of the subset aged by otolith analysis was less than the mean age offish at the estuary exit. But if the calculated age of ocean juveniles (from the regi'ession of age on FL) is used, during the calculated 13 d of ocean residence they gi-ew at about 0.6 mm/d and 0.5 g/d, well above rates while in the San Francisco Estuary. Growth rate estimates from the regressions of FL or weight on age for all juveniles from both the estuary and gulf were 0.16 mm/d and 0.029 g/d. These results agree MacFarlane and Norton: Physiological ecology of Oncorhynchus tshawytscha 251 Table 2 Stomach contents of juvenile chinook salmon frequency of occurrence percentage; and IRl is . '7rN is the numerical percentage; %V is the percent relative volume; %F0 is the the index of relative importance, {%N + %V)%FO. Location and prey species ''i-N 7rV TfFO IRI Location and prey species %N %V %FO IRI km 68, Chipps Island Ui=2l) km 26, San Pablo Bay («=20) continued Malacostraca Malacostraca Decapoda Decapoda Caridean shrimp 2.6 5.7 4.8 39.3 Crab megalopae 0.5 0.3 35.0 28.0 Crab megalopae 7.7 10.0 19.0 336.3 Mysidacea Mysidacea Unidentified 0.5 0.5 5.0 5.0 Unidentified 7.7 4.6 14.3 175.9 Cumacean Amphipoda Unidentified 61.7 17.9 50.0 3980 Gammaridea 2.6 5.7 4.8 39.8 Amphipoda Corophium spp. 30.8 26.4 33.3 1905 Gammaridea Eusirdae unidentified 2.6 5.7 4.8 39.8 Ampelisca abdita 1.6 1.2 10.0 28.0 Isopoda Corophium spp. 3.6 23.2 5.0 134.0 Gnnrimosphaero luta 2.6 0.6 4.8 15.4 Corophium Insecta spi/Ticorne 0.5 4.2 5.0 23.5 Hymenoptera Maera spp. 0.5 1.5 5.0 10.0 Unidentified 2.6 5.7 4.8 39.8 Unidentified 1.0 1.3 3.6 8.3 Homoptera Cirri pedia Aphid 2.6 0.3 4.8 13.9 Thoracic Diptera Barnacle cirri 1.6 1.3 10.0 29.0 Flies unidentified 2.6 1.1 4.8 17.8 Insecta Culicidae 2.6 2.6 4.8 25.0 Coleoptera Unidentified 7.7 6.3 19.0 266.0 Unidentified 1.0 1.1 5.0 10.5 Unidentified 10.3 9.7 14.3 286.0 Hemiptera Algae Unidentified 0.5 0.5 5.0 5.0 Unidentified 5.2 11.4 9.5 157.7 Homoptera Unidentified 7.7 4.0 4.8 56.2 Flatidae 0.5 0.3 5.0 4.0 km 46. Carquinez Strait (?!=8t Diptera Malacostraca Unidentified 1.6 1.0 15.0 39.0 Mysidacea Lepidoptera 1.3 1.0 5.0 11.5 Acanthomysis spp. 3.9 17.0 12.5 257.5 Orthoptera 0.5 0.3 5.0 4.0 Unidentified 2.0 6.0 12.5 100. Unidentified 10.4 12.7 25.0 577.5 Amphipoda Polychaeta Gammaridea 3.9 3.0 12.5 86.3 Phyllodocida Cumacea 5.9 5.0 12.5 136.3 Nereidae 1.0 4.7 10.0 57.0 Copepoda Unidentified 1.6 8.2 15.0 147.0 Calanoida Pisces Eucalanus califomicus 5.9 15.0 12.5 261.3 Unidentified larvae 0.5 2.4 5.0 14.5 Insecta Unidentified 0.5 1.6 5.0 10.5 Coleoptera Algae Unidentified 2.0 14.0 12.5 200.0 Unidentified 8.8 8.7 30.0 10.5 Hemiptera Unidentified 0.5 0.3 5.0 4.0 Hesperocorixa spp. 72.5 1.0 12.5 918.8 km 3. Central Bay (^ = 10) Unidentified 2.0 20.0 12.5 275.0 Crustacean Pisces LInidentified — 1.2 10.0 — Unidentified 2.9 19.0 12.5 273.8 Malacostraca km 26, San Pablo Bay in =20 ) Decapoda Crustacean Crab megalopae 1.2 0.1 10.0 13.0 Unidentified — 2.9 5.0 — continued 252 Fishery Bulletin 100(2) Table 2 (continud) Location and prey species %N ^V %F0 IRl Location and prey species %N %V %FO IRI km 3, Central Bay 220 cm, were investigated by using chi-square (X" > goodness-of-fit and analysis-of-variance (AN OVA) meth- ods. A chi-square test for equal proportions was used to de- termine if there were differences in the number of strand- ings between years. An ANOVA was used to determine if there were differences in the number of strandings be- tween months. To determine if there were seasonal trends in the strandings, the data were stratified into gi'oups of three months representing four seasons: January-March (winter). April-June (spring). July-September (summer), and October-December (fall) (Fig. 2). Expected number of strandings for each season was determined by aver- aging over the 5-year period. To determine whether the stranding pattern for any given year deviated significantly from the "norm," we compared each years seasonal num- ber of strandings with the expected seasonal number by using a chi-square goodness-of-fit test. An ANOVA was performed to determine if there was a difference in the number of strandings between seasons. A chi-square good- ness-of-fit test was used to determine seasonal trends be- tween zones. A chi-square goodness-of-fit test was used to determine if there was any difference in the proportion of male and female bottlenose dolphin strandings. A chi-square test for trend was used to test the hypothesis that there would be a downward trend in the number of animals that were of unknown sex due to increased training of stranding net- work volunteers in determining the sex of bottlenose dol- phins. A chi-square goodness-of-fit test was used to deter- mine if there was a difference in the number of strandings of females >220 cm between seasons. 25 20 22 ro 15 E 10 E 19 ~\ 17 17 14 11 2 D U U U U u _ 1 3 16 2 3 12 Jan Feb Mar Winter Apr May Jun Spring Months Jul Aug Sep Oct Nov Dec Summer Fall Figure 2 Number of bottlenose dolphin strandings in South Carolina for each month and season of the year from 1992 to 1996 (n = 153). The numbers over each bar represent the number of strandings reported for that month. Results Yearly trends From 1992 to 1996, 153 bottlenose dolphin strandings were reported along the coast of South Carolina. The number of strandings each year ranged from a low of 28 in 1992 to 33 in 1993 (.v=30.6) (Table 1); there was no significant differ- ence among years (X" test for equal proportions, P=0.968). Prior to 1992 the highest number of bottlenose dolphin reported stranded for one year was 17 in 1991 (the year the network was formed), excluding the unusual mortality eventof 1987 1/! =60). Monthly trends Over the five-year period, the greatest number of reports ( 22, or 1AA'''< ) of bottlenose dolphin strandings occuiTed during July and the least in January (/;=2) and October (/!=3) (Fig. 2). Tliere was no yearly differences in the total number of strandings by month fi-om 1992 to 1996 (ANOVA, P=0.172). Seasonal trends The highest number of strandings occurred in spring («=53, ZA.&^ I and the lowest number of strandings were recorded in winter (/;=26, 17.0'^^f ). Strandings during the years 1992, 1994. 1995. and 1996 did not deviate from the expected pattern (x^ goodness-of-fit; P=0.994. 0452, 0.379, and 0.062, respectively), but the seasonal pattern in 1993 was significantly different (P=0.016). This was due in large part to the high number of strandings (/( = 14) in the fall, when we expected the number to be less than seven. The ANOVA analysis indicated that the mean number of strandings differs significantly between seasons (P=0.021 ). McFee and Hopkins Murphy: Stiandings of Tuisiops liuncatti'i off Soutli Carolina 261 Table 1 Summarv of results of human interaction from evaluat ions of hottlenost dolphins stranded in South Carol ina from 1992 to 1996. "t^ialT wounds" refer to puncture wounds made by the g; ff,a Ions ''igid pol e with sh; rp point! s) used to speai fish or retrieve fishing gear. 1992 1993 1994 1995 1996 Total Total dolphins stranded 28 33 31 32 29 153 Human and fishery interactions Rope marks 2 0 0 3 4 9 Flukes cut off and mutilations 3 1 1 0 1 6 Boat strike 2 0 1 2 0 5 Blunt-object trauma 0 2 0 1 0 3 Net marks 0 0 0 0 1 1 Gaff wounds 0 1 0 0 0 1 Total 7 4 2 6 6 25 No human interaction 15 21 16 16 15 83 Human interaction could not be determi ned (CBD) 6 8 13 10 8 45 Percent of human or fishery interaction. ; 31.8% 16.0% 11.1%' 23.1%' 28.6% 27.3% (n=22) (n=25) (n = 18) (n=22) (/)=21) (/( = 108l ' t^alculated from total number of .strandinf^.s minu.s CBD, Given that 1993 vi'as an unusual year in the stranding pattern, this year was excluded from the analysis. Specif- ically, the number of strandings was found to be higher in the spring than in other seasons (contrast analysis, P=0.012). The majority of strandings occurred in the southern half of the state, zones 2 (/!=67: 43.8%) and 3 (n=6h 39.9'7f). Seasonally, in zone 1 (n=25), 72.0% of its strandings oc- curred in fall and winter, whereas in zones 2 and 3, the majority of their strandings occurred in spring and sum- mer, 68.7% and 70.5%, respectively. Seventy-three percent 1 11 = 11) of the bottlenose dolphins that stranded in the southern half of zone 1 did so between October to April. The difference in seasonal patterns of strandings between zones was significant (X" goodness-of-fit; P=0.003). Gender The total number of stranded bottlenose dolphins with known sex was 115. The sex ratio for 1992-96 was 1.00:0.89, females («=61) to males (n=54), not significantly different from parity (X" test of association; P=0.979). A significant decrease in the proportion of unknown gender occurred during the period 1992-96 (X' test for trend; P=0.012) because of an increase in the number of animals examined in necropsy. Length classes The total number of stranded bottlenose dolphins with known length was 138. Based on length-at-age data from known bottlenose dolphins (Readetal., 1993) and stranded bottlenose dolphin data from Texas (Fernandez and Hohn, 45 40 35 30 25 20 - 15 10 5 0 ■ Unknown ID Females ■ Males _JL JX I II III IV V Length classes Figure 3 The number of strandings for males, females, and for bottle- nose dolphin of unknown sex in each length-class stratum from 1992 to 1996 in South Carolina (class I=neonates; class II<185 cm; class 111=186-200 cm; class IV=201-240 cm; class V>240 cm). 1998) the length data were stratified into five classes: class I (neonates — defined as a newborn having a folded dorsal fin or flukes or with umbilical remnants |or with both physical features]); class II (<184 cm, young of the year); class III (185-200 cm— calves); class IV (201-240 cm, mostly physically immature, especially females); and class V (>240 cm, mostly mature) (Fig. 3). Males and females were distributed proportionately and evenly across the length classes with the exception of two classes: class III and class IV (Fig. 4). In class III, males 262 Fishery Bulletin 100(2) dominated (83.3'7f ). Females were more prevalent in class rV {GG.T^'i). Males showed the lowest numbers of strand- ings in class III (n=5) and highest numbers in class V (/(=20). Neonates Neonates represented 19.6^/(- (;(=27) of the total number (;) = 138) of strandings of dolphins with known length, ranging from 13.3'^^ in 1993 to 24,1'f in 1994 and were found in every month of the year, except March (Fig. 4). Twenty (74.1%) neonates stranded during the spring (« = 14) and summer (n=6) months. June had the greatest number of strandings in=7), followed by May («=4) and November (;? =4). Thirteen of the 27 neonates {48.19i ) were <100 cm. Twelve of these stranded during the spring and fall months. More female neonates (1.3:1.0) stranded in South Carolina than males, though this difference was not significant (X" test of association; P=781). Twenty-four (88.9'7t) of the neonates stranded in zones 2 (?! = 13) and 3 (n=ll). Neonates were found dead in the in- ner waterways (?! = 18) and along the outer beaches (;?=9). Twelve of the dead neonates found in the inner watei-ways were retrieved while they were floating. Females ^220 cm Females found at a length that showed them capable of being reproductively mature (i.e. >220 cm) (Odell, 1975; Mead and Potter, 1990) represented approxi- mately 507( (?i=30) of the total number (/!=61i of females stranded. The proportions of females >220 cm stranded each year were similar, with the exception of those for 1992. where only one out of seven females was this length. However, this finding may be biased, except for 1994, because of the number of animals >220 cm that were of unknown sex. The proportion of strandings of females >220 cm in each season was statistically significant (X" goodness-of-fit; P=0.011). A large proportion of the lengths of female bottlenose dolphin stranded during winter (40'~( i and spring (40%) were >220 cm compared with lengths for summer (6.7%) and fall (13.3%). Human interaction The total number of stranded bottlenose dolphins where either human interaction or no human interaction could be determined was 108. Twenty-five bottlenose dolphin strandings, averaging five per year, showed evidence of human interaction. Eighty-three showed no signs of human interaction and 45 could not be determined (Table 1). Inci- dents of net entanglements, made evident by rope or line marks, net (mesh) marks, and mutilations, accounted for 16 of the human interaction cases. Incidence of confirmed human interaction on bottlenose dolphins was highest from March to July (/? = 18). Rope or line marks were more prevalent from February through May (n=8). The ratio of males (?i = 10) to females (/; = 11) was 1:1 in the number of positive human interactions, but there were differences in the length class and types of interaction between the sexes. Of the five males that were involved with net entangle- ments, four were less than 218 cm. Of the eight females associated with entanglements, seven were greater than 210 cm and six of these were >220 cm. Eighty-eight percent Jan Feb Mar I Apr May Jun I Jul Aug Sep I Oct Nov Dec Winter I Spring I Summer I Fall Months Figure 4 Total number of strandings of neonatal bottlenose dol- phins in each month froml992 tol996 1/1=27). Diagonally lined bo.\es represent those neonates <100 cm in length. of reported human interactions occurred in zones 2 (/) = 10) and 3 (n=12). Preliminary analysis of stomach contents from bottlenose dolphins stranded on account of human interaction in our study showed that the majority of ani- mals had full stomachs with shrimp or fish remains (or both)(McFee, personal obs.). Discussion Despite the establishment of an organized marine mammal stranding network in the southeastern United States since 1990, there has been little published on basic data from stranded bottlenose dolphuis other than from reports that can be found as "gray literature." Results from our study indicated the value of analyses of strandings and produced three main findings: 1) the northern portion (zone 1) of the state reported significantly more bottlenose dolphin strandings between November and March 2 ) neonatal bot- tlenose dolphin strandings occurred with more frequency between May and July and 3) evidence of human interac- tion as the cause, or contributing factor, in the deaths of some bottlenose dolphins. Several hypotheses regarding stock structure of Atlan- tic bottlenose dolphins have been proposed (Hohn, 1997). One hypothesis is that a single coastal migi'atory stock mi- grates seasonally from Long Island, New York, to the cen- tral east coast of Florida (Scott et alM. The other hypoth- esis is that multiple bottlenose dolphin stocks exist that include 1) year-round residents with small home ranges, 2 ) seasonal residents with large home ranges, or 3 ) migra- tory groups with long-range movements (Hohn, 1997). Bottlenose dolphins begin to leave Virginia in mid-Oc- tober and are mostly absent by mid-November (Swingle, ^ Scott, G. P., D. M. Burn, and L. J. Hansen. 1988. The dolphin die-off: long-term effects and recovery of the population. Proc. of the Oceans '88 Conf. m; p. 819-823. Unpubl, manu.scnpt. Southeast Fisheries Science Center, 75 Virginia Beach Dr, Miami, Florida 33149. McFee and Hopkins Murphy: Strandings of Tursiops truncatus off Soutfi Carolina 263 1994; Barco et al., 1999). At about the same time, large numbers of dolphins begin to appear along the "Grand Strand" in northern South C'arolina (zone 1) in October and peak in early November, according to bottlenose dol- phin sighting data collected during photo-identification studies (Wiung^). During the 1987 bottlenose dolphin die- off, 52 bottlenose dolphin strandings were reported in South Carolina from October through December (Wang et al., 1994). Densities of bottlenose dolphins during a one- year aerial survey of waters from the shore to the Gulf Stream showed the greatest numbers of sightings in fall 1982 (concentrated in the Carolinas), and in winter 1983 (concentrated in northern Florida) (Wang et al., 1994). Stranding patterns may reflect the abundance of an- imals. Although large numbers of dolphins occur year- round in South Carolina, there appears to be a peak in strandings in the late fall (November) which would coin- cide with data from Myrtle Beach (Young^), Charleston (Zolman, 1996), and Hilton Head Island (Petricig, 1994) in which greatest abundance of dolphins occurred in late fall. Water temperature, distribution of prey, and use of coastal shrimp trawlers have been implicated as reasons for dol- phin movements and abundance in certain areas (Kenney, 1990: Mead and Potter, 1990; Brager et al., 1994; Fertl, 1994). The late fall increase in the number of strandings in South Carolina could be due to the increased numbers of dolphins from any one of the migratory stocks suggested in the above hypotheses. Zone 1, in particular, provided evidence that a portion of the strandings is from a coastal bottlenose dolphin migratory stock or stocks. The north- ern half of zone 1 is known as the "Grand Strand" which extends from N. Myrtle Beach to Murrells Inlet (approxi- mately 59 km). This area is highly populated and animals coming ashore here are found and reported regardless of the season of the year. Coverage in the southern half of zone 1 (approximately 78 km) tends to be high from May to September when the beaches are monitored for sea tur- tle nesting and hatching, but low during October to April. However, the majority of strandings occurred during the latter time period. This may suggest an influx of bottle- nose dolphins migrating through zone 1 from October to April, either from the north or south. We would expect bottlenose dolphin mortality to be simi- lar to that for terrestrial mammals ( Ralls et al., 1980): high neonatal and first-year mortality and high adult mortal- ity, and an even distribution of mortality among males and females. If stranding data reflect natural mortality pat- terns, our results and other studies (Hersh and Duffield, 1990; Hersh et al., 1990; Wells and Scott, 1990; Fernandez and Hohn, 1998) are consistent with mortality patterns suggested for terrestrial mammals. Further, the percent- age of stranded bottlenose dolphin neonates (19.6'^) was intermediate when compared with that of previous studies (observations in Sarasota, Florida, 36.8'+ [Wells and Scott, 1990], and Indian/Banana River System, Florida, 11.2% IHersh et al, 19901 ), but similar to that of Texas (20. 0'^, (Fernandez and Hohn, 1998] I. We can only assume that ^ Young, R. 1998. Personal commun. Coastal Carolina Uni- versity, P.O. Box 1954, Conway, SC 29526. mortality during the first year of life is high for bottlenose dolphins regardless of geographical location. Age and ovarian analysis of stranded bottlenose dol- phins >220 cm (Odell, 1975; Mead and Potter, 1990) are necessary to determine whether these animals are sex- ually mature and whether the seasonal patterns noted above correlated with a seasonal reproductive cycle. Sea- sonal reproduction cycles are complex and not well studied in the South Carolina bottlenose dolphin population but have been demonstrated where adaptations to local envi- ronmental conditions may influence seasonal reproductive cycles (Urian et al., 1996). Over large geographic regions, bottlenose dolphins ex- hibit year-round calving cycles, but within small geograph- ic regions there may be a higher degree of local reproduc- tive seasonality (Urian et al., 1996). A unimodal seasonal distribution of neonate bottlenose dolphin strandings was noted from Sarasota, Florida, and along the Texas coast, although peak neonatal strandings occurred in different months of the year — May and March, respectively (Urian et al., 1996; Fernandez and Hohn, 1998). A bimodal sea- sonal distribution was noted for the east coast of Florida in the Indian River Lagoon (Urian et al., 1996). In Sara- sota, Florida births have been noted in every month of the year (Urian et al., 1996). Although sample size over the five-year period for our study was too small to esti- mate significance of trends, our results showed a unimodal distribution and a peak number in June. However, more data may show a bimodal distribution of bottlenose dol- phin neonatal strandings because of a second peak that occurred in November. These peaks do not appear to be a function of effort because the majority of neonate strand- ings occurred on the banks of inland waterways or the neonates were found as floating bodies. The number of neonates in the Stono River estuary, Charleston, South Carolina, peaked in the fall, during a 15-month photo- identification study (Zolman, 1996). Further, all four neo- nate bottlenose dolphins stranded in South Carolina in November were <100 cm; therefore these animals may have been aborted near-term fetuses. The determination of human interaction as the cause of mortality for bottlenose dolphins is an important role of the marine mammal stranding networks and can in- fluence management decisions. For example, the Marine Mammal Protection Act (MMPA), as amended in 1994, required that annual stock assessment reports for each stock of marine mammals be prepared. One of the items to be addressed in these reports was a description of com- mercial fisheries that interact with each stock and the level of mortality caused on each stock by each fishery (Waring et al., 1999). The level of mortality each fishery contributes to a stock, in turn, is essential in determining potential biological removal (PBR) estimates for the stock and the subsequent classification category that regulates each fishery (Waring et al., 1999). The current PBR for Atlantic coastal bottlenose dolphins is 25 (Waring et al., 1999). In a study on the American shad iAlosa sapidis- sima ) fishery in South Carolina from 1994 to 1995, no hu- man interactions were shown to be a cause of bottlenose dolphin mortality in the area of fishery effort (McFee et 264 Fishery Bulletin 100(2) al., 1996). As a result, the South CaroUna shad fishery re- tained its original classification as a category-Ill fishery (i.e. unlikely to take marine mammals in the course of op- eration) as described in the MMPA amendments of 1988. Bottlenose dolphin mortality due to human interactions is variable along the eastern United States and Gulf of Mexico (Wang et al., 1994). Incidents of human interaction in South Carolina were also variable over our five-year study period. We believe that the number of bottlenose dol- phins in our study showing positive human interaction is a minimum because determination of human interaction cases is difficult to assess owing to a lack of trained per- sonnel, the decomposition of some carcasses, and the pre- sumption that some interactions do not leave any physical evidence. It was a rare occurrence to have gear attached to the carcass; therefore, determination of human interac- tion was usually made by obsei-ving external marks such as cross-hatched lines or lines imprinted by the fishing gear. Human interaction as a cause or contributing factor in a dolphin's death can include fishery interactions (crab pots, trawls, etc. ), boat collisions, gun shot wounds, environ- mental contaminants (agricultural run-off, pesticide use, oil spills ). These interactions can result in acute ( drowning in a net) or chronic (environmental contaminants) death, show physical evidence (net marks) on the body or none at all. The percentage of human interaction cases observed in South Carolina was low compared with those in North Car- olina strandings (>35'7( in some years; Wang et al., 1994; FR, 1997). Resident bottlenose dolphins in South Carolina appear to be exposed to different fishing operations than do bottlenose dolphins that migi-ate through or inhabit North Carolina waters. Net marks were the most common obser- vation (10.5 animals per year) of human interaction cases in North Carolina (FR, 1997), whereas in South Carolina only one presumed net-caught animal was obsei-ved over a five-year period. In South Carolina incidents of entangle- ments as evidenced by rope or line marks are puzzling. At this time, it is highly speculative as to which fishery in South Carolina may be responsible for the incidence of en- tanglements associated with rope or line marks. There is evidence to suggest that relationships exist be- tween gender and lengths of various species of cetaceans involved with human interaction (Perrin et al., 1994; Cox et al., 1998). In our study small male bottlenose dolphins and female bottlenose dolphins >220 cm showed evidence that they were subject to human interaction. One study found that females with calves spent more time feeding at shrimp boats than did lone animals (Fertl, 1994). Food intake for lactating females can increase dramatically (Cockroft and Ross, 1990). Although heavy-feeding behav- ior may be energetically beneficial, it may also be costly to both the calf and mother by exposing them to fishing gear and predation. In summary, the stranding data collected for bottlenose dolphins in South Carolina from 1992 to 1996 provides baseline information for the demographics, life history studies, and management concerns for comparing future stranding rates of bottlenose dolphins in South Carolina. Although it cannot be definitively stated that stranding rates coincide with a portion of a migratory stock, strand- ings in the northern portion (zone 1) do increase during a period of greater dolphin abundance. More years of data will further elucidate the seasonal reproduction distribu- tion for bottlenose dolphin. Finally, the detection of human interaction as a cause or contributing factor in the deaths of some bottlenose dolphins in South Carolina has demon- strated the need to continue the effort to report these inci- dents for management purposes. Acknowledgments The authors would like to acknowledge the numerous South Carolina marine mammal stranding volunteers, past and present, who have contributed their time in the noti- fication, recovery, transportation, and sample collection of stranded marine mammals. Without their dedication, our knowledge of marine mammals would be greatly reduced. We would also like to thank Sylvia Galloway, Pat Fair, Larry Hansen, David Whitaker, and the reviewers and the anonymous referees of the Fishery Bulletin; Keith Bangerter, Lori Schwacke, and Laura Kracker for help with statistical and spatial analyses; the Armed Forces Institute of Pathology, Washington, D.C.; and finally coun- ty, state, law enforcement, town administration person- nel, and the numerous concerned citizens who helped in many ways in the success of the South Carolina stranding network. This work was made possible through NOAA's responsi- bility under the Marine Mammal Health and Stranding Response Act (1993), and under a Letter of Authorization issued by the National Marine Fisheries Service to the South Carolina Department of Natural Resources. Fund- ing to support the South Carolina marine mammal strand- ing network comes from the South Carolina Endangered Wildlife Fund. Literature cited Barco, S. G., W. M. Swingle, W. A. McLellan. R N. Harri.s, and D. A. Pabst, 1999. Local abundance and distribution of bottlenose dol- phins {Tursiops truncatus) in the nearshore waters of Vir- ginia Beach, Virginia. Mar Mamm. Sci. 15{2):394-408. Brager, S., B, Wursig, A. Acevedo, and T. Henningsen. 1994. Association patterns of bottlenose dolphins (Tursiops truncatus) in Galveston Bay, Texas. J. Mamm. 7.5(2):431- 437. Brown, P. J. 1977. Variations in South Carolina coastal morphology. South. Geol. 18(4): 249-264. Cockroft, V. G. and G. J. B. Ross. 1990. Observations on the early development of a captive bottlenose dolphin calf In The bottlenose dolphin (S. Leatherwood and R .R. Reeves, eds.), p. 461-478. Aca- demic Press, Inc., San Diego. CA. Cox, T. M., A. . Read, S. Barco, J. Evans, D. P Gannon, H. N. Koopman, W. A. McLellan, K. Murray, J. Nicolas, D .A. Pabst, C. W. Potter W. M. Swingle, V. G. Thayer K. M. Touhey and A. J. Westgate. 1998. Documenting the bycatch of harbor porpoises, Pho- McFee and Hopkins Murphy Strandings of Twsiops tivncatus off South Carolina 265 toena phococna. in coastal gillnct fishiTii's from stranded carcasses. Fish. Bull. 96:727-734. FR (Federal Register). 1993. 50 CFR. part 216. Taking and iniijorting of marine mammals; depletion of the coastal migi'atorv stock of bot- tlenose dolphins along the U.S. Mid-Atlantic coast — final rule, vol.58, no. 64, p. 17789-17791. 1997. 50 CFR, part 229. Proposed list of fisheries, vol. 62, no. 101, p. 28657-28669. Fernandez, S., and A. A. Hohn. 1998. Age, growth, and calving season of bottlenose dolphins, Titrsiopa tntiicntii.s. ofT coastal Te.xas. Fish. Bull. 96:357- 365. Fertl, D. 1994 Occun-ence patterns and behaviour of bottlenose dol- phins iTiirsiops triincalus) in the Galveston Ship Channel, Texas. Tex. J. Sci. 46(4):299-317. 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 (8. Leather- wood and R. R. Reeves, eds. ), p. 129-139. Academic Press, Inc., San Diego, CA. Hersh, S. L., D. K. Odell, and E. D. Asper. 1990. Bottlenose dolphin mortality patterns in the Indian/ Banana River System of Florida. In The bottlenose dol- phin (S. Leatherwood and R. R. Reeves, eds.), p. 155-164. Academic Press, Inc., San Diego, CA. Hofman, R. J. 1991 . History, goals, and achievements of the regional marine mammal stranding networks in the United States. In Marine mammal strandings in the United States: proceedings of the second marine mammal stranding workshop, Miami, Florida, Dec. 3-5 (J. E. Revnolds and D. K. Odell. eds. ), p. 7-15. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 98. Hohn, A. A. 1997. Design for a multiple-method approach to determine stock structure of bottlenose dolphins in the mid-Atlantic. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-401, 22 p. 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, Inc., San Diego, CA. McFee, W. E., D. L. Wolf, D. E. Parshley, and P. A. Fair. 1996. Investigation of marine mammal entanglement asso- ciated with a seasonal coastal net fishery. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-386, 104 p. Mead, J. G., and C. W. Potter 1990. Natural history of bottlenose dolphins along the cen- tral Atlantic coast of the United States //; The bottlenose dolphin (S. Leatherwood and R. R. Reeves, eds. ), p. 165-195. Academic Press, Inc., San Diego, CA. Odell, D. K. 1975. Status and aspects of the life history of the bottle- nose dolphin, Tursiops Iriincatus, in Florida. J. Fish. Res. Board Can. 32:105,5-10.58. Perrin, W. F., G. P. Donovan, and J. Barlow (eds.). 1994. Report of the workshop on mortality of cetaceans in passive fishing nets and traps. The International Whal- ing Commission (IWC), Cambridge (special issue 15):6-71. Petricig, R. O. 1994. Population and behavior patterns of bottlenose dolphins in Bull Creek, South Carolina. In Coastal stock(s) of Atlan- tic bottlenose dolphins: status review and management (K. R. Wang, P. M. Payne, and V. Thayer [compilersl l, p. 56-57. U.S. Dep. Commerce, NOAA Tech. Memo. NMFS-OPR-4. Ralls, K., R. L. Brownell, and J. Ballau. 1980. Differential mortality by sex and age in mammals, with specific reference to the sperm whale. Rep. Int. Whal. Comm. (special issue 2):233-243. Read, A. J., R. S. Wells, A. A. Hohn, and M. D. Scott. 1993. Patterns of growth in wild bottlenose dolphins. Tiir- siops truncatiis. J. Zool. Lond. 231:107-123. Swingle, M. 1994 What do we know about coastal bottlenose dolphins in Virginia? In Coastal stocks of Atlantic bottlenose dol- phins: status review and management (K. R. Wang, P. M. Payne, and V. Thayer [compilers]), p. 34-40. U.S. Dep. Commerce, NOAA Tech. Memo. NMFS-OPR-4. Urian, K. W., D .A. Duffield, A. J. Read, R. S. Wells, and E. D. Shell. 1996. Seasonality of reproduction in bottlenose dolphins, Tursiops truncatus. J. Mamm. 77l2):394-403. Wang, K .R., P M. Payne, and V. G. Thayer. 1994. Coastal stock(s) of Atlantic bottlenose dolphin: status review and management. U.S. Dep. Commerce, NOAA Tech. Memo. NMFS-OPR-4, 121 p. Waring, G. T, D. L. Palka, P J. Clapham, S. Swartz, M. C. Rossman, T. V. N. Cole, K. D. Bisack, and L. J. Hansen. 1999. U.S.Atlantic marine mammal stock assessments. LIS. Dep. Commer. NOAA Tech. Memo. NMFS-NE-116, 182 p. Wells, R. S., and M. D. Scott. 1990. Estimating bottlenose dolphin population parameters from individual identification and capture-release tech- niques. Rep. Int. Whal. Comm. (special issue 12):407-415. Zolman, E. S. 1996. Residency patterns, relative abundance and popula- tion ecology of bottlenose dolphins in the Stono River Estu- ary, Charleston County, South Carolina. M.S. thesis, Univ. Charleston, Grice Marine Biological Laboratory, Charles- ton, SC. 266 Abstract— Growth parameters were estimated for porbeagle shark tLamna iiasus) in the northwest Atlantic Ocean on the basis of vertebral annuli. A total of 578 vertebrae was analyzed. Annuli were validated up to an age of 1 1 years by using vertebrae from recaptured oxy- tetracycline-injected and known-age sharks. Males and females grew at sim- ilar rates until the size of male sexual maturity, after which the relative growth of the males declined. The gi-owth rate of the females declined in a similar manner at the onset of maturity. Growth cui-ves were consistent with those derived from tag-recapture analyses iGROTAG) of 76 recaptured fish and those based on length-frequency methods with mea- surements from 13,589 individuals. Von Bertalanffy growth curve parameters (combined sexes) were L., = 289.4 cm fork length. A' = 0.07 and /„ = -6.06. Maxi- mum age, based on vertebral band pair counts, was 25 and 24 years for males and females, respectively. Longevity cal- culations, however, indicated a maxi- mum age of 45 to 46 years in an unfished population. Validated age and growth of the porbeagle shark {Lamna nosus) in the western North Atlantic Ocean Lisa J. Natanson Joseph J. Mello National Marine Fisheries Service 28 Tarzwell Dr Narragansett, Rhode Island 02882 E-mail address (for L J Naianson) Lisa Natansonia'noaa gov Steven E. Campana Marine Fish Division Bedford Institute of Oceanography PO Box 1006 Dartmouth, Nova Scotia Canada B2Y 4A2 Manuscript accepted 22 August 2001. Fish. Bull. 100:266-278 (2002). The porbeagle iLamna nasiis) is a large pelagic shark in the family Lam- nidae that occurs in the coW, temper- ate waters of the North Atlantic, South Atlantic, and South Pacific oceans. The species extends from Newfoundland to New Jersey in the western North Atlan- tic (Castro, 1983), and from Iceland and the western Barents Sea to Morocco and the Mediterranean in the eastern North Atlantic (Compagno, 1984). Directed commercial fisheries for porbeagle have existed in the western North Atlantic in both U.S. and Canadian waters since the early 1960s (Campana et al.M, The fishery collapsed in 1967, apparently from overfishing. Canadian fishermen maintained low and apparently sus- tainable catches in the 1970s and 1980s, which allowed the stock to rebuild. A renewed fishery for porbeagle began in both the United States and Canada in the 1990s (Campana et al.M. Accurate age determinations are nec- essary for both the assessment and management of the porbeagle shark be- cause they form the basis for calcula- tions of growth and mortality rates, age at maturity, age at recruitment, and estimates of longevity. Aasen (1963), in an early study of porbeagle growth, generated a growth curve for the west- ern North Atlantic population based on analyses of length frequencies from a single year and on vertebral readings of one fish. However, he did not pro- vide any independent confirmation of the accuracy of his age estimates. Fran- cis and Stevens (2000) used length-fre- quency analysis to estimate the growth of juvenile porbeagles in the South Pa- cific. Both of these studies indicated that porbeagle grow relatively rapidly through the first year of life, but only minimal information has been avail- able for older fish. In view of the history of the porbea- gle fishery and the need for accurate biological information for management of this species, an intensive U.S. -Cana- dian cooperative research program was initiated in 1999 to obtain detailed life-history and population data. This study reports on one portion of this pro- gram, specifically, the use of vertebrae to determine age and growth. The ver- tebral growth readings were validated as annuli on the basis of recaptures of tetracycline-injected and known-age sharks and verified by comparison with growth curves based on tag-recapture and length-frequency analyses. ' Campana, S., W. Joyce, L. Marks. P. Hurley, L. J. Natanson, N. E. Kohler, C. F. Jensen, J. J. Mello, and H. L. Pratt Jr 2000. The rise and fall (again) of the porbeagle shark population in the Northwest Atlan- tic. Unpubl. manuscr. Marine Fish Divi- sion, Bedford Institute of Oceanogi'aphy, P.O. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2. Natanson e! a\ Age and growth of Lamnus nasus in the western North Atlantic 267 Materials and methods Vertebral aging Vertebral samples from porbeagle sharks were obtained between 1966 and 1999 on board commercial and research vessels. The majority of samples idT'i) were collected after 1990 on Canadian commercial longline vessels. Sam- pling took place in U.S. and Canadian waters between Massachusetts (NE U.S.) and the Grand Banks (off south- ern Newfoundland) and all individuals were treated as belonging to the same stock on the basis of tagging data (Campana et al.-). Multiple vertebrae were removed from the area just above the branchial chamber wherever pos- sible; except on commercial vessels where samples were obtained closer to the head. Vertebrae were then stored frozen or in VO'^f ETOH until processing. Only samples that had measured fork length (FL — tip of the snout to the fork in the tail, over the body) or total length (TL — tip of the snout to a point on the horizontal axis intersecting a perpendicular line extending down- ward from the tip of the upper caudal lobe to form a right angle, over the body; Kohler et al., 1995) were used. All lengths reported in this document are over-the-body FL unless othei-wise noted. Total length in centimeters (cm) can be converted to FL cm by using the regression equa- tion (Campana et al.M: FL = 0.885(rLi + 0.99 [n = 361 ;--=0.99]. One vertebra from each sample was removed for processing. 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 sagit- tal plane and the resulting "bow-tie" sections were stored individually in capsules in TOVf ETOH. Each section was digitally photographed with a MTI CCD 72 video camera attached to a SZX9 Olympus stereo microscope by using reflected light. Magnification depended on the size of the section and varied from 4x to 12. 5x. Band pairs (consisting of one opaque and one translucent band) were counted and measured from the images by using Image Pro 4 software (Media Cybernetics, 1998). Measurements were made from the midpoint of the isthmus of the full bow-tie to the middle of the opaque growth bands at points along the internal edge of the corpus calcareum (Fig. 1). The vertebral radius (VR) of each centrum was measured from the midpoint of the isthmus to the distal margin of the intermedialia along the same diagonal as the band measurements. The identity of the birth band in the vertebra was con- firmed through comparison of the birth band radius (BR) measurements to vertebral radius measurements of late- term embryos, early young-of-the-year, and late age-0 fish. - Campana, S., L. Marks. W. Joyce, P. Hurley, M. Showell, D. Kuika. 1999. An analytical assessment of the porbeagle shark iLamna nasus) population in the northwest Atlantic. Cana- dian Stock Assessment Secretariat Research Document 99/158. 57 p. Marine Fish Division, Bedford Institute of Oceanogra- phy, P.O. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2. The late-term embryo vertebral samples were obtained from the South Pacific porbeagle population, courtesy of Malcolm Francis.'' The assumptions were made that verte- bral growth ;/( utero was comparable between the two pop- ulations and that measurements of the rehydrated dried South Pacific vertebrae were similar to those from wet NW Atlantic preparations. The relationship between VR and FL was calculated in order to confirm the interpretation of the birth band and to determine the best method for back-calculation of size- at-age data. The FL to VR relationship was curvilinear; therefore, the data were In-transformed before linear re- gression. The Fraser-Lee equation of the In-transformed data was derived for back calculation: ln(FL„ ) = h + (InlFL,, \-h)(\n radius,, } {\n radius,) where a = age; b = intercept from the FL-VR regression; and c = capture. Validation The accuracy of the vertebral band pair counts as annual indicators was determined by using both known-age recap- tures and recaptures of oxytetracycline (OTC)-injected and tagged individuals. Vertebrae from young of the year ( YOY) sharks whose FL was measured at both tagging and recapture were used for known-age analyses. Band pair counts were compared with time at liberty to determine band pair periodicity. One hundred and fifty-five porbea- gles of various lengths were also tagged and injected with 25 nig/kg of OTC (senior author, unpubl. data) Returned vertebrae from these sharks were examined with reflected UV light for the OTC mark. The number of band pairs distal to the OTC mark were then compared with the number of years at liberty. Data analysis Aging bias and precision of annulus counts were exam- ined by using age-bias plots and the coefficient of varia- tion (Campana et al., 1995). Two readers independently counted 100 vertebral sections from which the pair-wise age-reader comparisons were generated. Von Bertalanffy growth functions (VBGF) were fitted to the length-at-age data by using the following equation (von Bertalanffy, 1938): LJl .g-K"-WI) where L, = predicted length (cm) at age /; L = mean theoretical maximum fork length; K = a growth rate parameter (per yr); and tg = the theoretical age (yr) at zero length. ■' Francis, M. 2000. Personal commun. National Institute of Water and Atmospheric Research P.O. Box 14-901, Wellington, New Zealand. 268 Fishery Bulletin 100(2) ^?\ Figure 1 Photop'aph of a vertebral section from a porbeagle estimated to be 15* years old. Insert shows higher magnification view of narrow bands at edge. Scale bar = 1 mm. White dots are on annuli. The VBGF was calculated by using the nonlinear regres- sion function in Statgraphics. Locally weighted least squares regression (LOESS) curves were fitted to the FL vs. age data for each sex by using Statgraphics (Manguis- tics, 1997). Length frequency Length-frequency data were obtained from the Canadian International Observer Program operating primarily on the Scotian Shelf, and some data from the Grand Banks. Although the entire data set ( 1986-98) was analyzed, only data from the most complete year ( 1991 ) were used for the final analysis. Monthly length-frequency histograms were developed for eight months of 1991 for modal analysis. Calculations of mean fork length and annual growth rate for ages 0 and 1 were based on the first two modes of these data, which were easily distinguished and tracked across months. MULTIFAN (Fournier et al., 1990) was used to esti- mate the VBGF parameters from the 1991 length-frequen- cy data. The model analyzes multiple length-frequency distributions by using a maximum likelihood method to estimate the number of age classes present and VBGF pa- rameters L_ and K. An initial systematic search was con- ducted based on user-supplied K values and age classes. Constraints were placed on the estimates of length at age for the first two age classes. MLILTIFAN allows the user to start with a generalized search and then add parameters to further refine the model. The initial search included es- timates of A' ranging from 0.05 to 0.25 and age classes of 11 through 20. The hypotheses tested were the following: 1 ) constant length standard deviation for all age classes; 2) variable length standard deviation for all age classes; 3) constant length standard deviation for all age classes with seasonal growth; and 4) variable length standard devia- tion for all age classes with seasonal growth. A model in- corporating constant length standard deviation was fitted first with the additional parameters added sequentially. Results from the four models were compared by using log- likelihood tests following Fournier et al. ( 1990) and Fran- cis et al. ( 1999). The von Bertalanffy growth parameter /q was estimated from the equation Natanson et a\ Age and growth of Lamnus nmtJi in the western North Atlantic 269 where o tn = ^ = the age estimated by MULTIFAN for the youngest age class at the time it first appeared in the length-frequency samples; and = the time elapsed in years between the theoret- ical birthday and the first appearance of the youngest year class in the samples. The theoretical birthday was defined as 1 April based on the gestation period and the time of mating (Jensen et al.^). The first appearance of the youngest age class in the samples was 1 July. The model with seasonal growth components required the use of a modified von Bertalanffy equation to incorpo- rate the amplitude and phase of the seasonal growth: L. = LJ 1 khere -K'((-(„+(»/2;rlsin(2;r((12/+l>/12)-02) /)j = amplitude; and &2 = (MULTIFAN phase) -^ t^ (from Eq. 1) (1) The Gulland and Holt ( 1959) and Francis (1988a) mod- els were used to generate VBGFs from the tag-recapture data. The Gulland and Holt ( 1959) method uses graphical interpretation of the recapture data to produce estimates of L_ and K. Specifically, annual growth rate (cm/jT) was plotted 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. With this method, a coeffi- cient of variation of growth variability iv), the mean and standard deviation of measurement errors (m and s), and outlier contamination (p), are estimated as well as growth rates at two user-selected lengths (a and /3). The reference lengths, a and /3, were chosen to lie within the range of tagged individuals. The form of the von Bertalanffy equa- (2) tion becomes M Pea - ag,i ga-Sp -L, 1 + g„ - gp a-p (3) Tag-Recapture analysis Data from three independent tagging studies from the western North Atlantic Ocean were combined for tag and recapture analysis. In the 1960s, 542 porbeagles were tagged and 53 recaptured as part of a Noi-wegian study of the unfished population. In 1994 through 1996, the Canadian Department of Fisheries and Oceans (DFO) con- ducted a tagging program in which 256 porbeagles were tagged and 25 recaptured. Between 1979 and 1999, mem- bers of the National Marine Fisheries Services (NMFS) Cooperative Shark Tagging Program tagged 1034 and recaptured 119 porbeagles. Sharks were tagged and recap- tured by biologists and commercial and recreational fish- ermen in the United States and Canada and by biologists in the Norwegian study. All measurements were con- verted to FL by using the morphometric conversions reported in Campana.' Where Norwegian measurements were reported as Aasen's (1963) total length, they were converted to FL with the equation: FL = Q.93TL. Only those sharks reliably measured at the time of tagging and recapture were used in the analyses. Reliability was based on prior knowledge of the individual's expertise in measuring the shark or on detailed questioning of those indi- viduals as to the method used. The majority of sharks were measured by NMFS biologists or their representatives. ^ Jensen, C, L. J. Natanson, H. L. Pratt Jr., N. E. Kohler, and S. Campana. 2001. The reproductive biology of the por- beagle shark, Lamna nasus, in the western North Atlantic Ocean. Unpubl. manuscript. Apex Predators Program, NMFS, 28 Tarzwell Dr, Narragansett, RI 02882. where Lj = length at tagging; AL and AT = increments in length and time, respecitively; and g^, and gp = mean annual growth rates at the arbitrary lengths a and /3. The simplest model, with minimal parameters (a and /3), was used initially with additional parameters added to suc- cessively increase model complexity. Significant improve- ment in the model results were determined by using log likelihood ratio tests (Francis, 1988a). The modeling was carried out by using a Solver-based spreadsheet in MS Excel (Simpfendorfer''). The value of t^ cannot be estimated from tagging data alone; rather it requires an estimate of absolute size at age, such as size at birth, and was calculated with the VB- GF by solving for fg such that :f-Kl/A:)[ln{(L. -L,)/L,]], (4) where L, = known length at age (size at birth); The tfy values were calculated from an average size at birth of 67 FL cm (Aasen, 1963) with t = 0. Longevity Several methods were used to estimate longevity. The oldest fish aged from the vertebral method provides an initial value, but is likely to be underestimated in a fished population. Taylor (1958) defined the life span of a teleost species as the time required to attain 95'7( of ^ Simpfendorfer. C. 2000. Unpubl. data. Mote Marine Labo- ratory, 1600 City Island Park, Sarasota, FL 33577. 270 Fishery Bulletin 100(2) the L . Using a wide range of species, Hoenig (1983) calculated the relation- ship between longevity, ^„,„j-, and the natural mortality rate, M, needed to attain one percent of initial abundance in an unfished population as ln(M)= 1.44 0.982 ln(^„„_,). However, a relationship based on other species need not be used (Campana et al.^). Assuming a constant instan- taneous natural mortality, M, in an unfished population, the following equa- tion applies: Ln (proportion offish that survive to As with Hoenig 1 19831, this equation was evaluated at a value of 0.01 for the proportion offish that survive. Results Vertebral aging 250 200 Mean radius of birth mark (n=575) 150 s , Smallest g I free-living 100 50 Size at birtfi 10 15 Vertebral radius (mm) 20 25 Figure 2 Relationship between vertebral radius and fork length for male and female por- beagles. The solid diamond is the mean vertebral radius of the smallest free-living specimens (n=2), the solid circle is the mean vertebral radius of the largest embryos (n=3). The horizontal line represents the size at birth and the vertical line represents the mean radius of the birth mai'k in sharks less than 1.50 cm FL. Vertebral samples from 578 porbeagles were used in our analysis; 283 were males, 292 were females, and 3 were of unknown sex. All vertebrae had distinct band-pair patterns (Fig. 1). The birth mark was indicated by a slight change in angle of the centra and was often the most pronounced first band pair. Subsequent annuli consisted of a pair of alternating opaque and translucent bands that crossed the entire centrum, except in the oldest sharks. Band-pair width decreased with age, narrowing substantially in the oldest individuals (Fig. 1). The FL-VR relationship was curvilinear 'Fig. 2i, and al- though the In-transformed relationship was not complete- ly linear, it was preferable. Therefore, regressions were calculated based on the ln(FL)-ln(VTl) relationship where ln(FL) = 0.88 ^ In(VR) -^ 2.96 |r-=0.94./;=5751. There was no significant difference between the regres- sions of males and females (ANCOVA, P>0.01). The identity and location of the birth band was confirmed through comparison of the BR of all individuals to the VR of YOY and embryos. The mean VR of three late-term em- bryos ranging in size from 56.1 to 58 cm FL (mean VR=4.3 mm, 01=0.57) was slightly less than the mean BR value of the total sample (mean BR=5.4 mm, CI=0.03 mm, /)=578l. The mean BR those of two early YOY (67.7 and 69.2 cm FL) were also similar to those of the total sample (mean VR=5.3 mm, CI=1.27 mm). The VR of the age-0 individuals ranging in size from 76.5 to 100 cm FL (mean VR=6.2 mm, Cl=0.31 mm,/i = 16) was slightly higher (Fig. 2). The placement of the mean BR between the VR of both the YOY and embrvos in- dicates that the birth ring was identified correctly. The birth ring radius increased slightly with increasing FL likely be- cause of an increase in the length of the isthmus of the how- tie section in larger sharks. Validation Known-age recaptured porbeagles and OTC-injected recap- tured porbeagles returned with vertebrae confirmed the accuracy of the band-pair counts as indicators of age. Six porbeagles tagged as YOY were recaptured after three to five years at liberty (Table 1 ). Of the four sharks that were recaptured in the spring, all had translucent material at their gi'owing edge. In contrast, the two sharks recaptured in November had a broad opaque zone at the gi'owing edge. In each case, the vertebral band-pair counts matched the expected counts based on time at liberty (Fig. 3). Vertebrae from six OTC-injected sharks were returned after 0.02 to 2.5 years at liberty. All vertebrae showed a distinct fluo- rescent mark indicating that the OTC was incorporated within six days (0.02 years) of injection (Fig. 3). Two of the OTC-injected sharks were at liberty for over one year and were used for validation. In both cases, the expected number of growth bands was deposited on the vertebrae between the date of injection and the date of recapture (Table 1). The shark at liberty for 2.5 years had 3 full bands after the OTC mark, the last band having formed just prior to capture. The OTC-injected shark at liberty for 1.5 years was an adult (189.8 cm FL) at tagging and was aged, by vertebrae, to be 11 years at recapture. This Natanson et al : Age and growth of Lomnus nasus in the western North Atlantic 271 Table 1 Tag-recapture fork length at data for the OTC-injected recapture. md known -age recaptured poi beagles w th vertebrae. TFL = fork length at tagging, RFL = Sample number Sex TFL (cml RFL (cm) Date tagged Date recaptured Years at liberty Growth (cm) No. of bands past birth mark Known age LN207 F 104.0 178.0 10 Nov 1993 23 Apr 1999 5.4 74.0 6 553 M 85.0 160.0 1 Nov 1994 9 Nov 1999 5.2 75.0 5 206 M 100.0 153.0 24 Oct 1994 27 May 1999 4.6 53.0 4 556 F 97.0 154.0 17 Nov 1995 19 Nov 1999 4.0 57.0 4 185 F 89.7 148.5 17 Nov 1995 29 Mar 1999 3.4 58.8 4 183 F 94.2 145.0 17 Nov 1995 23 Mar 1999 3.3 50.8 4 No. of bands past OTC mark OTC-injected LN 171 M 118.0 156.0 16 Sep 1996 30 Mar 1999 2.5 38.0 3 172 M 189.8 187.0 27 Sep 1997 29 Mar 1999 1.5 -2.8 1 + 473 F 104.0 117.0 15 Apr 1999 21 Nov 1999 0.6 13.0 — 184 M 147.5 152.0 21 Oct 1998 30 Mar 1999 0.4 4.5 — 204 M 105.0 95.0 23 Mar 1999 23 May 1999 0.2 -10.0 — 173 F 96.5 96.0 26 Sep 1997 2 Oct 1997 0.02 -0.5 — shark had one full band and 83'^'f growth of the next band, based on the size of the last full band. This, along with the YOY known-age individuals, confirmed annulus formation and our band interpretation from birth to 11 years of age. Sharks older than age 1 1 were assumed to have been aged correctly due to a similar interpretation of the bands. Data analysis Comparisons of counts between the two readers indicated no appreciable bias (Fig. 4). The coefficient of variation for age 1+ sharks was approximately 157c. In the absence of bias, this level of precision was considered acceptable; thus the counts generated by one reader for the entire set of vertebrae were used for the analyses. Length-at-age data showed that males and females grow at similar rates until approximately 170 cm FL. at which point the relative growth rate of the males declines (Fig. 5). The change in relative gi-owth between the sexes coin- cides with the size and age of male maturity (Jensen et al.^). Von Bertalanffy gi-owth functions fitted to the verte- bral band-pair count data suggested that males attain a smaller maximuin size than females (Table 2). The growth rate of females also declined at size at maturity (approxi- mately 218 cm FL; Jensen et al.'*). The considerable over- lap in size at age between the sexes indicated that the difference in growth rate is minor; therefore subsequent comparisons were made for the sexes combined. Tag-Recapture analysis A total of 76 porbeagles was recaptured with sufficient information for tag-recapture analysis. Time at liberty ranged from 0.02 to 6.0 years and size at tagging ranged from 78 to 204 cm FL. Sharks were tagged and recaptured in all months of the year. Tagging effort was fairly evenly distributed throughout all months, whereas most recap- tures were made between March and May (539'f ). For both tagging and recapture, January and February were repre- sented by the least data OVc and 1% for tagging and recap- ture, respectively). More tags were released in December than were recaptured ( 12% of the tags but only 2% of the recaptures). Most tagged sharks were small (74% <150 cm FL) because the majority were opportunistically tagged onboard commercial fishing vessels. Data from 54 sharks at liberty greater than 0.9 years were used in the Gulland and Holt (1959) analysis, whereas all individuals were used for GROTAG (Francis 1988a). The results of the likelihood ratio tests with GROTAG (Francis, 1988a) demonstrated that the more complex non- linear model with five of the six parameters included was the best fit for these data (model 3, Table 3). The high value of ,s suggests a lack of sufficient information for GROTAG (Francis, 1988a) to distinguish between growth variabil- ity and measurement variability (Francis and Mulligan, 1998). The mean annual growth rates at FL= 95 cm and 150 cm were 19.21 cm/yr and 9.52 cm/yr, respectively (Fig. 6). Von Bertalanffy estimates from the Gulland and Holt (1959) and GROTAG (Francis, 1988a) methods produced similar results (Tables 2 and 4). Length frequency Analysis of modal length-frequency progressions verified the size at age and growth rate of age-0 and age-1 indi- viduals (Fig. 7). Age-0 fish entered the fishery in July with 272 Fishery Bulletin 100(2) Figure 3 Vertebral sections from recaptures of three known-age porbeagles and one OTC-injected porbeagle. Annuli are indicated, as is the birth mark. Scale bar = 1 mm. a mean length of 85 cm FL and grew to a mean length of 98 cm FL by December. Age-1 individuals had a mean length of 106 cm FL in April. 113 cm FL in July, and 123 cm FL in December, resulting in an annual growth of 25 cm/yr between December and December. Although larger length modes were occasionally visible, only the age-0 and age-1 modes were clear and unambiguous throughout the year. The MULTIFAN models that best fitted the 1991 data were the most complex, having variable standard devia- tions in length and variable seasonal growth. The data with sexes combined had 18 age classes, whereas males and females had 16 and 15 age classes, respectively. The MULTIFAN L . and A' von Bertalanffy parameters fell out- side the 95% confidence intei-vals for the tagging and ver- tebral studies, although the tg values did not (Table 2). These differences are reflected in the VBGF curves as com- pared with the other methods (Fig. 8). The annual growth rate calculated from the MLILTIFAN data was consistent with that of the vertebral and tag-recapture analyses at 150 cm FL (Fig 6). However, the reliability of the MULTI- FAN results is questionable given the large number of age classes in the population. The standard error estimates calculated by MLILTIFAN were not reported because they were unrealistically low (Francis and Francis, 1992; Fran- cis and Mulligan, 1998). Longevity The maxmium ages based on vertebral band pair counts were 25 and 24 years for males and females, respectively. These ages likely underestimate longevity, given the long- Nalanson et a\ Age and growth of Lamnus nosus in the western North Atlantic 273 20 1 1 / CV=0.15 * y / 18- <> • /< > 16- p / — A\ S 1"- ,/ >> y < A L 01 — r.. . y ^ lOH Xn — 0) o 8- j^X — —I - a> <>y 'J Lf o> 6- A'' 4- ^^ ^ -" 2- lA^ -*-^ y n=31161779483331176222l21 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Age from reader 1 (years) Figure 4 Age bias graph for pair-wise comparison of porbeagle vertebral counts from two independent age readers. Each error bar represents the 95''>'r 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. Table 2 Von Bertalanffy growth function parameters ind 959r confidence intervals calculatec by using vertebral, tag-rec£ pture, and length- frequency methods. Confidence intervals (CI) are not shown for the MULTIFAN parameters. 0,, 0., are seasonal parameters and /, is the time (years 1 between the theoretical birthday and recruitment to th e fishery for the MULTIFAN analyses Method L,. A' 'n n ! li,., /■, Vertebral Combined CI 289.4 ±16.5 0.066 0.010 -6.06 0.71 576 Male 257.7 0.080 -5.78 283 CI ±15.6 0.015 0.92 Female 309.8 0.061 -5.90 291 CI ±26.2 0.013 0.93 MULTIFAN Combined 320.1 0.054 -5.33 13,589 0.95 0.76 0.25 Male 280.2 0.076 -4.42 7142 0.95 0.92 0.25 Female 419.0 0.039 -5.45 6269 0.95 0.89 0.25 Gulland and Holt (1959) Combined CI 212.5 ±31.7 0.172 0.089 -2.21 54 GROTAG Combined 204 0.194 -2.05 76 term fishery for this species. Taylor's (1958) method, the age at which 95% of the L.. is reached, provided a value of 26 years. However, more reahstic estimates of longev- ity take advantage of estimated mortality rates. Given M=0.10 (Campana et al.'), Hoenig's (1983) equation based on multiple species resulted in a longevity estimate of 45 years. The estimate of Campana et al.^ was species-inde- pendent and was calculated by assuming a constant instan- taneous rate of mortality = 0.10 in an unfished population. The resulting longevity estimate was 46 years. Each of these methods assumes that M is constant throughout the lifetime of a fish, whereas in fact, it probably increases in sexually mature or senescent fish. Any such increase would result in a lower estimate of longevity. Based on prelim- 274 Fishei^ Bulletin 100(2) Table 3 Log-likelihood function values and parameter estimates for four growth models fitted to porbeagle tagging data by using GROTAG (Francis 1988a). For a significant (P<0.05l improvement in fit, the inti •eduction of one extra parame- ter must increase A by at least 1.92 (Francis 1988al. ' indi- | cates fixed par ameters. Symbol (unit! Model Parameter 1 2 3 Log hkelihood A -292.04 -291.97 -285.45 Mean growth ^c^(cn-iyyi ) 19.12 17.92 19.21 rates ^i.^o'c"i/y r) 9.47 8.88 9.52 Measurement ,s ( cm 1 10.48 10.92 9.43 error in (cm) 0'^ 0.99 1.00 Outliers P 0* 0* 0.07 300, Fork length (cm) 8 8 o jj ''•""• 0 c ) 10 20 30 Vertebral age (yr) Figure 5 Porbeagle growth data based on vertebral band counts. LOESS cui-ves have been fitted to the data by sex. Open circles represent males, open triangles represent females. inary results suggesting an increase in female natural mortality rate (to 0.20) at the age of sexual maturity (Campana"), longevity would be estimated at 29 years. Discussion Age and growth studies of lamnoids have often been con- founded because of the continued debate over the peri- odicity of band-pair formation. Parker and Stott ( 1965) first Table 4 Size at age for the porbeagle iLamiia nasii.s) calculated from von Bertalanffy equations based on tag -recapture. length-frequency, and vertebral dat a. Size (cm. FL) Tag-recapture Gulland Age and Holt ( years 1 Vertebral MULTIFAN (1959) GROTAG 0 95 82 67 67 1 107 94 90 91 2 119 106 109 111 3 130 117 126 127 4 140 128 139 141 5 149 138 151 152 6 158 148 161 161 7 166 157 169 169 8 174 165 176 175 9 181 174 181 180 10 188 181 186 184 11 195 189 190 188 12 201 195 194 191 13 206 202 197 193 14 212 208 199 195 15 217 214 201 197 16 221 220 203 198 17 226 225 205 199 18 230 230 206 200 19 233 207 201 20 237 208 201 ^ Campana, S. 2001. L'npubl. data. Marine Fish Division. Bedford Institute of Oceanography, P.O. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2 suggested that two growth band pairs formed each year (biannual band-pair deposition) in their study of the bask- ing shark iCetorhuius maximiisy Pratt and Casey (1983) assumed biannual band-pair deposition for shortfin mako shark, Isiniis oxyriiichus. based on consistency with length- frequency and tag-recapture analyses. Branstetter and Musick ( 1994 ) also suggested biannual band-pair deposition for the sand tiger shark, Carcharias taiiruti, based on mar- ginal increment analysis (MIA) and examination of aquar- ium-reared sharks. Cailliet et al. (198.3, 1985) assumed annual band-pair deposition for Pacific coast shortfin mako and white sharks, Carcharodon carcharias, based on mar- ginal increment analysis. Wintner and Cliff (1999) stated that they could not determine band periodicity using mar- ginal increment analysis in the white shark off the coast of South Africa, although one OTC-injected recapture sug- gested annual deposition. With the exception of Winter and Cliff (1999), direct validation of band periodicity, such as by OTC injection or by known-age tag-recaptures, has not previously been reported in lamnids. Although several stud- ies have attempted validation with MLA (Branstetter and Musick, 1994; Wintner and Cliff, 1999), this technique is not Natanson et a\: Age and growth of Lamnus nasiis in the westein North Atlantic 275 30 - • Annualized tag-recapture growth/year (Gulland and Holt, 1959) *, o IVIodal progression * 25 - * Vertebral t 20 1 ^'- "to P 10 K * * * -•— GROTAG ♦ ♦ »t» ^ — - MULTIFAN < 5 50 100 - 150 200 250 Fork length (cm) Figure 6 Comparison of the annual growth rate of the porbeagle shark by using multiple aging methods. 200" -IJ- Apr 100* AqjuJ -pi -T 1 lll-n-i May 220 * j_ -T 110 - r -T In o- 400 - 0) it , Jun 200 - ' n rT -| 0 _^^jWy "h-H 80 - p. Jul 40 - n - -1 0 - =4lJ Th->^ 50 10 0 40 -i 0 A Sep 1 ffTHfTff^ 1 TlTT-l-,_r-, -| Oct a. Hr ill- Tkl pL. Nov t-T - -ThTh-f -l \ Dec rf 1- -|_r -yr 1 70 90 110 130 150 170 190 210 70 90 110 130 150 170 190 210 Fork lengtti (cm) Figure 7 Monthly progressions of age-0 and age-1 length-frequency modes collected by observers in the 1991 Scotian Shelf fishery. 276 Fishery Bulletin 100(2) 250 200 - 150 100 50 SW New Zealand (Francis and Stevens, 2000) Tag-recapture (GROTAG) Australia (Francis and Stevens. 2000) . Known age samples (n=6) 0 Length-frequency mode (n=1) O OTC iniected samples (n=2) 10 15 Age (years) 20 25 Figure 8 Von Bertalanffy growth cur^-es generated from vertebral data, GROTAG, and MULTIFAN seasonally oscillating parameters, as compared with the length-frequency mode, OTC recaptures, and known-age recaptures I validated i. Included for comparison are the von Bertalanffy gi'owth cur\'es of Aasen 1 19631 and Francis and Stevens (2000). well suited to slow growing species because the narrowness of the bands at the margin makes it difficult to objectively determine marginal growth. In the present study, we validated annual band-pair pe- riodicity up to age 11 using recaptures of both OTC-inject- ed and known-age porbeagles. Our data clearly indicated that the vertebral band pairs are deposited annually and that the translucent zone is deposited between November and April. Validation of an annual frequency of band-pair forma- tion confirms Aasen's (1963) interpretation of the verte- bral growth zones in the porbeagle. Our size at birth gen- erated from the vertebral bands (95 cm FL), however, was unrealistically high. This may have been due to our sam- ple of age-0 fish being biased towards the faster growing, larger fish that were recruited first into the fishery. The early portion of our growth cui've, corresponding to ages 0 and 1, may therefore have been overestimated (Fig. 8). Francis (1988b) suggested that growth cui-ves derived from age-length and length-increment data were not di- rectly comparable and that the comparison of growth rates at length was more appropriate. The growth rates at Lj^,, were similar for all methods, verifying the growth rate at this size. However, the overall growth curves from the dif- ferent methods were also similar (Fig. 8). The tag-recap- ture curve shows a more reasonable early growth than the vertebral curve but levels off well below the obsei-ved ma.x- imum size. The lower L and higher A' for the tag-recap- ture method was expected because of the different deriva- tion of the parameters and the absence of recaptured old sharks (Francis. 1988b). The K and ?y parameters derived from MULTIFAN are close to those obtained by using ver- tebral ages, and the scatter in the age readings overlaps the MULTIFAN von Bertalanffy curve (Figs. 5 and 8). Al- though the MULTIFAN L . value was slightly higher than that of the vertebral value (Table 2), a difference at the upper end of these curves was not unexpected because length-frequency models are generally considered unreli- able for the older age classes where the modes are not easily defined (Francis and Francis, 1992; Francis, 1997; Francis et al., 1999). Maturity occurs in the porbeagle at 8 and 13 years of age (174 and 218 cm FL, males and females, respectively; Jensen et al.^i. Growth for both sexes is similar up to the size of male maturity, whereupon, the male growth rate is reduced. Females continue to grow rapidly until the on- set of maturity, at which point their growth slows as well (Fig. 5). Owing to this change in growth rate, males reach a smaller maxiinum size than females; however, the over- all growth rate for both sexes is not substantially differ- ent. Additionally, the vertebral L generated for females is higher than what is being observed in the fishery, suggest- ing that it has been overestimated and that the combined curve is more appropriate. Aasen" (1963) also found no dif- ference in the growth rate between the sexes although the basis for his conclusion is questionable. Aasen, O. 1961. Some observations on the biology of the por- beagle shark iLamna nasus, L). ICES, CM. Copenhagen 1961, Near Northern Seas Committee (109):l-7. Natanson et a\- Age and growth of Lamnus nasus in the western North Atlantic 277 Aasen (1963) relied extensively on length-frequency modes to estimate the growth of porbeagle. Although his modes were similar to ours, his interpretation of the age-1 mode differed. Our data indicated that age-0 porbeagle average 85 cm FL in July. Aasen (1963) interpreted this same mode (91 cm FL) as age 1+, thus shifting his ages by one year. Neither Aasen's ( 1963) modal distribution (his Fig. 4) nor ours, supports the contention that this first mode is age 1+. His classification of these fish as age 1+ was based on size at birth and his opinion that the smallest measured fish were from the age-0 group. There is, however, no mode at this small size. Any age-0 fish born in April and caught between July and September I his sampling period) would certainly be larger than the birth size. Therefore, we feel that these fish represented the faster growing age-0 fish that were large enough to be caught with commercial longline gear. Francis and Ste- vens (2000) also used length-frequency analysis to esti- mate the growth rate of porbeagles in the South Pacific. Although their modes were once again comparable to ours, their age l-i- fish were similar in size to our age-O-i- fish (Fig. 8). In their view, this first mode represented slow growing age-1 individuals rather than fast growing age-0 individuals. As an alternative explanation, we suggest that the first mode in both the southwest and northeast New Zealand samples represents YOY that have grown during the 4-month sampling period, thus accounting for the apparent absence of individuals close to a birth size. This alternative explanation would also explain why the modal analysis of the Australian sample shows a strong peak at birth and subsequent modes that are similar to ours. If correct, our interpretation of the Francis and Ste- vens (2000) data would bring their estimates of size at age and gi-owth rate in line with ours. Of course, the compari- son of growth rates from such widely separated stocks is difficult, and its value questionable. Longevity estimates for the porbeagle indicate that they may live for more than 40 years. The maximum time at lib- erty for any tagged porbeagle is 13 years (Stevens, 1990). The length of this shark when tagged was approximately 120 cm TL (107 cm FL; l-i- years); it was recaptured at an estimated 225 cm FL (age 14-1-) which would correspond to an age of 16-i- years according to our vertebral growth cui've. This is substantially less than the oldest observed age from vertebrae (25) and the estimates from the Hoenig ( 1983 ) and Campana et al." methods of 45 and 46 years, respectively. The growth rate and longevity of the porbeagle are simi- lar to those of other lamnids. Wintner and Cliff ( 1999) cal- culated a K value of 0.065 for the white shark from the east coast of South Africa, and Cailliet et al. (1985) esti- mated a if value of 0.058 for the same species off the coast of California. Both estimates are very similar to the K value of 0.066 calculated for porbeagle in our study. Short- fin mako K values have been estimated at 0.072 (Cailliet et al, 1983) and 0.266 (Pratt and Casey 1983); however the Pratt and Casey value was based on the assumption that two band pairs were deposited annually. Longevity estimates have ranged between 27 years for the Califor- nia white shark (Cailliet et al., 1985) and 45 years for the shortfin mako in the Pacific (Cailliet et al., 1983). Comprehensive age and growth studies of pelagic sharks are difficult to implement because many species are highly migi-atory and are caught sporadically as part of seasonal fisheries. Thus, aging studies of pelagic sharks have usu- ally been less rigorous than desired, despite the oft-repeat- ed call for age validation (Beamish and McFarlane, 1983; Cailliet et al., 1986, Cailliet, 1990). Previous studies on pe- lagic species such as the blue shark (Stevens, 1975; Cail- liet etal., 1983), white shark (Cailliet etal., 1985), thresher (Cailliet et al., 1983), .shortfin mako (Cailliet et al, 1983), pelagic thresher (Liu et al., 1999), oceanic whitetip shark (Lessa et al., 1999), and porbeagle (Aasen, 1963; Francis and Stevens, 2000), have included analyses of ages deter- mined by vertebral or length-frequency analyses (or by both methods), but none of the age interpretations were validated. Wintner and Cliff (1999) used vertebral counts and had one OTC-injected recapture but were unable to provide validation or consistency with other methods. Pratt and Casey (1983) aged the shortfin mako by using four methods (temporal analysis of length-month information, tag-recapture data, length-frequency data, and vertebral band counts) but could not validate their age interpreta- tions. The conclusion that band pairs were deposited bian- nually was based on vertebrae from four tag-recaptures and consistency between methods. Skomal ( 1986) aged the blue shark in the western North Atlantic using a combina- tion of vertebral, length frequency and tag-recapture meth- ods for verification. Although Skomal ( 1986) had two OTC- injected recaptures, they provided conflicting results for validation. The present study is the first that has used validated vertebral band-pair counts in conjunction with length-frequency and tag-recapture analyses to provide consistent and accurate age estimates for a pelagic shark species. We suggest that a similar approach would be use- ful in studies of other pelagic shark species. Acknowledgments We thank Clearwater Fine Foods, Karlsen Shipping, the Atlantic Shark Association, and Stephanie Jane, Inc. for providing access to their fishing vessels. We also thank Andy Kingman, Christopher Jensen, and Warren Joyce for collecting samples. Malcolm Francis kindly provided verte- brae from porbeagle embryos as well as much appreciated knowledge on the procedures associated with MULTIFAN. Colin Simpfendorfer's assistance and spreadsheet were invaluable during the use of GROTAG. Nancy Kohler and Sabine Wintner provided invaluable comments on the manuscript. We are indebted to the thousands of fisher- men who voluntarily tag and return tags to us and thus make tagging programs possible. Literature cited Aasen, O. 1963. Length and growth of the porbeagle (Lamna nasus, Bonneterre) in the North West Atlantic. Fisk. Skrift. Ser. Havund. 13(6):20-37. 278 Fishery Bulletin 100(2) Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fish- eries biology. Trans. Am. Fish. Soc. 1 12:735-743. Branstetter, S. and J. A. Musick. 1994. Age and growth estimates for the sand tiger in the north- western Atlantic Ocean. Trans. Am. Fish. Soc. 123:242- 254. CailHet, G.M. 1990. Elasmobranch age determination and verification: an updated review. In Elasmobranchs as hving resources: advances in the biology, ecology, systematics. and status of the fsheries (H. L. Pratt Jr., S. H. Gruber. and T. Tanmchi. 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 gi-owth of blue, Prionace glauca. common thresher, Alopias rulpinus, and shortfin mako, Isiirus oxynnchus, sharks from California waters. In Proceedings of the international workshop on age determination of oceanic pelagic fishes: tunas, bill- fishes, and sharks (E. D. Prince and L. M. Pulos, eds.), p. 179-188. U.S. Dep. Commer,Tech. Rep. NMFS 8. Cailliet, G. M., L. J. Natanson, B. A. Welden, and D. A. Ebert. 1985. Preliminary studies on the age and growth of the white shark, Carcharudon carcharias, using vertebral bands. Mem. S. Calif Acad. Sci. 9:49-60. Cailliet, G. M., R. L. Radtke. and B. A. Welden. 1986. Elasmobranch age determination and verification: a review. In Indo-Pacific fish biologv" proceedings of the second internation conference on Indo-Pacific fishes (T Uyeno, R. Arai, T Taniuchi and K. Matsuura. eds. >, p. 345- 360, Ichthyol. Soc. Japan, Tokyo. Campana, S. E., M. C. Annand, and J. I. McMillan. 1995. Graphical and statistical methods for determining the consistency of age determinations. Trans. Am. Fish. Soc. 124:131-1,38. Castro, J. I. 1983. The sharks of North American waters. Texas A&M Univ Press, College Station, TX, 180 p. Compagno, L. J. V. 1984. FAO species catalogue. Sharks of the world: an anno- tated and illustrated catalogue of shark species known to date. Part 1: Hexanchiformes to Lamniformes. FAO Fish Synop. 125, vol. 4, 250 p. Fournier, D. A., J. R. Sibert, J. Majkowski, and J. Hampton. 1990. MULTIFAN a likehhood-based method for estimat- ing gi'owth parameters and age composition from multiple length frequency data sets illustrated using data for south- ern bluefin tuna iTIuinnus niaccoyuK Can. J. Fish. Aquat. Sci. 47:301-317. Francis, M. P. 1997. Spatial and temporal variation in the growth rate of elephantfish (Callorhinchufi milii). NZ J. Mar. Freshwa- ter Res. 31:9-23. Francis, M. P., and R. I. C. C. Francis. 1992. Growth rate estimates for New Zealand rig iMuatc- [lis lenticiilatiis). Aust. J. Mar. Freshwater Res. 43:1157- 1176. Francis, M. P., and K. P. Mulligan. 1998. Age and growth of New Zealand school shark, Galeo- rhinusgaleus. NZ J. Mar. Freshwater Res. 32:427-440. Francis, M. P., K. P. Mulligan, N. M. Davies, and M. P. Beentjes. 1999. Age and growth estimates for New Zealand hapuku. Potyprion oxygeneios. Fish. Bull. 97:227-242. Francis, M. P. and J. D. Stevens. 2000. Reproduction, embryonic development, and growth of the porbeagle shark, La/(i;ia nasiis, in the southwest Pacific Ocean. Fish. Bull. 98:41-63. 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 gi'owth parameters estimated from tagging and age-length data comparable? Can. J. Fish. Aquat. Sci. 45:936-942. GuUand, 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. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mortality rates. Fish. Bull. 81:898-903. Kohler, N. E., J. G. Casey and P. A. Turner 1995. Length-weight relationships for 13 species of sharks from the western North Atlantic. Fish. Bull. 93(21:412- 418. Lessa, R., F. M. Santana, and R. Paglerani. 1999. Age, growth and stock structure of the oceanic whitetip shark, Carcharhinus longimanus. from the southwestern equatorial Atlantic. Fish. Res. 42:21-30. Liu, K. M., C. T Chen, T H. Liao, and S. J. Joung. 1999. Age, gi'owth and reproduction of the pelagic thresher shark, Alopias pelagictis in the Northwestern Pacific. Copeia 1999(11:68-74. Media Cybernetics. 1998. Image Pro 4 software. Media Cybernetics, Silver Spring, MD. Parker, H. W. and F. C. Stott. 1965. Age, size and vertebral calcification in the basking shark, Cetorhinus maximus (Gunnerus). Zool. Meded. 40134): 305-319. Pratt, H. L., Jr, and J. G. Casey 1983. Age and gi-owth of the shortfin mako. Imiriis oxyrin- chiis, using four methods. Can. J. Fish. Aquat. Sci. 40(11): 1944-1957. Skomal, G. 1986. Age and growth of the blue shark, Prionace glauca, in the North Atlantic. Master's thesis, LTniv. Rhode Island, Kingston, RI, 82 p. Statgraphics. 1997. Statgraphics Plus, version 3. Manguistics.Inc, Rock- ville, MD. Stevens, J. D. 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. 1990. Further results from a tagging study of pelagic sharks in the north-east Atlantic. J. Mar Biol. Assoc. U.K. 70: 707-720. Taylor, C. C. 1958. Cod growth and temperature. J. Cons. Int. Explor Mer 23:366-370, von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquiries on growth laws II). Hum. Biol. 10:181-213. Wintner, S. P, and G. Cliff 1999. Age and growth determination of the white shark, Car- charodon carcharias. from the east coast of South Africa. Fish. Bull. 97(1 ):153-169. 279 Abstract— An ecosystem approach to tisluTR's management requires an un- derstanding of the impact of predatory fishes on the underlying prey resources. Defining trophic connections and mea- suring rates of food consumption by apex predators lays the gi-oundwork for gaining insight into the role of pred- ators and commercial fisheries in influ- encing food web structure and ecosys- tem dynamics. We analyzed the stomach contents of 545 common dolphinfish iCoryphaena hippurus) sampled from 74 sets of tuna purse-seine vessels fish- ing in the eastern Pacific Ocean lEPO) over a 22-month period. Stomach full- ness of these dolphinfish and digestion state of the prey indicated that diel feeding periodicity varied by area and may be related to the digestibility and energy content of the prey. Common dol- phinfish in the EPO appear to feed at night, as well as during the da.vtime. We analyzed prey importance by weight, numbers, and frequency of occurrence for five regions of the EPO. Prey impor- tance varied by area. FUnngfishes. epi- pelagic cephalopods. tetraodontiform fishes, several mesopelagic fishes. Aiixis spp., and gempylid fishes predominated in the diet. Ratios of prey length to pred- ator length ranged from 0.014 to 0.720. Consumption-rate estimates averaged 5.6''c of body weight per day. Stratified by sex. area, and length class, daily rations ranged up to 9.6'< for large males and up to 19.8'5 for small dolphinfish in the east area (0-15°N. lll°W-coastline). Because common dolphinfish exert sub- stantial predation pressure on several important prey groups, we concluded that their feeding ecology provides im- portant clues to the pelagic food web and ecosystem structure in the EPO. Food habits and consumption rates of common dolphinfish (Coryphaena hippurus) in the eastern Pacific Ocean Robert J. Olson Inter American Tropical Tuna Commission 8604 La Jolla Shores Dnve La Jolla, Calilornia 92037-1508 E-mail address rolsoniaiattcorg Felipe Galvan-Magana Centre Interdisciplinario de Ciencias Mannas Institute Politecnico Nacional Apdo Postal 592 La Paz, B,C S , Mexico Manuscript accepted 3 October 2001. Fish. Bull. 279-298 (2002). Dolphinfishes ^Coryphaena hippurus and C. equiselis) are abundant, wide- ranging, epipelagic predators in tropi- cal and subtropical oceans (Palko et al., 1982 ). They support important commer- cial, artisanal, and recreational fisher- ies in several regions (Beardsley, 1967: Oxenford and Hunte. 1986; Patterson and Martinez. 1991; Campos et al.. 1993; Norton and Crooke. 1994; Lasso and Zapata. 1999). Dolphinfishes are also a large component of the bycatches of the tuna purse-seine and longline fisheries in the Pacific Ocean (Lawson. 1997; lATTC. 1999 ). They are commonly found near natural and artificial float- ing objects (Kojima, 1956; Hunter and Mitchell. 1966; Gooding and Magnu- son, 1967; Wickliam et al., 1973 ). a trait which facilitates their capture. Calls have been issued for developing an ecological approach to fisheries man- agement, taking greater note of species interactions and underlying ecosystem dynamics (FAO, 1995: Larkin, 1996; Mangel et al., 1996: Botsford et al.. 1997). Removal of predator biomass by commercial fishing represents a "top- down" disturbance of the system. Selec- tive exploitation of apex predators can have profound effects on pelagic ecosys- tems because of the removal of preda- tion pressure ( Essington et al.. in press ) and because of top-down, trophic-cas- cade effects (Shiomoto et al.. 1997; Es- tes et al.. 1998; Verheye and Richard- son, 1998). An understanding of how top-down processes infiuence the dy- namics of marine communities derives from a basic understanding of the tro- phic connections and rates of food con- sumption of the predators. Although four studies have provided limited da- ta on the food habits of dolphinfishes in coastal areas of the eastern Pacific Ocean (EPO) (Hida, 1973; Campos et al.. 1993; Aguilar-Palomino et al., 1998; Lasso and Zapata. 1999 ). little is known of the predation dynamics of dolphin- fishes over the majority of their oceanic habitat. Common dolphinfish (C. hippurus) are renowned for their rapid rates of growth and metabolism. In Hawaiian waters, common dolphinfish have attained av- erage lengths of 120 cm and weights of 12.5 kg at 12 months of age (Uchi- yama et al., 1986). Standard metabolic rates of common dolphinfish are compa- rable to those of yellowfin (Thunnus al- bacares) and skipjack iKatsuwonus pe- kunis) tunas (Benetti et al.. 1995). All three species have large surface areas and thin blood-water interfaces in their gills, morphological features that permit high oxygen diffusion capacity and ele- vated metabolic rates (Brill, 1996). High energy requirements imply that preda- tors like dolphinfish can account for im- poi1;ant amounts of tertiary production removed from an ecosystem (Essington et al., in press), but rates of food con- sumption by dolphinfish in nature have not been measured. The objectives of our study were to define the trophic relations of the com- 280 Fishery Bulletin 100(2) mon dolphinfish, and their prey in both coastal and oceanic areas of the EPO and to provide prehminary estimates of their daily rates of food consumption. Materials and methods 30 N 20 10 0 10 20 O Floating-object sets a Sctiooifisti sets X Dolphin sets North 140 The dolphinfish were caught by tuna purse- seine vessels of Colombian, Mexican, Pan- amanian, and Venezuelan registry from December 1992 through September 1994. The fish were caught as bycatch of the purse-seine fishery for tunas associated with dolphins, with floating objects, and as unassociated schools ("schoolfish"). In dol- phin sets, the net is deployed around aggre- gations of primarily yellowfin tuna and spotted or spinner dolphins {Stenella atten- uata or S. longii-ost/'ig) (or both dolphin spe- cies) after a high-speed chase by speedboats. Floating-object sets are made by encircling flotsam, commonly tree parts and artificial fish-aggi-egating devices (FADs), and associ- ated fauna with the purse seine, usually in the early morning. Schoolfish are detected by seabird activity and disturbance of the water surface caused by the fish swimming just below. The species com- position and size and age distribution of the fauna are distinctly different for the three aggregation types and fishing strategies (Hall, 1998). Stomach samples Common dolphinfish stomach samples were taken at sea by observers of the Inter-American Tropical Tuna Com- mission (lATTC). The purse-seine sets yielding the dol- phinfish samples were distributed across the geographical range of the EPO tuna fishery at that time (Fig. 1). We obtained samples from 74 purse-seine sets over a 22-month period: 61 sets (82'7f) were made on floating objects; 4 sets were made on dolphins; and 9 sets were made on unassociated tuna. On board the vessels, the observers measured the fork length (mm) of each dol- phinfish, determined the sex if possible, and excised and immediately froze the stomachs. In the laboratory, we thawed the stomachs and visually estimated the stomach fullness as a percentage of the stomach capacity. Then, we identified the stomach contents to the lowest taxon pos- sible, weighed them to the nearest gram, and enumerated them when individuals were recognizable. The counts of paired structures, such as cephalopod mandibles and fish otoliths, were divided by two to estimate numbers of prey. We categorized the digestion state of the prey: 1 = intact or nearly intact; 2 = soft parts partially digested; 3 = whole or nearly whole skeletons without flesh (or comparable state for nonfish taxa); and 4 = only hard parts remaining (primarily fish otoliths and ceph- alopod mandibles). We measured the length, or maxi- mum dimension of individual prey to the nearest mm. West 130 120 Souttiwest 80 W Figure 1 Locations where common dolphinfi.sh samples were caught by three types of purse-seine sets. We stratified the data into the five areas shown. if sufficiently intact. For cephalopods, we recorded the mantle length excluding tentacles. Identifying the prey depended on the digestion state of the remains. We used the following keys to identify fish prey in digestion state 1: Jordan and Evermann (1896), Meek and Hildebrand ( 1923), Parin ( 1961 ), Miller and Lea (1972), Thomson et al. ( 1979), Allen and Robertson ( 1994), and Fischer et al. ( 199.5b and 1995c). When the fishes were digested to state 2 or 3 we used taxonomic keys of ver- tebral characteristics (e.g. Clothier, 1950; Monod, 1968; Miller and Jorgenson, 1973) and compared skeletons of whole fishes collected in the EPO. We identified the crus- tacean prey from exoskeleton remains using the keys of Garth and Stephenson (1966), Brusca (1980), and Fischer et al. (1995a). We identified cephalopod prey from man- dible remains (Clarke, 1962; Iverson and Pinkas, 1971; Wolff 1982; Clarke, 1986). The fish collections at Scripps Institution of Oceanography and the Natural History Mu- seum of Los Angeles County, and the cephalopod collection at the Santa Barbara Museum of Natural History were used to compare and validate prey identifications. Data analysis We analyzed the diet data by calculating three diet indices for each prey taxon. We calculated gravimetric impor- tance of the prey {9fW) as percentages of the total prey weights, numerical importance (9(N) as percentages of total counts, and frequency of occurrence as the number of dolphinfish stomachs that contained a particular prey. We calculated percent occurrence {"tO) as a percentage of all the dolphinfish sampled, regardless of whether their stomachs contained food. We present these three indices by prey taxon in detailed tables, summarized at several Olson and Galvan Magaha: Food habits and consumption rates of Co/yphaena hippurus 281 levels of taxonnmic resolution. To facilitate analysis, we also grouped the prey laxa by order (e.g. Tetra- odontiformes), family (e.g. Carangidae), genus (e.g. Auxis spp. ), or functional group (e.g. flyingfishes). Because the three diet indices provide different in- sights into predation habits, we applied a graphical representation of these measures, proposed by Cortes (1997), to help interpret the data. We made three-di- mensional scatter plots of '/i O, "^ W, and '7(N for all samples and for the data pooled by sampling area to help evaluate the degree of dominance of particular prey and the feeding strategy (generalized vs. special- ized) of the dolphinfish. Although we measured the three components of the index of relative importance (IRI) (Pinkas et al., 1971), we did not calculate IRI values because the index is dependent upon the taxo- nomic resolution of the prey (Hansson, 1998). Also, for a predator that consumes a large size range of prey (see heading "Prey size," below), the IRI is overly in- fluenced by numerous small prey. We examined diel feeding characteristics by stratify- ing the data according to stomach fullness of the pred- ator and digestion state of the prey. The scheme for grouping the data, patterned after Calliet ( 1976 ), is dia- grammed in Figure 2. Prey in digestion states 1 and 2 were categorized as from "recent" feeding events, whereas prey in states 3 and 4 were categorized as from "previous" feed- ings. These two strata were further subdivided according to stomach fullness. Prey from stomachs <50'7f full were cate- gorized as "low" fullness or empty, whereas those from stom- achs >50'7f full were categorized as "high" fullness (Fig. 2). We plotted the percent occurrence of the prey items in these tour digestion and fullness strata by area and the time of day the sets were made: "early morning" (05:12-09:00), "late morning"(09:01-12:00), "early afternoon"( 12:01-15:00), and "late afternoon" (15:01-18: 16 hours). We fitted regression trees (Breiman et al., 1984) to the gravimetric data for each prey group to detect statisti- cally important differences by area and dolphinfish size. Regression trees are well suited for detecting and extract- ing important relations and complex interactions in multi- variate ecological (Death and Fabricius, 2000) and fisher- ies data (Walters and Deriso, 20001. We used a two-step process. For both steps, the %W of each prey group in the stomach contents was the response variable. For the first step, defining area strata (see next paragraph), we used latitude and longitude as the predictor variables. For the second step, modeling the importance of area and dol- phinfish size in explaining variation in the %W for each prey group, we used area designations (north, west, east, southwest, and southeast) and fork length as the predictor variables. We used the tree functions in S-Plus (MathSoft Inc., 1999) and cross-validation to prune fully grown trees so that only important splits remained. Prediction errors were used as pruning criteria (Breiman et al., 1984; De'ath and Fabricius, 2000). We stratified the data by area (Fig. 1) according to two criteria. Latitude divisions at 15°N and 0° were based on the spatial and seasonal heterogeneity of the purse-seine sets that provided the samples. All the sets sampled from 3&4 1 &2 Previous low and empty Previous high Recent low Recent high 0-50° 51-100° Stomach (ullness Figure 2 Schematic diagram showing prey digestion-state and predator stomach-fullness criteria for four categories used to analyze diel feeding characteristics of common dolphinfish in the eastern Pacific Ocean. May through November each year were made north of the equator, and most sets sampled during December through April were made south of the equator. Also, the regression trees indicated that latitude and longitude were important in explaining the variability in the gravimetric data for several prey taxa. Epipelagic-cephalopod taxa were most important in the diet of the common dolphinfish caught east of 82°41'W and south of 1° 46'S, and flyingfishes were most important in the diet of those caught west of 81°W. Therefore, we stratified the data from south of the equa- tor into "southwest" and "southeast" areas separated at 82°41'W (Fig. 1). Similarly, we stratified the data from samples collected between 0° and 15°N into "west" and "east" areas divided at 111°W because a regression tree fit- ted to the gravimetric data for the Tetraodontiformes indi- cated that this meridian was important in explaining vari- ation in '7(W for this taxon. Tetraodontiformes were more important in the diet of the fish caught west of 111°W. Consumption rates We employed a method described by Olson and Mullen (1986) to calculate preliminary estimates of daily rates of food consumption by common dolphinfish. The model pre- dicts feeding rate ( r, grams per hour) by dividing the mean weight of the stomach contents per predator (W, grams) by the integral (A. proportion x hours=hours) of the func- tion that best fits experimental gastric evacuation data. For a predator that consumes a variety of prey that are evacuated at different rates. ,=0 ^ where subscripts / refer to each of/ prey types. (1) 282 Fishery Bulletin 100(2) Table 1 Numbers of common dolphinfish sampled by size, sex, and area with undigested food in empty stomachs or trace amounts of hard parts (in parentheses). their stom achs (first nurr ber) and with Fork length (mm) Sex Area Total no. of dolphinfish Males Females Undetermined sex North West East Southwest Southeast 417-650 2(14) 14(39) 4(5) 0(11 0(3) 4(24) 16(21) 0 (9) 20 (58) 651-800 12(211 48(47) 8(3) 1(0) 2(91 14(11) 37(47) 14 (4) 68 (71) 801-950 24(29) 51(70) 11(7) 11 (Hi 13(13) 43(60) 19 (22) 86 (106) 951-1100 26(25) 15(21) 3(1) 0(2) 10(101 19(19) 15 (16) 44 (47) 1101-1250 9(6) 10(6) 2(11 2(0) 10(51 5(11 4 (7) 21 (13) 1251-1770 6(1) 1(1) 0(11 1 (1) 2(1) 4 (1) 7 (3) Total 79(96) 139(1841 29(18') 3(1) 13(25) 52(64) 122(150' 56 (59) 246 (299') ' Including one dolpliinfish with no length or sex data. Daily meal is /• multiplied by 24 h for fish that feed both day and night. Daily ration is daily meal expressed as a percent of body weight. We estimated the body weight of each dolphinfish from the length, according to the rela- tionships of Lasso and Zapata (1999): where M ■■ L ■- M = aL\ body weight (g); and fork length (cm). (2) They estimated that a = 0.0406. 0.0420, and 0.0224, and b = 2.6588, 2.6328. and 2.78 for males, females, and common dolphinfish of undetermined sex, respectively. We esti- mated daily consumption rates for dolphinfish of six size strata by sex and area. The time-course of gastric evacuation has not been ad- equately described for dolphinfishes. This fact prevents a rigorous analysis of consumption rates using stomach-con- tents data. However, we provide preliminary, first-order es- timates of daily rations of common dolphinfish because this information is important for analyses of ecosystem effects of fishing (see "Discussion" section). A preliminary experi- ment indicated that the gastric evacuation rate for squid tissue by juvenile dolphinfish is comparable to that for squid by yellowfin tuna. Five juvenile common dolphinfish passed pellets of moist squid tissue through the digestive tract in 6-8 h at 27"C (Suzuki. 1992). The fastest gastric evacuation times for squid (Loligo opalescens) voluntarily ingested by yellowfin tuna (mean L=36.2 cm) in the labo- ratory were about 8 h at 23.5-25.5"C (Olson and Boggs. 1986). We assume, therefore, that gastric evacuation rates measured for yellowfin tuna are adequate for estimating daily rations of dolphinfish. We assigned values of A (Eq. 1) for squid (4.48), mackerel iScomber japonicus) (5.29). smelt (Hypomesus pretiosus) (4.12). and nehu iStolepho- rits purpureus) (2.24). and the mean for four experimental food species (3.77) determined by Olson and Boggs (1986) to the various prey taxa of common dolphinfish, as they did for yellowfin tuna. We omitted the data for the trace hard parts (cephalopod mandibles and fish otoliths), which apparently accumulate in the stomachs, when calculating consumption rates because of the possibility that these re- mained from predation on previous days. Results Stomach samples were obtained from 545 dolphinfish: 175 males, 323 females, and 47 of undetermined sex. Two hundred and forty-six specimens had fresh or partially digested food remains in their stomachs. 274 had empty stomachs, and 25 had only trace amounts of digestion- resistant hard parts (cephalopod mandibles and fish oto- liths). We present the sample sizes by sex and area in Table 1 and the detailed prey-composition data by area in Tables 2-4. We analyzed the prey composition data by se.x but did not discover important differences or trends. Few samples were obtained in the north and west areas. Therefore, we briefly summarize those data here and in Tables 2-4 and do not include them in the detailed treat- ments of diel feeding periodicity, diet measures by area, and size-specific predation. Only four common dolphinfish were sampled in the north area, three from sets on schoolf- ish and one from a set on dolphins. Three of the stomach samples contained food and one was empty A large dol- phinfish ( 1149 mm) had recently eaten a large squid iSthe- noteuthis oualaniensis) in the early morning. Another large fish (1239 mm) had a full stomach containing 7 fresh Cory- phaena equiselis in the late afternoon. Both fish had been collected from schoolfish sets. A smaller dolphinfish (768 mm) from a dolphin set in the late afternoon had remains of various taxa. principally C. equiselis, flyingfishes. and galatheid red crabs {Pleuruncodes plampes) in advanced states of digestion. In addition, a 564-mm dolphinfish from a schoolfish set was found to have an empty stomach. Thirty-eight common dolphinfish were sampled in the west area from 4 sets on floating objects between 05:40 and Olson and Galvan Magana Food habits and consumption rates of Coryphaena hippurus 283 Table 2 Taxonomic composition in 'rVVof the p rev of common dolphinf sh 1 1(1 m five areas of the eastern Pacific Ocean (Fig. 1) and from all | aroas combined. Weights do not include hard parts i cephalopod mar dibl es and fi sh otoliths . Prey category codes are as follows: EC = | epipelagic cephalopods, MC = mesope agic cephalopods, Cr = Crustacea, MiF = miscellaneous fishes, MsF = mesopelagic fishes, F = flyingfishes, Co = Coryphaenidae, Ca = Carangidae, G = Gempylidae, A = Auxis spp., YT = yellowfin tuna, N = Nomeidae, T = Tetraodontiformes. A, is the integral of the function fitted to experimental gastric evacuation data and is used to calculate | consumption rates (Eq. 1). Area codes are N = north W = west E = east, SW = southwest. SE = southeast. Assumed Area Taxon Category A, N W E SW SE All Phylum MoUusca 4.48 16.35 2.54 9.48 21.69 82.73 32.20 Class Cephalopoda 4.48 16.35 2.54 9.48 21.69 82.73 32.19 Order Teuthoidea 4.48 16.35 2.54 9.48 21.66 81.84 31.91 Family Enoploteuthidae EC 4.48 * * * * Abraliopsis falco EC 4.48 * * * * Family Mastigoteuthidae MC 4.48 0.15 0.01 0.04 MaMigoteuthis spp. MC 4.48 0.15 0.01 0,04 Family Ommastrephidae EC 4.48 16.35 2..54 5.47 21.40 81.84 30.72 Dnsidicus gigas EC 4.48 2.54 0.78 5.08 70.73 19.27 Sthenoteuthis oualaniensis EC 4.48 16.35 4.69 16.. 32 11.11 11.45 Family Onychoteuthidae EC 4.48 3.26 0.09 0.91 Onychoteuthis banksii EC 4.48 3.26 0.09 0.91 Onychoteuthis spp. EC 4.48 * :i: Family Pholidoteuthidae EC 4.48 * * Ph olidotcu th IS bosch man i EC 4.48 * * Family Thysanoteuthidae EC 4.48 * 0.17 0.07 Thysanoteuthis rhombus EC 4.48 * 0.17 0.07 Order Octopoda EC 4.48 * * 0.02 1.16 0.29 Family Tremoctopodidae EC 4.48 1.16 0.28 Tremoctopus violaceus EC 4.48 1.16 0.28 Family Argonautidae EC 4.48 * * 0.02 * 0,01 Argonauta spp. EC 4.48 * * 0.02 * 0.01 Family Bolitaenidae MC 4.48 * * * Japetella diaphana MC 4.48 * * * Phylum Anhropoda 3.37 1.39 3.26 0.01 0.94 Class Crustacea Cr 3.37 1.39 3.26 0.01 0.94 Order Decapoda Cr 3.37 1.39 3.26 0.01 0.94 Family Galatheidae Cr 3.37 1.39 0.06 Pleuroncodes planipes Cr 3.37 1.39 0.06 Family Penaeidae Cr 3.37 0.61 0.16 Family Portunidae Cr 3.37 2.65 0,71 Portunus xantusii Cr 3.37 2.65 0.71 Phylum Chordata 4.12 82.26 97.46 87.24 78.30 17.00 66.84 Class Osteichthyes 4.12 82.26 97.46 87.24 78.30 17.00 66.84 Order Clupeiformes MiF 4.12 0.08 0.02 Family Engraulidae MiF 4.12 0.08 0.02 Order Stomiiformes MsF 2.24 0.48 4.53 2.01 Family Phosichthyidae MsF 2.24 0.48 4.53 2.01 Vinciguerria lucetia MsF 2.24 0.48 4.53 2.01 Order Myctophiformes MsF 2.24 0.07 4.26 3.11 ^- 0.97 Family Myctophidae MsF 2.24 0.07 4.26 3.11 * 0.97 Benthosema panamense MsF 2.24 * * continued 284 Fishery Bulletin 100(2) Table 2 (continued) Area Assumed Taxon Category A, N W E SW SE All Myctophuin aurolatcrnatum MsF 2.24 0.97 0.26 Myctophum nitidulum MsF 2.24 * ^ Symbolophonis spp. MsF 2.24 4.26 2.14 * 0.71 Order Lampridiformes MsF 5.29 3.04 1.27 Family Trachipteridae MsF 5.29 3.04 1.27 Desmodesma polysticturn MsF 5.29 3.04 1.27 Order Beloniformes 1.59 26.50 49.58 31.64 9.29 29.58 Family Belonidae MiF 5.29 0.26 1.34 0.39 Strongylura spp. MiF 5.29 0.26 1.34 0.39 Family Exocoetidae F 4.12 16.14 30.77 28.43 7.95 22.48 Cheilopogon furcatus F 4.12 0.66 0.27 Cheilopogon spilonotopterus F 4.12 0.12 0.05 Cheilopogon spp. F 4.12 10.96 0.53 3.59 2.94 2.68 Exocoetus monocirrhus F 4.12 19.19 0.69 1.09 5,69 Exocoetus volitans F 4.12 5.18 5.34 17.64 1..58 9.31 Exocoetus spp. F 4.12 0.92 5.04 1.81 2.77 Hirundichthys speculiger F 4.12 3.17 0.53 0.98 Prognichthys spp. F 4.12 1.22 0.47 0.52 Family Hemiramphidae F 4.12 1..59 10.36 18.55 3.21 6.71 Oxyporhainphus mtcroptenis F 4.12 1.59 10.36 18.55 3.21 6.71 Order Perciformes 80,11 32.60 36.32 5.76 28.90 Family Echeneidae MiF 3.77 2.48 1.03 Rhoin boch irus osteoch ir MiF 3.77 2.48 1.03 Family Corvphaenidae Co 5.29 80.11 1.43 2.36 0.20 5.08 Coryphaena equiselis Co 5.29 80.11 0.01 3.67 Coryphaena hippurus Co 5.29 1.23 0.51 Family Carangidae Ca 5.29 0.34 5.48 2.37 Naucrates ductor Ca 5.29 0.34 3.88 1,71 Family Gempylidae G 5.29 16.53 3.93 0.01 6,07 GeiJjpyluti serpens G 5.29 16.53 3.93 0.01 6,07 Family Scombridae MiF 5.29 13.18 15.97 5.54 11,50 Acanthocybium solandri MiF 5.29 0.04 0.01 Auxis spp. A 5.29 10.15 15.78 9.29 Euthynnus Imealus MiF 5.29 1.50 0.40 Thunnns spp. YT 5.29 1.49 0.19 5.54 1.80 Family Nomeidae N 2.24 1.12 6.11 2.84 Cubiceps pauciradiatus N 2.24 1.12 6.11 2.84 Order Tetraodontiformes T 3.77 0.49 66.70 1.38 2.77 3.63 Family Balistidae T 3.77 0.21 38-78 1.22 Xantichthys mento T 3.77 38.78 1.21 Family Ostraciidae T 3.77 25.99 0.81 Lactorin diaphanum T 3.77 25.99 0.81 Family Tetraodontidae T 3.77 0.28 1.93 1.38 2.77 1.60 Lagocephalus lagocephatus T 3.77 0.28 1.93 1.38 2.77 1.60 Unidentified fishes MiF 3.77 1.95 0.47 Total prey weight (gl 1443 985 8459 13,118 7523 31,527 * Only trace quantities of hard parts were present Olson and Galvan-Magana: Food habits and consumption rates of Coiyphaena hippurus 285 Table 3 Taxononiic composition in '",N of tho proy ol" common (iolphiiifish f'ron five areas of the eastern Pacific Ocean (Fig. 1) and from all areas conibinod. Numbers include hard p; irts icephal opod mandibles ind fish otoliths) divided by two. N jmbers are not complete for some taxa because sometimes prey could not be enumerated due to being in an advanced digestion state. Area codes are N = | nortli, W = west, E = east, SW = southwest, SE = southeast. Taxon Area N W E SW SE All Phylum Mollusca 2.17 12.36 24.46 19.97 87.35 37.43 Class Cephalopoda 2.17 12.36 24.46 19.97 87.35 37.37 Order Teuthoidea 2.17 7.87 24.21 16.93 85.25 35.27 Family Enoploteuthidae 1.12 3.62 0.23 1.62 Abraliopsia falco 1.12 3.62 0.23 1.62 Family Mastigoteuthidae 0.24 0.29 0.18 Mastigoteuthis spp. 0.24 0.29 0.18 Family Ommastrephidae 2.17 6.74 4.60 11.87 85.01 28.20 Dosidicus gigas 6.74 2.91 9.99 82.20 26.23 Sthenoteuthis oualaniensis 2.17 1.69 1.74 2.81 1.92 Family Onychoteuthidae 18.16 0.72 4.79 Onychoteuthis banksii 18.16 0.58 4.73 Onychoteuthis spp. 0.14 0.06 Family Pholidoteuthidae 0.29 0.12 Ph olidoteu th is bosch in a n i 0.29 0.12 Family Thysanoteuthidae 0.24 0.14 0.12 Thysanoteuthis rhombus 0.24 0.14 0.12 Order Octopoda 4.49 0.24 3.04 2.11 2.10 Family Tremoctopodidae 0.94 0.24 Tremoctopus violaceus 0.94 0.24 Family Argonautidae 2.25 0.24 2.75 1.17 1.62 Argonaiita spp. 2.25 0.24 2.75 1.17 1.62 Family Bolitaenidae 2.25 0.29 0.24 Japetella diaphana 2.25 0.29 0.24 Phylum Arthropoda 69.57 12.38 0.14 5.05 Class Crustacea 69.57 12.38 0.14 5.05 Order Decapoda 69.57 12.38 0.14 5.03 Family Galatheidae 69.57 1.92 Pleuroncodes planipes 69.57 1.92 Family Penaeidae 1.69 0.42 Family Portunidae 10.65 2.63 Portunus xantiisii 10.65 2.63 Phylum Chordata 2.17 87.64 62.95 79.88 12.65 57.31 Class Osteichthyes 2.17 87.64 62.95 79.88 12.65 57.31 Order Clupeiformes 0.48 0.12 Family Engraulidae 0.48 0.12 Order Stomiiformes * 41.97 17.37 Family Phosichthyidae * 41.97 17.37 Vinciguerria lucetia * 41.97 17.37 Order Myctophiformes 2.17 57.30 14.77 0.14 6.83 Family Myctophidae 2.17 57.30 14.77 0.14 6.83 Benthosema panamense 16.85 0.90 continued 286 Fishery Bulletin 100(2) Table 3 (continued) Taxon Ai •ea N W E SW SE All Myctophum aurolaternatum 3.39 0.84 Myctophum nitidulum 3.15 0.78 Symbolopbonis spp. 40.45 8.23 0.14 4.25 Order Lampridiformes 0.58 0.24 Family Trachiptendae 0.58 0.24 Desmodesina polystic/um 0.58 0.24 Order Beloniformes 4.35 11.24 36.08 29.52 10.77 24.61 Family Belonidae 0.24 0.23 0.12 Strongylurn spp. 0.24 0.23 0.12 Family Exocoetidae 6.74 19.13 26.34 10.54 18.68 Cheilopogon furcatus 0.29 0.12 Cheilopogon spilonotopterus 0.29 0.12 Cheilopogon spp. 2,25 0.48 1.74 2.34 1.56 Exocoet us monocirrhiis 8.96 0.87 0.47 2.69 Exocoetus volitans 4.49 5.33 13.31 3.51 7.96 Exocoetus spp. 1.94 9.55 3.98 5.45 Hirundichthyf! speculiger 1.45 0.23 0.42 Progiiichthys spp. 0.97 0.29 0.36 Family Hemiramphidae 4.35 4.49 16.71 3.18 5.81 Oxyporha mph us muroptcrus 4.35 4.49 16.71 3.18 5.81 Order Perciformes 17.39 10.41 6.80 1.64 6.29 Family Echeneidae 0.29 0.12 Rhombochirus osteocbir 0.29 0.12 Family Coryphaenidae 17.39 0.24 0.72 0.47 0.96 Coryphaena equiselis 17.39 0.14 0.54 Coryphaena hippurus 0.29 0.12 Family Carangidae 0.97 1.88 1.02 Naucrates ductor 0.97 1.45 0.84 Family Gempylidae 3.63 1.01 0.47 1.44 Gempylus serpe?is 3.63 1.01 0.47 1.44 Family Scombridae 1.69 1.88 0.70 1.38 Acanthocybium solandn 0.24 0.06 Auxis spp. 0.97 1.74 0.96 Euthynnus luteal us 0.24 0.06 Thuiiniis spp. 0.24 0.14 0.70 0.30 Family Nomeidae 3.87 1.01 1.38 Cubiceps pauciradiatus 3.87 1.01 1.38 Order Tetraodontiformes 4.35 19.10 1.21 0.72 1.74 Family Balistidae 2.17 2.25 0.18 Xantichthys men to 2.25 0.12 Family Ostraciidae 15.73 0.84 Lactoria diaphanum 15.73 0.84 Family Tetraodontidae 2.17 1.12 1.21 0.72 0.72 Lagocephalus lagocephalus 2.17 1.12 1.21 0.72 0.72 Unidentified fishes 0.14 0.23 0.12 Total number of prey 46 89 412 691 427 1665 ■ Indix'idiial pri'v could nut bf enumerated Olson and Galvan Magana: Food habits and consumption rates of Coryphaena hippums 287 Table 4 Taxoiiomic composition in "cO of the pre.\ areas comhined. The occurrence of hard W = west. E = east, SW = southwest. SE • of conuiKin parts (ceph; = southeast. ddlphinfish from lopod mandibles five areas of the eastern Pacific and fish otoliths) was included Ocean ( Fig. Area codes 1) and from all are N = north. Taxon Area N W E SW SE All Phyhim Mollusca 25.0 15.79 15.52 14.34 45,22 21.28 Class Cephalopoda 25.0 15.79 15.52 13.97 45.22 21.10 Order Teuthoidea 25.0 10.53 15..52 12.13 43.48 19.45 Family Enoploteuthidae 2.63 0.37 0.87 0.55 Abraliopsis falco 2.63 0.37 0.87 0.55 Family Mastigoteuthidae 0.86 0.74 0,55 Mastigoteuthis spp. 0.86 0.74 0.55 Family Ommastrephidae 25.0 7.89 8.62 10.29 43.48 16.88 Dosidicus gigas 7,89 6.03 7,35 37.39 13.39 Slhenoteuthis oualaniensis 25.0 4.31 2.57 8.70 4.22 Family Onychoteuthidae 9.48 1.84 2.94 Onychoteuthis hanksii 9.48 1.47 2.75 Onychoteuthis spp. 0.37 0.18 Family Pholidoteuthidae 0.74 0,37 Pholtdoteuthis boschmani 0.74 0.37 Family Thysanoteuthidae 0.86 0.37 0.37 Thysanoteuth is rhombus 0.86 0.37 0.37 Order Octopoda 7.89 0.86 3.68 5.22 3.67 Family Tremoctopodidae 3.48 0.73 Tremoctopus i-iulaceus 3.48 0.73 Family Argonautidae 5.26 0.86 3,68 1.74 2.75 Argonauta spp. 5.26 0.86 3.68 1.74 2.75 Family Bolitaenidae 2.63 0.74 0..55 Japetella diaphaim 2.63 0.74 0,55 Phylum Arthropoda 25.0 8.62 1.10 2,75 Class Crustacea 25.0 8.62 1.10 2,75 Order Decapoda 25.0 7.76 0.37 2.02 Family Galatheidae 25.0 0.18 Pleiironcodes planipes 25.0 0.18 Family Penaeidae 4.31 0.92 Family Portunidae 3.45 0.73 Poi'tiinus xantusii 3,45 0.73 Phylum Chordata 50.0 36.84 43,10 42.28 25.22 38,53 Class Osteichthyes 50.0 36.84 43.10 42.28 25.22 38.53 Order Clupeiformes 0,86 0.18 Family Engraulidae 0.86 0.18 Order Stomiiformes 0.86 2.21 1.28 Family Phosichthyidae 0.86 2,21 1.28 Vinciguerria lucetia 0,86 2.21 1.28 Order Myctophiformes 25.0 10.53 6,90 0.37 2.57 Family Myctophidae 25.0 10.. 53 6.90 0.37 2,57 Benthosema panamense 2.63 0,18 continued 288 Fishery Bulletin 100(2) Table 4 (continued) Taxon A •ea N W E SW SE All Myctophum aurolaternatum 6.03 1.28 Myctophum nitidulum 0.86 0.18 Synibolophurus spp. 7.89 3.45 0.37 1.47 Order Lampridifornies 0.74 0.37 Family Trachipteridae 0.74 0.37 Desmodesma polystictum 0.74 0.37 Order Beloniformes 25.0 13.16 36.21 32.35 20.00 29.17 Family Belonidae 0.86 0.87 0.37 Strongylura spp. 0.86 0.87 0.37 Family Exocoetidae 13.16 25.86 28.68 19.13 24.77 Cheilopogon furcatus 0.74 0.37 Cheilopogon spthmotopterus 0.37 0.18 Cheilopogon spp. 5.26 1.72 4.04 4.35 3.67 Exocoetus monocirrhus 12.07 0.74 0.87 3.12 Exocoetus vo/itarif: 7.89 6.90 18.01 6.09 12.29 Exocoetus spp. 2.59 6.99 9.57 6.06 Hiriindichthys speculiger 1.72 0.87 0.55 Progiuchthy!< spp. 1.72 0.74 0.73 Family Hemiramphidae 2.5.0 2.63 24.14 6.99 8.99 Oxyporhamphus micropterus 25.0 2.63 24.14 6.99 8.99 Order Perciformes 50.0 17.24 11.40 5.22 10.83 Family Echeneidae 0.37 0.18 Rhombochirus osteochir 0.37 0.18 Family Coryphaenidae 50.0 0.86 1.84 1.74 1.83 Coryphaena equiselis 50.0 0.37 0.55 Coryphaena lilppurus 0.74 0.37 Family Carangidae 1.72 1.84 1.28 Naucrates ductor 1.72 1.47 1.10 Family Gempylidae 7.76 1.47 0.87 2.57 Gempylus serpens 7.76 1.47 0.87 2.57 Family Scombridae 5.17 4.41 2.61 3.85 Acanthocybiuni solandri 0.86 0.18 Auxis spp. 2.59 4.04 2.57 Euthynnus Imeatus 0.86 0.18 Thunnus spp. 0.86 0.37 2.61 0.92 Family Nomeidae 6.90 1.84 2.39 Cubiceps pauciradiatus 6.90 1.84 2.39 Order Tetraodontiformes 25.0 18.42 3.45 1.47 2.94 Family Balistidae 25.0 5.26 0.55 Xantichthys mento 5.26 0.37 Family Ostraciidae 10,53 0.73 Lactoria diaphanum 10.53 0.73 Family Tetraodontidae 25.0 2.63 3.45 1.47 1.83 Lagocep halu s lagocep halus 25.0 2.63 3.45 1.47 1.83 Unidentified fishes 0.74 0.87 1.10 Total number of samples 4 38 116 272 115 545 Olson and Galvan Magaha: Food habits and consumption rates of Coryphaena hippuius 289 08:34 hours; 1.'^ fish had stomachs with undigested food remains, 23 had stom- achs that were empty, and 2 had stom- achs that contained only trace amounts of hard parts. One of the two dolphin- fish in the fi51-800 mm size group con- tained remains of flyingfishes and the other contained tetraodontid puffers. In the larger group (801-950 mm). 11 dol- phinfish ate mostly triggerfishes (Balis- tidae), boxfish (Ostraciidae), and flying- fishes. In addition, 3 fish <650 mm and 2 fish in the 951-1100 mm class had empty stomachs or contained only trace amounts of hard parts. Diel feeding periodicity Early a m Late am Early p.m. Late p.m E 84 4 21 7 Early a.m. Late am Early p.m. Late p.m Early am Late am. Early p.m Late p.m -J Early a. m Late a.m Early p.m Late p.m Early am Late am Early p m Late p m Although common dolphinfish are thought to be visual predators that feed primarily in the daytime (Massutf et al., 1998), our data suggest that they also feed at night. In the areas where suffi- cient sample sizes were obtained (east, southwest, and southeast), an average of about IS*/? of the dolphinfish caught in the early morning contained food classi- fied in the "previous-high" category (Fig. 3). Many of these prey were flyingfishes, cephalopods, dolphinfishes, wahoo, and snake mackerel in digestion states 3 or 4 and were found in stomachs that were over 50% full. Prey of these or similar taxa were found to be completely evacu- ated from the stomachs of yellowfin tuna in about 6-18 h (Olson and Boggs, 1986). If dolphinfish gastric evacuation rates are on the order of those of yellowfin tuna (see "Discussion," and "Consump- tion rates" sections), these prey would have been ingested during the night. We examined the time of day that the dolphinfish fed, by area. In the east area, the data indi- cated peak feeding activity in the early morning and ear- ly afternoon, although few samples were obtained in the late morning and late afternoon ( Fig. 3, "recent high"). The most important prey taxa by percent biomass in the early morning were flyingfishes (549t}, snake mackerel (Gem- pylidae, 18'7r), and epipelagic cephalopods ( 10%). The most important prey taxa in the early afternoon were frigate or bullet tunas iAuxis spp., 37%) (or a combination of both), pelagic portunid crabs (27% ), and Thunnus spp. (yellowfin and bigeye tunas, 15%). In the southwest area, feeding appears to have occurred throughout the day (Fig. 3). The frequency of observations in the two "recent" categories combined was highest in the early afternoon (4170 and lowest in the late afternoon (22'7r). The highest proportion of empty stomachs and those containing only residual hard parts occurred in the late af- ternoon, followed by the early and late morning. The flving- East - RH m^ East - RL a East - PH -3 East - PL&E SW 132 59 66 14 _l_ SE 81 16 18 0 Southwest - RH Southwest - RL m Southwest - PH Southwest - PL&E ^ e: Southeast - RH Southeast - RL Southeast - PH i Southeast - PL&E :m :m 25 50 75 too 75 100 Percent occurence Figure 3 Percent occurrence of the prey items of common dolphinfish corresponding to four fullness-digestion categories (defined in Fig. 2) in three areas of the east- ern Pacific Ocean (Fig. 1) and by the time of day the sets were made. RH = recent high, RL = recent low, PH = previous high, PL&E = previous low and empty. Early a.m. = 05:12-09:00, Late a.m. = 09:01-12:00, Early p.m. = 12:01-1.5:00, Late p.m. = 1.5:01-18:16 hours. Sample sizes for each time-area stratum are shown at the top. Data for the north area are not shown because the sample size was too small, nor for the west area because only one time stratum was represented. fishes dominated in the diet between 09:00 and 15:00 hours (64% and 57% by weight), and 71% of the diet of dolphinfish caught in the late afternoon was epipelagic cephalopods. The important prey in the early morning were more varied and comprised frigate and bullet tunas (Au.xis spp., 28%), flyingfishes (18%), epipelagic cephalopods (17%), and meso- pelagic fishes (primarily Vinciguerria lucetia, 15%). In the southeast area, although no samples were ob- tained in the late afternoon, the data suggested peak feed- ing activity in the early afternoon (Fig. 3). The frequency of observations in both "recent" categories in the early af- ternoon summed to 42% of the total, compared with only 25% in the late morning. Most of the empty stomachs were from fish captured in sets made before noon. The epipelag- ic cephalopod Dosidicus gigas dominated in the diet dur- ing all time periods. Fifteen percent of the stomach con- tents of the common dolphinfish caught before 09:00 hours were small yellowfin tuna. 290 Fishery Bulletin 100(2) J) 40 20 All (n = 545) 75 50 25 % Weight Southwest (n = 272) 50 25 % Weight East(n= 116) fiOi ■' a) o g 40 * 3 O ^20 -.■'''-■ sS y " 0 100 75 50 % Weight ' Southeast (n= 115) 50 25 % Weight Figure 4 Three measures of diet importance for 13 prey categories (defined in Table 2) in three areas of the eastern Pacific Ocean (Fig. 1 1. and for all samples pooled. MsF = mesopelagic fishes, F = flyingfishes. EC = epipelagic cephalopods, G = Gempylidae, A = Auxis spp. Diet measures by area We present three measures of diet importance, '^rW, '^iN. and %0. in Figure 4 for all samples pooled, and by area. We included the data for trace quantities of hard parts (cephalopod mandibles and fish otoliths) in the analysis of %N and ^^O, but not o{''}W. The graphical representation mdicates that, when all data are pooled, most components of the diet appear to be quite rare (close to the origin of the graph. Fig. 4, all I. However, when the data are examined by area, the diet proves to be more varied. Flyingfishes and epipelagic cephalopods were clearly the dominant prey. Overall, fly- ingfishes were eaten by more of the dolphinfish (29'~f oc- currence) than any other category, followed by epipelagic cephalopods C2V-i occurrence). Because 274 stomach sam- ples were totally empty, these are equivalent to 58'7f and 42% occurrence, respectively, in the stomachs that con- tained food or hard parts. Prey counts contributed more than prey biomass to the apparent importance of mesope- lagic fishes and epipelagic cephalopods in the diet. This is partly due to the accumulation of digestion-resistant hard parts of these two taxa in the stomachs. Area was an important source of variation in the '7 W of three prey groups. The regi-ession tree for epipelagic cepha- lopods indicated that 25'/ of the apparent variation in the %W of that prey was explained by area (southeast vs. oth- ers). Area was also an miportant predictor of fiyingfish pre- dation; 15'f of the apparent variation in the 'iW was ex- plained by area ( north, west, and southeast vs. other areas ). The regression tree for Tetraodontiformes indicated that 4Kf of the apparent variation in predation on that taxon was explained by area (west vs. others). The gravimetric im- portance of the 10 other prey taxa could not be modeled by regression trees (i.e. the trees pruned back to the overall mean %W for those prey) owing to their infrequency in the diet or low sample size ( or to both ). Nevertheless, we pres- ent our results by area to illustrate the substantial spatial variability of the diet of common dolphinfish in the EPO. In the east area, the stomachs of 116 common dolphin- fish were sampled from 1 dolphin set and 22 floating-ob- ject sets. Sixty-two of these stomachs were empty and 2 contained only trace hard parts (SS.C^ of the females, 52.5'^i of the males). The flyingfishes were the dominant prey in the east area in terms of all three indices (Fig. 4, east). Flyingfishes were eaten by 3591 of the dolphinfish sampled and comprised 49*^/ of the total weight of the stomach contents. Epipelagic cephalopods were also eaten by substantial numbers of dolphinfish (16% occurrence), and accounted for 24% of the prey counts. The epipelagic cephalopods were numerically important, but less so by weight, because their mandibles resist digestion and may accumulate in the stomachs over time. The other prey taxa were fairlv rare in the east area. Olson and Galvan Magana: Food habits and consumption rates of Coryphaena hippuivs 291 100 80 60 40 20 H North i m !• ' • "I Epi. cephalopods [""jj Gempylidae I ] Meso. cephalopods fT^^ Auxis spp. ^^S Crustacea .t-j-j ., ,. . . !SS= ^pr'jj Yellowfin tuna ^ 1 N/leso. fishes " „'„ F^ mm Nomeidae \f^ Flyingllshes '^^ R^^^ /-— ,„t,^„„,H.,« [ 1 Tetraodontilormes Lv^ Coryphaentdae I 1 ^SS <^3rangidae | IVIisc. fishes 417-650 801-950 1101-1250 417-650 801-950 1101-1250 651-eOO 951-1100 1251-1770 651-600 951-1100 1251-1770 Fork length (mm) Figure 5 Gravimetric composition of prey groups (defined in Table 2), excluding trace quantities of hard parts, in the stomachs of 545 common dolphinfish sampled in five areas of the eastern Pacific Ocean (Fig. 1 ) versus dolphinfish size. Sample sizes for each size-area stratum are given in Table 1. In the southwest area, the stomachs of 272 common dol- phinfish were sampleci from 2 dolphin sets, 5 schoolfish sets, and 22 floating-object sets. One hundred and thirty- six of the samples were empty and 14 contained only trace hard parts (60.1'%^ of the females, 47.6% of the males). Fly- ingfishes and epipelagic cephalopods were found in 329?^ and 14''f of the dolphinfish sampled, and also ranked first and second in biomass, respectively (Fig. 4, southwest). The dietary importance of both, as well asAuxis spp., was determined more by weight than by numbers of individu- als, whereas the contrary was true for mesopelagic fishes. All the other prey categories were relatively rare in the diet according to all three measures. In the southeast area, the stomachs of 115 dolphinfish were sampled from 1 schoolfish set and 13 floating-object sets. Fifty-two of the stomachs were empty and 7 con- tained only trace hard parts (44. S"^* of the females, 61.5^r of the males). The epipelagic cephalopods were, by far, the dominant component in the diet in this area (Fig. 4, south- east). They were present in 45'7f of the stomachs sampled. Biomass and counts contributed about equally to the over- all importance of epipelagic cephalopods in the diet. Fly- ingfishes also occurred in many of the samples (19%), but their contribution to the diet by weight and numbers was overshadowed by the epipelagic cephalopods. Four other diet categories were rare. Size-specific predation We present the prey composition in ^W. excluding trace quantities of hard parts (cephalopod mandibles and fish otoliths), by six dolphinfish size strata for five areas (Fig. 5). Sample sizes for each stratum are given in Table 1. Considerable variability in the diet was apparent for common dolphinfish of different sizes. However, the statis- tical importance of size could be detected by the regression trees for only two prey groups because of the rarity of the other prey in the diet or because of small sample sizes. The regression tree for flyingfishes indicated that 7% of the ap- parent variation in the % W of these prey was explained by dolphinfish size. Size was also an important predictor of predation on Tetraodontiformes, explaining 9% of the apparent variation in the 9(W of these prey. Although dol- phinfish size was not an important predictor of predation on the other prey categories, we present the feeding data by size strata within each area because we believe this variability is biologically important. In the east area, the flyingfishes were the most impor- tant prey gi'oup overall for dolphinfish of the three small- est size classes and of the largest size class. For the two size classes between 951 and 1250 mm, frigate and bullet tuna [Auxis spp.), snake mackerels (Gempylidae), and epi- pelagic cephalopods were dominant (Fig. 5, east). 292 Fishery Bulletin 100(2) 700 600 500 E - 400 300 200- 100 A Cephalopod prey D Crustacean prey • Fish prey 400 600 800 1000 1200 1400 Dolphinfish fork length (mm) 1600 1800 07 ra 06 c (D 03 05 04 CL 1, 03 c a g- 02 0.1 ■ B fi Cephalopod prey n Crustacean prey • Fish prey • .. • • • . • •^ f •^•T ** — i" ■ — V r- 400 600 800 1000 1200 1400 1600 1800 Dolphinfish fork length (mm) Figure 6 The length or maximum dimension of the prey (A) and pi-ey relative sizes (Bi versus dolphinfish size for three prey groups found in the stomach contents of common dol- phinfish in the eastern Pacific Ocean. The lines represent the data smoothed with a smoothing spline, excluding the point for the largest dolphinfish. In the southwest area, few large common dolphinfish were sampled (Table 1). The flyingfishes were important prey for the four smallest size groups (up to 1100 mm). Epipelagic cephalopods were important for the smallest and two largest size groups. Auxis spp. represented over 46% of the diet of dolphinfish between 651 and 800 mm fork length (Fig. 5, southwest). In the southeast area, the epi pelagic cephalopods domi- nated the diet (52-90% ) of dolphinfish of all five size strata (Fig. 5, southeast). The flyingfishes comprised most of the remaining prey composition in the 651-800 and 801-950 mm categories, and Thuruiiis spp. (assumed to be mostly yellowfin tuna) comprised most of the remaining in the 951-1100 and 1101-1250 mm categories. Prey size We present the sizes of cephalopod, crustacean, and fish prey found in the dolphinfish stomach contents in Figure 6A. Intact prey ranged from 14 to 650 mm in length, and averaged 160 mm overall. The data smoothed with a smoothing spline showed an increasing trend of prey size with dolphinfish size. The maximum prey sizes increased gradually for dolphinfish up to about 760 mm, then abruptly to a maximum of about 600 mm for dolphinfish of approximately 1000 mm. This pattern was largely con- sistent for fish and cephalopod prey. Crustacean prey were small, but were consumed by a wide size range of dolphin- fish. " . Olson and Galvan Magana Food habits and consLimption rates of Coiyphaena hippurus 293 10 c 10 o >> ra 5^ Q 10 -Females (n=323) ■ - Males (n=175) 417-650 651-800 801-950 951-1100 1101-1250 1251-1770 B -Wesl(n=38) — »— East (n=116) - ■- Southwest (n=271) -^<- Southeast (n=1 15) 417-650 651-800 801-950 951-1100 1101-1250 1251-1770 Fork length (mm) Figure 7 Mean (±1 SE) daily rations for six size strata of common dolphinfish. for all samples pooled by sex (A) and by area (B). Sample sizes are given in Table 1. Rations for some strata were not plotted if sample sizes were small (i.e. n<5). The ratios of prey length to predator length ranged from 0.014 to 0.720, and averaged 0.177 (Fig. 6B). The length ratios, smoothed with a smoothing spline, showed a slight decreasing trend with dolphinfish size. Consumption rates The values of A, ( Eq. 1 ) chosen for the various prey taxa are listed in Table 2. We estimated a daily ration of 5.6 ±0.56'X (mean ±1SE) of body weight per day for all common dol- phinfish samples pooled. Mean ration estimates stratified by sex ranged from 0.4% of body weight per day for males of the smallest size group to 9.6% for males of the largest size group (Fig. 7A). Except for the 801-950 mm stratum, mean rations increased with dolphinfish size. Ration esti- mates were comparable for females and males, except for those in the smallest size stratum. The estimate for females in the 1251-1700 mm class was excluded from Figure 7A owing to a low sample size (n=2). We present mean (±1 SE) daily ration estimates strati- fied by area in Figure 7B, if sample sizes were five fish or more. The calculations revealed that the 25 fish of the 651-800 mm group from the east area had ingested large amounts of food. These dolphinfish accounted for the high rations overall for both males and females of that size class (Fig. 7A). Except for that group and perhaps the 651-800 mm group in the west area, the ration estimates were comparable for all size classes in all areas (Fig. 7B). The number of empty stomachs did not unduly influence the consumption estimates by area. The percent of empty 294 Fishery Bulletin 100(2) stomachs (or containing only hard parts) ranged from 51% to 66% for the southeast, southwest, east, and west areas, respectively. Discussion Four previous studies of the food habits of common dol- phinfish have been conducted in the EPO. Our results are difficult to compare meaningfully with those because they used different analytical techniques and had lower sample sizes. Larger sample sizes typically demonstrate greater trophic diversity (Manooch et al., 1983). Hida ( 1973) exam- ined the stomachs of seven dolphinfish (two C. hippurus and five C. equiselis) caught at about 4°N-119°W. The largest component of the diet was flyingfishes (33%), fol- lowed by cephalopods (22%). Campos et al. (1993) sam- pled an unspecified number of C. hippurus caught off the Pacific coast of Costa Rica by experimental longline. Our results were similar to theirs in that flyingfishes was the most important component of the diet, and in both stud- ies snake mackerel (Gempylidae) were found in the stom- achs. Our samples from the east area, which encompasses the coastal waters of Costa Rica, contained more prey diversity, including mesopelagic fishes, portunid crabs, and penaeid shruiips. Campos et al. (1993) identified tuna of unknown species in the diet, and we found small amounts of predation on Thunnus spp. (yellowfin or bigeye tuna) in the east and southeast areas. Aguilar-Palomino et al. (1998) reported on the food habits of 500 C. hippurus caught by sport hook-and-line fishing in a small area at the tip of the Baja California peninsula, Mexico. They determined that the cephalopod Dosidicus gigas was the most important component of the diet, followed by the red crab Pleuroncocles pla/iipes, a triggerfish, a flyingfish, and Auxis spp. Their results, however, could not be quantita- tively compared with ours because they presented only the index of relative importance (IRI) (Pinkas et al., 1971), but not its three components (% W, %A', and %0). For example, it is impossible to determine if the apparent importance of Dosidicus gigas in their study was due to the presence of fresh biomass (i.e. high %W). the accumulation of numer- ous, small mandibles from previous meals (i.e. high %A^ and/or %0), or both. Likewise, the IRI indicated that small red crabs were important in the diet, but this index can be overly influenced by numerous small prey. We sampled only four dolphinfish in the north area near the south end of the Baja California peninsula and found some of the same prey taxa reported by Aguilar-Palomino et al. ( 1998). Lasso and Zapata ( 1999) analyzed the stomach contents of 228 C. hippurus using nonstandard methods. Their results were presented only by large categories, fishes, mollusks, and crustaceans. Feeding periodicity Our analysis of diel feeding periodicity suggests that, al- though common dolphinfish may be primarily visual pred- ators (Massuti et al., 1998), they also feed at nighttime. Fish prey, such as flyingfishes, dolphinfishes, wahoo, and snake mackerel, would need to be ingested during night- time hours to reach digestion states 3 or 4 before 09:00 hours the next morning, as we obsei^ved in our study, unless gastric evacuation rates are much faster than expected. Shcherbachev (1973) concluded from the presence of par- tially digested flyingfishes, myctophid fishes, and squids in the stomachs of C. hippurus. and Crustacea in C. equise- lis that dolphinfishes feed around the clock in the Indian Ocean. Rothschild ( 1964) described active feeding on flying- fishes and myctophids at night by C. hippurus in the cen- tral Pacific. Massuti ot al. (1998) found that almost half of the stomachs of C hippurus sampled at sunrise contained mesopelagic prey. Massuti et al. (1998) and Oxenford and Hunte (1999) also concluded that common dolphinfish feed at night, as well as during the day Our data analysis by area revealed an apparent rela- tionship between the principal prey taxa in the diet and feeding periodicity. In the southwest and southeast areas, recently eaten cephalopods were in the stomachs through- out the daytime sampling period, although no samples were taken in the southeast in the late afternoon (Fig. 3). In contrast, in the east area the dolphinfish preyed mostly on fishes (flyingfishes, Ai/.v/s spp., and gempylids) that re- quire more time than cephalopods for gastric evacuation (Olson and Boggs, 1986). In the east area, recently eaten prey were present only in the early morning and early af- ternoon, and the food remains in the late morning and late afternoon were in advanced stages of digestion. These re- sults suggest that foraging activity may have been influ- enced by the digestibility and energy content of the avail- able prey. Cephalopods are typically low in energy content, whereas fishes store lipids in the musculature and viscera and have higher energy densities (Cummins and Wuy- check, 1971). Grove et al. (1978), Flowerdew and Grove (1979), and Jobling (1981) demonstrated that low-energy foods are emptied from the stomach more rapidly than foods of higher caloric content. Elevated lipid content in natural organisms is thought to have a retarding effect on gastric evacuation iFange and Grove, 1979). In yellow- fin tuna, gastric evacuation rates were inversely correlat- ed with total lipid content of four food organisms (Olson and Boggs, 1986). Apparently, the dolphinfish in the south- west and southeast areas spent more time foraging to ful- fill their energy requirements than the dolphinfish in the east area. Diet considerations Our study indicated that only two prey groups, flyingfishes and epipelagic cephalopods, were dominant in the diet of common dolphinfish in the EPO (Fig. 4). Diet differences we attributed to spatial stratification also had a seasonal component. Of the stomach samples obtained north of the equator in the north, west, and east areas, most (85%) of those that contained fresh food, were caught from May through November The trends described for common dolphinfish in the southwest and southeast areas may also have been attributable to seasonality of the prey from December through April because all the dol- phinfish in these areas were caught during these months. Olson and Galvan Magana: Food habits and consumption rates of Coryphaena hippurus 295 The marked differences in food habits with predator size are noteworthy. For all areas combined, the general trend was for increased predation on cephalopods and decreased predation on flyingfishes as the dolphinfish grew larger. The exception was the smallest size group (417-650 mm). These dolphinfish ate 50% cephalopods and 40% flying- fishes by weight. Zavala-Camin ( 1986) also found greater predation on cephalopods by large dolphinfish (>850 mm) than by smaller specimens off Brazil. Piscivores are known to feed selectively according to prey body size (Tonn et al., 1992). Maximum prey size is deter- mined by the mouth gape of the predator (Magnuson and Heitz, 1971; Hambright, 1991 ), and minimum prey size was correlated with the gap width between the gill rakers for a variety of tunas, mackerels, and dolphinfishes (Magnuson and Heitz. 1971). Preference for the largest prey a preda- tor can ingest is supported on theoretical grounds (Ivlev, 1961; Harper and Blake, 1988), but a sui-vey of studies ex- amining prey-size selectivities of piscivorous fishes showed a consistent pattern of selection for small prey ( Juanes, 1994). In our study, the common dolphinfish of all sizes in- gested small prey (Fig. 6A). Ratios of prey size to predator size for piscivores tend to average 0.2-0.3 (Juanes, 1994). In our study, the dolphinfish ingested prey that averaged slightly smaller, 17% of their length. A few dolphinfish ate prey that were greater than the maximum reported for pi- scivores, 50% of their length (Juanes, 1994). Consumption rates The method w-e employed for estimating daily rates of food consumption was judged by Cortes (1997) to be among the two most appropriate methods for top predators. How- ever, applying gastric evacuation rates derived for yellow- fin tuna to estimate daily rations of common dolphinfish requires justification. Suzuki's (1992) obser\'ations of gut- evacuation times of small dolphinfish do not unequivo- cally justify our assumption that gastric evacuation rates are comparable to those of yellowfin tuna, which were based on food passage through the stomach alone. How- ever, energetics requirements suggest that gastric evac- uation times for dolphinfish, at the same temperature, would be on the order of those for yellowfin. Our hypoth- esis is supported by the similar standard metabolic rates iif common dolphinfish and yellowfin (Benetti et al., 1995; Brill, 1996 ). Brill ( 1996 ) argued that high rates of digestion are consistent for high-performance fishes like tunas, bill- fishes, and dolphinfishes and demonstrated that dolphin- fishes share several characteristics of high-performance physiology with tunas and billfishes. Other, similar-size teleost fishes require about five times as long as yellowfin and skipjack to evacuate a meal (Magnuson, 1969; Olson and Boggs, 1986). Until gastric evacuation rates of dol- phinfish are measured, we are confident that our first- order estimates are adequate approximations of daily rations of common dolphinfish in nature. Our ration estimates for common dolphinfish are great- er than those for yellowfin tuna of comparable size, esti- mated by the same method (Olson and Boggs, 1986; Olson and Mullen, 1986). Our estimates are consistent with the observation that, although standard metabolic rates and lo- comotion costs are comparable for these species, common dolphinfish have greater growth rates than yellowfin (Uchi- yama et al., 1986; Wild, 1986) and may require more energy for growth. In summary, dolphinfish are an important component of the pelagic food web in the EPO, and as such, their feeding ecology provides clues to the underlying ecosystem struc- ture. Clearly, C. hippurus imparts predation pressure on cephalopods, flyingfishes, and other prey that are shared by a suite of predators ( Juhl, 1955; King and Ikehara, 1956; Blunt, 1960; Perrin et al., 1973; Nakamura, 1985; Olson and Boggs, 1986; Robertson and Chivers, 1997; Markaida and Sosa-Nishizaki, 1998). This predation pressure lends support to the hypothesis that cephalopods and flyingfish- es are abundant or have high ratios of production to bio- mass (P/Bi (or both! in the EPO. This hypothesis is based on 1) high consumption rates on these prey, indicated by the stomach contents in our present study, 2) high P/B of dolphinfish (Oxenford, 1999), 3) high consumption rates of cephalopods and flyingfishes by other predators (cited above), and 4) high P/B of those predators (Boggs, 1989; lATTC, 1999). This analysis illustrates the importance of diet studies for providing ecological insights. Our study provides key data for implementing ecosys- tem analyses based on food-web models (Christensen and Pauly, 1992; Walters et al., 1997). For example, Ecopath with Ecosim (www.ecopath.org) requires data on both the diet compositions and consumption rates of predators. Ac- cordingly, we summarized the prey data by several levels of taxonomic resolution and functional groups for dolphinfish sampled at multiple spatial scales and size classes. These data help lay the groundwork for a community- and ecosys- tem-level approach to fisheries management in the EPO. Acknowledgments We thank Richard Rosenblatt and H. J. Walker for access to the fish collection of the Scripps Institution of Oceanogra- phy, Robert Lavenberg and Jeff Siegel for access to the fish and otolith collections of the Natural History Museum of Los Angeles County, and F. G. (Eric) Hochberg of the Santa Barbara Museum of Natural History for reviewing iden- tifications of cephalopod beaks. Robert Pitman (NMFS, Southwest Fisheries Science Center) provided information on flyingfishes, and Edward Brinton (Scripps Institution of Oceanography) identified some crustaceans. George Wat- ters. Inter- American Tropical Tuna Commission (LATTC), helped us with the regression-tree analysis and the graph- ics. We thank the personnel and observers of the LATTC in Colombia, Mexico, Panama, and Venezuela for collecting stomach samples. Julio Martinez (lATTC, Cumana, Ven- ezuela) identified some of the prey. This manuscript was improved by reviews of William Bayliff (LATTC), George Walters (LATTC), and three anonymous reviewers. The second author (FGM) thanks the lATTC and the Institute Politecnico Nacional, Comision de Operacion y Fomento de Actividades Academicas for fellowships to participate in this project. 296 Fishery Bulletin 100(2) Literature cited Aguilar-Palomino, B., F. Galvan-Magana, L. A. Abitia-Cardenas, A. F. Muhlia-Melo, and J. Rodriguez-Romero. 1998. Aspectos alimentarios del dorado Coryphaena hippii- rus Linnaeus. 1758 en Cabo San Lucas, Baja California Sur, Mexico. Cienc. Mar. 24(3):253-'265. Allen, G. R., and D. R. Robertson. 1994. Fishes of the tropical eastern Pacific. Univ. Hawaii Press, Honolulu. HL 332 p. Beardsley, G. L. 1967. Age, growth, and reproduction of the dolphin, Cory- phaena hippiirus, in the Straits of Florida. Copeia 1967: 441-451. Benetti, D. D., R. W. Brill, and S. A. Kraul Jr 1995. The standard metabolic rate of dolphin fish. J. Fish Biol. 46:987-996. Blunt, C. E. Jr. 1960. Observations on the food habits of longline caught bigeye and yellowfin tuna from the tropical eastern Pacific 1955-1956." Calif Fish Game 46( 1 1:69-80. Hoggs, C. H. 1989. Vital rate statistics for billfish stock assessment. In Proceedings of the second international billfish symposium (R. H. Stroud, ed.), p. 225-233. National Coalition for Marine Conservation, Savannah, GA. Botsford, L. W., J. C. Castilla. and C. H. Peterson. 1997. The management of fisheries and marine ecosystems. Science (Wash. D.C.) 277:509-515. Breiman. L.. J. H. Friedman, R. A. Olshen, and C. J. Stone. 1984. Classification and regression trees. Chapman and Hall, New York, NY', 358 p. Brill, R. W. 1996. Selective advantages conferred by the high perfor- mance physiology of tunas, billfishes. and dolphin fish. Comp. Biochem. Physiol. 11.3A(1 1:3-15. Brusca, R. C. 1980. Common intertidal invertebrates of the Gulf of Cali- fornia. Univ. Arizona Press, Tucson, AZ, 513 p. Cailliet, G. M. 1976. Several approaches to the feeding ecology of fishes. In Fish food habits studies (C. A. Simenstad and S. J. Lipovsky. eds.), p. 1-13. Washington Sea Grant, Astoria, OR. Campos. J. A.. A. Segura. O. Lizano. and E. Madrigal. 1993. Ecologia basica de Coryphaena hippurus ( Pisces: Cory- phaenidae) y abundancia de otros grandes pelagicos en el Pacifico de Costa Rica. Rev Biol. Trop. 41(3):783-790. Christensen, V., and D. Pauly. 1992. Ecopath II — a software for balancing steady-state models and calculating network characteristics. Ecol. Model. 61:169-185. Clarke, M. R. 1962. The identification of cephalopod "beaks" and the rela- tionship between beak size and total body weight. Bull. Br. Mus. (Nat. Hist.) Zool. 8( 101:419-480. Clarke, M. R., ed. 1986. A handbook for the identification of cephalopod beaks. Oxford Unw. Press, New York, NY. 273 p. Clothier, C. R. 1950. A key to some southern California fishes based on ver- tebral characters. Calif Dep. Fish Game Fish Bull. 79:1- 83. Cortes, E. 1997. A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elas- mobranch fishes. Can. J. Fish. Aquat. Sci. 54:726-738. Cummins, K. W., and J. C. Wuycheck. 1971. Caloric equivalents for investigations in ecological energetics. Mitt. Int. Ver. Theor. Angew. Limnol. 18:1- 158. Death, G., and K. E. Fabricius. 2000. Classification and regression trees: a powerful yet simple technique for ecological data analysis. Ecology 81( 11): 3178-3192. Essington, T. E., D. E. Schindler, R. J. Olson, J. F Kitchell, C. H. Boggs, and R. Hilborn. In press. Alternative fisheries and the predation rate ofyellow- fin tuna {Thiinnus a/bacares) in the eastern Pacific Ocean. Ecol. Appl. Estes. J. A., M. T. Tinker, T. M. Williams, and D. F Doak. 1998. Killer whale predation on sea otters linking oceanic and nearshore ecosystems. Science (Wash. D.C.) 282:473- 476. Fange, R., and D. Grove. 1979. Digestion, /n Fish physiology, vol. VIII: Bioenergetics and growth (W. S. Hoar and D. J. Randall, eds), p. 161-215. Academic Press, New York, NY. FAO (Food and Agriculture Organization). 1995. Code ofconduct for responsible fisheries. FAO, Rome, 41 p. Fischer, W., F. Krupp, W. Schneider, C. Sommer, K. E. Carpenter, and V. H. Niem. 1995a. Guia FAO para la identificacion de especies para los fines de la pesca. Pacifico centro-oriental. Volumen I: Plantas e invertebrados. FAO ( UN. Food and Agriculture Organi- zation) 1:1-646. 1995b. Guia FAO para la identificacion de especies para los fines de la pesca. Pacifico centro-oriental. Volumen II: Ver- tebrados— Parte 1. FAO (UN. Food and Agriculture Orga- nization) 11:647-1200. 1995c. Guia FAO para la identificacion de especies para los fines de la pesca. Pacifico centro-oriental. Volumen III: Ver- tebrados— Parte 2. FAO (U.N. Food and Agriculture Orga- nization) 111:1201-1813. Flowerdew. M. W.. and D. J. Grove. 1979. Some observations of the effects of body weight, tem- perature, meal size and quality on gastric emptying time in the turbot, Scophthalmus ma.ximus (L.) using radiography J. Fish Biol. 14:229-238. Garth, J. S., and W. Stephenson. 1966. Brachyura of the Pacific coast of America, Brachy- rhyncha: Portunidae. Allan Hancock Monogr. Mar Biol. 1, 1.54 p. Gooding, R. M., and J. J. Magnuson. 1967. Ecological significance of a drifting object to pelagic fishes. Pac. Sci. 21(41:486-497. Grove, D. J., L. G. Loizides, and J. Nott. 1978. Satiation amount, frequency of feeding and gastric emptying rate in Salnio gaiirlneri. J. Fish Biol. 12:507- 516. Hall, M. A. 1998. An ecological view of the tuna-dolphin problem: impacts and trade-offs. Rev. Fish Biol. Fish. 8:1-34. Hambright, K. D. 1991. Experimental analysis of prey selection by large- mouth bass: role of predator mouth width and prey body depth. Trans. Am. Fish. Soc. 120:500-508. Hansson, S. 1998. Methods of studying fish feeding: a comment. Can. J. Fish. Aquat. Sci. 55:726-738. Olson and Galvan Magaha Food habits and consumption rates of Coiyphaena hippurus 297 Harpor. D. C... and K. W. Blake. 1988. EntTRctics of piscivorous predator-prey interactions. J.Theor. Biol. 134:59-76. Hida, T. S. 1973. Food of tunas and dolphins (Pisces: Scombridae and Coryphaenidae) with emphasis on the distribution and biol- ogy of their prey Stoh'phorus buccaneeri (Engraulidae). Fish. Bull. 7ia):13.5-143. Hunter, J. R.. and C. H. T. Mitchell. 1966. Associations of fishes with flotsam in the offshore waters of Central America. Fish. Bull. 66( 1 ): 13-29. lATTC (Inter-American Tropical Tuna Commission). 1999. Annual report of the Inter-American Tropical Tuna Commission, 1997. Inter-Am. Trop. Tuna Comm., La Jolla, CA, 310p. Iverson, L. K., and L. Pinkas. 1971. A pictorial guide of beaks of certain eastern Pacific cephalopods. Calif Dep. Fish Game Fish Bull. 152:83- 105. Ivlev, V. S. 1961. Experimental ecology of the feeding of fishes. Yale Univ. Press, New Haven, CT, 302 p. Jobling, M. 1981. Dietary digestibility and the influence of food compo- nents on gastric evacuation in plaice, Pteuronectes platessa L. J. Fish Biol. 19:29-36. Jordan, D. S., and B. W. Evermann. 1896. The fishes of North and Middle America: a descriptive catalogue of the species offish-like vertebrates found in the waters of North America, north of the Isthmus of Panama. Smithsonian Institution. Bull. U.S. Natl. Mus. 47, 3313 p. [Reprinted 1963, T.FH. Publications, Jersey City, NJl. Juanes, F. 1994. What determines prey size selectivity in piscivorous fishes? In Theory and application in fish feeding ecology (D. J. Stouder, K. L. Fresh, and R. J. Feller, eds.), p. 79-100. Univ. South Carolina Press, Columbia, SC. Juhl. R. 1955. Notes on the feeding habits of subsurface yellowfin and bigeye tunas of the eastern tropical Pacific Ocean. Calif Fish Game 41( 1 1:99-101. King, J. E., and I. I. Ikehara. 1956. Comparative study of the food of the bigeye and yellow- fin tuna in the central Pacific. Fish. Bull. 57(108):61-85. Kojima, S. 1956. Fishing for dolphins in the western part of the Japan Sea II. Why do the fish take shelter under floating materi- als? Bull. Jpn. Soc. Sci. Fish. 21( 101:1049-10.52. Larkin, P. A. 1996. Concepts and issues in marine ecosystem manage- ment. Rev. Fish Biol. Fish. 6:139-164. Lasso, J., and L. Zapata. 1999. Fisheries and biology of Coryphaena hippiirua ( Pisces: Corj'phaenidae) in the Pacific coast of Colombia and Panama. Sci. Mar. 63(3-4):387-399. Lawson, T A. 1997. Estimation of bycatch and discards in central and western Pacific tuna fisheries: preliminary results. South Pac. Comm., Oceanic Fish. Prog. Int. Rep. 33:1-30. Magnuson, J. J. 1969. Digestion and food consumption by skipjack tuna (Kat- suwonus pelamis). Trans. Am. Fish. Soc. 98(3):379-392. Magnuson, J. J., and J. G. Heitz. 197 1 . Gill raker apparatus and food selectivity among mack- erels, tunas, and dolphins. Fish. Bull. 69(2):361-370. Mangel, M., L. M. Talbot, G. K. Meffe, M. T Agardy, D. L. Alverson, J. Barlow, D. B. Botkin, G. Budnwski, T. Clark, J. Cooke, et al. 1996. Principles for the consei-vation ofwild living resources. Ecol.Appl.6(2):338-.362. Manooch, C. S. I., D. L. Mason, and R. S. Nelson. 1983. Food and gastrointestinal parasites of dolphin, Cory- phaena hippurus, collected along the southi'astern and gulf coasts of the United States. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFC 124:1-36. Markaida, U, and O. Sosa-Nishizaki. 1998. Food and feeding habits of swordfish.Xip/jias g/adius L., off western Baja California. In Biology and fisheries of sviordfish, Xiphias gladius. Papers from the international symposium on Pacific swordfish, 11-14 December 1994, Ensenada, Mexico (I. Barrett, O. Sosa-Nishizaki, and N. Bartoo, eds.), p. 245-259. U.S. Dep. Commer., NOAA Tech. Rep. NMFS-142. Massuti, E., S. Deudero, P. Sanchez, and B. Morales-Nin. 1998. Diet and feeding of dolphin ( Coiyphaena hippurus ) in west- em Mediterranean waters. Bull. Mar. Sci. 63(2):329-341. MathSoft Inc. 1999. S-Plus 2000 Guide to statistics, volume I: Data analy- sis. MathSoft, Seattle, WA, 637 p. Meek, S. E., and S. F Hildebrand. 1923. The marine fishes of Panama. Field Mus. Natl. Hist. (Zool. Ser.), vol. 15. 1045 p. Miller, D. J., and R. N. Lea. 1972. Guide to the coastal marine fishes of California. Calif Dep. Fish Game Fish Bull. 157:1-249. Miller, D. L., and S. C. Jorgenson. 1973. Meristic characters of some marine fishes of the west- ern Atlantic Ocean. U.S. Fish Wildlf Serv. Fish. Bull. 71(1): 301-312. Monod, J. G. 1968. Le complexe urophore des poissons teleosteens. Memories de LTnstitute Fundamental D' Affrique Noire 81: 1-705. Nakamura, I. 1985. FAO species catalogue. Vol. 5: Billfishes of the world: an annotated and illustrated catalogue of martins, sail- fishes, spearfishes and swordfishes known to date. FAO Fisheries Synopsis 125, 65 p. Norton, J. G., and S. J. Crooke. 1994. Occasional availability of dolphin, Coryphaena hippu- rus, to southern California commercial passenger fishing vessel anglers: observations and hypotheses. Calif Coop. Oceanic Fish. Invest. Rep. 35:230-239. Olson, R. J., and C. H. Boggs. 1986. Apex predation by yellowfin tuna iThunnus alba- cares): independent estimates from gastric evacuation and stomach contents, bioenergetics, and cesium concentra- tions. Can. J. Fish. Aquat. Sci. 43(9):1760-1775. Olson, R. J., and A. J. Mullen. 1986. Recent developments for making gastric evacuation and daily ration determinations. Environ. Biol. Fish. 16: 183-191. Oxenford, H. A. 1999. Biology of the dolphinfish (Coryphaena hippurus) in the western centra! Atlantic: a review. Sci. Mar. 63(3^): 277-.301. Oxenford, H. A., and W. Hunte. 1986. A preliminary investigation of the stock structure of the dolphin, Coryphaena hippurus. in the western central Atlantic. Fish. Bull. 84(2):451-460. 298 Fishery Bulletin 100(2) Oxenfoid, H. A., and W. Hunte. 1999. Feeding habits of the dolphinfish iCoryphaena hippii- rus) in the eastern Caribbean. Sci. Mar. 63(3-4):303-31.5. Palko, B. J.. G. L. Beardsley. and W. J. Richards. 1982. Synopsis of the biological data on dolphin-fishes, Cory- phaena hippurus Linnaeus and Coryphafna equisclis Lin- naeus. U.S. Dep. Commer., NOAA Tech. Memo NMFS Circ. 443:1-28. FAO Fish. Synop 130. Parin, N. V. 1961. The basis for the classification of the flying fishes (family Oxyporhamphidae and Exocoetidae). U.S. Natl. Mus., NMFS Transl. 67, 104 p. Patterson, K. R.. and J. Martinez. 1991. Exploitation of the dolphin-fish Coryphaena hippurus L. off Ecuador: analysis by length-based virtual population analysis. Fishbyte 9(2):21-23. Perrin, W. F, R. R. Warner, C. H. Fiscus, and D. B. Holts. 1973. Stomach contents of porpoise, Stenella spp., and yel- lowfin tuna, Thunnus albacares. in mixed-species aggrega- tions. Fish. Bull. 71(4):1077-1092. Pinkas, L., M. S. Oliphant, and I. L. K. Iverson. 1971. Food habits of albacore, bluefin tuna, and bonito in California waters. Calif Dep. Fish Game Fish Bull. 152: 1-105. Robertson, K. M., and S. J. Chivers. 1997. Prey occurrence in pantropical spotted dolphins, Stenella attenuata. from the eastern tropical Pacific. Fish. Bull. 95:334-348. Rothschild, B. J. 1964. Observations on dolphins iCoryphaena spp. i in the central Pacific Ocean. Copeia 1964(21:445-447. Shcherbachev, Y. N. 1973. The biology and distribution of the dolphins (Pisces, Coryphaenidaei. J. Ichthyol. 13(2):182-191. Shiomoto, A.. K. Tadokoro. K. Nagasawa, and Y. Ishida. 1997. Trophic relations in the subarctic North Pacific eco- system: possible feeding effect from pink salmon. Mar Ecol. Prog. Ser 150:75-85. Suzuki, E.Y 1992. Protein accretion and bioenergetics in dolphin fish (Coryphaena hippurus). M.S. thesis, Univ. Hawaii, Manoa. HI, 146 p. Thomson, D. A., L. T. Findley, and A. N. Kerstitch. 1979. Reef fishes of the Sea of Cortez. John Wiley and Sons, New York. NY, 302 p. Tonn, W. M., C. A. Paszkowski, and I. J. Holopainen. 1992. Piscivory and recruitment: mechanisms structuring prey populations in small lakes. Ecology 73:951-73958. Uchiyama, J. H., R. K. Burch, and S. A. Rraul Jr 1986. Growth of dolphins, Coryphaena hippurus and C. equiselis, in Hawaiian waters as determined by daily incre- ments on otoliths. Fish. Bull. 84(1):186-191. Verheye, H. M., and A. J. Richardson. 1998. Long-term increase in crustacean zooplankton abun- dance in the southern Benguela upwelling region (1951- 1996): bottom-up or top-down control? ICES J. Mar. Sci. 55:803-807. Walters, C, V. Christensen, and D. Pauly. 1997. Structuring dynamic models of exploited ecosystems from trophic mass-balance assessments. Rev. Fish Biol. Fish. 7:139-172. Watters, G., and R. Deriso. 2000. Catch per unit of effort of bigeye tuna: a new analysis with regi'ession trees and simulated annealing. Inter-Am. Trop. Tuna Comm. Bull. 21(81:527^571. Wickham, D. A., J. W. Watson Jr, and L. H. Ogi-en. 1973. The efficacy of midwater artificial structures for attract- ing pelagic sport fish. Trans. Am. Fish. Soc. 102(3):563- 572. Wild, A. 1986. Growth of yellowfin tuna. Thunnus albacares, in the eastern Pacific Ocean based on otolith increments. Bull. Inter-Am. Trop, Tuna Comm. 18:421-482. Wolff G. A. 1982. A beak key for eight eastern tropical Pacific cephalo- pod species with relationships between their beak dimen- sions and size. Fish. Bull. 80(2):1-14. Zavala-Camin. L. A. 1986. Conteiido estomocal e distribui?ao do dourado Cory- phaena hippurus e ocorrencia de C. equiselis no Brasil (24''S- 33°S). Bol. Inst. Pesca, Brazil 13(2):5-14. 299 Abstract-Not catches from 1985-86 1(1 1991-9.") at Fivers Island. North Caro- lina, indicated that glass-ccl stage Ainer- ican eels {AngiiilUi rostrata) were re- cruited to the estuary from November to early May. with peak numbers in January. February, and March. There was no declining trend in recruitment over the years of sampling. Except for one year, there was no clear seasonal decrease in mean length. But shorter glass eels were older than longer glass eels, as judged by age within the glass eel growth zone of the otolith, suggest- ing that smaller fish took longer to arrive. The mean age of glass eels col- lected from the lower estuary and a freshwater site 9.5 km upriver differed by 8.4 d (36.2 vs. 44.6, respectively). Outer increments (30-35) of the oto- lith gi-owth zone of glass eels from North Carolina were significantly wider than corresponding increments of oto- liths from New Brunswick. Mean total ages of North Carolina, New Jersey, and New Brunswick elvers were 175.4. 201.2, and 209.3 d, corresponding to mean lengths of 55.9, 60.9. and 58. 1 mm TL. respectively. The mean durations of glass-eel growth zones (44.6, 62.3, and 69.8) were in close agreement with those from previous studies, but total ages were not. This suggested that per- haps some finer (leptocephalus stage) increments were not detected by light microscopy, differences occurred in sea- sonal increment deposition, or absorp- tion of the otolith material may have taken place during metamorphosis, ren- dering the aging of larvae inaccurate. Judging from the long recruitment period and seasonal uniformity in both mean age and length found in our study, the spawning period of American eels may be somewhat more protracted than previously considered. Recruitment season, size, and age of young American eels (Anguilla rostrata) entering an estuary near Beaufort, North Carolina Perce M. Powles Biology Department Trent University Peterborough Ontano, K9J 7B8, Canada E-mail address: ppowlesiq trenlu ca Stanley M. Warlen Center for Coastal Fisheries and Habitat Research National Ocean Service, NOAA 101 Pivers Island Road Beaufort, North Carolina 28516-9722 Manuscript accepted 16 August 2001. Fish. Bull. 100:299-306 (2002). The life history of the American eel {An- guilla rostrata ) is both fascinating and mysterious. It spawns in the Sargasso Sea and the young willow leaf-shaped leptocephalus lai-va disperses over great distances before it metamorphoses into a transparent glass eel. This stage reaches the estuaries of North America where it typically ascends rivers and streams as an elver, becoming progressively more pigmented as the run progresses. This stage is followed by the juvenile or yellow eel, and finally by the silver or maturing adult (Able and Fahay, 1998). Eel physiology and ecology are still not completely understood, although much has been learned since the early 1900s. Schmidt (1922) discovered the general area of spawning in the Sargasso Sea, and Kleckner et al. 1 1983 ) further demar- cated breeding to an area defined by thermal front boundaries. Comparini and Rodino (1982) used electrophoretic analysis on American and European eel (Anguilla anguilla) leptoeephali to con- firm that both spawn in the Sargasso Sea. Based on oceanographic surveys and entrance times into rivers, estimates of migration time to North America of young A. rostrata range from 250 days to one year, corresponding with a peak spawning time of February and March (McCleave et al., 1987; Haro and Kreuger, 1988). Many studies have described seasonal sizes of European glass eels and elvers entering estuar- ies, but similar long-term data sets for the American species are uncommon. Elvers enter estuaries along the coast of North America progressively from south to north, becoming larger with increas- ing latitude (Haro and Ki-euger, 1988), and for many (but not all) areas, sizes have been recorded. We document the recruitment period and variations in its seasonality for the Beaufort, NC, region for the 1985-86 to 1994-95 seasons. Michaud et al. (1988) documented the sizes but not the ages of American eel elvers entering a Quebec river Jes- sop ( 1998) used large samples from the Nova Scotian elver fishery to show sea- sonal declines in length and weight and an increase in pigmentation of newly arrived elvers. Wang and Tzeng (1998) used SEM (scanning electron micros- copy) and elemental analysis of oto- liths to determine ages of elvers taken from Haiti, north to Nova Scotia. They also estimated the length in days of the leptocephalus and glass-eel stages from the same sites. Helfman et al., (1984) described the length range of glass eels entering a Georgia river in February. Although most earlier work- ers (e.g. Vladykov, 1966; Eldred, 1968, and others) did not have long-term da- ta series, their samples are useful for comparisons with current data. 300 Fishery Bulletin 100(2) Long-term records of estuarine recruitment of early life- history stages of eels are particularly relevant at this time, when reduction in juveniles (but not elvers) to the upper St. Lawrence River area in North America has been noted (Castonguay et al., 1994a), and when catches of adults and recruitment of elvers are declining in Europe (Castonguay et al., 1994b). We therefore present subsamples of ages along with our catch per effort (densities) for glass eels. Earlier studies of eel otoliths used SEM to determine ag- es, but we used light microscopy for comparison with such studies. Castonguay (1986) aged A. rosti-ata leptocephali captured in the Gulf of Maine, and his largest individual of 49 mm TL (which would shrink during metamorphosis) was 130 d. However, only the daily growth increments in the elver gi'owth zone (area after the "transition" or "el- ver" check) have been validated (Martin, 1995). He vali- dated daily increments in the field over a one-month pe- riod. Lecomte-Finiger (1992) aged otoliths of A. anguilla elvers from several European locations, as did Antunes and Tesch (1997). The latter suggested that daily incre- ments oi A. anguilla may be underestimated because of the "diffuse" (their term) nature of the metamorphosis zone. More recently, Cieri and McCleave (2000) suggested that there are a number of inconsistencies when growth zones of leptocephali are compared with the same zone in the otoliths of glass eels and juvenile American eels. Both radii and ages were less in the older fish stages than in the leptocephali. The possibility of resorption of otolith material during metamorphosis (and subsequent oblitera- tion of increments) led them to discourage the use of lar- val eel otoliths for aging or back-calculating life history events until these discrepancies could be explained. Al- though the present authors could not validate increments in the leptocephalus stage or metamorphosis stage (these stages metamorphose before reaching North Carolina), we collected samples to temporally validate daily increments in the glass-eel stage. We tested the hypotheses that 1 ) the increment number for glass eels at an upi'iver (freshwa- ter) location would be greater than that for glass eels at a lower (estuarine) sampling site by a difference (in days) equal to the distance divided by the sum of the swimming speed and tidal transport, and that 2) increment widths for corresponding glass-eel portions of the otolith for fish from the Bay of Fundy (New Brunswick sample) would be narrower than for fish of warmer water (i.e. North Car- olina! because the warmer waters would produce faster growth which, in turn, could produce wider otolith incre- ments. A New Jersey sample was also aged to represent an intermediate geogi-aphic position between North Carolma and New Brunswick. The objectives of this study were li to describe the 1985-95 annual recruitment of glass eels to the estuary near Beaufort; 2) to present the variations in seasonal densities of glass eels: 3) to analyze the seasonal length frequency by weekly intei-vals, and 4) to compare mean ages of glass eels at three locations along the eastern coast of North America (North Carolina, New Jersey, and New Brunswick). We also tested two hypotheses relevant to val- idating daily deposition of otolith growth increments in the glass-eel stage of Anguilla rostrata. Methods Collections Glass eels were collected from two sites in North Caro- lina: 1 ) in the lower Newport River at Fivers Island, about 2 km inside Beaufort Inlet: and 2) at Black Creek, a small tributary of the Newport River, at the entrance to a mill- pond, about 9.5 km upriver from Beaufort. North Caro- lina. The samples at Fivers Island were collected over the 10-year period 1985-95 with a 1x2 m neuston net fitted with 945-um mesh and flow meter and suspended from a bridge, except during 1985-86. when 60-cm bongo nets were used. Details of the sampling protocol are given in Warlen (1994). Samples were always collected at night during midflood tide, from November to April. Density of fish in the catches was standardized as the mean number per 100 nr'. Additional samples were collected by dip net from Black Creek just below a small dam leading into a millpond (22 February and 20 March 1994). Samples were also obtained from two areas north of North Caro- lina: 1 ) Little Egg Inlet, New Jersey, at the site of the Marine Field Station, State University of New Jersey, Rutgers.Tuckerton, NJ (39°30'N. 74°14'W) on 21 Febru- rary and 9 March 1994 (with a 1-m plankton net): 2) elvers were captured from the Lepreau River, at the vil- lage of Lepreau, New Brunswick, Canada, on the Bay of Fundy (45°00'N, 66°20'W) on 6 June 1994. This sample was collected with a 1-mm mesh dip net and was pre- served in 70% ethanol. All other samples were preserved in 950r ethanol, and then preserved in fresh 70% ethanol after 24 h. Aging methods All eels were measured (TL to the nearest 0.1 mm) with a Vernier caliper and saggital otoliths were removed, washed, dried, and mounted on glass slides with thermal cement, sulcus groove down. The otoliths were ground with a series of wet grit papers (no. 600 to 1200i until the core was visible, then polished with diamond paste, after which they were aged by using an oil-immersion lens from 1200 to 3100x. A subsample of 10 otoliths was also sectioned (transverse) and pi-epared for reading, and ages were compared with ages from whole otoliths from the same fish. Sectioned otoliths had about 4% more rings in the outer edge. Otolith increments were read from the hatching ring outwards, as suggested by Lecomte-Finiger ( 1992 ) and Tzeng ( 1996 ). Umezawa and Tsukamoto ( 1991 ) found that no rings were formed during the incubation period (5 days) of A. japonica. To maintain consistency with other authors, we therefore added five days to our ages. An optical imaging system was used to count incre- ments and to measure increment widths and the distance from the core to the first-feeding (exogenous) ring, etc. The lowest increment width (distance between adjacent dark rings) recorded was 0.326 p, a distance which was probably close to the resolution minimum of our micro- scope. Incremental counts were made of the leptocephalus growth zone, the metamorphic zone, the glass-eel growth Powles and Warlen Recruitment season, size, and age of Angutlla rostrata entering an estuary near Beaufort 301 zone, and from the "elver" check (transition mark) to the edge of the otolith (see Cieri and McCleave, 2000). Total ages were esti- mated as the age from the hatching ring to the elver mark. Glass-eel gi'owth zone ages (increments from the outer boundary of met- amorphic zone to elver check) were read as in Wang and Tzeng ( 1998). Age studies The mean ages in the glass-eel growth zones of eels taken at Pivers Island in February were compared with mean ages in glass-eel growth zones of eels taken at an upper estu- ary site, the Black Creek Mill Pond entrance (also in February). The difference in mean age presumably reflected the time required to travel the distance upriver to the collec- tion site. The total ages and the ages found in the glass-eel growth zone were compared with those for small and larger glass eels arriving during the recruitment season. Glass eels were divided into length groups (from 46.1-48.0 TL to 58.1-60.0) and correspond- ing age frequencies were assigned to each length group. A linear regression was then fitted to the data to assess whether or not smaller fish were older or younger than the longer members of their recruitment class. Statistical analysis Seasonal trends in the weekly length distributions were examined by linear regression (SAS Institute, 1996) for each sampling (recruitment) season, and for the whole 10-year period ( 1985-86 to 1994-95). The differences in otolith increment widths in the early (proximal) glass-eel growth zone (increments 10-15) and later ( mid-distal ) glass-eel growth zone ( increments 30-35 ) from the New Brunswick and North Carolina samples were examined by analysis of variance (ANOVA). Age dis- tributions of glass eels captured by month were compared by using the F-test. Statistical significance was accepted at P < 0.05. 15 - E _^-— -•^ /\ "=768 /^""^^^ o • — ' ^^""v^ X \ / o ^^""-^f \ j' S- ^° \ ^^^ *U) \ ^f c o "D V / 0) w / \ / m 5 to \ / CO \ / O 1 1 I I L_ 1 1 !.__ 1 1 _i 1985/86 86/87 87/88 88/89 89/90 90/91 91/92 92/93 93/94 94/95 Recruitment year Figure 1 Mean density of Angutlla rostrata glass eels (no. of eels/100 m''l caught each year, from 1985-86 to 1994-95, between November and April at Pivers Island, Beaufort, North Carolina. again (F.j=4.88; slope 2.04, 7-2=0.78, P=0.113), approaching the previous mean level. Seasonal recruitment Although glass eel recruitment sometimes occurred over a 5 1/2 month period (mid-November to early May), most recruitment of young eels to the estuary occurred from December through April (Fig. 2). Although there was some variation among years within the recruitment period, the largest catches usually occurred in February and March. Peak catches varied from early in the season (1985-86) to late in the season (1987-88). In 1993-94, densities were highest in mid-January and mid-March. In (1990-91), the year with the lowest catches, almost all of the catch occurred in December and January, with virtually none afterwards. Wlien seasonal changes in length of glass eels among re- cruitment years were analyzed (Table 1), the slope of the regression for all 10 years (Fig. 3) was positive, but not sig- nificant (Fj=19.50, P=0.001). Four seasons had negative regression slopes and slopes for the other years were posi- tive, three of which were significant. Results Annual recruitment The total density of glass eels in the Beaufort estuary varied considerably between 1985 and 1995, but there was no significant trend (Fj,= 1.605, adj r'=0.123, slope=-0.031, P=0.246) among the 10 consecutive recruitment years in the Beaufort estuary (Fig. 1). Densities were highest in 1988-89 and 1993-94 ( 13.5-14.0 eels/100 m') and lowest in 1990-91 (1.5 eels/100 m^i, a ninefold difference. Apart from the low of 1990-91, all other years had total densi- ties of more than seven eels/100 m*. After the 1990-91 low, the total annual density of elvers at Beaufort increased Age studies The mean age observed in the glass-eel growth zones of eels collected in February from the Pivers Island and Black Creek sites differed significantly, by eight days (F^= 6.65 l,P=0.003i. The Pivers Island eels averaged 36.2 d (52"9 ±0.9 mm TL) for their as yet uncompleted glass-eel phase, and the Black Creek elvers averaged 44.6 d (55.9 ±0.9 mm TL). The distance between the two sites was about 9.5 km. In respect to otolith growth, the mean widths of the first 10-15 daily growth increments in the glass-eel growth zone (Table 2) were not significantly different (F2=1.363, P=0.185) between the northern eels (New Brunswick) and the southern eels (North Carolina). However, both sam- ples of otoliths from North Carolina indicated that mean 302 Fishery Bulletin 100(2) 94/95 - ' ' O e ^ o oO© ° oQo O o O ^; OOTlCX) O o " " " 93/94 - " o ^ oo°ooo cQ^IBdo °q0^o oO ^ ^ " " " 92/93 " - © ^ 'O'O^ -O' cOoOoQoOoooO^ ' ' 91/92 T n z z z z ' ooooO°OoOC(j)0° oO°OooO" ^ 90/91 ^ c "o'2ooOoOoOO'''''22gzzzznnnn 1 89/90 - n 2 z ^o©' 'QOo°Oo(3j jf^ /OO' ' ' ' ' " " cr 88/89 - no.®'' oO°0^ ooOoO0(I^~^(JDoO' O" " 87/88 - z o z o o o ' ' ' ' ^ooo<3o^°oOqGD° " " 86/87 - " " ' . o o o ' 0 oT^ o ' 0(Tj(; yP'-^^^C^' J n o " " 85/86 - - - ^GO^o^ ^ -gOQ ^dX) - - - " Nov Dec Jan Feb Mar Apr May Collection month Figure 2 Total densities o{ Anguilla rostrata glass eels (no. of eels/100 m^) by week during recruitment seasons, from 1985-86 to 1994-95. Larval density is proportional to circle area: smallest density (0.034) at week 2 of season 1990-91; largest density (3.690) at week 17 of 1986-87 season, n = no sampling and c = sampling with zero catch. Table 1 Linear regression coefficient < for total length m collec ion week for glass -eel stages of American eel taken at Pi vers Is and. Beau- fort. North C arolina, in each of 10 recruitment seasons. with 95'^'f confidence intei-vals for slopes and in tercepts. P IS the probability of far ing to 1 eject the H„ All years: Y= 52.15 -f 0.088A' /■2=0.026). Recruitment year n Adjusted r- Intercept Slope F P 1985- -86 34 0.19 48.92 ±3.48 0.24 ±0.15 1.58 0.245 1986- -87 61 0.03 50.87 ±2.94 0.15 ±0.14 2.87 0.095 1987- -88 73 0.11 50.63 ±2.49 0.19 ±0.12 10.21 0,002 1988- -89 81 0.02 52.65 ±2.41 0.09 ±0.07 2.27 0.136 1989- -90 48 0.00 54.45 ±2.80 -0.09 ±0.08 1.03 0.315 1990- -91 15 0.07 53.18 ±5.24 -0,01±0.39 0.00 0.9,58 1991- -92 60 0.33 48.82 ±1.70 0.27 ±0.10 30.66 0.001 1992- -93 60 0.01 54.11 ±4.00 -0.07 ±0.07 1.37 0.247 1993- -94 18 0.02 54.94 ±1.48 -0.09 ±0.09 3.70 0.057 1994 -95 132 0.01 52.97 ±1.55 0.07 ±0.08 2.40 0.124 incremental widths (1.717 and 1.732 ) in the later stage of the glass eel (increments 30-35), were wider (F.,=3.93 , P=0.004) than the con-esponding mean width (1.314 ) for New Brunswick otoliths, as hypothesized, suggesting fast- er growth in the southern waters. The youngest (mean 168 d) Ainerican eel elvers were found at the Pivers Island site (Table 3), whereas elvers at Black Creek (just north of Pivers Island) were not sig- nificantly different at age 175 d (F.j=1.125, P=0.376). Nei- ther elvers from New Jersey (201 d) nor those from New Powles and Warlen: Recruitment season, size, and age of Anguilla rostrata entering an estuary near Beaufort 303 Table 2 Mean increment widths ( microns! with 95'7r confidence limits in the glass- eel growth zone o{ Anguilla rostrata otoliths from three difTcrent sampling sites and the number showing the "elver" or "transition ' mark. Number in sample {n ) applies to both increment groups. Means without a letter in common (y and z) were significantly different (P <0.05). Width (;/) of Width (;i ) of Presence of elver mark Sample site n 10-15th mcrement 30-35th increment (%) Pivers Island, North Carolina 77 1.425 ±0.114 z 1.717 ±0.189 z 0 Black Creek, North Carolina 27 1.330 ± 0.169 z 1.732 ±0.180 z 93 Lepreau, New Brunswick 43 1.232 ±0.168 z 1.341 ±0.163 y 98 Table 3 Means (±SD) for total age, glass ages were significantly differen -eel age, and total length (TL) for young to between locations ( F-test, P<0.05 ). ee\s. Anguilla rostrata collected in our n = number of otoliths analyzed. study. Only glass-eel Location Date n Total age (d) Glass eel age (d) Length (TL mm) Pivers Island, North Carolina Dee-Apr 1985- 94 77 167.2 ±16.9 36.2 ±3.2* 52.9 ±2.4 Black Creek, North Carolina 2-23 Feb 1994 25 175.4 ±12.6 44.6 ±1.6* 55.9 ±3.1 Little Egg Inlet, New Jersey Feb-Mar 1994 22 201.2±16.1 62.3 ±8.8* 60.9 ±3.0 Lepreau, New Brunswick 6 June 1994 43 209.3±18.1 69.8 ±10.5* .58.1 ±3.8 Table 4 Regression of collection week ( 1-26) against total age in days (with 959; confidence limits) and regi'ession of glass-eel growth zone age (outer metamorphic zone to elver check on otolith) against total length ( mm) for glass eels collected at Pivers Island, North Carolina. The sampling period (recruitment season) extended from November to mid-May, 1985-86 to 1994-95. n = number of otoliths analyzed. n Slope Adjusted r'~ Intercept P Mean age Collection week against total age 89 0.339 0.053 Glass eel growth zone age against total length 89 -0.311 0.499 170.2 0.976 173.1 ±2.7 77.5 0.014 60.6 ±1.7 Brunswick ( 209 d ) were significantly older than those from North Carolina (^2=1. Ill, P=0.407; F2=1.140, P=0.373) with respect to total age. Although New Jersey fish ap- peared to be longer (60.9 mm TL) than New Brunswick el- vers (58.1 mm TL), the difference in mean length was not significant (F._;=1.283, P=0.266). Although total mean age (d) counts were not signifi- cantly different between sites (Table 3), glass-eel age (the portion between metamorphosis zone and elver mark of otolith) differed between all three sites (NC vs. NJ, F.,= 7.713, P=0.005; NJ vs. NB, F2=1.622, P=0.091). FurtheV more, when glass-eel ages were regressed against lati- tude, the relationship was significant (slope=0.303, adj. r-=0.83.5, P=0,028). In hght of Jessop's (1998) finding of a decrease in length over the recruitment season for elvers in the Bay of Fun- dy waters, we re-examined our age data by month and length. Could it be that even though our lengths showed no significant seasonal decrease, shorter fish were, on av- erage, older than longer fish? The regression of the capture date against total mean age suggested an increase in age over time (Table 4), but the relationship was weak (adj. /■^=0.053). Furthermore, an F-test revealed no difference in incremental total age between any month. But total age is apparently not to be trusted (Cieri and McCleave, 2000). When we regressed mean length against only the glass- eel zone age, it revealed just the opposite relationship: a negative slope (Table 4), with a good relationship (adj. r'^=0A99}. The smaller fish during the recruitment period were indeed older, as suggested by Jessop ( 1998). The lon- ger fish, on average, had fewer increments in their glass- eel growth zone than the shorter fish, suggesting that the latter had taken longer to reach Beaufort estuary from the Sargasso Sea. 304 Fishery Bulletin 100(2) Jan Feb Mar Collection monttn Figure 3 The regression of total length on week of capture for Anguilla rostrata glass eels collected at the Pivers Island bridge during 10 recruitment seasons from 1985-86 to 1994-95. Boundary lines represent 95*^ confidence limits. Discussion Glass eels were recruited to the Beaufort estuary over a 7-nionth period from November to early May during the ten years from 1985-86 to 1994-95. Peak recruitment occurred in February-March of most years. Catch distribu- tions were usually skewed, either positively or negatively, and clumped. Variation in estuarine recruitment time may have been due to differences in oceanic transport rates for larvae, and perhaps to slight differences in times of peak spawning, a well-known phenomenon in many other fish species. Able and Fahay ( 1998 ) also found that most glass eels recruited to a New Jersey estuary from January to June (in 1991 to July I, with peaks usually in February and March. In Nova Scotia the main fishery for elvers in 1997 extended from late April to mid-August (Jessop. 19981. The recruitment period for the two southernmost localities and Nova Scotia thus overlapped in April and May: as the North Carolina and New Jersey recruitment period ends, the northern Bay of Fundy area typically commences. Although there was a large variation in the numbers of eels caught among years, there was no indication of an overall reduction in recruitment. Castonguay et al. (1994b), and Marcogliese et al. (1997) have suggested a number of possible reasons for a decline of European eels, but such a decline was not reflected in our catches of elvers at Beaufort. Jessop (1997) also concluded that there had not been a decline in elvers recruited to Atlantic coastal waters of the Canadian maritime provinces over the past decade (last year 1996). The lengths of elvers captured in the tidal net at Pivers Island fell within the expected latitudinal length range reported by Haro and Kreuger (1988) and were similar to the 49-56 mm TL length range for glass eels from a river in Georgia (Helfman et al., 1984). Mixed elvers and small juveniles (not aged) in the Great Bay area of New Jersey, from 1986 to 1994 averaged 60-90 mm TL (Able and Fahay, 1998). Elvers collected by Jessop ( 1998 ) in 1997 in the Bay of Fundy waters ranged from 52.0 to 70.0 mm TL (considerable variation was evident among rivers of New Brunswick and Nova Scotia. ). Even farther to the north (Quebec), Michaud et al. (1988) reported lengths of elvers for 1987 from the St. Lawrence River estuary to ex- tend from 50 to 70 mm TL. Comparing our data with that of other studies, and allowing for 2.8'7f shrinkage in length of our eels (unpubl. data) due to alcohol preservation (our samples), as opposed to 1-2"^ shrinkage in 6-10'/( forma- lin (Jessop, 1998; Stobo, 1972), we found that there was a remarkable consistency in the clinal length ranges from north to south. Some inconsistency was expected because we compared studies in which sampling was done at both river mouths and estuaries, where it is known that as- cent up rivers is usually dependent upon temperature or stream level (Martin, 1995; Jessop, 1998). In a study of the Bay of Fundy waters, Jessop (1998) showed that as the season progressed, lengths of young eels entering most, but not all, rivers declined to a maxi- mum of 7'^(. The length of recruiting European eel elvers also declined seasonally (Cantrelle, 1981). Yet our com- bined length-frequency plot of all fish collected over the ten-year period showed a nonsignificant regression. It may be that only elvers to the north of our location show a de- finitive decrease in lengths over time or that our samples were too small to confirm the phenomenon. Powles and Warlen; Recruitment season, size, and age of /^Dyu/Z/o /os^/o/o entering an estuary near Beaufort 305 The difference in estimated mean age of 8 days from glass-eel growth zones at two North Carolina locations about 9.5 km apart implies that upriver movement was about 1 km a day, much of which could be tidal. We do not know whether or not this rate of movement is reasonable or tjrpical. The mean age from our glass-eel growth zones on oto- liths of glass eels agreed with the ages determined from other glass eels collected in North Carolina (Wang and Tz- eng, 1998). Finally, the apparent total ages and glass-eel ages from North Carolina, New Jersey, and New Brunswick support the hypothesis that eels are progressively older to the north, and are consistent with their greater distance from the Sargasso Sea. Total age estimates of glass eels can be compared with those from two other studies of American eel. Our average total age of glass eels from Black Creek (175 d) was 14% older than the 153 d that Budimawan (1996) found for the same location, date, and year. On the other hand, estimated total ages of glass eels were considerably less (20-22'>; ) than the ages obtained by Wang and Tzeng (1998) for eels from the same general areas (but different dates). Because ages from glass-eel growth zones (as opposed to total age) were in very close agreement with those of Wang and Tzeng (1998), one assumption for this inconsistency in total age is a dif- ference in intei-preting age in either the leptocephalus or metamorphic zone. It is also possible that increments <0.3 p wide may not have been distinguishable by light microscopy, which could account for our ages being younger than those of Wang and Tzeng ( 1998), who used SEM. Antunes and Tesch 1 1997) have also suggested that these two methods of otolith observation may produce different results. It is conceivable that increments in the leptocephalus growth zone may be laid down in fewer numbers in some seasons, but it seems unlikely that low water temperatures would cause a cessa- tion of otolith increment deposition, because the Gulf Stream top 200-m layer (the assumed habitat of leptocephali ) main- tains an annual temperature above 16°C (Stommel, 1965). Absorption of increments during metamorphosis ( Cieri and McCleave, 2000) is possible, but our studies could not shed any light on this proposed phenomenon because all lepto- cephali had metamorphosed before reaching North Caroli- na. Obviously, there is consistent agreement among authors in incremental counts of glass-eel growth zones, which are wide and clear, and therefore easy to count. Regardless of the discrepancy in age estimates of larval eels, evidence from the rate of transport of other species in the Gulf Stream suggests that once they have reached the Gulf Stream, it is possible for leptocephali to reach the continental shelf of North Carolina in about 110 days. A NOAA (National Oceanic and Atmospheric Administra- tion) sea drifter released in 1996 west of the western edge of the Sargasso Sea in 1996 and travelling just inside the western front of the Gulf Stream at 100 m depth reached the North Carolina outer continental shelf in 50 days. Its release site appears to overlap stations where leptocephali of A. rosti-ata were captured (Wippelhauser et al., 1985; Kleckner and McCleave, 1985). Larvae of several fish spe- cies from Florida waters are known to reach the area off- shore North Carolina in 25-30 d (Hare and Ahrenholz).! For example, larval Atlantic menhaden (Brevoortia tyrcin- nux) that spawned offshore from Beaufort in an area close to the path of the sea drifter reached the estuary near Beaufort, North Carolina, in an average of 60 d (Warlen, 1994). This estimate of across-continental shelf transport, when added to the 50 d of travel for the sea drifter, gives an elapsed time of 110 d. This estimate of transport duration may prove to be useful in the future. McCleave et al. (1998) surmised that leptocephali of the European eel (virtually identical in physical structure to A. rostfata ) do not actively .swim toward land during their migration to Europe. Granted that the age increments in the larval zone of the otolith of American eels may not ac- curately depict early life-history duration (Cieri and Mc- Cleave, 2000), the consistency of recruited lengths and glass-eel ages over the years from 1985 to 1995 is note- worthy. In addition, the annual recruitment seasons and sizes observed in coastal North Carolina and New Jersey regions (Able and Fahay, 1998) reveal a remarkably consis- tent sequential latitudinal pattern. It appears that current data on seasonal recruitment of young eels should contin- ue to be used to examine temporal relationships between geographical areas and oceanic transport times from the spawning area (Sargasso Sea) to the estuaries. Acknowledgments We thank Betsy Laban for sectioning and for instruction in processing initial samples of otoliths. Beverley Powles removed and prepared the remainder of our otoliths. Dick Peterson, Department of Fisheries and Oceans, St, Andrews, procured our New Brunswick sample. Kenneth Able furnished the New Jersey sample. Dean Ahrenholz provided advice and aging facilities at the Beaufort Labo- ratory. We also thank Trent University for research support and facilities. The (then) Director of the Beaufort Labora- tory, Bud Cross, offered facilities to the senior author, and Director Don Hoss suggested valuable comments on the manuscript. A number of other staff members, were very helpful: in particular Curtis Lewis (figures), David Colby ( statistics ), Peter Hanson ( SAS ), Patti Marraro ( literature), Jon Hare (seabed drifter, oceanographic current patterns), Jeff Govoni (manuscript), and Allyn Powell (anguillid biol- ogy). Martin Castonguay offered helpful comments, as did Raymonde Lecomte. Yves Desaunay provided information on the "transition ring" in the European eel, and Jim McCleave offered many obsei-vations and suggestions. We also thank the anonymous referees of the manuscript. Literature cited Able, K. W., and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the Middle Atlantic Bight: chapter 6. Anguilla rostrata (Le- ' Hare,J.,andD. Ahrenholtz. 1996. Personal comniun. Center for Coastal and Fisheries and Fisheries Habitat Research, 101 Pivers Island Road, Beaufort, NO 28516-9722. 306 Fishery Bulletin 100(2) Sueur I. American eel. Rutgers Univ. Press, New Bruns- wick, NJ, 342 p. Aiitunes, C, and F.-W Tesch. 1997. A critical consideration of the metamorphosis zone when identifying daily rings in otoliths of European eel, Anguilla anguilla (L.). Ecol. Freshwater Fish 6:102-107. Budimawan. 1996. Contribution a la connaissance des phases marines du developpement larvaire de cinq especes d'anguilles atlan- tiques et indo-pacifique: etude otolithometrique. These de doctoral de I'Universite de Perpignan, Perpignan, France. Cantrelle, I. 1981. Etude de la peche et de la migration des civelles i An- guilla anguilla) dans I'estuaire de la Gironde. These 3 cycle, Universite Pierre et Marie Curie (Paris VI). Castonguay, M. 1986. Growth of American and European eel leptocephali as revealed by otolith microstructure. Can. J. Zool. 65: 875-878. Castonguay, M., P V. Hodson, C. M. Caillard, M. S. Eckersley. J. D. Dutil. and G. Vesseault. 1994a. Why is recruitment of the American eel, Anguilla rostrata. declining in the St. Lawrence River, and Gulf? Can. J. Fish. Aquat. Sci. 51:479-488. Castonguay. M., P. V. Hodson, C. Moriarty, K. Drinkwater, and B. M.Jessup. 1994b. Is there a role of ocean environment in American and European eel decline? Fish. Oceanogr. 3(3):197-203. Cieri, M. D., and J. D. McCleave. 2000. Discrepancies between lar\'al and juvenile otoliths of the American eel (Anguilla rostratat: is there something fishy going on at metamorphosis? J. Fish. 57:1189-1198. Comparini, A., and E. Rodino. 1982. Electrophorctic evidence for two species of Anguilla leptocephali in the Sargasso Sea. Nature 287:43.5-437. Eldied, B. 1968. Larvae and glass eels of the American freshwater eel, Anguilla rostrata (LeSueur, 18171, in Florida waters. Fla. Board Conserv. Mar. Res. Lab. Leaflet. Ser. 4, No. 9:1-4. Haro. A. J., and W. H. Kreuger. 1988. Pigmentation, size, and migration of elvers (Anguilla rostrata (LeSueur)) in a coastal Rhode Island Stream. Can. J. Zool. 66:2528-2533. Helfman, G. S., E. L. Bozeman, and E. B. Brothers. 1984. Size, age, and sex of American eels in a Georgia river. Trans. Am. Fish. Soc. 113:132-141. Jessop, B. M. 1997. The biological characteristics of, and efficiency of dip- net fishing for, American eel elvers in the East River, Ches- ter, Nova Scotia. Diadromous Fish Division Doc. 97-01. Department of Fisheries and Oceans, Canada. Jessop, B. M. 1998. Geographic and seasonal variation in biological char- acteristics of American eel elvers in the Bay of Fundy area and on the Atlantic coast of Nova Scotia. Can. J. Zool. 76:2172-2185. Kleckner, R. C, J. D. McCleave, and G. S. Wippelhauser. 1983. Spawning of American ee\. Anguilla rostrata. relative to thermal fronts in the Sarga.sso Sea. Env. Biol. Fishes 9, No. 3/4:289-293. Kleckner, R. C, and J. D. McCleave. 1985. Spatial and temporal distribution of American eel larvae in relation to North Atlantic Ocean current systems. Dana 4:67-92. Lecomte-Finiger, R. 1992. Growth, life history and age at recruitment of Euro- pean glass eels (Anguilla anguilla ) as revealed by otolith microstructure. Mar. Biol. 114:205-210. Marcogliese, L. A., J. Casselman, and P. V. Hodson. 1997. Dramatic declines in recruitment of American eel (Anguilla rostrata ) entering Lake Ontario — long-term trends, causes and effects. Http://www.cciw.ca/eman-temp/reports/ publications/nm97_abstracts/part-22.htm (last viewed Febru- ary 14, 2001). Martin, M. H. 1995. Validation of daily growth increments m the otoliths of Anguilla rostrata iLeSueuri elvers. Can. J. Zool. 73:208- 211. McCleave, J. D., R. C. Kleckner. and M. Castonquay 1987. Reproductive sympatry of American and European eels and implications for migration and ta.xonomy Am. Fish. Soc. Symposium 1:286-297. McCleave, J. D., R J. Brickley, K. M. O'Brien, D. A. Kistner, M. W. Wong, M. Gallagher, and S. M. Watson. 1998. Do leptocephali of the European eel swim to reach continental waters? Status of the question. J. Mar Biol. Assoc. U. K. 78:285-306. Michaud, M., J.-D. Dutil, and J. J. Dodson. 1988. Determination of the age of young American eels, Anguilla rostrata. in freshwater, based on otolith surface area and microstructure. J. Fish. Biol. 32:179-189. SAS.Institute, Inc. 1990. SAS/GRAPH software, version 6, first edition. SAS Institute, Inc, Cary, NC, 664 p. Schmidt, J. 1922. The breeding places of the eel. Phil. Trans. R. Soc. London 6,211:179-211. Stobo, W. T 1972. Effects of formalin on the length and weight of yellow perch. Trans. Am. Fish. Soc. 101:362-364. Stommel, H. 1965. The Gulf Stream: a physical and dynamical de.scrip- tion. 2"'' ed. Univ. California Press, Berkeley and Los An- geles, CA, 248 p. Tzeng, W-N. 1996. Effects of salinity and ontogenetic movement on stron- tium:calcium ratio in otoliths of the Japanese eel. Anguilla japonica Temminck & Schlegel. J. Exp. Mar. Biol. Ecol. 199: 111-122. L'mezawa, A., and K. Tsukamoto. 1991. Factors influencing otolith increment formation in Japanese eel, Anguilla Japonica T. & Schlegel elvers. J. Fish Biol. 39:211-223. Vladykov, V 1966. Remarks on the American eel (Anguilla rostrata LeSueur). Sizes of elvers entering streams; the relative abundance of adult males and females; and present eco- nomic importance of eels in North America. Verb. Int. Verein. Limnol. 16:1007-1017. Wang. C-H.. and W-N. Tzeng. 1998. Interpretation of geogi'aphic variation in size of Amer- ican eel, Anguilla rostrata elvers on the Atlantic coast of North America using their life history and otolith ageing. Mar. Ecol. Prog. Ser. 168:35-43. Warlen, S. M. 1994. Spawning time and recruitment dynamics of lan'al Atlantic menhaden, Biwoortia tyrannus. into a North Car- oHna estuary Fish. Bull. 92:420-433. Wippelhauser. G. S., J. D. McCleave, and R. C. Kleckner. 1985. Anguilla rostrata leptocephali in the Sargasso Sea during February and March 1981. Dana 4: 93-98. 307 Abstract— Triennial lidltoni tnnvl sur- \ (•>• data from 1984 to 1996 were used to I'valuati' changes in the summer distri- bution of walleye pollock in the western and central Gulf of Alaska, Ditfercnces between several age groups of pollock were evaluated. Distribution was exam- ined in relation to several physical char- acteristics, including bottom depth and distance from land. Interspecies associa- tions were also analyzed with the Bray- Curtis clustering technique to better iHiderstand community structure. Oiu' ri'sults indicated that although the pop- ulation numbers decreased, high con- centrations of pollock remained in the same areas during 1984-96. However, there was an increase in the number of stations where low-density pollock con- centrations of all ages were observed, which resulted in a decrease in mean population density of pollock within the GOA region. Patterns emerging from our data suggested an alternative to Mac- Call's "basin hypothesis" which states that as population numbers decrease, there should be a contraction of the pop- ulation range to optimal habitats. During 1984-96 there was a concur- rent precipitous decline in .Steller sea lions in the Gulf of Alaska. The results of our study suggest that decreases in the mean density of adult pollock, the main food in the Steller sea Hon diet, combined with slight changes in the distribution of pollock (age-1 pollock in particular) in the mid-1980s, may have contributed to decreased foraging effic- iency in Steller sea lions. Our results support the prevailing conceptual model for pollock ontogeny, although there is evidence that substantial spawning may also occur outside of Shelikof Strait. Changes over time in the spatial distribution of walleye pollock (Theragro chalcogramma) in the Gulf of Alaska, 1984-1996 Michiyo Shima School of Aquatic and Fishery Sciences University of Washington 1122 Boat St NE Seattle, Washington 98105 E mail address mshimatOu Washington edu Anne Babcock Hollowed Alaska Fishenes Science Center 7600 Sand Point Way NE BinC15700 Seattle, Washington 981 15 Glenn R. VanBlaricom Washington Cooperative Fish and Wildlife Research Unit School of Aquatic and Fishery Sciences University of Washington Box 355020 Seattle, Washington 98195 Manuscript accepted 21 September 2001. Fish. Bull. 100:307-323 (2002). The incorporation of space into ecolog- ical models has increased our knowl- edge of the competition and co-existence among species (Steinberg and Kareiva, 1997). It has also shown that the dis- tributions of most organisms are not homogeneous throughout their range, but rather occur as patches. Ajialyses of predator-prey interactions and food- web dynamics are enhanced by consid- ering the habitat association of species. A first step towards this goal is to describe the spatial distributions of the major players in an ecosystem. The purpose of our study was to ex- amine changes in walleye pollock (T/ier- agra chalcogramma (Pallas, 1814)) dis- tribution. Walleye pollock (hereafter called "pollock") are a dominant spe- cies in the Gulf of Alaska (GOA). Pol- lock are the major prey of many species of flatfish and marine mammals and have been the predominant species in Steller sea lion [Eumctopias jiibatus (Schreber, 1776)) diets since at least the mid-1970s (Pitcher, 1981; Merrick and Calkins, 1996). Numbers of Stell- er sea lions have declined significantly in recent decades in the western and central GOA. Examination of tempo- ral and spatial changes in the abun- dance and population structure of pol- lock is important to understanding the decline in sea lion numbers. Existing and proposed management decisions to protect Steller sea lions include tempo- ral and spatial changes to pollock fish- ing quotas. The GOA pollock population peaked in the early 1980s as a result of sever- al strong year classes in the late 1970s (Fig. lA). Since then, the population has had fewer strong year classes. Mod- eled estimates of total biomass of the population declined in the early 1980s and leveled off in the 1990s (Fig. IB). Bottom trawl survey biomass estimates (since 1984) have not reflected the de- cline in the early 1980s (Fig. 10). In- stead, they showed that the demersal fraction of the population changed lit- tle. Compared with other gadoid stocks worldwide, the GOA pollock stock has been lightly exploited; average landings have constituted less than 10% of the stock biomass during the last 20 years (Shima etal,, 2000). In this study we document changes in the spatial distribution and popula- tion density of juvenile and adult pol- 308 Fishery Bulletin 100(2) lock in the western and central GOA (defined as the area between 147°W and 170°W), from 1984 to 1996. Catch per unit of effort (CPUE) (number of fisli/km'-) was used as a measure of density throughout our analysis. We also examine changes in species associations and the distribution of pollock. Analysis is limited to the summer months because the source of our pollock data were sum- mer research surveys. 4000 - 3000 - 2000 1000 L1I1IJIIII....I1..II....I1 S-" 4000 o 3000 2000 g 1000 x-^^ \^''^ 4000 3000 2000 ^ 1000 Methods Data source Data were collected from triennial bottom trawl surveys in the GOA conducted by the Alaska Fisheries Science Center of the U.S. National Marine Fisheries Service (RACE divi- sion trawl sui-vey relational database, MundellM. The use of bottom trawl data to assess juvenile pollock abundance and distribution has many prece- dents. In the Bering Sea, bottom trawl sui-vey data have been used to determine estimates of age-1 pollock, which were used as indices of recruitment (Wespestad-). Others have used the same data to evaluate the relationship between juvenile pollock and the movement of the ice edge in the Bering Sea (Wyllie-Echever- ria and Wooster, 1998). Bottom trawl estimates have also been used to explore the relationships between different age groups of pollock in She- likof Strait (Schumacher and Kendall, 1991). Brodeur and Wilson (1996) modeled the depth distribution of pollock in the GOA during their first year and showed that age-1 pollock were found near bottom. Acoustic surveys of spawn- ing pollock in Shelikof Strait have detected age-1 pollock in both nearbottom and pelagic layers. These data suggest that some age-1 pol- lock may reside off bottom. Therefore, juvenile pollock data from bottom trawl surveys are used as an index rather than an absolute mea- sure of abundance (Bailey and Spring, 1992; Guttormsen and Wilson^). Survey design Triennial bottom trawl sui-veys were conducted during the summer months (May-Sep) of 1984, 1987, 1990, 1993, and 1996. These surveys have consistently been conducted in a strat- ified random sampling pattern (Martin and Clausen. 1995; Munro and Hoff 1995; Stark and Clausen, 1995; Martin, 1997). The GOA is divided into 49 strata according to water depth I I I I I .=»*'' .or ,.-\cP\.^\.-^ Year Figure 1 Data for walleye pollock in Gulf of Alaska, 1969-1997: pollock recruit- ment (A), biomass estimates from stock assessment analysis (B), and biomass estimates from triennial bottom trawl sui-veys IC). Data are from Dorn, M. W., A. B. Hollowed. E. Brown, B. Megrey. C. Wilson, and J. Blackburn. 2000. Walleye pollock. In Stock assessment and fishery evaluation report for the 2000 Gulf of Alaska groundfish fishery. 60 p. North Pacific Fishery Management Council. P.O. Box 10:3136. Anchor- age. Alaska 99.510. • Mundell. G. 1999. Personal commun. Alaska Fisheries Science Center, NMFS/NOAA, 7600 Sand Point Way NE, Seattle, WA 98115. - Wespestad. V. 199,5. Walleye pollock. //; Plan team for groundfish fisheries of the Bering Sea/ Aleutian Islands stock assessment and fishery eval- uation report for the groundfish resources of the Bering Sea/Aleutian Islands region as projected for 1996, 35 p. North Pacific Fishery Management Council, PO. Box 103136, Anchorage. AK 99510. 3 Guttormsen, M. A., and CD. Wilson. 1999. Echo integration-trawl survey results for walleye pollock in the Gulf of Alaska during 1998. In Stock assess- ment and fishery evaluation report for the 1998 Gulf of Alaska groundfish fishery, 25 p. North Pacific Fishery Management Council, 605 W 4"' Avenue, Suite 306, Anchorage, Alaska 99501. Shima et a\ Spatial dislributlon of Theragm chalcogramma in the Gulf of Alaska 309 Ji Kadiak Island Kodiak ■ area 630 Shumagin - area 610 Gulf of Alaska Figure 2 Example of the spatial coverage of the triennial bottom trawl survey for a sample year. Dots indicate station locations. and area boundaries defined by the North Pacific Fisheries Management Council (http://www.fakr.noaa.gov/rr/figures/ fig3.pdf). Allocation of sampling effort within each stra- tum was based on coefficients of variation, mean CPUE, and sampling densities for all fish species from data col- lected on previous triennial surveys (see Cochran, 1977). Sampling density within a stratum depended on the antic- ipated fish density of that particular stratum (Martin and Clausen, 1995; Stark and Clausen, 1995; Martin, 1997). Stations were prioritized within each stratum so that in the event of complications, sampling density would remain controlled. The total survey area was reduced by 7% after 1987 when stations deeper than 500 m were eliminated (Stark and Clausen, 1995). In the interest of consistency, only data from stations less than 500 m in bottom depth were used in our analyses. Station density was spread over a wide range; thus the survey data were most suit- able for analyses of geographic distribution and commu- nity composition on a large scale (Fig. 2). For each survey a Nor'eastern trawl (NT) with a 1.25 inch codend liner was used. The liner helped retain juve- nile pollock (Brown, 1986). The nylon and polyethylene NTs had net dimensions of 18.3 m wide by 4.7 m high and 18.3 m by 5.5 m, respectively. In 1984 and 1987 a Jap- anese bottom trawl (JBT) was also used. The horizontal opening of the JBT ranged from 19 to 30 m, the vertical opening from 3.2 to 3.3 m (Brown, 1986). CPUE from the JBT were corrected (using Tables 28 and 31 in Munro and Hoff, 1995) to account for differences between the catch- ability of the NTs and the JBT. Data analyses and statistical considerations Analyses were conducted on total pollock catch as well as on four size groups of pollock: <150 mm (age-0 or young- of-the-year); 150-230 mm (age-1 juveniles); 230-330 mm (age-2 juveniles); and >330 mm (adults). The length-at-age categories were based on a histogram of pollock lengths from a historical database of all NMFS research surveys conducted during the summer. The catchability of age-0 pollock may not have been as high as that for the other juvenile age classes and thus supports the interpretation of all juvenile data as index values (Bailey and Spring, 1992; Guttormsen and Wilson, 1999). Comparable numbers of length measurements were taken every triennial year but measurements were not taken at every station. Thus, anal- yses of different age groups were based on subsets of the data. The abundance offish in a specified age category was influenced by year-class strength (Table 1; Megrey et al., 1996; Hollowed et al.^). Strong year classes were present in 1984 (age 0), 1990 (age 2), and 1996 (age 2). Hollowed, A. B., E. Brown, B. Megrey, and C. Wilson. 1996. Walleye pollock. In Stock assessment and fishery evaluation report for the 1997 Gulf of Alaska groundfish fishery, 64 p. North Pacific Fishery Management Council, 605 W 4"' A. venue, Suite 306, Anchorage, AK 99501. 310 Fishery Bulletin 100(2) Table 1 Summary of year class strength of juvenile walleye pollock in bottom trawl survey years. 1984 1987 1990 1993 1996 AgeO Strong Weak Weak Weak Weak Age 1 Weak Weak Average Weak Weak/Average Age 2 Weak Strong/Average Strong Weak Strong Marine surveys typically yield data sets that are highly variable and contain a substantial proportion of zero catches, particularly when the data set is broken down into age groups (Stefansson, 1996). Thus, in some cases the data were analyzed in terms of presence and absence of pollock (hereafter termed "occurrence"). Because non- zero data often follow a lognormal distribution (Penning- ton, 1996), the lognormal transformations (log(CPUE-i-l)) of nonzero density data were used during our analyses. Changes in the depth distribution of pollock were an- alyzed by lOG-m bottom depth intei-vals, corresponding to the depth stratification used in the triennial surveys. Changes in geogi-aphic distribution of CPUE were eval- uated by GOA area boundaries, which again correspond- ed to the stratification design used during the triennial sui-veys. In our study, three areas were taken into con- sideration: "Shumagin" ( 1.59°-170°W, area 610), "Chirikof" (154°-159°W, area 620), and "Kodiak" ( 147°- 154° W, area 630). The distribution of pollock in relation to distance from land was evaluated in 20 nautical mile (nnii) increments by using GIS software (ESRI, Inc., 1996). However, the 20-nmi increments crossed the triennial sui-vey strata boundaries, which may have resulted in the disruption of the stratified random sampling scheme. By addressing pollock occurrence or density at each station, instead of biomass, we avoided the need for stratifying the data after collection. Pollock usually reside at depths less than 300 m within the shelf and slope regions of the GOA (NPFMC"'). Thus, the relatively narrow shelf region would limit the offshore distribution of pollock. Initial exploratory analyses involved three-dimensional contingency tables to assess the relationship between the occurrence of pollock and year, bottom depth, geographic region, and distance from land. In cases where the three- dimensional analyses rejected the null hypothesis of inde- pendence, more detailed partial contingency tables were constructed for the occurrence of pollock against each in- dividual variable. Othei' studies suggested that bottom depth and geographic location explain most of the vari- ability in fish distribution data (e.g. Overholtz and Tyler, 1985; Jav, 1996). ^ NPFMC (North Pacific Fishery Management Council). 1998. Essential fish habitat assessment report for the groundfish resources of the Gulf of Alaska region. NPFMC, 60.5 West 4"' Avenue, Suite 306, Anchorage, AK 99.501, 117 p. Single-factor ANOVAs were performed on the nonzero CPUE data to examine the sources of variation in the da- ta. The In(CPUE-i-l) of the nonzero data was used as the dependent variable among years. A separate ANOVA was calculated for each category of bottom depth, geographic region, and distance from land. Examination of the population density data revealed that for most age groups, the number of stations with a low density (<1000 fisli/km-) of pollock increased whereas the number of high-density stations remained relatively stable. This finding suggested that the distribution of suitable habitat for the demersal fraction of the popula- tion might have been expanding. Therefore, we examined the distribution of low-density concentrations of pollock separately in relation to the three physical character- istics across the survey years. Although ocean bottom temperature (OBT) was not measured at every survey station, available data were used to help interpret our re- sults. Mean bottom temperature was averaged over bot- tom depth intervals for each year to provide a general ovei'view (Fig. 3). Additional information on OBT has been summarized by S. Hare (http://www.iphc.washington.edu/ staff/hare/html/papers/OBT/obt.html ) from a variety of da- ta sources and averaged over five-year periods. Interspecies associations were examined with Bray- Curtis clustering techniques (Boesch, 1977; Walters and McPhail, 1982). Such classification techniques are useful in generating hypotheses about community structure, which may then be used to aid management actions (Cor- mack, 1971). Because the surveys used in the cluster analyses were conducted during the summer months, the results could not be extrapolated to other seasons. Trien- nial survey data were log-transformed and clustered by using the group average fusion strategy. For the species clusters, the top thirty species by weight were chosen in addition to the four age groups of pollock. After careful evaluation of all the cluster dendrograms, a common dis- similarity level (dissimilarity coefficient (A)=27) was cho- sen for all years as the level at which the most clearly defined clusters occurred throughout the triennial survey years. Diversity indices were calculated for all five sui-vey years. The top ten species by number of fish were com- bined for all years to make one list of species for which diversity was analyzed. Simpson's diversity index was cal- culated both in terms of richness and evenness (Simpson, 1949; Tokeshi, 1993). Richness was interpreted to mean "effective number of species," whereas evenness was un- Shima et al : Spatial distribLition of Themgra cha/cogramma in tine Gulf of Alaska 311 12, ^ 12 D 10 ' 10 8 y 8 6 4 t { i , =1 * 4 1 i ' 2 2j 0 n 200-300 300-500 0-100 100-200 200-300 300-500 0-100 100-200 1 12 B ^2 E temper 00 o ■ 10 8- Average bottom i 4 2- n - { 1 I 0-100 100-200 200-300 300-500 0-100 100-200 200-300 300-500 12- ^ 10- T 8- , 6 4^ ^ ^ • ■ . 2- 0 0-100 100-200 200-300 300-500 Bottom depth bins (m) Figure 3 Average bottom temperature by bottom depth category from data collected during the triennial survey (A)1984,(B)1987,(C)1990, (Dl 1993, lEi 1996. Error bars mdicate 1 SD. derstood to be the "distribution of numbers offish amongst those species." Confidence inter\'als were calculated by jack-knifing the diversity index (Magurran. 1988). Results All pollock Three-dimensional contingency tables indicated whether occurrence of pollock was mutually independent of all combinations of bottom depth, geographic region, dis- tance from land, and year. The three-dimensional con- tingency tables for all three null hypotheses of mutual independence resulted in rejection of the null hypothesis (P<0.001). The null hypotheses considered for the three- dimensional and partial contingency tables are listed in Table 2. A summary of the distribution of stations within categories of the three physical characteristics is given in Table 3. All the hypotheses of partial independence on combinations of two of the physical characteristics could be rejected (P< 0.001). When each physical characteristic was tested separately, the only hypothesis that could not be rejected (a=0.05) was the independence of pollock occur- rence against geographic region (P=0. 097, Table 2). Graphs of pollock occurrence versus each of the physical charac- teristics revealed distinctive patterns (Fig. 4). The graphs indicated that pollock are most frequently observed in the 100-200 m bottom depth interval and the 0-20 nmi dis- tance category (Fig. 4, A and B). However, this interpre- tation may be an artifact of the percentage of stations sampled within each category (Table 3). For all years, the 312 Fishery Bulletin 100(2) gi'eatest number of stations were sampled in the 100-200 m bottom-depth category and in the 0-20 nmi distance- from-land category. The results were standardized by taking the proportion of stations within each category that had a positive oc- currence of pollock ( Fig. 5 ). These graphs showed that for each bottom depth category, the proportion of positive oc- currences had increased over the years. The proportion of stations where pollock were present was consistently high in the 0-20 nmi category (Fig. 5). Farther from land, the proportion of positive stations increased between 1984 and 1996. In all three geogi'aphic regions, there was an increase in the proportion of stations where pollock were present. Table 4 lists the number of hauls with pollock occur- rence and summarizes the CPUE values for the nonzero data. Results from the single-factor ANOVAs calculated for the nonzero density data are summarized in Table 5. Significant differences were observed among most years. Table 2 Contingency table analysis of walleye pollock occurrence. Type of table P-value Three-dimensional tables of mutual independence Null hypothesis Pollock occurrence, year, and bottom depth are mutually independent P< 0.001 Pollock occurrence, year, and distance from land are mutually independent P< 0.001 Pollock occurrence, bottom depth, and region are mutually independent P< 0.001 Pollock occurrence, region, and distance from land are mutually independent P< 0.001 Three-dimensional tables of partial independence Null hypothesis Pollock occurrence is independent of year and bottom depth P< 0.001 Pollock occurrence is independent of year and distance from land P< 0.001 Pollock occurrence is independent of bottom depth and region P< 0.001 Pollock occurrence is independent of region and distance from land P< 0.001 Two-dimensional tables of partial independence Null hypothesis Pollock occurrence is independent of year P< 0.001 Pollock occurrence is independent of bottom depth P< 0,001 Pollock occurrence is independent of region P=0.097 Pollock occurrence is independent of distance from land P< 0.001 except for the Chirikof region and the distances from land ofO-20 and 40-60 nmi. Adult pollock The initial three-dimensional contingency tables of mutual independence between adult pollock occurrence and the three physical characteristics proved significant. Subse- quent contingency analyses revealed significant results for all partial independence tests except for the two-dimen- sional test for independence between occurrence of adult pollock and geographic region (P=0.58). Examination of all the positive data showed that the mean of the log-transformed CPUE decreased and the variance increased over time for bottom depths 100-200 m in each geogi-aphic area (Table 6). The mean CPUE also decreased over time for stations <40 nmi from land but there was no trend in the changes in variance (Table 6). The ANOVA confirmed that changes in the distribution of pollock from 1984 to 1996 were significant (a=0. 05) among years at all depth intei-vals, all geographic areas , and at a distance from land <60 nmi (Table 7). Thus, although the number of stations where pollock could be encountered in- creased (e.g. the low-density distribution data), the mean adult pollock density decreased and became more variable between 1984 and 1996 (Fig. 5, Table 6). Changes in the distribution of adult pollock were fur- ther examined by sorting the data into seven CPUE bins and by making histogi'ams of the frequency of stations within each bin (Fig. 6). The histogram revealed that over the years, the occurrence of stations with zero adult pollock CPUE decreased, whereas stations with low-den- sity (CPUE <1000 fish/km-) concentrations of pollock in- creased. Higher density stations did not exhibit any par- ticular trend. To better understand observed changes in the distribu- tion of low-density concentrations of adult pollock, all the stations were characterized by bottom depth, geographic location, and distance from land (Fig. 7). The number of stations that had low-density concentrations of adult pol- lock increased in all habitat categories over the years. The proportion of stations within each bottom depth bin with low-density concentrations of adult pollock more than dou- bled from 1990 to 1996. Comparable dramatic increases were obsei-ved with regard to distance from land and geo- graphic region. Juvenile pollock Contingency table analysis for juvenile pollock occurrence resulted in significant (P<0.001) results across all ages for tests of mutual independence among all the physical char- acteristics. Two-dimensional contingency analyses also resulted in significant results. It was difficult to discern any chronological trends in the mean and variance of all the positive data for juvenile pollock. Minor fluctuations in juvenile pollock density may have been related to interan- nual differences in year-class strength (Table 1). Histograms similar to those for adult pollock were made for all three juvenile age groups of pollock, where juvenile Shima et a\ Spatial distribution of Theragra chalcogiamma in the Gulf of Alaska 313 I4(}l) 12(HI KKK) «IK) 600 400 2(I<1 0 B It Int l(K)-2(X) :00-30() .iOO-5«) Bottom depth (m) 0-20 20-40 40-60 Distance from land (nmi) 1400 -, £) 1200 - JllljJni JnJni Year Sliuma^'in Chinknt Geograptiic region Figure 4 Tfie number of stations at which walleye pollock occurred by physical characteristic; (A) bottom depth (ml; (B) distance from land inmi); (C) year; and (D) geographic region. Table 3 Summary of the percentage of stations within each bin for the three types of physical parameters discussed in the text. The number in parentheses indicates the total number of stations. Physical category Percentage of stations in category by year 1984(7491 1987 (648) 1990(5341 1993(6161 1996(6131 Bottom depth (ml 0-100 29.9 33.7 23.7 32.8 35.4 100-200 46.1 53.7 59.1 50.9 43.2 200-300 16.5 9.6 13.8 12.0 14.7 300-500 7.4 3.0 3.4 4.3 6.7 Geographic location Shumagin 33.2 28.5 26.7 29.3 33.7 Chirikof 27.5 32.5 30.8 29.5 30.9 Kodiak 39.3 40.0 42.5 41.2 35.4 Distance from land category (nmi) 0-20 28.9 39.3 36.2 38.1 40.6 20-40 27.2 25.7 25.5 27.4 23.9 40-60 19.2 16.9 20.4 16.6 14.3 60+ 24.7 18.1 17.9 17.8 21.1 pollock density was parsed into bins and plotted (Fig. 8). The same trend seen in the adults was also seen with the juvenile pollock. As the years progressed, there was an in- crease in the proportion of stations that had low-density (<1000 fish/kni-) concentrations of all juvenile age groups of pollock. For age-0 and age-1 pollock, there was about a tenfold difference in the proportion of stations with low- density concentrations after 1987 (Fig. 8 ). Age-2 pollock 314 Fishery Bulletin 100(2) Table 4 Summary information of walleye pollock occurrence and CPUE by ti-iennial sui-vcy year CPUE V alues are given in number of fisli/km- and summarize only the nonzero positive tows. Year Number of hauls - CPUE Total Pollock present Minimum Maximum Mean SD 1984 672 375 11.48 397079.10 9759.45 31564.10 1987 603 412 13.85 602901.72 9640.16 43282.89 1990 506 386 16.77 130009.85 6866.74 15946.86 1993 583 454 19.03 218130.41 4994.91 15947.79 1996 593 497 .32 22 239402.07 5434.23 18829.21 Table 5 Single-factor ANOVAs of InlCPUE + 11 of positive walleye pollock catches among years by 3ottom depth category, 1 geographic region, an 1 distance fi ■om land. SS = sum of squares; MS = mean square. SS df MS F P Bottom depth (m) 0-100 73.60 4 18.40 3.37 0.01 100-200 91.04 4 22.76 4,24 <0.001 200-300 77.22 4 19.31 6.47 <0.001 300-500 19.00 4 4.75 3,19 0,02 Geographic area Shumagin 84.68 4 21.17 3,94 <0,001 Chirikof 24.29 4 6.07 1,16 0,33 Kodiak 185.06 4 46,27 10.47 <0.001 Distance offshore ( nm I 0-20 23.80 4 5,95 1.21 0.30 20-40 71.94 4 17,99 3.51 0.01 40-60 29.94 4 7,49 1.66 0.16 60+ 56.66 4 14.17 3.99 <0.001 also exhibited this trend but not to the same degree as age-0 and age-1 pollock. Stations with low-density concentrations of juvenile pol- lock were examined with respect to bottom depth, geo- graphic area, and distance from land (Fig. 9). For all ages of juvenile pollock there was a marked difference between 1984-87 and the 1990s. A gi'eater proportion of stations with age-0 pollock was observed at bottom depths <200 m than at deeper depths, especially in 1996. Age-0 pollock occurrence increased in the Chirikof and Shumagin re- gions before also increasing in the Kodiak region in 1996. Age-0 pollock also occurred more frequently at distances <60 nmi from land. In 1990, the proportion of stations with low-density con- centrations of age-1 pollock increased the most at bottom depths >200 m, before the same increasing trend was al- so observed at shallower bottom depths in 1993 and 1996 I E1I>IS4 ■ 1 1)87 DIWO BIW1 aiWh U)li-200 2(lll-.l(Kl Bottom deptti (m) 1 1 ^ zn 1 ^ 1 1 E Shumagin fhirikot Geographic region = B ^ ^1 — 1 ^ 1 :()-4)) 4()-M) Distance from land (nmi) Figure 5 The proportion of stations within each category of physical characteristic where walleye pollock were found . (Fig. 9). The increase was observed across all geographic regions and at all distances from land. Shima et a\ Spatial distribution of Theragra chalcogramma in the Gulf of Alaska 315 Table 6 Mean and variance estimates of ln(CPUE+l I for 1984-96 for nonz.ero adult pollock data by pli ysical charactt ristic. The number of | observations is given in parentheses. Pbysical characteristic Mean Variance 1984 1987 1990 1993 1996 1984 1987 1990 1993 1996 Bottom depth m) 0-100 8.05 (24) 8.59(37) 7.05(23) 6.73(74) 6.41(108) 3.45 2.80 3.87 7.71 5.38 100-200 8.84(121) 8.66(110) 8.17(158) 6.74 (204) 6.20(192) 2.77 2.43 2.56 3.81 4.23 200-300 7.82(52) 8.14(24) 7.68(60) 6.58(64) 5.88(82) 2.40 0.70 1.59 1.68 1.69 300-500 7.02(1) 0 (1) 7.43(1) 5.28(12) 4.70(30) 0 0 0 0.30 0.88 Geographic area Shumagin 8.80(63) 8.45(37) 7.40(33) 6.63(1011 6.32(121) 2.65 2..35 2.30 5.54 3.72 Chirikof 8.09 (57) 8.62(49) 7.91 (52) 6.65(83) 5.74(125) 3.64 2.47 2.65 3.23 3.71 Kodiak 8.47(78) 8.62(821 8.22(1.34) 6.91 (138) 6.42(138) 2.53 2.23 2.50 3.94 4.21 Distance from land inmi) 0-20 8.49(141) 8.67(140) 8.04(161) 7.02(236) 6.40(269) 2.90 2.15 2.76 4.42 4.28 20-40 8.50(42) 7.76(19) 7.59(57) 5.79(96) 6.12(86) 3.57 2.33 1.93 2.75 1.36 40-60 8.16(13) 8.66(10) 5.31(16) 6.23(20) 8.00(40) 1.88 3.46 2.20 2.52 2.24 60+ 5.38(2) 5.95(2) 8.26(8) 6.66(10) 5.66(7) 4.54 4.55 2.. 53 4.42 2.55 Q19$4 ■1987 □ 1990 11993 B1996 §5^ S 5 CPUE (number of fish/km2) Figure 6 Porportion of stations for different density levels of adult wal eye pollock. There was a general increase over time of low-density concentrations of age-2 pollock with respect to bottom depth and geographic region (Fig. 9). The increase in the proportion of low-density concentrations of age-2 pollock was the greatest 20-60 nmi from land until 1996, when the proportion of stations with low-density concentrations increased the most >60 nmi from land. Community composition Results of clustering by species by using the triennial bottom trawl data are presented in Table 8. The top 30 species or species types (e.g. adult and juvenile pollock) Table 7 Single-factor AN OVAs of ln(CPUE-i-l of positive adult walleye pollock catches among years by bottom depth category, geographic region. and distance from land. SS = sum of squares; MS = mean squ are. SS df MS F P Bottom depth l m ) 0-100 162.23 4 40.56 7.55 <0.001 100-200 872.70 4 218.17 65.99 <0.001 200-300 196.38 4 49.09 30.76 <0.001 300-500 37.80 4 9.45 13.04 <0.001 Geographic area Shumagin 345.66 4 86.42 22.87 <0.00] Chirikof 448.03 4 112.01 34.19 <0.001 Kodiak 453.24 4 113.31 35.09 <0.001 Distance (nmi I 0-20 758.52 4 189.63 53.61 <0.001 20-40 417.66 4 104.41 40.42 <0.001 40-60 174.97 4 43.74 18.59 <0.001 60+ .30.82 4 7.70 2.48 0.071 by weight over all sui^ey years were chosen. Adult pol- lock were consistently associated over the years most closely with flathead sole (Table 8). The cluster containing adult pollock usually included 3-6 other species of impor- tance, either commercially or in terms of abundance. These included arrowtooth flounder {Atheix'sthes stomias (Jordan 316 Fishery Bulletin 100(2) |QI984 ■ 19871 D 1 t)9(l Blvwl GW% ! 100-200 Bottom depth (m) B g a. 20^0 nmi Distance from land (nmi) 0.2 - o^^ Shumagin Chinkof Kodiak Geographic region Figure 7 Proportion of stations where low-density concentrations I < 1000 fish/km- 1 of adult walleye pollock were obsei'ved, binned by bottom depth (A), distance from land (B), and geographic region (C). and Gilbert, 1880)), Pacific halibut iHippoglossus stenole- pis Schmidt, 1904), Pacific cod iGadux macrocephalus Tile- sius, 1810), sablefish (Anoplopoma fimbria (Pallas, 1814)), Dover sole (Microstomus pacificus (Lockington, 1879)), and rex sole (Glyptocephalus zachirus Lockington, 1879). This small cluster was usually isolated from the larger main cluster (i.e. high dissimilarity between the two clusters). Adult pollock were also clustered separately from the juve- nile pollock age groups. In all years except 1987, age-1 and age-2 pollock were clustered with each other, whereas age-0 pollock were clustered separately. In 1987, age-0 and age-1 pollock clustered together, although not as closely associated as ages 1 and 2 in the other years. Until 1996, age-0 pollock clustered with Pacific sleeper shark (Som- BI984 ■ 1987 D1990 ■ 1993 B1996 CD ^> CD c= C5 C2 CD C^ — — C=y CD _JiB — CD — CD C3 CD CD =^ ^ CD CD CD — — C=^ CD ■ — CD ; CPUE (number of fish/km^) Figure 8 Proportion of stations with different levels of juvenile wall- eye pollock CPUE; age 0 (A), age 1 (B), and age 2 (C). niuaus pacificus Bigelow and Schroeder, 1944), Aleutian skate (Bathyraja aleutica (Gilbert, 1896)), or silvergray rockfish iSebastes brevispinis (Bean, 1884)). In the larger cluster that contained all the juvenile pollock age groups, it was common to find Pacific herring iClupea pallasii (Cuvier and Valenciennes, 1847)) in all years. Cluster analysis by station of the triennial data result- ed in clusters that fell into clean zoogeogi'aphic groups that followed depth contoiu-s. Maps of clusters identified seven groups of species that could be tracked in the GOA throughout most years (Table 9). Groups 1, 2, and 5 were present only in 4 out of the 5 triennial survey years. In 1984, groups 2 and 5 were absent whereas in 1990, group 1 was absent. The species within each cluster were listed m order of dominance. Although many of the clusters had Shima el al Spatial distiibution of Thciagia cha/cogramma in the Gulf of Alaska 317 ((.5 11,4 (1.2 II I 0 0 5 DA 0,3 o: II I 0 a 1984 m\w7 D1990 a 1991 H 1996 0-100 100-200 200+ B s^« 0-100 100-200 200+ 0-100 100-200 200+ Bottom depth (m) 0,5 0,4 0,3 n: 11 1 0.5 0,4 03 I) 2 60+ 0-20 20-40 40-60 60+ 0-20 20-40 40-60 60+ Distance from land (nmi) Siluinagm Kodiak Shumagin Ctiinkof Kodiak Shumagin Chinkof Kodiak Geographic region Figure 9 Proportion of stations with low-density concentrations (<1000 fisli/lim-) of juvenile pollock by bottom depth (first column), distance from land (second column), and geographic region (third column) for age 0 (A), age 1 (B), and age 2 (C) juveniles. the same main species in common, as listed in Table 9, differences in dominance distinguished the various clus- ters. Adult pollock were most dominant in groups 1, 3, and 6, which were either made up of nearshore or deep shelf stations (Table 9). Mean values of several environ- mental variables in these station groups in which pollock were found indicate that bottom depth and temperature varied (Table 9). Shallow water stations (groups 1 and 2) were grouped into warm (6.4°C) and average (5.7°C) bot- tom temperature clusters (Table 9). As expected, deeper water stations tended to be characterized by cooler tem- peratures. Juvenile (ages 0-2) pollock were found mostly in the nearshore stations. Diversity Separate diversity indices were calculated for habitat between 0-100 m bottom depth and 100-200 m bottom depth. In the 0-100 m habitat, there was a slight decrease in richness from 9-10 in the 1980s to 7-9 effective number of species in the 1990s (Fig. lOA). Calculations of even- ness for the same data indicated that there had been little change in the evenness component of species diversity in the 0-100 m habitat (Fig. IOC). Histograms of the propor- tion (in terms of number offish) contributed by each of the top species indicated that the decrease in richness might partially be explained by an increase in the predominance of pollock and a decrease in the presence of rockfish spe- cies (Fig. 11,A-E). Richness in the 100-200 m category increased from .5 to 10 effective numbers of species (Fig. lOB). The predomi- nance of pollock and arrowtooth flounder in the 1980s had shifted to include eulachon iThaleicht/iys pacificiis (Gi- rard, 1858)) and Pacific ocean perch iSebastes alutus (Gil- bert, 1890); Fig. 11, F-J). Evenness was relatively consis- tent throughout the survey years (Fig. lOD). 318 Fishery Bulletin 100(2) Discussion A major result of our study was that for all age gi-oups of pollock, the number of stations where pollock occurred at low densities (<1000 fish/km-) increased during 1984-96 while the mean density decreased. This pattern suggests that the pollock population was increasing its range, a characteristic often seen in growing populations that is consistent with the "basin hypothesis" iMacCall, 1990). MacCall hypothesized that at low-densities, marine fish will occupy habitats that are optimal for sui-vival. As pop- ulations grow, however, some portions of the populations will expand into locations of less suitable habitat quality. Our analyses of the data indicated that the demersal portion of the pollock population was stable but that the overall (pelagic and demersal) pollock population declined Table 8 Species associated with walleye pollock according to Bray- Curtis cluster analyses of triennial sui-vey data. The spe- cies are listed in order of association. Adults AgeO Ages 1 and 2 Flathead sole Pacific sleeper shark Atka mackerel Sablefish silvergray rockfish eulachon Dover sole lingcod Pacific herring Rex sole longnose skate bigmouth sculpin Arrowtooth flounder sharpchin rockfish lingcod Pacific halibut Pacific herring Chinook salmon Pacific cod during the study (1984-96, Fig. 1. B and C). Year-class strength was only sporadically strong and therefore would not account for the sustained increased positive occurrence by station during the early- to mid-1990s. Under MacCall's hypothesis, the range of the population should have been stable or contracting. Thus, patterns emerging from our da- ta are not entirely consistent with MacCall's model. Analysis of CPUE of all positive tows revealed that there was a decrease in mean CPUE over the years from 1984 to 1996. The data showed an increase in low-density concentrations of adult pollock stations across all catego- ries but in particular at bottom depths 200-300 m and in the Chirikof and Shumagin regions. We conclude that adult pollock had expanded into deeper water in the 1990s but that the expansion had resulted in a decrease of adult pollock in high-density stations, combined with decreased mean density of adult pollock throughout the region. We hypothesize that the expansion of pollock was due to an increase in the suitability of habitats for adult pollock, possibly caused by a spread in the distribution of pollock forage, during a period of stable or decreasing pollock pop- ulation trends. If predators of pollock rely on high-den- sity patches, the decreased mean density of pollock may have negative ramifications for the successful foraging of top predators. We refer to this as the "forage density hy- pothesis"; i.e. habitat suitability has changed, perhaps on a local scale, such that there has been an expansion of the overall population distribution. However, the expan- sion, combined with decreasing population numbers, had caused the density of pollock patches to decrease below a threshold at which top predators can successfully for- age. The underlying assumption here is that the predators need patches of high prey density rather than a uniform distribution of average or low prey density. Analysis of trends in the distribution and abundance of juvenile age groups is more difficult because data for Table 9 Description of consistent station groups designation "flatfish" includes all flatfish dard deviations are given in parentheses ound as a result of cluster except arrowtooth flounder analyses of the triennial bottom trawl survey data. The gi^oup Mean values are given for the environmental variables. Stan- Group no. Location Bottom depth (m) SSTi^C) Temperature at depth (°Cl Main species 1 Nearshore 1 8.5.9 (32.1) 9.7 (2.4) 5.7 (4.9) Age 2 and Adult pollock. Flatfish 2 Nearshore 2 8.3.7 (38.4) 9.4 (2.4) 6.4 (1.7) Adult pollock. Flatfish. Pacific cod 3 Shallow shelf 119.9 (47.7) 9.0 (2.5) 5.8 (1.9) Arrowtooth flounder. Adult pollock. Pacific halibut. Pacific cod 4 Deep shelf 120.6 (31.9) 9.5 (2.21 5.5 (0.71 Arrowtooth flounder. Flatfish, Pacific cod. Adult pollock 5 Inner slope 167.0 1.50.3) 10.9 (2.2) 5.6 (0.8) Northern rockfish. Pacific ocean perch, AiTowtooth flounder 6 Middle slope 197.1 (80.4) 10.1 (2.9) 5.3 (0.7) Adult pollock. Pacific ocean perch. AiTowtooth flounder, Sablefish 7 Outer slope 412.9 (153.5) 10.7 (3.1) 4.9 (0.9) Giant grenadier. .Sablefish. Rockfish Shima et a\ Spatial distribution of Theiagra chakogramma in the Gulf of Alaska 319 these age groups are strongly influenced by interannual variations in year-class strength. However, spatial trends in distribution were consistent with those seen for adult pollock. For all ajje groups of juvenile pollock there was an uicrease in the proportion of stations where low densities occurred, and a decrease in the mean CPUE during the time period examined. The spatial expansion by a species should be reconciled with more detailed analyses of the characteristics of the area it occupied to determine whether changes in habitat suitability have occurred. Adult pollock are not expanding into areas with physical characteristics previously not as- sociated with pollock. Cluster analysis revealed that adult pollock are associated with average bottom temperatures between 5.2°C and 6.4°C. The bottom temperature data from field surveys revealed a consistent range over time of 5-6"C (Dorn et al.''). It was difficult to discern a trend from either our bottom temperature or the OBT data. Tempera- tures may have been slightly warmer in the 1980s than in 6 Dorn. M. W., A. B. Hollowed. E. Brown, B. Megrey, C. Wilson, and J. Blackburn. 1999. Walleye pollock. In Stock assess- ment and fishery evaluation report for the 2000 Gulf of Alaska groundfish fishery, 65 p. North Pacific Fishery Management Council, P.O. Box 103136, Anchorage, Alaska 99510. the 1990s. Temperature changes, together with the shoal- ing of the mixed layer depth (Polovina et al., 1995; Shima, 1996), may have caused a redistribution of prey, possibly contributing to changes in pollock distribution. Evidence of the importance of external forcing on the spatial range of fish species has been noted in other eco- systems. Movement of fish in response to environmental change was recorded in the Barents Sea when Atlantic cod (Gadus morhua Linnaeus, 1758) shifted westward as a consequence of cooler waters (entering from the east) across the region from 1977-81 (Loeng, 1989). Primary shifts in distribution, in response to temperature changes, were seen in younger age classes of fish. Because there were some species (e.g. haddock, Melanogrammus aeglefi- nus (Linnaeus, 1758)) that did not respond to changes in temperature, it may be that the movement of the fish may be dependent on the sensitivity offish prey to the temper- ature shifts (Shevelev et al., 1987). Cod (like pollock) feed mostly on planktonic organisms that may respond rapidly to temperature changes whereas haddock feed mostly on benthic species (Shevelev et al, 1987). External forcing may also include anthropogenic fac- tors. Domestication (U.S.) of the pollock fishery occurred during the period investigated (Megrey, 1989) with the re- sult that fishing operations shifted from at-sea processors 15 ^ A 15 B w III 1 CO C o "^ 5- 5 n * ' i ■ ■ 0 , , , , 1984 1987 1990 1993 1996 "*'' "'^ '''" 1993 1996 i C , D ■■ill 0 8 0.8 i ' i ■ ■ S 06 c c 0.6 £ 0.4 0.4 0.2 0.2 0 0 1984 1987 1990 1993 1996 '''«•' '^S^ 1990 1993 1996 Triennial survey year Figure 10 Simpson's diversity indices: richness for the 0-100 m bottom depth (A) and for the 100-200 m bottom depth (B) evenness for the 0-100 bottom depth (Ci and for the 100-200 m bottom depth iD), Error bars indicate ISD. 320 Fishery Bulletin 100(2) 03- ^ 02 03 B 02 ■ I ■-■ ■ 01 0 O 03 O 02 Q. O 0 03 02 01 0 L.I I A D Ij i I I J I ■Jl J .■J-^ J>'' o . J. Mammol. Smith. G. B., G. E. Walters. R A. Raymore Jr , and W. A. Hirschberger. 1984. Studies of the distribution and abundance of juvenile groundfish in the northwestern Gulf of Alaska, 1980—82: Part I: Three-year comparison. U.S. Dep. Commer, NOAA Tech. Memo. NMFS F/NWC-59, 100 p. Stark, J. W.. and D. M. Clausen. 1995. Data report: 1990 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer. NOAA Tech. Memo. NMFS-Af\SC-49, 221 p. Stefansson. G. 1996. Analysis of groundfish survey abundance data: com- bining the GLM and delta approaches. ICES J. Mar Sci. 53:577-588. Steinberg, E. K., and P. Kareiva. 1997. Challenges and opportunities for empirical evalua- tion of "spatial theory." In Spatial ecology: the role of space in population dynamics and interspecific interactions (D. Tilman and P. Kareiva, eds.), p. 318-332. Princeton Univ. Press, Princeton. NJ. Tokeshi, M. 1993. Species abundance patterns and community struc- ture. Advances in Ecological Research 24:111-186. Walters. G. E., and M. J. McPhail. 1982. An atlas of demersal fish and invertebrate community structure in the eastern Bering Sea: Part 1. 1978-81. U.S. Dep. Commer NOAA Tech. Memo. NMFS F/NWC-35, 122 p. Wyllie-Echeverria. T, and W. S. Wooster 1998. Year to year variations in Bering Sea ice cover and some consequences for fish distributions. Fish. Oceanogr. 7:159-170. Yang, M-S. 1993. Food habits of the commercially important ground- fishes in the Gulf of Alaska in 1990. U.S. Dep. Commer.. NOAA Tech. Memo. NMFS-AFSC-22, 150 p. 324 Abstract— In August and September of 1997 and 1998, we used SCUBA techniques to surgically implant Vemco V16 series acoustic transmitters in 6 greenspotted rockfish iSehastes chlo- rostictus) and 16 bocaccio (S. paucitipi- nis) on the flank of Soquel Canyon in Monterey Bay, California. Fish were captured at depths of 100-200 m and reeled up to a depth of approximately 20 m, where a team of SCUBA divers anesthetized and surgically implanted acoustic transmitters in them. Tagged fish were released on the seafloor at the location of catch. An array of recording receivers on the seafloor enabled the tracking of horizontal and vertical fish movements for a three-month period. Greenspotted rockfish tagged in 1997 exhibited almost no vertical movement and showed limited horizontal move- ment. Two of these tagged fish spent more than 90'^r of the time in a 0.58-km'- area. Three other tagged greenspotted rockfish spent more than 60'^t of the time in a 1.6-km- area but displayed frequent horizontal movements of at least 3 km. Bocaccio exhibited some- what greater movements. Of the 16 bocaccio tagged in 1998, 10 spent less than lO^'r of the time in the approx- imately 12-km- study area. One fish stayed in the study area for about 50'/{ of the study time. Signals from the remaining 5 fish were recorded in the study area the entire time. Bocaccio fre- quently moved vertically 10-20 m and occasionally displayed vertical move- ments of 100 m or greater Movements of bocaccio (Sebastes paucispinis) and greenspotted (S. chlorostictus) rockfishes in a Monterey submarine canyon: implications for the design of marine reserves Richard M. Starr University of California Sea Grant Extension Program 8272 Moss Landing Road Moss Landing, Calilornia 95039 E-mail address: Starnaimlml calstateedu John N. Heine Jason M. Felton Gregor M. Cailliet Moss Landing Manne Laboratones 8272 Moss Landing Road Moss Landing, California 95039 Manuscript accepted 19 September 2001. Fish. Bull, 100:324-337 (2002), Rockfishes (Sebastes spp, ) are an impor- tant component of the commercial and recreational fisheries on the U,S. west coast. Recent stock assessments con- ducted by the Pacific Fishery Manage- ment Council (PFMC) have indicated large population declines for several species of rockfishes (PFMC), For example, bocaccio (Sebastes paucispi- nis) abundance was estimated by Mac- Call et al.- to be 2-A^7c of pre-harvest levels, causing bocaccio to be formally designated as overfished by the U.S. Na- tional Marine Fisheries Sei-vice. This severe population decline prompted con- servation organizations such as the World Conservation Union to consider bocaccio to be at high risk of extinction (lUCN, 1996). Even as management reg- ulations have become more stringent, however, bocaccio populations continue to decline (MacCall et al.-'). As a result of these population de- clines, new management techniques, such as marine reserves, have been contemplated for west coast rockfishes (Yoklavich, 1998). Marine reserves cur- rently are being considered as supple- ments to traditional fishery manage- ment schemes in many places aroimd the world ( Agardy, 1997; Allison et al,, 1998), Marine reserves can serve as undisturbed areas for research, as re- gions designated for limited harvest, or as fishery exclusion zones where fishes can take refuge from exploitation (Mur- ray et al., 1999). They also can serve as a buffer for management trials and as sources for recruits to fisheries (John- son et al., 1999; Nowlis and Roberts, 1999). The effectiveness of marine re- serves for conservation of heavily fished species, however, is dependent upon the size, shape, and location of reserves and on rates of movement of the pro- tected species (Polacheck, 1990; DeMar- tini, 1993; Lauck et al.. 1998). In this respect, an understanding of the rates and directions of daily movements of rockfishes is vital to understanding the value of marine reserves for these spe- cies (Can- et al,, 1998; Starr. 1998). ' PFMC ( Pacific Fishery Management Coun- cil). 1999. Status of the Pacific coast gi-oundfish fishery through 1999 and rec- ommended acceptable biological catches for 2000: stock assessment and fishery evaluation. Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224. Portland, OR. - MacCall, A. D.. S. Ralston, D. Pearson, and E.Williams. 1999. Status of bocaccio off California in 1999 and outlook for the next millenium. In Appendix: status of the Pacific coast groundfish fishery through 1999 and recommended acceptable biologi- cal catches for 2000: stock assessment and fishery evaluation. Pacific Fishery Man- agement Council. 2130 SW Fifth Avenue, Suite 224. Portland. OR. Starr et al : Movements of Sebastes paudsplnis and S chlorostictus in Monterey submarine canyon 325 Few studies have described the movements of commeirially caught rockfishes in water dee|)er than 100 m. however, because of the difficulty in achieving high sui"vival rates for fish tagged at those depths. Most rockfishes have pliysoclis- tous swim bladders that expand with the reduced pressure as these fish are brought to the sur- face. The resulting barotrauma causes death for almost all fish when captured from waters deep- er than 20-30 m, rendering traditional tagging techniques ineffective. We developed techniques to surgically implant sonic transmitters at depth, thus reducing tag- ging mortality (Starr et al., 2000). These tech- niques enabled us to estimate the movements of two rockfishes with different life history charac- teristics. In 1997, we placed sonic transmitters in six greenspotted (S. chlorostictus) rockfish, a spe- cies presumed to be relatively sedentary (Yoklav- ich et al., 2000). In 1998. we tagged 16 bocaccio, a more mobile species (Love, 1996). In both years, we tracked the horizontal and vertical move- ments of tagged fish for a three-month period. Materials and methods Field procedures Our study site was located on the flank of the sub- merged Soquel Canyon in 100-250 m of water, approximately 20 km off shore in Monterey Bay, California (Fig. 1). Soquel Canyon contains steep sediment slopes and rock walls interspersed with 50-100 m wide benches, comprising soft sediment and rock outcrops (Yoklavich et al., 2000). Several of the rock outcrops located at the rim of the canyon are 10-20 ni high by 50-100 m long scarps with boulders at the bases of the linear rock walls. In 1997, we caught greenspotted rockfish using long- line fishing gear deployed from a research vessel. In 1998, we hired a commercial fisherman to catch bocaccio using modified trolling gear In both years, fishing lines were retrieved at about 20 m/min. Fishes were brought to a depth of about 20 m and held there for tagging. Divers then surgically implanted Vemco V16 (Vemco Ltd., Nova Scotia, Canada) sonic tags into the captured fish. Follow- ing surgery, tagged fish were placed in a recovery-release cage, then towed to the differential GPS location at which they were caught, whereupon the cage was lowered to the seafloor and fish were released. We used the Delta sub- mersible to verify that tagged fish were alive after release. A more complete description of tagging and underwater tracking procedures is provided in Starr et al. (2000). Vemco VR-20 receivers were moored on the seafloor for the duration of the study as a means of tracking tagged fish. The positions of receiver deployment spanned the dis- tribution of release locations of the tagged fish. To increase the amount of positional information available from the receiver data, we placed the tagged fish and receivers Figure 1 Fish-release locations and tag numbers, receiver locations, current meter location, expected signal detection range, and resulting receiving zones for the 1997 study of greenspotted rockfish. along the side of a submarine canyon that stretched north- east to southwest. On 7 October 1997, we placed three re- ceivers on a ledge at a depth of about 160 m along the wall of the canyon (Fig. 1). The receivers spanned a distance of 1500 m and recorded signals for 6 minutes out of every half-hour We retrieved them on 5 January 1998. In 1998, we deployed two receivers on surface buoys from 17 August through 10 September because tagging operations commenced one month before the submersible was scheduled to deploy the underwater moorings. These receivers were placed near the locations at which fish were released, in the northeastern portion of the Soquel Canyon study area (site 1, Fig. 2) and in the southwestern portion of the study area (site 2, Fig. 2). On 16 September 1998, we placed six receivers at various depths about 1000 m apart (Fig. 2) and retrieved them on 30 December 1998. Two of the receivers were located in about 100 m of water on the relatively flat substrate above the canyon rim. The other four receivers were placed in deeper water on flat benches or gently sloping walls below the canyon rim. These receiv- ers recorded signals for 12 minutes every hour In both years, battery life of the tags exceeded the length of time the receivers were moored. Signal detection range of the moored receivers was about 800 m (Starr et al. 2000). Thus, if a signal was re- corded at a particular time by only one receiver, we knew 326 Fishery Bulletin 100(2) t Receiver locations • Fish release locations I 1998 study site 1 II 1998 study site 2 Signal detection Meters n range / Zone 3 ^"^[""""^s* ^°"^ \' ,-----. ', ""'34/ I v2one2\ / / Zone 5 ^V -^-'fi / *--_l'^ / /\ ^ i«^^ -3 Rcv'2 / 1 D„„<: * .. / ^-CVl? / 0 400 800 v,"' ,,,- --1*.' ' I ^^ 1" >?, Zone 4 Rcvr4 18.20 — - w '^one 6 4FICvr6 I 100 m \, I \ 36 48 0' - Figure 2 Fish-release locations with tag numbers, receiver locations, current meter loca- tion (moored with receiver 11, expected signal detection range, and resulting receiving zones for the 1998 field study of bocaccio. Table 1 Patterns of signal receptions that define "receiving zones. Signals •ecorded concurently by the listed combinations of receivers were assigned a receiving zone that represents an appi oximate location of a tagged fish in a specified time per iod. 1997 1998 Receivers (by receiver number) Receiving Receivers (by receiver number) that recorded signals zone no. that recorded signals Receiving zone no. 1 1 1 1 1,2 2 1,2 2 2 3 2 3 2.3 4 2.3 3 3 5 2,3,4 2,4 3 3.4 4 4.5 5 5.6 6 3 3 3 4 4 5 5 5 6 the tag was located within 800 m of that receiver and outside the range of detection of other receivers. If a sig- nal was recorded at the same time by two or more receiv- ers, we knew the tag was somewhere within the intersec- tion of the circles that represented the overlap of detection range for the respective receivers. We defined the combi- nations of intersections or exclusions of overlapping 800-m detection ranges as "receiving zones." In 1997. five such receiving zones were detected (Table 1). We labeled the northeasternmost receiving zone (signals only recorded by receiver 1 ) as zone 1. Zone numbers increased to the south- west. Because we knew the depths of all the gi-eenspotted rockfish tracked in 1997. we were able to refine estimated locations of a tag to the area in each receiving zone that was between the 100-300 m isobaths (Fig. 1). In 1998, the same method provided 13 combinations of overlapping detection ranges — too many to permit the pat- terns of fish movement to be understood easily Conse- quently, we gi'ouped combinations of overlapping detection ranges into six receiving zones by their spatial distribution Starr et a\ Movements of Sebastes pauaspinis dnd S chlomstictus in Monterey submarine canyon 327 (Table 1). This enabled us to use signal location and tag depth data to more effectively estimate fish positions and movements. Again, the most northeastern receiving zone (signals only recorded by receiver 1) was labeled as zone 1; zone numbers increased to the southwest (Table 1, Fig. 2). Each year we placed an S-4 current meter with record- ing thermometer and salinometer on a mooring near the seafloor to determine if changes in current, salinity, or temperature affected fish movement. In 1997, the current meter was located in 100 m of water on the shelf about 400 m away from receiver 2 (Fig. 1 ). In 1998, the current meter was moored with receiver 1, and was located in 100 m of water (Fig. 2). Data analysis Receivers logged the tag number, date, time of day, acoustic frequency, and tag depth each time a signal was detected. Receivers also recorded signal strength, noise, gain, and error messages provided by the receiver software. Data collected by the moored receivers were downloaded as text files and imported into a database for analysis. Tag depth was plotted versus time of signal reception for each tag and each receiver. Differences in fish depths by time of day were analyzed with ANOVA and Scheffe's F-test post-hoc analysis (Sokal and Rohlf 1997). In 1997, we grouped signals into half-hour time inter- vals to compare signal receptions between receivers. In 1998, we grouped signals into hourly intei-vals. We labeled each inten'al a time "bin" and standardized bin numbers among all receivers. Thus, any signal recorded by a re- ceiver in a given time period (bin) could be directly com- pared with signals from other receivers in similar time bins. In 1997, the study included 4309 half-hour time bins. In 1998. the study included 2535 hourly time bins. Each signal was thus assigned a time bin and a receiv- ing zone according to the time interval of signal reception and the combination of receivers recording signals from that tag number. Treating each receiving zone number as a numeric rank enabled us to use a simple average of rank to identify the predominant receiving zone that a fish oc- cupied in the time bins that occurred during a week (336 weekly bins in 1997 and 118 weekly bins in 1998). A differ- ence in average ranking among weeks indicated the fish had moved; the value of the average indicated the direc- tion in which the fish moved. For each tag, a chi-square test of heterogeneity was used to test for differences in average ranking among weeks. In 1997. week 2 of the study was the first week for which there were transmis- sions from all tagged fish; thus week 2 was used to repre- sent the expected fish distribution by receiving zone for the chi-square test. In 1998. we used week 1 to represent the expected distribution. We also used the unplanned test of homogeneity of replicates tested for goodness of fit (So- kal and Rohlf, 1997) to determine if the average location of a fish (average rank) was similar between weeks. This method enabled us to group weeks in which a fish was in a similar location. In addition to generating weekly distri- butions, we used the ranking system to plot semihourly (1997) or hourly (1998) movements of tagged fish. 15 Oct 1997 15 Nov 1997 15 Dec 1997 Figure 3 Movements of tagged greenspotted rockflsh (tag-2 and tag-6 fish) across study area in 1997. as depicted by changes in receiving zones that were derived from pat- terns of signal receptions. Results Greenspotted rockfish In 1997. we tagged six greenspotted rockfish, ranging in total length from 35 to 39 cm (Starr et al., 2000). Lea et al. (1999) reported that greenspotted rockfish in this size range are 11-15 yr old and probably mature. The three moored receivers recorded signals throughout the study period from all tags except tag 1. Signals from tag 1 were recorded for 18 hours, then not again until 67 days later. The total number of transmissions recorded from each tagged fish ranged from 156 to 24,132 (Starr et al., 2000). Except for tag-1 fish, tagged greenspotted rockfish ex- hibited two patterns of relatively small horizontal move- ments. Tag-2 and tag-6 fish remained primarily in the re- ceiving zone in which they were released (zone 4) and exhibited few cross-zone movements (Fig. 3, Table 2). Re- ceivers recorded transmissions from each of these tags in 99'7f of the time bins. More than 569f of the time, signals originated in receiving zone 4, and 94'7f of the time sig- nals originated from receiving zones 3 or 4 (Table 2), an area comprising 58 ha, or a linear distance along a ledge of 1200 m. Wlien the fish moved out of zones 3 or 4, they most often moved southwest, towards the mouth of Soquel Canyon (towards Zone 5 in Fig. 3). Chi-square and post- hoc analyses indicated similarity in the pattern of signals received from the tags for most weeks, as indicated by the weekly average rank of receiving zone (Fig. 4). 328 Fishery Bulletin 100(2) 6 - ^-^ Tag 2 -*~ Tag 6 -ir- Tag 10 Tag 3 Tag 7 5 - 4 - — ~~___^— -..-__----^ ^^==>e^^i=— ■-* . . - ^ *- 3 - y — -*s<-- >v A-^ X 2 - / ~^^--t::-^=^^^^^N^ /^ 1 - 1 1 1 1 1 1 1 I 1 1 1 2 3 4 5 6 7 8 Week Oct 1997 I Nov 1997 11 12 13 Dec 1997 I Figure 4 Distribution of tagged greenspotted rockflsh in 1997 as depicted by the weekly average rank calculated from receiving zones which were derived from patterns of signal receptions. Table 2 Percentage of half-hour time bins by receiving zone in which signals were recorded from tagged greenspotted rockfish from 7 October 1997 through 5 January 1998. None = the percentage of time bins in which no signal was recorded. Tag numbei' Zone number 1 2 3 4 5 None 1 0.1 0.2 0.4 0.3 0.1 98.8 2 0.0 0.4 37.9 56.2 4.4 1.0 3 2.0 3.6 26.5 28.2 14.1 25.6 6 0.0 0.2 26.4 67.3 5.1 1.0 7 11.6 9.6 24.2 8.1 5.2 41.2 10 11.2 12.7 31.6 6.9 2.6 35.0 Zone area (hal 53.3 11.4 3.9 54.0 36.4 Tag-3, tag-7, and tag- 10 fi.sh moved greater distances, and made relatively frequent short-term movements out of the zone in which they were originally tagged and re- leased (Fig. 5, Table 2). For these tags, 26-41% of the time bins contained no signals. The maximum time recorded for a tag in a single receiving zone was 32';f of the time bins, and at least three zones were needed to account for 609f of the signals received from each tag. All three tagged fish showed evidence of moving across the entire study zone, a linear distance of 2940 m, and an area of about 1.6 km^. Chi-square and pos/-Aoc analyses indicated there Table 3 Frequency of time lapse between signals after a time bin in which no receivers recorded signals from tagged greenspot- ted rockfish in 1997. NA = no signal was recorded for 67 davs. Max. Tag number 0-1 h 1-5 h 5-10 h >10h time (h) 1 NA NA NA NA 2 7 1 0 0 1.5' 3 368 102 7 2 12.5' 6 6 0 0 0 1' 7 253 107 14 16 30 10 296 101 19 6 27.5 Signals from these fish were not recorded for the first 17 li that receivers were in place and are not included in this table. Signals may not have been recorded because of electronic interference from boats in the area that prevented receivers from recording signals. The other two fish i tag-7 and lag- 10 fishl were released a week after receivers were in place. were significant (P<0.05) differences in the weekly aver- age rank of receiving zone for tags 3, 7, and 10 (Fig. 4). Transmissions from these tags were frequently recorded in all receiving zones (Fig. 5) and the fish also moved out of the study area for short time periods. The maximum time between recorded signals from any of these tags was 30 h. More than 90% of the time, the interval between recorded signals was 5 hours or less (Table 3). Stair et a\ Movements of Sebastes paucispinis and S chlorostictus in Monterey submarine canyon 329 5- 4- 3- 2 - 1 5 - 4 - 3 - 2 Tag 3 Tag? Tag 10 15 Oct 1997 15 Nov 1997 15 Dec 1997 Figure 5 Movements of tagged greenspotted rockfish (tag-3, tag-7, and tag- 10 fish) across the study area in 1997 as depicted by changes in receiving zones which were derived from patterns of signal receptions. All tagged greenspotted rockfish, except tag-1 fish, showed little vertical movement, and 99% of depth trans- missions from each tag were within ±3 m (see Fig. 6 for an example of this pattern). This distance is effectively the range of tidal variation and the error associated with the depth sensors in the sonic tags. Some of the fish occa- sionally made short-term movements to deeper locations. The only exception to this pattern was tag-1 fish that was tracked for 18 hours, lost, then heard again 67 days later It exhibited vertical movements of about 90 m in the hours before it left the study area (Fig. 7). Bocaccio In 1998 we tagged 16 bocaccio, ranging in length from 35 to 58 cm (Starr et al., 2000). Ten of the bocaccio we tagged were larger than the size at 50% maturity reported by Gunderson et al. (1980). The receivers placed on surface buoys in August and early September recorded continu- ously; the total number of transmissions recorded from each of the tags in that time bin ranged from 0 to 9531 (Starr et al., 2000). The total number of signals recorded by the six subsurface receivers that were deployed from 16 Septem- ber through 30 December ranged from 0 to 19,213 (Starr et al., 2000). Ten of the tagged bocaccio spent little time in the 12-km- study area (Fig. 8). Signals from three tags were recorded only within a few days after tagging and were not heard again (tags 13, 24, 26), either because the fish left the study area or the tags failed to send signals. Lengths of these fish ranged from 47 to 51 cm (Starr et al., 2000). Transmissions from five tags were recorded in the study area for only 1-4 weeks (tags 3, 4, 10, 12, 21). Lengths of these tagged fish ranged from 35 to 52 cm. Three fish appeared to leave the study area and return two weeks to a month later (tags 17, 20, 25). Lengths of these fish ranged from 45 to 55 cm. Signals from the remaining five tags (tags 7. 9, 14, 18, 27) were recorded in the study area the entire time. Lengths of these fish ranged from 47 to 58 cm. Signals from 10 tags were recorded in less than 10% of the time bins in any zone (Table 4). Signals from one tag 330 Fishery Bulletin 100(2) 140 - Tag 3 180 - E 220 - ■ ■ - - ••• ••- . 15 Oct 1997 1 Nov 1997 15 Nov 1997 1 Dec 1997 Figure 6 Depth distribution olsignals received from tag 3 in 1997. Tag 3 is presented as an example of the depth distributions observed from all tagged greenspotted rockfish. except those of tagl. 1 1 1 1 1 1 \ 1 1 1 2100 22 00 23 00 00.00 0100 02 00 03 00 04 00 05 00 06 00 21 Oct 1997 22 Oct 1997 Figure 7 Depth distribution of signals received from tag-1 greenspotted rockfish in 1997 were recorded about SO'X of the time, and transmissions from the remaining five tags were recorded in more than 80% of the time bins. The six fish that remained in the study area 50% of the time or more stayed in a small part of the study area (Table 4». Signals from each of four tags (tags 9, 14, 17, 27) were recorded almost exclusively only where they were released, in receiving zones 4 or 5. These receiving zones comprised an area of 168 ha and 201 ha, respectively. Signals from the other two fish (i.e. tags 7, 18) were recorded primarily in receiving zones 3 and 4, an area of about 400 ha. Chi-square and post-hoc analyses indicated all fish stayed primarily in the same receiving zones for the time they were in the study area (Fig. 9). Although the six fish that stayed in the study area stayed primarily in one or two receiving zones, they often exhibited small movements. Plots of zone numbers of re- corded signals indicated that two of the fish (tags 7, 18) frequently moved across all receiving zones (Fig. 10). Ad- ditionally, cross-talk from receivers 5 and 6 indicated that tag-27 fish made frequent short-term movements (Starr et al., 2000). Tagged fish also occasionally left the study area. Except for tag 17. which was not recorded in the study area for 27 days, the maximum time interval between re- corded signals from any of these tags was 57 h. More than 90*:?^ of the time, the interval between recorded signals was 5 h or less (Table 5). Starr et a\ Movements of Sebasles paucispinis and S. chloiostictus in Monterey submarine canyon 331 Eight of'the 16 tags transmitted information about depth. We recorded signals from five of the eight tags for only short time periods after the fish were re- leased. Two of these five fish (tag-3 and tag-12 fish) moved vertically to within 15 m of the surface 9-12 hours after tagging, then returned below a depth of 90 m for a 12-15 day period before signals were lost (see Fig. 11 for an example of tag-3 fish movements). A third fish (tag-4 fish) fluctuated ±10 m around a depth of 90 m for a week, moved vertically to a depth of 14 m, and returned to a depth of about 90 m for another three weeks before signals ceased. A fourth tag (tag 13) was recorded for 3 days and exhibited frequent fluctuations of 20 m in depth before signals ceased. Data from the remaining three fish containing depth transmitters were recorded throughout the study. Depth transmissions from tag 14 varied less than 3 m, whereas depth transmissions from tag 7 (Fig. 12) and tag 9 (Fig. 13) indicated cyclical ver- tical movements of ±10-20 m, and occasional deep- er dives. The greatest variation in depth was exhib- ited from tag 7. That fish made several dives of 100 m; once it moved from 100 m to 220 m and back in a 20-h period. Tag-7 and tag-9 fish also demonstrat- ed a diurnal periodicity in vertical movements. Fish were more active and higher during the day and less active and deeper at night (Fig. 14). The AN OVA and Scheffes F test statistics indicated significant differ- ences in both depth and change in depth between dawn, day, dusk, and night hours (Table 6). Environmental parameters fluctuated but did not appear to be related to fish movements. Salinity and temperature data from the S-4 current meter at 100-m depth fluctuated within 24-h periods, but there were no obvious trends within or between years. Salinity averaged 33 ppt in 1997 and 34.5 ppt in 1998. Water temperature fluctuated from 10 to 15°C in 1997, whereas in 1998 it was more consis- tent, ranging from 9 to 11°C. Current speed and direction fluctuated in what appeared to be a tidal basis (Shea and Broenkow, 1982), but since there was no obvious relation- ship to recorded movements, this relationship was not ex- plored in more detail. Discussion Fish movements The in situ tagging procedures we developed alleviated many problems associated with surface tagging and provided means for tracking deeper-water rockfishes. We expected gi'eenspotted rockfish to move only small distances because of their affinity to seafloor habitats such as overhangs and crevices (Stein et al.. 1992; Yoklavich et al., 2000). Our work confirmed that gi-eenspotted rockfish are relatively seden- tary. The greenspotted rockfish we tagged with depth trans- mitters moved less than ±3 m vertically during the study, except for a few occasions when a fish swam down 20-40 26 24 13 • 25 20 17 21 12 10 4 3 27 18 14 9 7 dlDOtlKJCD o o 17 Aug 17 Sep 1 17 Oct 1998 17 Nov 17 Dec Figure 8 Dates of signal receptions in study area from sonic transmitters implanted in bocaccio in 1998. Circles represent individual occur- rences of recorded signals; solid lines represent almost continu- ous occurrences of recorded signals. Dashed lines on the graph are shown to separate the groups of tagged fish discussed in the text. m, only to return to its original depth within 2 hours. This small variation in depth displayed by the tagged greenspot- ted rockfish indicated that these fish most likely do not leave canyon wall habitats to feed. Love (1996) indicated that they eat mainly small invertebrates, but also cephalopods and fishes. It is possible that prey in the water column is advected toward them along the canyon wall (Isaacs and Schwartzlose, 1965; Genin et al., 1988). They do not appear to migrate vertically to feed on scattering layer organisms as do other rockfishes such as yellowtail rockfish (Sebastes flavidus) off Oregon (Perewa et al., 1969). Horizontal movements of the tagged gi'eenspotted rock- fish were on the order of hundreds of meters to a few ki- lometers. Receivers 2 and 3 each recorded cross-talk from signals of the same two tags (tags 2, 6) at several different times, indicating that these fish alternately swam close to one receiver, then at a later time swam close to a sec- ond receiver that was more than 300 m away (Starr et al., 2000). Although most movements were contained well within the 3-km long study area, half of the fish made so- journs out of the study area for short time periods. We think these small, short-term movements represent forag- ing activity along the canyon ledge. 332 Fishery Bulletin 100(2) 6 - 5 - -6 6 C: 6 C: A 6 !^ ft A— —6 A 4 - 3 - •5 - A — -P t 1 — ^ — ♦ -♦ — >\.* -^^-c 1 - Tag 7 Tag 9 -*- Tag 1 7 o Tag 25 -D- Tag 18 -r^ Tag 27 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I Oct 1998 Week I Nov 1998 I Dec 1998 Figure 9 Distribution of six tagged bocaccio in 1998 as depicted by the weekly average rank calculated from receiving zones derived from patterns of signal receptions. Table 4 Percentage of hou riy receiving bins in which signal s from tagged bocaccio were recorded from 16 September through | 30 December 1998 . None = the percentage of time bins in | which no s ignal was recorded. Zone number Tag number 1 2 3 4 5 6 None 3 0.0 0.0 0.0 0.0 0.0 0.0 100.0 4 0.0 0.0 0.0 0.0 0.0 0.0 100.0 7 3.3 3.8 41.5 35.3 1.7 0.7 13.7 9 0.0 0.0 0.0 0.2 67.5 17.6 14.7 10 0.0 0.0 0.0 0.0 0.0 0.0 100.0 12 0.0 0.0 0.2 0.4 0.1 0.1 99.2 13 0.0 0.0 0.2 0.4 0.1 0.1 99.2 14 0.0 0.0 0.1 98.0 0.5 0.0 1.4 17 0.0 0.0 0.0 50.4 0.0 0.0 49.6 18 2.0 8.7 29.6 39.3 2.0 0.0 18.4 20 0.0 0.0 0.2 0.0 0.5 0.1 99.3 21 0.0 0.0 0.4 0.4 1.3 0.0 97.9 24 0.0 0.0 0.0 0.0 0.1 0.0 99.9 25 0.3 0.0 5.1 0.0 0.0 0.0 94.7 26 0.0 0.0 0.0 0.0 0.0 0.0 100.0 27 0.0 0.0 0.0 0.0 99.7 0.0 0.3 Zone area (hai 124.0 75.5 232.2 168.0 201.0 132.4 15 Sept 1998 15 Oct 1998 Figure 10 Movements of tag-7 and tag-18 bocaccio across the study area in 1998 as depicted by changes in receiving zones derived from patterns of signal receptions. Starr et a\ Movements of Sebastes pauaspinis and 5 ch/orostictus in Monterey submanne canyon 333 24 Aug 5 Sep 9 Sep 1998 Figure 11 Depth distribution of signals received from tag-3 bocaccio in 1998. Tag-12 bocaccio displayed a similar pattern. We expected bocaccio to move more than greenspotted rockfish. Bocaccio are frequently caught in midwater trawl nets and are considered more mobile than greenspotted rockfish (Love, 1996). Hartman (1987) reported that ju- venile bocaccio moved a maximum of 148 km over two years in tag-recapture studies in southern California. In this respect, young bocaccio are similar to yellowtail rock- fish iSebastes flavidus) that have exhibited movements on the scale of hundreds of kilometers in tag-recapture stud- ies conducted in Alaska and British Columbia (Stanley et al., 1994). In Stanley et al.'s studies, IS^c of the tag recover- ies were within 25 km of the release point, but maximum movement observed was 250 km for Canadian yellowtail rockfish, and 1400 km for Alaskan fish. Not all studies have shown that yellowtail rockfish move great distances, however Using ultrasonic telemetry, Pearcy ( 1992) tracked them from the surface intermittently for pe- riods of several days and reported that tagged fish gener- ally stayed within 2 km of the capture or release location. After a period of 13 days in 1990, 11 of 12 tagged fish were detected within 300 m of the capture site, even though some fish had been displaced. A month after release, eight of 12 tagged fish were within 1.4 km of the capture location. Pearcy ( 1992) also suggested that the tagged fish exhib- ited site fidelity to a pinnacle habitat. Pearcy 's work, com- bined with results from other displacement studies (Carl- son and Haight, 1972; Hallacher, 1984; Matthews, 1990; Heilprin, 1992), suggests that several species of rockfish possess homing ability. Half of the bocaccio we tagged ei- ther stayed in the study area during the entire time of the study, or left and returned, suggesting some site fidel- Table 5 Frequency of time lapse between recorded signals after a time bin in which no receivers recorded signals from tagged bocaccio in 1998. Only tagged fish that remained in the study area for more than 50% of the time are shown. Tag number 0-1 h 1-5 h Max. .5-10 h >10h time (hi 7 115 30 6 3 57 9 152 66 1 2 28 14 14 2 0 0 1 17 143 86 10 5 660 18 201 86 2 1 12 27 2 0 0 0 1 ity. Evidence of site fidelity was also provided by the chi- square analyses of fish location in receiving zones. The analyses indicated that bocaccio remained in the same ar- ea over the course of a week, despite the evidence of fre- quent movements out of a receiving zone on a daily basis. Although our study was designed to evaluate the resi- dence time of bocaccio in a discrete area, and not to quan- tify the maximum distance bocaccio move, the relatively small percentage of tagged bocaccio that stayed in our 12-km- study area during the entire study period indicated that bocaccio may also move large distances. The results of our study tend to reinforce the hypothesis presented 334 Fishery Bulletin 100(2) 60 80 - 100 120 - 5. 140 - Q 160 180 - 200 - 220 1 15 Aug 1 15 Sep Tag 9 1 5 Oct 15 Nov 1 5 Dec 1998 Figure 13 Depth distribution of signals roccivud liom tag-9 bocaccio in 1998. (MacCall-) that there are two types of obsei-ved movements gle individuals and the occurrence of smaller animals in of bocaccio: pelagic and "refugial." They suggested, from mid-vi^ater trawl catches, that there may be an ontogenic observations of the large sizes of bocaccio observed as sin- shift from pelagic to refugial habits. They hypothesized Starr et a\: Movements of Sebastes paucispinis and S ch/orostictus in Monterey submarine canyon 335 that subadults are more mobile than adults, and as the fish increase in size, they become more sedentary. From tag transmissions and submersible sui-veys, we observed both pelagic and refugial behavior (about 25% of the tagged ani- mals exhibited refugial behavior). The two largest tagged bocaccio were among the five fish that moved the least. The narrow size range of most of the tagged fish, however, pre- cluded an analysis of movements by fish length. Tagged bocaccio made frequent small vertical move- ments that were associated with time of day. The magni- tude of the vertical movements matched the vertical relief of the habitats used by tagged fish. During submersible operations, we tracked tagged fish in small schools as they moved along rock scarps that were 10-20 m high and 100-200 m long (Starr et al., 2000). These scarps were of- ten at, or just below, the rim of the submarine canyon. The depth transmissions from tags indicated that fish were at the top or just above the rock habitats during the day, and lower and more sedentary at night. We attribute the in- creased activity during the day to the fact that bocaccio are visual predators (Love, 1996) and are thus more apt to forage during the day. Four of the eight fish containing depth trans- mitters made rapid vertical movements. Three of the bocaccio rose vertically to spend a short time near the surface, then returned to depths from which they came. These fish all remained only a few weeks in the study area. A fourth fish made a deep dive to 220 m and back to 100 m in less than a day (Fig. 12). The purposes of these dives are un- known, but such dives reinforce our observations during tagging operations that bocaccio can modi- fy the volume of air in their swim bladder. Pearcy ( 1992 ) obsei-ved that yellowtail rockfish also are ca- pable of discharging air from their swim bladders in a relatively short time. For some tags, there were short time intei-vals when signals were not recorded. This may have been caused by the canyon topogi-aphy or by fish behavior. The complex topography of Soquel Can- yon and the behavior of the rockfishes may have prevented the receivers from receiving all possible transmissions from tagged fish. The hard, steep walls that undulate along the side of the canyon (Yoklavich et al., 2000) occasionally cause echoes and an original transmission to coincide, thus preventing the receivers from recording a valid signal. Other signal lapses may also have been due to the tendency of rockfish to take shelter under rocks or ledges. From the surface, we occasionally heard weak signals that were unusually high- pitched and that sounded "tinny." We have experienced the same type of reception when tracking shallow-water rock- fishes that took shelter under a rock or ledge. Signals re- turn to full strength and timbre when the fish leaves the crevice. Implications for marine reserves Tag-recapture studies of shallow water (<100 m) demersal rockfishes typically have indicated very little movement; Table 6 Results of Scheffc's F post- hoc test iP values) were used to identify differences in fish depth between different times of the day (i.e. between dawn and day Idawn-dayl, dawn and dusk [dawn-dusk], etc. I or diel differences in depth and change in depth of signal receptions from tagged bocaccio (tag-7 and tag-9 fish) in 1998. Tag? Tag 9 Depth Change in depth Depth Change in depth Dawn-day 0.0097 0.0194 <0.0001 <0.0001 Dawn-dusk 0.2840 0.277 0.0040 0.0004 Dawn-night 0.5045 <0.0001 <0.0001 <0.0001 Day-dusk 0.7455 0.9031 0.7.312 0.9998 Day-night <0.0001 0.0065 0.1010 <0.0001 Dusk-night 0.0007 0.0156 0.0440 <0.0001 90 95- 100 130 o Tag 7 depths / \ Dayligtit curves \ \ 1 \ \ 1 22 Oct 23 Oct 24 Oct 25 Oct 26 Oct 27 Oct 28 Oct 1998 Figure 14 Depth transmissions recorded for tag-7 bocaccio plotted with day- light curve for the period 22-28 October 1998. only one species in ten studies exhibited long-term move- ments greater than 3 km (Stanley et al. 1994; Lea et al., 1999). Results from these studies support the idea that small harvest refugia may effectively protect nearshore rockfishes. This would be fortuitous if almost all near- shore rockfishes have small home ranges, because most of the marine reserves in the eastern Pacific Ocean are very small and have been created without regard to typical movements of species (McArdle, 1997; Starr, 1998; Yoklav- ich, 1998). Our results indicate a considerable short-term varia- tion in the movements of individual greenspotted rockfish, which may be masked by the long-term nature of tag-re- capture studies. Although almost all fish may remain in a small area over the course of several years, an individual 336 Fishery Bulletin 100(2) fish may occasionally move longer distances. This raises a question about the efficacy of marine resei-ves that are designed to account for modal distributions of fish based on traditional tag-recapture studies. The efficacy of a marine resei-ve is directly related to its size and shape and the movements of the protected spe- cies (Polacheck, 1990; DeMartini, 1993; Nowlis and Rob- erts, 1999). Thus, without estimates of both the range and frequency of movements of the target species, it is difficult to predict the effectiveness of a reserve for consei-ving fish- es (Carr and Raimondi, 1998). Infrequent foraging excur- sions out of a reserve, for example, could potentially ne- gate the value of a reserve if its purpose is to act as a harvest refugium. Conversely, small home ranges could preclude a "spillover" effect and diminish a reserve's abil- ity to enhance local fisheries, if that were the goal. Small reserves may appear to protect species that move little, but if protected fishes occasionally move greater distances, as did the tagged bocaccio and gi'eenspotted rockfishes in our study, then larger resei-ves may be needed to encom- pass 90*^^ or more of the typical monthly movements of a species. Models that incorporate movement into the theo- retical design and evaluation of marine resei-ves may thus be strengthened by using the probability or percentage of time an animal actually remains in the resei-ve boundar- ies rather than modal distributions. The periodic departure of tagged fish from their center of activity suggests that a consei-vative strategy for the design of marine reserves would be to include a buffer zone to account for these infre- quent sojourns. Only one-fourth of the tagged bocaccio spent more than 80% of their time in the 12-km- study area. Yet the Soquel Canyon study area was two times larger than the mean size and four times larger than 70% of marine resei-ves in California that regulate fishing in any way (McAi'dle, 1997 ). Such movement will require that marine resei^ves intended to effectively protect bocaccio will need to be much larger than most current reserves. Until the full ex- tent of the range and periodicity of movements is known, however, it will be difficult to adequately design a system of marine reserves to protect bocaccio stocks. After tagging, greenspotted rockfishes returned to rock habitats on the ledges at the side of Soquel Canyon and ex- hibited small horizontal and vertical movements. Based on the signal receptions for our three-month period, the size of a marine resei-ve would need to have a diameter of 3 km to account for 95% of the typical movements of the tagged greenspotted rockfish. Bocaccio. however, returned to high- er relief habitats near the canyon rim after release and ex- hibited much greater horizontal and vertical movements. An effective reserve for bocaccio, therefore, would need to be larger than the 12 km- area encompassed by our study. Acknowledgments A project of this complexity requires the help of many people. We especially want to thank Korie Johnson, who was an integi-al part of the research team in the first year of this study. We thank the people who helped us catch fish, especially Dempsey Bosworth from the FV Beticia, and Lee Bradford, Joe Bizzarro, Kate Stanbury, and other students and staff of Moss Landing Marine Laboratories. We thank the people who provided the ships and submers- ibles, and their crew, including MLML ship operations staff Wayne Kelly, the crew of the CV Cavalier, the pilots of the submersible Delta, Dave Slater and Chris Ijames, and the crew of the RV Point Lobos and Ventana, and ROV pilots. Funding for this project was provided by the West Coast and Polar National LTndersea Research Center, LTni- versity of California Sea Grant Extension Program, and Moss Landing Marine Laboratories. We also thank the three anonymous reviewers of the manuscript. Literature cited Agardy, M. T. 1997. Marine protected areas and ocean consei-vation. Aca- demic Pres.s, San Diego, CA. 244 p. Allison, G. W., J. Lubchenco, and M. H. Carr. 1998. Marine re.serves are necessary but not sufficient for marine conservation. Ecol. Appl. 8 (supplement): S79- S92. Carison, H. R. and R. E. Haight. 1972. Evidence for a home site and homing of adult yel- lowtail rockfish, Sehastes flavidus. J. Fish. Res. Board Canada 29:1011-1014. Carr, M. H., and P. T. Raimondi. 1998. Concepts relevant to the design and evaluation of fishery reserves. In Marine harvest refugia for West Coast rockfish: a workshop (M. M. Yoklavich, ed.), p, 27-31. U.S. Dep. Commer., NOAA/NMFS/SWFSC Tech. Memo. 255, La Jolla. CA. Carr, M. H., L. Ziobro.T. Houngan, D. Ito, D. McAi-dle, L. Morgan, W. Palsson, R. Parrish. S. Ralston. R Reilly. J. Sladek Nowlis, and R. Starr. 1998. Working gi'oup on design considerations. In Marine har\'est refugia for West Coast rockfish: a workshop (M. M. Yoklavich, ed. I, p. 143-148. U.S. Dep. Commer, NOAA/ NMFS/SWFSC Tech. Memo. 255, La Jolla, CA. DeMartini, E. E. 1993. Modeling the potential of fishery reserves for manag- ing Pacific coral reef reserves. Fish. Bull. 91:414-427. Genin, A., L. Haury, and P. Greenblatt. 1988. Interactions of migrating zooplankton with shallow topogi-aphy: predation by rockfishes and intensification of patchiness. Deep-Sea Res. 35:151-175. Gunderson, D. R., P. Callahan, and B. Goiney. 1980. Maturation and fecundity of four species ofSebastes. Mar Fish. Rev 421.3-41:74-79. Hallacher, L. E. 1984. Relocation of original territories by displaced black- and-yellow rockfish, Sebastes chiysomelas, from Carmel Bay, California. Calif Fish Game Bull. 70(31:158-162. Hartman, A. R. 1987. Movement of scorpionfishes (Scorpaenidae: Sebastes and Scorpaena) in the southern California Bight. Calif Fish and Game Bull. 73l2):68-79. Heilprin, D. J, 1992. The role of olfaction in the homing ability of the blue rockfish, Sebastes mystinus, in Carmel Bay, Califor- nia. M.S. thesis. Moss Landing Marine Laboratories, San Jose State Univ., San Jose, CA, 63 p. Starr et a\ Movements of Scbastcs pauaspinis and S chlorostictus in Monterey submarine canyon 337 Isaacs, J. D., and R. A. Schwailzlose. 1965. Migrant sound scatterers: interaction witli tlio son floor. Science 150 (Wash. D.C.):1810-181,i. Il'CN (International L'nion for the Conservation of Nature and Natural Re.sources). 1996. Marine Fish and the lUCN red list of threatened ani- mals. Report of the workshop held in collaboration with WWF and lUCN at the Zoological society of London from April 29"'-May 1st. 1996. Institute of Zoology, Zoological Soc. London. London, England. Johnson, D. R., N. A. Funicelli, and J. A. Bohnsack. 1999. EfTectiveness of an existing estuarine no-take fish sanctuary within the Kennedy Space Center. Florida. N. Am. J. Fish. Manag 19:436-453. Lauck.T, C. W. Clark, M. Mangel, and G. R. Munro. 1998. Implementing the precautionary principle in fisheries management through marine reserves. Ecol. Appl. 8(1) ( supplement ):S72-S78. Lea, R. N., R. D. McAllister, and D. A. VenTresca. 1999. Biological aspects of nearshore rockflshes of the genus Sebastes from Central California, with notes on ecologically related sport fishes. Calif Fish Game Fish Bull 177, 109 p. Love, M. 1996. Probably more than you want to know about the fishes of the Pacific Coast. Really Big Press, Santa Barbara, CA, 381 p. Matthews, K. R. 1990. An experimental study of the habitat preferences and movement patterns of copper, quillback, and brown rock- fishes {Sebastes spp.). Environ. Biol. Fish. 29:161-178. McArdle, D. A. 1997. California marme protected areas Publication T-039, California Sea Grant College System, La Jolla, CA, 268 p. Murray, S., R. Ambrose, J. Bohnsack, L. Botsford, M. Carr, G. Davis, P. Dayton, D. Gotshall, D. Gunderson, M. Hixon, et al. 1999. No-take reserves: sustaining fishery populations and marine ecosystems. J. Am. Fish. Soc. 24( 1 1 ): 1 1-25. Nowlis, J., and C. M. Roberts. 1999. Fisheries benefits and optimal design of marine re- serves. Fish. Bull. 97:604-616. Pearcy, W. G. 1992. Movements of acoustically-tagged yellowtail rockfish Sebastes flavidus on Heceta Bank, Oregon. Fish. Bull. 90:726-735. Pereyra, W. T., W. G. Pearcy, and F E. Carvey Jr 1969. Sebastes flavidus, a shelf rockfish feeding on mesope- lagic fauna, with consideration of the ecological implica- tions. J. Fish. Res. Board Can. 26:2211-2215. Polacheck, R. 1990. Year around closed areas as a management tool . Nat- ural Resource Modeling 4(3):327-353. Shea, R. E., and W. W. Broenkow. 1982. The role of internal tides in the nutrient enrichment of Monterey Bay, California. Estuar Coast. Shelf Sci. 15: 57-66. Sokal. R. R., and F J. Ruhlf 1997. Biometry. The principles and practice of statistics in biological research, third ed. W.H. Freeman and Co., New York, NY, 887 p. Stanley. R. D., B. M. Leaman, L. Haldorson, and V. M. O'Connell. 1994. Movements of tagged adult yellowtail rockfish, Se- bastes flavidus. off the west coast of North America. Fish. Bull. 92:65.5-663. Starr. R. M. 1998. Design principles for rockfish reserves on the U.S. West Coast. In Marine harvest refugia for West Coast rockfish: a workshop ( M. M. Yoklavich. ed. ). p. 50-63. U.S. Dep. Commer, NOAA/NMFS/SWFSC Tech. Memo. 255, La Jolla, CA. Stan-, R. M., J. N. Heine, and K. A. Johnson. 2000. In situ techniques for tagging and tracking rockflshes. N. Am. J. Fish. Manag 20:597-609. 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. Yoklavich, M.M.(ed. I. 1998. Marine harvest refugia for west coast rockfish: a work- shop. U.S. Dep. Commer., NOAA/NMFS/SWFSC Tech. Memo. 155, La Jolla, CA, 159 p. 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 rockflshes in a submarine canyon: an example of a natural refuge. Fish. Bull. 98:625-641. 338 Abstract— Two bycatch reduction de- vices ( BRDs 1 — the extended mesh funnel (EMF) and the Florida fisheye (FFE)— were evaluated in otter trawls with net mouth circumferences of 14 ni, 17 m, and 20 m and total net areas of 45 m-. Each test net was towed 20 times in parallel with a control net that had the same dimensions and configuration but no BRD. Both BRDs were tested at night during fall 1996 and winter 1997 in Tampa Bay Florida. Usually the bycatch was composed principally of finfish (44 species were captured); horseshoe crabs and blue crabs sea- sonally predominated in some trawls. Ten finfish species composed 92'^i of the total finfish catch; commercially or rec- reationally valuable species accounted for I'i of the catch. Mean finfish size in the BRD-equipped nets was usually slightly smaller than that in the con- trol nets. Compared with the corre- sponding control nets, both biomass and number of finfish were almost always less in the BRD-equipped nets but nei- ther shrimp number nor biomass were significantly reduced. The differences in proportions of both shrimp and finfish catch between the BRD-equipped and control nets varied between seasons and among net sizes, and differences in finfish catch were specific for each BRD type and season. In winter, shrimp catch was highest and size range of shrimp was greater than in fall. Sea- son-specific differences in shrimp catch among the BRD types occurred only in the 14-m, EMF nets. Finfish bycatch species composition was also highly seasonal; each species was captured mainly during only one season. How- ever, regardless of the finfish composi- tion, the shrimp catch was relatively constant. In part as a result of this study, the State of Florida now requires the use of BRDs in state waters. Efficiency of bycatch reduction devices in small otter trawls used in the Florida shrimp fishery Philip Steele Theresa M. Bert Kristine H. Johnston Sandra Levett Florida Fish and Wildlife Conservation Commission Florida Manne Research institute 100 8th Avenue S.E. St Petersburg, Flonda 33701-5095 E-mail address (for T M Bert, contact author) theresa bertiqfwc state fl us Manuscript accepted 13 September 2001. Fish. Bull. 100:338-350 (2002). Commercial fishermen use a variety of gears to harvest shrimp in southeast- ern U.S. waters, but they have predom- inantly used the otter trawl since the 1940s. The otter trawl is an unselective gear that commonly has an associated catch of untargeted organisms (e.g. fin- fish, miscellaneous invertebrates) that are referred to as "bycatch." Numerous definitions for the term " bycatch" have been proposed (Allsopp, 1982; Caddy, 1982: Saila, 1983 ). The most comprehen- sive, suggested by Alverson et al. ( 1994 ), refers to nontargeted species retained, sold, or discarded for any reason. An estimated average of 27.0 million metric tons (t) (range=17.9-39.5 mil- lion t) of bycatch are discarded annu- ally by the worlds marine fishing fleets (Alverson et al, 1994). Shellfish fisher- ies compose 14 of the top 20 fisheries worldwide in quantity of bycatch dis- cards (Alverson et al., 1994) and ac- count for 9.5 million t of discards an- nually. Because the harvest of bycatch often exceeds that of the targeted spe- cies, the issue of bycatch in marine fish- eries has become a global concern. In the southeastern United States, the penaeid shrimp fishery often ranks first in value of all fisheries for com- mercially harvested marine species. In 1996, total landings were 98 million kg and were valued at appro.xiniately $434 million, ex-vessel price (size-spe- cific price per unit volume paid to the fisherman for the catch ) (NMFSM. The Gulf of Mexico (referred to as "Gulf" in this study) shrimp fishery accounted for 909( of this volume and 87'^'i of this value. In U.S. waters, the Gulf and the southeast U.S. Atlantic (referred to as "South Atlantic") shrimp trawl fisher- ies ranked 5th and 9th, respectively Their ratios of kg finfish bycatch to kg shrimp were 10.3:1 for the Gulf, and 8.0:1 for the South Atlantic (Alverson et al., 1994). However, the Gulf and South Atlantic Fisheries Development Foun- dation (GSAFDF-') estimated that the ratio of finfish bycatch to shrimp harvest was 4.2:1 for the Gulf shrimp fishery and 2.8:1 for the South Atlantic shrimp fishery. Thus, using the more conserva- tive ratios reported by GSAFDF and the 1996 shrimp landings for the Gulf fish- ery ( 88 million kg) and the south Atlan- tic fishery i9.9 million kg: NMFS'), the estimated total finfish bycatch for these two fisheries is 370 million kg and 28 million kg, respectively. In 1996, approximately 11.3 million kg of shrimp were landed along the Florida Gulf coast and 1.8 million kg of shrimp were landed along the Flor- ida Atlantic coast (NMFSM. Ratios of finfish bycatch to shrimp for the Flori- da Gulf coast ranged from 2.3:1 (fish- ' NMFS (National Marine Fisheries Ser- vice). 1997. Bycatch in the southeast shrimp trawl fisherv A data summary report. National Marine Fisheries Ser- vice, Southeast Science Center, 75 Virginia Beach Drive, Miami, FL 33149. 197 p. - GSAFDF (Gulf and South Atlantic Fish- eries Development Foundation). 1997. Bycatch and its reduction in the Gulf of Mexico and South Atlantic shrimp fishery. Final report to the National Oceanic and Atmospheric Administration (award NA57 FF0285). GFSFDF, Suite 997, Lincoln Center. 5401 West Kennedy Boulevard, Tampa. FL 33609. 27 p. Steele et al Efficiency of bycatch reduction devices in small otter trawls in thie Flonda sfirimp fisfiery 339 ing depth >1() f'atiioms llnil) to '2.5:1 (fishing depth < 10 fm) (NMFS'i; thus total finfish bycatch can be estimated at 26.0-28.3 niilhon kg for the Florida Gulf shrimp fishery. With the finfish-to-shrimp ratio of of 2.8:1 for the South Atlantic fishery and the current landings information for Florida, the finfish bycatch for the Florida Atlantic shrimp fishery can he estimated at 5.0 million kg. Since 1990, con- siderable research has been conducted to characterize by- catch composition and to develop methods to reduce by- catch in the Gulf and the South Atlantic shrimp fisheries (Nance, 1992. 1993: GSAFDF-'; Nichols et al.^; NMFS^). In addition, numerous fishery-independent sui-veys exam- ining bycatch characterization and the efficiency of by- catch reduction devices (BRDs) have been conducted by state and private organizations throughout the southeast- ern U.S.A. (Burrage et al.''). In 1990, the Florida Marine Fisheries Commission (FM- FC; now Florida Fish and Wildlife Conservation Commis- sion) began to develop a shrimp fishery management plan that included a mandate to reduce the bycatch of total fin- fish biomass in shrimp trawls by 50'^'f . Responding to this policy decision, a bycatch-characterization study of the inshore Florida shrimp fishery was conducted statewide (Coleman et al.~; Coleman et al.'*). Field studies compar- ing the efficiencies of two types of BRDs (Florida fish eye [FFEI, large-mesh extended-mesh funnel (EMF) ) in otter trawls and rollerframe trawls were also conducted (Conti- nental Shelf Associates Inc.^; Coleman and Koenig'"; Cole- man et al."). The issue of bycatch in the Florida shrimp trawl fishery has exacerbated conflicts between conservationists and rec- reational and commercial fishermen over the allocation of marine resources. Relevant issues include the following: 1) the high mortality rates of economically important juvenile finfish caught in shrimp trawls, which could reduce har- vestable finfish stocks; 2) the high mortality rates of non- harvested species caught in shrimp trawls, which could al- ter the overall health of the marine environment; and 3) the perceived waste of bycatch species that are discarded. This controversy was partly responsible for the passage of a Florida constitutional amendment (Article X, Section 16) that reduced the size of shrimp trawl nets used in the coastal shrimp fishery to 500 sq. ft. (45 m-) of mesh area per net and limited the number of nets to two per vessel. In previous studies conducted in Florida to examine the efficiency of BRDs in shrimp trawls (Coleman and Koe- nig'"), net sizes gi'eatly exceeding that authorized by the amendment were tested. The goal of our study was to test how efficiently the FFE and EMF excluded finfish in small otter trawls (overall mesh area=45 m-) of various mouth- perimeter sizes. This information can be used by fisheries managers when considering the use of BRDs in inshore and nearshore shrimp fisheries. * GSAFDF ( Gulf and South Atlantic Fisheries Development Foun- dation). 1993. Organization and management of a Gulf of Mexico and south Atlantic Ocean fishery bycatch management program (year 21. Final report to National Marine Fisheries Service (award NA37FD0032). GSAFDF, Suite 997, Lincoln Center, 5401 West Kennedy Boulevard, Tampa, FL 33609. 65 p. ^ Nichols. S.. A. Shah. G. J. Pellegi-in Jr., and K. Mullin. 1990. Updated estimates of shrimp fleet bvcatch in the offshoi'e waters of the U.S. Gulf of Mexico, 1972-1989. Report to the Gulf of Mexico Fishery Management Council, The Commons at River- gate 3018 US.'Highway 301 N., Tampa, FL 33619. '' NMFS (National Marine Fisheries .Service). 1995. Coopera- tive research program addressing finfish bycatch in the Gulf on Mexico and south Atlantic shrimp fisheries: a report to Con- gress. National Marine Fisheries Sei-vice, Southeast Fisheries Center, Southeast Regional Office, 9721 Executive Center Drive, St. Petersburg FL 33702, 68 p. •^ Burrage, D. D., S. G. Branstetter, G. Graham, and R. K. Wallace. 1997. Development and implementation of fisheries bycatch monitoring programs in the Gulf of Mexico. Report to the U. S. Environmental Protection Agency (report MX-994717-95-0). Mississippi State LIniversity, P.O. Box 5325, Mississippi State. MS 39762, 103 p. ■ Coleman, F C, C. C. Koenig, and W. F Herrnkind. 1991. Survey of the Florida inshore shrimp trawling bycatch and preliminary tests of bycatch reduction devices. First annual report to the Florida Department of Natural Resources. National Marine Fisheries Service MARFIN gi-ant NA37FF0051. Insti- tute for Fishery Resource Ecolog>', Florida State Univ., Tallahas- see, FL 32306,25 p. 8 Coleman, F C, C. C. Koenig, and W. F Herrnkind. 1992. Survey of the Florida inshore shrimp trawling bycatch and preliminary tests of bycatch reduction devices. Second annual report to the Florida Department of Natural Resources. National Marine Fisheries Service, MARFIN Grant NA37FF0051. Insti- tute for Fishery Resource Ecology, Florida State Univ.. Tallahas- see. FL 32306.21 p. Materials and methods Tampa Bay is located on the west-central coast of Florida (Fig. 1) and is the largest open-water estuary in the state (Lewis and Estevez. 1988). The bay is a subtropical estu- ary that has patches of fringing seagrass meadows (Lewis et al., 1981), but fine sand is the predominant seabottom type (Brooks'-). Gear specifications Conventional semiballoon otter trawls (Fig. 2) are used to harvest pink shrimp (Farfantepenaeus duorarum) in Tampa Bay. Otter trawls are typically used on unvege- tated, sandv-bottom areas. We tested the effectiveness of '■'Continental Shelf Associates, Inc. 1992. Commercial food shrimp fishery impacts on by-catch in the lower St. Johns River, Flonda. Draft final report C-7238. Continental Shelf Associ- ates, Inc., 759 Parkway Street, Jupiter. FL 33477. 35 p. '"Coleman. F C. and C. C. Koenig. 1994. Flonda inshore shrimping: experimental analysis of bycatch reduction. Final report. National Marine Fisheries Service. MARFIN grant NA37FF0051. Institute for Fishery Resource Ecology, Florida State Univ., Tallahassee, FL 32306,63 p. " Coleman, F.C., P. Steele, and W.Teehan. 1996. Use of bycatch reduction devices in small trawls of sizes set by the net ban. Final Report. Florida Department of Environmental Protec- tion contract MR081. Florida Dep. Environ. Protection. 100 8'*' Avenue S.E., St. Petersburg, FL 33701, 75 p. '- Brooks, H. K. 1974. Geological oceanography. In Summary of knowledge, eastern Gulf of Mexico (J. I Jones, R. R Ring, M. O. Rinkel, and R. E. Smith, eds), p. IIEl-50. Fla. State Univ. Syst. Inst. Oceanogr., St. Petersburg, FL. 340 Fishery Bulletin 100(2) two BRDs, the FFE and the EMF (Fig. 3) in three sizes of otter trawls in Tampa Bay during October-Decem- ber 1996 (fall) and February-April 1997 (winter). Both BRDs are standard devices that have been rec- ommended by NMFS, are used by the commercial fishing sector, and have been extensively tested in off- shore and inshore waters throughout the southeast- ern United States (Murray et al., 1992: Watson et al., 1993; Rogers et al, 1997; Coleman et al." Christian et al.'-'; McKenna and Monaghan''*). The otter trawl dimensions were as follows: 1) 14-m net: mouth circumference = 14.0 m, float-line length = 6.3 m, lead-line length = 6.9 m; 2) 17-m net: mouth circumference = 17.0 m, float-line length = 6.9 m, lead- line length = 7.8 m; 3) 20-m net: mouth circumfer- ence = 20.0 m, float-line length = 8.1 m, lead-line length = 9.4 m. The nets were of appropriate lengths to conform to the 45-m- total-mesh-area rule. Net pe- rimeters were chosen after consultation with commer- cial shrimpers and personnel from the NMFS Labora- tory Hai-vesting Section, Pascagoula, Mississippi. All net bodies were constructed of no. 9 twine and had a stretch-mesh size of 3.8 cm; the tailbag was constructed of no. 18 twine and had a stretch-mesh size of 3.2 cm. The FFE was constructed of 13-mm-diameter stainless steel rods. It had an overall length of 30 cm and a 15-cm x 15-cm opening to allow fish to escape. The FFE was mounted at the top center of the tailbag at 70'7( of the distance be- tween the tie-off rings and the beginning of the codend (Watson et al., 1993; Christian et al.'- ), creating an area of reduced water flow directly behind the FFE, which would allow fish to escape. The EMF had an overall length of '■'' Christian, P. A., D. L. Harrington, D. R. Amos, R. G. Overman, L. G. Parker, and J. B. Rivers. 199.3. Final report on the re- duction of finfish capture in south Atlantic shrimp trawls. Final report to National Marine Fisheries Service (award NA27FD 0070-01). Univ. Georgia, 71.5 Bay Street. Brunswick, GA 31520, 83 p. " McKenna, S. A., and J. P Monaghan Jr 1993. Gear develop- ment to reduce bycatch in the North Carolina trawl fisheries. Completion report to Gulf and South Atlantic Fisheries Develop- ment Foundation (cooperative agreement NA90AA-H-SK052I. North Carolina Div Mar. Fish.. 3441 Ai-endell Street. Morehead City NC 28557, 79 p. Lead line Coden(j Bag tie Figure 2 Components of a semiballoon otter trawl equipped with a super shooter turtle excluder device (TED). The TED is required in all shrimp nets in Florida. Figure 1 Tampa Bay, Florida. Hatched region shows sampling area. 121 cm and a circumference of 120 meshes; it consisted of a web funnel (3.5-cm stretch-mesh size) surrounded by a larger-mesh "escape" section (21-cm stretch-mesh size) held open by a plastic-coated hoop. One side of the funnel was extended to form a lead panel that created an area of reduced water flow on the back side of the funnel, similar to that created by the FFE. To conform to federal regulations, each net was equipped with a turtle excluder device (TED) placed near the mouth of the tailbag (Fig. 2). The standard super-shooter TED consisted of a metal grid of seven aluminum bars with a 9-cm interbar distance; the grid was set at a 45° angle to direct turtles downward toward the escape opening (Wat- son et al., 1993). Sewn in front of the TED was a section of webbing (3.2-cm stretch-mesh) that formed an acceler- ator funnel (Fig. 3A), which increased the velocity of wa- ter and entrained organisms both through the TED and into the net tailbag. The tailbag section and the combined TED and accelerator-funnel section could be zipped to any trawl body, regardless of size. The zipper ensured random pairing of trawl body and tailbag and enabled the experi- mental and control nets to be easily exchanged through- out the project. The BRDs and TEDs used during this project were approved by the NMFS Labora- tory Harvesting Section, Pascagoula, Mississippi. Both types of BRDs were tested in each net size. For each net size, one net of a matched pair was equipped with either the FFE or the EMF and served as the experimental net and the other, unal- tered net served as the control. In the experimen- tal net, the FFE or EMF was installed behind the TED-accelerator funnel section. The net with the BRD was deployed off a randomly chosen side of the boat and its paired control net was deployed simultaneously off the other side in a double-rig trawl towed from 3.5-m outriggers. Each net was spread by two 123-cm x 62-cm wooden trawl doors linked bv a tickler chain. Steele et aL; Efficiency of bycatcti reduction devices in small otter trawls in tfie Florida shrimp fishery 341 Side view Accelerator funnel TED Codend Figure 3 Stylistic diagram of the bycatch reduction devices used in this study: I A) Control net equipped with an accelerator funnel in front of the TED; (B) net with Florida fisheye (FFEi, the device is inserted into the tailbag behind the TED; (C) net with large-mesh extended-mesh funnel (EMF) inserted directlv behind the TED. Sampling protocol Our sampling protocol was established in consultation with representatives from the NMFS Pascagoula Laboratory and the FMFC. Coleman and Koenig^ estaWished that TEDs did not work as finfish excluder devices in inshore waters; therefore we did not test their exclusion efficiencies. Sampling was conducted aboard a 35-ft, diesel-pow- ered, Bruno & Stillman trawler boat, modified with out- riggers. The nets were deployed and retrieved with a hy- draulic powered system. Prior to instalHng and testing the BRDs, we equipped all pairs of nets of each size with the combined TED and accelerator-funnel sections and tested them for comparable catchability. To test each BRD type in each size of net. we conducted twenty paired tows at night diuing a three-week sampling period in each season. Each pair of nets was towed 10 times within a two-week time period. To minimize any jx)tential bias inherent to a particular net or side of the boat, the two nets of each pair were switched to opposite sides of the boat after 10 tows were completed. All paired nets were towed in water depths of 3.5 to 5.0 m for 30-min bottom time at an average speed of 2.5 kn; speed was determined through use of the glob- al positioning system (GPS). All trawling was conducted in ar- eas where the commercial shrimp fishery operates. The catches from the paired nets (BRD and control) were maintained separately and were sorted onboard the vessel. After each tow, the shrimp, finfish, invertebrates (horseshoe crabs, portunid crabs, sponges, tunicates) and "trash" (seagrass, rocks, shells, anthropogenic debris, etc.) from each net were separated. The large invertebrates (horseshoe crabs, blue crabs, etc.) and trash were weighed separately, the invertebrates were counted, and both the invertebrates and the trash were discarded. The total catch of shrimp and finfish from each net was weighed separately. The shrimp were counted, sex was determined for 10 randomly chosen individuals, and their carapace lengths (CL) were measured to the nearest 0.1 mm. These measurements from the 20 replicate tows were combined to obtain length-frequency distributions for the shrimp. The remaining bycatch, composed of finfish and small in- vertebrates, was weighed. If the total weight of the by- catch was less than or equal to 4.5 kg, the entire sample was kept; if the weight of the sample exceeded 4.5 kg, a subsample weighing a total of 4.5 kg + 20% of the total bycatch weight was kept. All species of vertebrates and in- 342 Fishery Bulletin 100(2) vertebrates from each bycatch sample or subsample were identified; finfish that could not be identified onboard were labeled and returned to the laboratory for identification. All individuals of each finfish species were counted and the finfish bycatch sample or subsample was weighed. To obtain an estimate of the size-frequency distribution for each species of finfish, we measured the standard length (SL) to the nearest 1 mm of 20 randomly selected individu- als of each species from each tow and combined the mea- surements from the 20 replicate tows. If fewer than 20 in- dividuals were caught in a tow, all individuals captured in that tow were measured. All weights were standardized to grams per minute towed to estimate CPUE (bioniass). All counts of individu- al species were standardized to number per minute towed (NPUE). ysis of variance (ANOVA). We then used the least-squares difference (LSD) post hoc test to locate the significant dif- ferences. Differences between the net with the BRD and its paired control net in the size-frequency distribution of the 10 most abundant fish species were assessed by using the Kolmogorov-Smirnov two-sample test. To deter- mine the percent reduction or increase in the biomass and number of each of the top ten finfish species, we compared the BRD-equipped nets with the control nets by using the untransformed mean CPUE and NPUE data for shrimp and finfish and the total number of individuals subsam- pled. Percent reduction for either CPUE or NPUE was then calculated (from Rogers et al., 1997) as Percent difference = Statistical analyses Statistical analyses followed Sokal and Rohlf ( 1995). Para- metric statistics were applied when the data conformed to the parametric assumptions of normality (Shapiro-Wilk test) and homogeneity of variances (Levene's test). Vari- ables that did not conform to parametric assumptions were transformed to log (biomass or number) + l. Non- parametric statistics were employed only after appropri- ate methods were deemed unsuccessful in transforming the data to meet parametric assumptions. Both paramet- ric and nonparametric statistical analyses were completed by using the STATISTICA software package (Statsoft Inc, 1999). Using ^tests, we evaluated the performance of the paired nets prior to the addition of the BRDs and com- pared the catchability of the BRD-equipped net to its con- trol. Because we used a paired-tow design for field testing, we analyzed each net size and type of BRD separately: net sizes and BRDs were not directly compared with each other but were always compared with the controls. The ability of a BRD-equipped trawl to retain shrimp while reducing bycatch was assessed by comparing the CPUE (biomass) and NPUE of finfish and shrimp and by compar- ing the CPUE and the NPUE of the 10 most abundant finfish species in the BRD-equipped net with CPUE and NPUE data for its paired control net. The CPUE and NPUE of shrimp caught, calculated as described above, were based on actual weight and numbers of shrimp caught in each trawl. Wlien the bycatch was subsampled, the finfish biomass or number was estimated using the formula Finfish biomass or number = Finfish subsample CPUE or NPUE xTotal bvcatch weight Subsample weight Because our sampling period ranged over two seasons, we considered the interactive effects of season and net type (BRD-equipped or control) for each net size by using anal- ( CPUE or NPUE of BRD net - CPUE or NPUE of control net) x 100 CPUE or NPUE of con trol net Results No significant differences were found in total weight of the finfish or shrimp catch between nets of equal size prior to the addition of the BRDs. Similarly, the total weight of finfish or shrimp was not significantly affected by trawl position. The standardized mean ratio of finfish bycatch to shrimp biomass for all control net sizes combined was 5.3:1 (range 2.9:1-11.3:1), The standardized mean ratio for the BRD-equipped nets (3.8:1; range 2,5:1—4.9:1) was not significantly different but was substantially lower than that of the control nets. CPUE and NPUE In contrast to results with the control nets, there were no significant differences in either biomass or number of shrimp captured in the 17-m net or the 20-m net equipped with either BRD (Table 1). In winter, the biomass and the number of shrimp captured in the 14-m net equipped with either BRD were significantly lower than these quantities captured in the corresponding control net iFFE: P=0.025; EMF: P=0.008). On the contrary, both the biomass and number of finfish were significantly and notably lower in most of the BRD-equipped nets than they were in the con- trol nets (for significant differences, P range for the FFE= 0.025-0.001 and P range for the EMF= 0,027-<0.001 ). The only exception was in the number of finfish caught in winter by nets equipped with either BRD. Significant seasonal differences always occurred in shrimp CPUE (FEE: P<0.001 for all tests; EMF P range: 0.003-<0.001 ) and nearly always occurred in shrimp NPUE (FEE P range: 0.003-<6.001; EMF P range: 0.002-0.001; exception: NPUE for the 17-m EMF-equipped net) and ac- counted for most of the variation in CPUE obsei'ved for each net size. Nearly all significant differences in shrimp CPUE and NPUE between seasons were due to a larger catch of shrimp in winter The only significant interactive Steele et a\ Efficiency of bycatcfi reduction devices in small otter trawls in [he Florida shrimp fisfiery 343 Table 1 Comparison of percentage difTerences in shrimp and finfish biomass CPUE) and number (NPUE) from the 14-m , 17-m, and 20-m 1 nets equipped with the Florida fisheye (FFE and extended mesh funnel (EMF) bycatch reduction devices (BRDs). CPUE is the | mean weip;ht ( jrams) caught per minute towed, and NPUE is the mean num ber of ir dividuals caught 3er minu te towed. Signifi- cance levels of <0.05 arc di Mioted by asterisks Sh imp 1 ish CPUE NPUE CPUE NPUE BRD control ' diff. BRD control '■^diff BRD control 'r difT. BRD control 'i difr. FFE Fall 1996 14-m net 29 31 -6 2 2 -5 99 154 -35* 3 4 -34* 17-m net 21 21 4 2 1 7 70 105 -33* 2 3 -41* 20-m net 17 16 6 1 1 11 176 196 -11 3 5 -39 Winter 1997 14-m net 36 45 -20* 2 2 -16* 113 132 -14 4 4 3 17-m net 55 59 -6 4 4 -5 119 124 -4 12 12 5 20-m net 42 38 11 2 9 14 114 130 -12 7 6 28 EMF Fall 1996 14-m net 32 33 -5 2 2 -5 90 168 -46* 3 5 -39* 17-m net 23 25 -9 2 2 -11 60 151 -60* 2 4 -60* 20-ni net 12 12 0 1 1 0 116 174 -33* 3 5 -28* Winter 1997 14-m net 42 60 -29* 2 3 -25* 106 130 -18 7 6 28 17-m net 31 39 -18 2 2 -21 86 121 -28* 8 7 7 20-m net 25 22 18 1 1 17 133 169 -21 5 4 12 effect between season and BRD type occurred in shrimp CPUE for the 14-m, EMF-equipped net (P=0.027). Similarly, finfish CPUE differed seasonally for most net sizes iFFE P range: 0.009-< 0.001; EMF P for all tests: <0.001; exceptions; 14-m and 17-m, EMF-equipped nets), and NPUE differed seasonally for all net sizes (FFE P for all tests; <0.001; EMF P range: <0.001-0.000). However, the season in which the largest catch was har- vested differed between net sizes and between BRD types. Both finfish CPUE and NPUE were significantly higher during winter than during fall in the 14-m and 17-m FFE- equipped nets but finfish CPUE and NPUE were signifi- cantly lower during fall than during winter in the '20-m FFE-equipped net. For the EMF-equipped nets, finfish CPUE differed seasonally only in the 20-m nets: CPUE in winter was higher than in fall. Finfish NPUE values for the EMF-equipped nets were always significantly higher in winter (P for all tests; <0.001). Percent reduction Differences in the percentage of shrimp in the BRD- equipped versus the control nets varied with season, net size, and BRD type. Although many of these differences were not significant (Table 1), patterns in shrimp loss or retention were apparent. Other than the 17-m, FFE- equipped net in fall, the addition of a BRD to a 14-m or 17-m net resulted in a reduction in shrimp CPUE and NPUE, regardless of BRD type. Howe%'er, the reductions were sig- nificant only for the 14-m nets in winter. In contrast, shrimp CPUE and NPUE usually were slightly higher in the 20-m BRD-equipped nets than in the control nets. Finfish CPUE was always less in BRD-equipped nets than in control nets (Table 1). The reduction in finfish bycatch CPUE was nearly always significant in fall, and most reductions were dramatic (20-60%). Reductions in finfish NPUE also had a strong seasonal component. For all net sizes, finfish bycatch NPUE in the BRD-equipped nets was notably (and nearly always significantly) less than that in the control nets in fall, whereas more (but not significantly more) finfish were captured in the BRD- equipped nets than in the control nets in winter. Catch composition Most of the biomass in both the BRD-equipped and the control nets usually was composed of finfish (30-70%). The remainder of the catch consisted of shrimp (1.5-20%), horseshoe crabs (Limulus polyphemus) and blue crabs (Callinectes sapidus) (15-58%), and miscellaneous inver- tebrates such as ctenophores, portunid crabs, sponges, and gastropods (<25'>). Wlien the catch of arthropods (princi- pally horseshoe crabs) was large, the finfish catch was gen- erally small. The shrimp catch was relatively stable even when the bycatch composition fluctuated. Horseshoe crabs were the most abundant invertebrate bycatch species. A total of 2867 horseshoe crabs were caught during the two sampling seasons; largest catches occurred during fall. Although the catch of horseshoe crabs caught in the BRD-equipped nets was generally smaller 344 Fishery Bulletin 100(2) than the catch in the corresponding control nets, only in the 14-m FFE-equipped net and the 20-m EMF-equipped net was the number of horseshoe crabs caught significant- ly lower than the number of horseshoe crabs caught in the corresponding control nets (P=0.001 and P=0.05. re- spectively). Blue crabs were the second most abundant invertebrate bycatch species. A total of 544 blue crabs were caught during the two sampling seasons; the largest catches occurred during winter Although fewer blue crabs were caught in the BRD-equipped nets, only in the 14-m EMF-equipped net was the number of blue crabs signifi- cantly lower than that in the corresponding control net (P=0.005 for both seasons). A total of 44 species of finfish were caught during our study (Table 2). Numerically, ten finfish species composed more than 92^( of the total finfish count, and a single spe- cies, the leopard searobin iPrionotus scitiilus), composed over 40'? . Abundance differed gi'eatly between seasons for nearly all fishes (Table 2). The silver jenny {Eiicinnsto- inus gula), hardhead catfish (Arius felis), gafftopsail cat- fish (Bagre marinus), sand seatrout (Cynoscion arena/ius), and silver perch (Bairdiella chryi^oura) predominated in the catch during fall. These were replaced during winter by the leopard searobin (Prionotus scitulus), blaekcheek tonguefish iSyrriphurusplagiusa). southern kingfish (Men- ticirrhus americanus), pinfish (Lagodon rhomboides). and spadefish iChaetodipteriis faber). Ten of the finfish species that we captured are important to the recreational or commercial fishing sectors. These are the southern kingfish (Menticirrhus americanus), scaled sardine (Harengula jaguana), striped anchovy (Anchoa hepsetus). bay anchovy i Anchoa mitchelli), spot (Leiosto- mus xantluunis). spotted seatrout iCynoscion nehulosus), gulf menhaden tBrevoortia patronus), gidf flounder iPara- lichthys albigutta), pompano (Trachinotus carolinus), and permit (Ti-achinotus falcatiis). These species each account- ed for less than Vi of the total finfish count, except for the southern kingfish, which accounted for 4.6''( . For the species captured principally in fall, the overall proportion of the bycatch excluded by the 14-m and 17-m BRD-equipped nets was similar, and both sizes of nets tended to exclude high percentages of these fishes (Fig. 4). The 20-m BRD-equipped net was not as effective in reduc- ing the numbers of these species. For the species captured principally in winter, the efficiency with which the BRD- equipped nets excluded these fishes varied among net sizes and BRD types. For some species (e.g. the leopard searobin and blaekcheek tonguefish), BRD-equipped nets retained more individuals than did the corresponding control nets. Size distribution Shrimp size-frequency distributions for pooled trawls ( BRD- equipped net and its corresponding control ) had significant seasonal variation (P<0.001,?=16.1,df=2,416). In fall, mean carapace length was 23.4 mm (SD=4.7 mm) and the range was 11.2—40.4 mm, whereas in winter, the mean was 27.0 mm (SD=6.5 mm) and the range was 7.3—43.8 mm. Mean sizes of the 10 most abundant finfish species dif- fered significantly between the BRD-equipped nets and their corresponding controls in approximately '257( of the trawls with the FFE-equipped nets and in 30"r of the trawls with the EMF-equipped nets (Table 3). The differences in mean sizes of individuals were usually small regardless of statistical significance. Nevertheless, the ratio of compari- sons in which mean size of fish from BRD-equipped nets was smaller than that of fish from control nets to compari- sons in which the mean size offish from BRD-equipped nets was larger than that offish from control nets was 2:1 for the trawls with the FFE-equipped net and 3: 1 for the trawls with the EMF-equipped net. The only net size and BRD- type combination for which the mean size of individuals from the BRD-equipped net was always smaller than that from the control net was the 14-m FFE-equipped net. Discussion Shrimp catch Although most BRD-equipped nets retained less biomass and fewer numbers of shrimp than did their corresponding control nets, the difference in these measures between the BRD-equipped nets and their controls was significant only for the 14-m net in winter Indeed, shrimp biomass and number in the 20-m BRD-equipped net slightly exceeded biomass and number in the corresponding control net. In previous studies, researchers evaluating the efficiency of BRDs also found that the shrimp catch in BRD-equipped nets tended to be higher than in control nets. They attrib- uted the increase in shrimp catch in their BRD-equipped net to a greater net spread caused by a reduction in the amounts of bycatch and drag (Rogers et al., 1997; Coleman and Koenig'") and to an increase in the volume of water filtered through the net due to the position of the BRD (Christian et al.'''). The numbers of shrunp retained in all BRD-equipped nets and in nearly all control nets were greater in winter than in fall. In the Tampa Bay region, adult female shrimp migi'ate out of the bay to spawn during spring and fall and juvenile shrimp are recruited into the bay during sum- mer and winter lEldred et al., 1965). The increase in abun- dance and the larger size range of shrimp that we caught during winter support this finding. Finfish bycatch Overall, both BRDs proved to be highly effective in reduc- ing finfish bycatch without greatly reducing shrimp catch. The reduction in bycatch was usually significant in the 14-m and 17-m BRD-equipped nets. The mean ratio of fin- fish biomass to shrimp biomass in our BRD-equipped nets was within the range of ratios reported by others who tested the BRD-equipped nets in the Gulf (Alverson et al., 1994; GSAFDF-). Branstetter (1997) and Watson et aV^ Watson.J.,A. Shah, and D.Foster. 1997. Report on the status of bycatch reduction device (BRD) development. National Marine Fisheries Sei-vice, Mississippi Laboratories, P.O. Drawer 1207, Pascagoula. MS, 39568. Steele et al Efficiency of bycatch ledLiction devices in small otter trawls in the Florida sfirimp fishery 345 Table 2 Percentage contribution of individual finfish species subsampled from catches of otter trawls towed in Ta mpa Bay during fall 1996 | and winter 1997. All tows have been pooled and incorpoi ate both BRD-equippcd and control nets for all three trawl- net sizes. Seasonal percentages are calculated for each species. Total number of % % % Common name Species sampled fish (Total) (Fall) (Winter) Leopard searobin Prionotus scitutus 28,299 41.02 6.6 93.4 Silver jenny Eucinostomus gula 9134 13.24 86.2 13.8 Gafftopsail catfish Bagre marinus 6658 9.65 83.5 16.5 Blackcheek tonguefish Symphurus plagiusa 4613 6.69 35.6 64.4 Sand seatrout Cynoscion arenarius 3365 4.88 80.6 19.4 Southern kingfish Menticirrhus americanus 3193 4.63 44.0 56.0 Hardhead catfish Ariiis felis 2949 4.27 88.0 12.0 Silver perch Bairdiella chrysoura 2463 3.57 64.1 35.9 Pinfish Lagodon rhomboides 1503 2.18 0.8 99.2 Spadefish Chaetodipterus faber 1487 2.16 10.7 89.3 Bighead searobin Prionotus tribulus 759 1.10 12.9 87.1 Scaled sardine Harengula jaguana 642 0.93 59.4 40.6 Hogchocker Trinectes maculatus 538 0.78 37.7 62.3 Striped anchovy Anchoa hepsetus 524 0.76 25.5 74.5 Southern puffer Sphoeroides nephelus 346 0.50 5.2 94.8 Southern hake Urophycis floridana 337 0.49 0.0 100.0 Bay anchovy Anchoa mitchelli 320 0.46 1.3 88.7 Pigfish Orthopristis ch rysoptera 274 0.40 3.7 96.3 Inshore hzardfish Synodus foetens 237 0.34 76.4 23.6 Lined sole Achirus lineatus 160 0.23 26.8 73.2 Lookdown Selene vomer 138 0.20 100.0 0.0 Atlantic bumper Chloroscombrus chysiirus 135 0.20 97.7 2.3 Striped burrfish Chilomycterus schoepfi 128 0.19 2.3 97.7 Ocellated flounder Ancylopsetta quadrocellata 93 0.13 1.1 98.9 Spot Leiostomus xanthurus 89 0.13 0.0 100.0 Crested blenny Hypleurochilus geminatus 82 0.12 1.2 98.8 Rough silverside Membras martinica 78 0.11 24.4 75.6 Planehead filefish Monacanthus hispidus 76 0.11 15.7 84.3 Threadfin herring Opisthonema ogUnum 72 0.10 38.8 61.2 Scrawled cowfish Lactophrys qtiadricornis 64 0.09 3.2 96.8 Crevalle jack Caranx hippos 47 0.07 100.0 0.0 Sheepshead Archosargus prohatocephalus 46 0.07 97.8 2.2 Spotted seatrout Cynoscion nebulosus 43 0.06 100.0 0.0 Orange filefish Aluterus schoepfi 22 0.03 9.1 90.9 Gulf menhaden Brevoortia patronus 19 0.03 68.4 31.6 Gulftoadfish Opsanus beta 15 0.02 13.4 86.6 Harvestfish Peprilus alepidotus 10 0.01 50.0 50.0 Leather] acket Oligoplites saurus 6 0.01 100.0 0.0 Shrimp eel Ophichthus gomesi 5 0,01 80.0 20.0 Gulf flounder Paralichthys albigutta 5 0.01 40.0 60.0 Gulfbutterfish Peprilus burti 4 0.01 .50.0 50.0 Striped mojarra Diapterus plumieri 1 0.00 100.0 0.0 Pompano Trachinotus carolinus 1 0.00 100.0 0.0 Permit Trachinotus falcatus 1 0.00 100.0 0.0 346 Fishery Bulletin 100(2) ro 1400- 1200 1000 800 600 400 200 0 1000 800 - 600 400 • 200 -j Silvery jenny Hardhead caHish 0 - 1000- 800- 600 400- 200 0 1000 800- 600 400 200 67 -55 "! r: ; i -88 I- -98 -91 -90 — 1 1 r* il i . '1 - t ■fl lii -Iv Sand seatrout III -Si Ml •M^ FFE EMF O 6 o o '-" 5 Leopard searobin 5 1000 800- 600 400 200 : : +32 Blackcheek tonguetish Gatftopsail catfish =i "D 0 1000 800 ■ ' 4 ♦ 9 u ^ 600 o 400- 200 Southern ld with the control nets varied markedly between seasons. In fall, the number of fish in the BRD-equipped nets was nearly always much lower than the number in the corresponding control nets; but in winter, the number in the BRD-equipped nets was generally slightly higher than the number in the corresponding control nets. This increase was due to the sizable influx of juvenile leopard searobins in the finfish catch in winter. The long pectoral- fin rays on these fish became entangled in the nets (and in the BRD) and prevented the fish from escaping. The detailed analysis of species-specific change in num- bers of fish in the BRD-equipped nets compared with their corresponding control nets revealed additional in- teresting patterns. The number of silver Jennys was re- duced in all BRD-equipped nets, except in the 17-m BRD- equipped net during winter. The numbers of hardhead catfish, sand seatrout, silver perch, and southern kingfish were always reduced in the BRD-equipped nets except in the 20-m FFE-equipped net during fall. The number of leopard searobins and blackcheek tonguefish nearly al- ways increased in the BRD-equipped nets, regardless of season. With some exceptions, larger fish were more like- ly to escape than smaller fish, probably because swim- ming ability is positively associated with size in fishes (Wardle, 1993). However, fish (particularly large individu- als) of species with protruding bony scutes or long fin rays (e.g. gafftopsail catfish, leopard searobin. southern king- fish) became entangled in the nets and thus could not es- cape. The potential for large individuals of these types of fish to become entangled in the net may have increased because of the restricted net circumference, caused by the presence of the EMF. Thus, for these types of species, mean size of individuals retained in the BRD-equipped nets was frequently larger than mean size of individuals retained in the corresponding control nets. A number of factors other than morphological features, such as pointed, projecting body structures, influence the ability of fish to escape from trawl nets equipped with BRDs. The behavior of fish in response to trawls has been described as a combination of optomotor response and rheotactic reaction, both of which contribute to a fishes' ability to escape capture in a trawl (Watson'"). When am- bient light conditions and water clarity allow for sufficient contrast between the trawl and the background, many, but not all, fishes orient their heads toward the mouth of trawl and maintain swimming speeds comparable to the trawling speed. Thereby, a fish can align itself with the in- trawl current. This optomotor response is usually associat- " Watson, J. W. 1988. Fish behavior and trawl design; poten- tial for selective trawl development. In Proceedings of the world symposium on fishing gear and fishing vessel design. (S. G. Fox and J. Huntington, eds.). p. 2.5-29. Newfoundland and Labrador Institute of Fisheries and Marine Technology. P.O. Box 4920. St. Johns, Newfoundland AlC 5R3. ed with the well-developed lateral line system found in pe- lagic schooling species and is usually absent in demersal species (Pavlov, 1969). However, this response is consider- ably diminished during nighttime and in turbid water, and both of these conditions existed during our trawling. Thus, fishes with well-developed optomotor responses probably required additional stimuli to escape from our nets, even if they were in close proximity to an escape opening. Most of these fishes may have escaped the trawl through the BRD when the trawl speed was reduced during trawl haul-back (Watson'"). The rheotactic response allows demersal fish to detect turbulent water flows and associated pressure gradients through the lateral line even when substantial visual cues are not available (Wardle, 1993). Areas of dis- turbed water exist within a moving trawl, especially near objects such as BRDs, which interrupt water flow. Demer- sal fishes with well-developed rheotactic responses can sense these areas of reduced velocity, align themselves be- hind these areas, and eventually escape through the exit provided by the BRD while the trawl is being towed. The finfish species with the largest percentage reductions in numbers in our BRD-equipped nets compared with the corresponding control nets were demersal and most likely used this response to assist in their escape. BRDs and Fishery management Both the FEE and EMF are standard bycatch reduction devices recommended by NMFS and used by the shrimp industry. The effectiveness of these two BRDs in reducing finfish bycatch without greatly reducing shrimp catches has been well documented for the Gulf and the South Atlantic shrimp fisheries. (Captiva and Rivers, 1960; Gutherz and Pellegrin, 1988; Murray et al., 1992; Watson et al., 1993; Branstetter, 1997; Rogers et al., 1997; Cole- man et al."; Christian et al.''^; McKenna and Monaghan'"'; Watson et al.'^; our study). The FEE is now required in all shrimp trawls used in the federal Exclusive Economic Zone along the South Atlantic and in the Gulf The policy set forth by the EMEC in 1990 to reduce the overall finfish bycatch in the Florida shrimp fishery has been addressed in our study. In part as a result of this study, Florida is the first state bordering on the Gulf of Mex- ico to require the use of BRDs in state waters. BRDs not on- ly serve to conserve natural marine resources, in the Flor- ida shrimp fishery they provide additional benefits to the shrimp fishermen. Reducing bycatch decreases drag during tow times, which, in turn, lowers fuel consumption thereby reducing fuel costs, diminishes wear and tear on the trawl gear, decreases culling time by the deck crew, and produces a better shrimp product. From a cost-benefit perspective, BRDs clearly provide consei-vational, economic, and socio- logical benefits that far outweigh their actual costs. Acknowledgments This paper is a publication of the Florida Fish and Wild- life Conservation Commission and was funded in part by Cooperative Agreement number NA67FI0118 from the 350 Fishery Bulletin 100(2) National Oceanic and Atmospheric Administration. We are grateful to J. Watson and D. Stevens of the NMFS Hai-vesting Systems Section, Pascagoula, MS, for technical guidance. We also thank S. Butler, S. Geiger, C. Crawford, C. Lund, D. Merriman, D. Pierce, J. Smith, W.H. Teehan, and D. Winkelman for assistance in field sampling. We thank L. French, J. Leiby, and J. Quinn for editorial assis- tance and W. S. Arnold, W. H. Teehan, and the anonymous reviewers for providing constructive comments on vainous versions of the manuscript. Literature cited Allsopp, W. H. L. 1982. Use offish by-catch from shrimp trawling: future devel- opment. In Fishbycatch — bonusfromthesea.Reportofatech- nical consultation on shrimp bycatch utilization, p 29-41. IDRC (International Development Research Centre), Otta- wa, Canada. Alverson. D. L.. M. H. Freeberg. S. A. Murawski, and J. G. Pope. 1994. A global assessment of fisheries bycatch and discards. FAO Fish. Tech. Pap. 339. FAO, Rome, 233 p. Branstetter, S. 1997. Bycatch and its reduction in the Gulf of Mexico and South Atlantic shrimp fisheries. Gulf and South Atlantic Fisheries Development Foundation, Inc. Tampa, FL, 27 p. Caddy J. F 1982. Management of shrimp fisheries. In Fish by-catch — bonus from the sea. Report of a technical consultation on shrimp by-catch utilization, p. 120-124. IDRC (Interna- tional Development Research Centre), Ottawa, Canada. Captiva, F. J., and J. B. Rivers. 1960. Development and use of otter-trawling gear for red snapper fishing in the Gulf of Mexico, June 1957-May 1959. Commer Fish. Rev. 22(10):1-14. Eldred. B., J. Williams, G. T. Martin, and E. A. Joyce, Jr 196.5. Seasonal distribution of penaeid lai-vae and postlar- vae of the Tampa Bay area, Florida. Fla, Board Conserv. Mar Res. Lab.. Tech. Sen 44, 47 p. Gutherz, E J., and G. J. Pellegrin. 1988. Estimates of catch of red snapper, Luljanua cantpecha- nus, by shrimp trawlers in the U.S. Gulf of Mexico. Mar Fish. Rev. 50(1): 17-25. Lewis, R. R., Ill, and E. D. Estevez. 1988. The ecology of Tampa Bay, Florida; an estuarine pro- file. U.S. Fish Wildl. Serv. Biol. Rep. 85, 132 p, Lewis, R. R., Ill, R. C. Phillips, and R. Lombardo. 1981. Seagi'ass mapping project. Hillsborough County, Flor- ida Fla. Sci. 44(Suppl. 25l:13. Murray, J. D., J. J. Bahen, and R. A. Rulifson. 1992. Management considerations for by-catch in North Car- olina and southeast shrimp fishery. Fisheries 17(1 ):21- 26. Nance, J. M. 1992. Estimation of effort in the Gulf of Mexico shrimp fishery U.S. Dep. Commer, NOAA Tech. Memo., NMFS- SEFSC-300, 12 p. 1993. Effort trends for the Gulf of Mexico shrimp fishery. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-337, 37 p. Pavlov, D. S. 1969. The optomotor reaction of fishes. FAO Fish. Rep. 62(31:80.3-808. Rogers. D., B. Rogers, J. de Silva, and V. Wright. 1997. Effectiveness of four industry-developed bycatch reduc- tion devices in Louisiana's inshore waters. Fish. Bull. 95: 552-565. Saila, S. B. 1983. Importance and assessment of discards in commer- cial fisheries. FAO Fish. Circ. 765. FAO. Rome, 62 p. SokalR. R., andF J. Rohlf 1995. Biometry: the principles and practice of statistics in biological research, third ed. W. H. Freeman & Co.. New York, NY, 887 p. StatSoft, Inc. 1999. STATISTICA software. Statsoft, Inc., Tulsa, OK, 230 p. Wardle, C. S. 1993. Fish behaviour and fishing gear /n Behavior of tele- ost fishes 2"'* edition (J. T Pitcher) p. 614-643. Chapman & Hall, London Watson, J.. I. Workman, D. Foster, C. Taylor, A. Shah. J. Barbour, and D. Hataway 1993. Status report on the potential of gear modifications to reduce finfish bycatch in shrimp trawls in the southeast- ern United States 1990-1992. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-327, 131 p. 351 Abstract— Status of tho southoastorn lis. stock of red poi-fj>' (/1r;^'rws po^'n/sl was estimatt'ti Ironi (isherv-dependent and fisherv-indi>pt>ndi'nt data, 1972-97. Annual population numbers and fish- ing mortality rates at age were esti- mated from virtual population analysis (\TA) calibrated with fishery-indepen- dent data. For the VPA, a primary matrix of catch at age was based on age-length keys from fishery-independent samples; an alternate matrix was based on fish- en-dependent keys. Additional esti- mates of stock status were obtained from a surplus-production model, also cali- brated with fishery-independent indices of abundance. Results describe a dramatic increase in exploitation of this stock and con- comitant decline in abundance. Esti- mated fully recruited fishing mortality rate iFl from the primary catch matrix increased from 0. 10/yr in 197,5 to 0.88/yr in 1997, and estimated static spawn- ing potential ratio (SPR) declined from about 67% to about 18%. Estimated recruitment to age 1 declined from a peak of .3.0 million fish in 197.3-74 to 94,000 fish in 1997, a decline of 96.9%. Estimated spawning-stock bio- mass declined from a peak of 3.530 t in 1979 to .397 t in 1997, a decline of 88.8%. Results from the alternate catch matrix were similar. Retrospec- tive patterns in the VPA suggest that the future estimates of this population decline w-ill be severe, but may be less than present estimates. Long-term and marked declines in recruitment, spawning stock, and catch per unit of effort ( both fishery-derived and fishery-independent ) are consistent with severe overexploitation during a period of reduced recruitment. Although F prior to 1995 has generally been esti- mated at or below the current man- agement criterion for overfishing ^F equivalent to SPR=35% i, the recent spawning-stock biomass is well below the biomass that could support max- imum sustainable yield. Significant reductions in fishing mortality will be needed for rebuilding the southeastern U.S. stock. Severe decline in abundance of the red porgy (Pogrus pagrus) population off the southeastern United States Douglas S. Vaughan Michael H. Prager Center lor Coastal Fisheries and Habitat Research National Oceanic and Atmospheric Administration 101 Pivers Island Road Beaufort, North Carolina 28516 E-mail address (for D S Vaugfian) Doug Vaugfiancg'noaa gov Manuscript accepted 13 September 2001. Fish. Bull, 100:351-375 120021. New analyses of the red porgy (Pagrus pagrus) stock off the southeastern U,S, coast indicate that dramatic changes in age structure and population abun- dance have occurred over the last 25 years. The red porgy, a protogynous sparid also known as silver snapper and pink snapper, associates with reefs and is commonly found over irregular and low-profile hard bottoms at depths between about 20 and 200 m (Manooch and Hassler, 1978). The species has been an important component of the snapper-grouper complex in the coastal Atlantic Ocean off the southeastern United States, particularly off North and South Carolina, This study is the first complete as- sessment of red porgy in this region since 1986 (Vaughan et al., 1992 1. We in- troduce methodological improvements that strengthen analyses considerably, including fishery-independent data to calibrate the virtual population analy- sis (VPA), as recommended in a recent review of marine stock assessments (NRC, 1998). Extensive analyses were made of the sensitivity of major results (estimates of abundance and fishing mortality rate) to assumptions. Sen- sitivity analyses (in the broad sense) included separate VPAs on two catch matrices, one based on fishery-inde- pendent age-length keys and the sec- ond on fishery-derived keys; estimation with different assumptions about natu- ral mortality rate; a retrospective analy- sis; alternate treatments of zeroes in an abundance index; and the use of a sur- plus-production model to provide com- plementary estimates of stock abun- dance and management benchmarks. Robustness of major conclusions to these factors and relatively close agree- ment between estimates from catch- at-age models and production models strengthen the finding that the red por- gy stock was severely depleted at the close of 1997. Range, stock structure, and reproductive biology Red porgy have an extensive range — they are found off the southeastern U.S. Atlantic coast; in the Gulf of Mexico; off the South American Atlan- tic coast from Brazil to Argentina; off Portugal and Spain; in the Mediterra- nean Sea; off west Africa south to the Cape Verde Islands; and around the Azores, Madeira, and Canary Islands. The stock unit analyzed in our study includes fish from U.S. Atlantic waters off North Carolina (NO south of Cape Hatteras; South Carolina (SC); Georgia (GA); and the east coast of Florida (FL), This range expands that previously defined for the stock, which (Vaughan et al., 1992; Huntsman et al.M included fish only from waters off NC and SC. Within the current stock definition, red porgy have been most abundant from NC and SC waters. Tagging studies ' Huntsman, G. R., D. S. Vaughan, and J. C. Potts. 1993. Trends in population status of the red porgy Pagrus pagrus in the Atlan- tic Ocean of North Carolina and South Car- olina, USA, 1971-1992. South Atlantic Fishery Management Council, 1 Southpark Circle, Charleston, SC 29407. [Available from Beaufort Laboratorv, 101 Pivers Island Road, Beaufort, NC, 28516] 352 Fishery Bulletin 100(2) have shown neither long-range migrations nor extensive local movements of adult red porgy (Manooch and Hassler. 1978), nor has circumstantial or anecdotal information suggested such movements. Peak spawning occurs from March through April (Manooch, 1976). Red porgy eggs and larvae are pelagic, hatch 28 to 38 h after fertiliza- tion, and can sui-vive transport by ocean currents for 30 days or more (Manooch et al., 1981). Thus, the population off the U.S. Atlantic coast could in theory receive eggs or larvae from the Gulf of Mexico stock. However, because of the distances involved and the variability of ocean cur- rents, the likelihood of significant population mixing in this way seems small, and we adopted the stock boundar- ies described above. Red porgy attain maximum size slowly and live relative- ly long (an 18-year-old specimen is the oldest on record), but maturity occurs far younger Roumillat and Waltz- col- lected red porgy between 1979 and 1987 along the conti- nental shelf between Cape Fear, NC, and Cape Canaveral, FL, using trawl nets, traps, and hook-and-line gear Life history information was obtained from 7104 red porgy; 5820 otoliths were examined (including 134 from histori- cal or port samples), of which 5491 had discernable rings; estimation of sex composition was based on 6044 red por- gy. Mature gonads were found in 18.8'^r of females at age 1, 85.29f at age 2, 99.7'>r at age 3, and lOO^f at older ages. Red porgy are protogynous hermaphrodites. Thus, fe- males predominate at smaller size intei-vals, but males occur in all age groups. Age-specific sex ratios reported by Roumillat and Waltz- (their Table 6) were used in our analyses: 89'^f female at age 1, 91'^ at age 2, ll'/i at age 3, 67% at age 4, 59% at age 5 , 51% at age 6, 25% at age 7, and 21% at age 8. Methods Fishery-dependent data sources and adjustments In the study area (Cape Hatteras, NC, through the east coast of FL), three fisheries take red porgy: commercial, recreational, and headboat. Hook-and-line has been the most common commercial gear, but occasional significant landings are taken with trawls and traps. Trawling for red porgy has been banned since 1989 (Amendment 12 ISAFMC'^] ) The recreational fishery includes all recre- ational fishing from shore, from private boats and from charter boats (for-hire vessels that usually accommodate six or fewer anglers as a group). The headboat fishery (larger for-hire vessels that charge per angler) is sampled - Roumillat, W. A., and C. W. Waltz. 199:3. Biology of the red porgy Pagrus pagrus from the southeastern United States. Data report 1993 MARMAP, South Carolina Wildlife and Ma- rine Resources Department, P.O. Box 125.59, Charleston, SC 29422-2559. ' SAFMC (South Atlantic Fishery Management Council. 2000. Final, amendment number 12 to the fishery management plan for the snapper grouper fishery of the South Atlantic region. South Atlantic Fishery Management Council. Charles- ton. SC, 159 p. + appendices. separately, and for that reason is distinguished here from other recreational fisheries. Recreational and headboat fisheries, like the commercial fishery, use hook-and-line gear almost exclusively. Data sources for all three fisheries are described in our study and summarized in Table 1. Commercial fishery Landings statistics in weight for the commercial fishery, 1972-97, were obtained from the NMFS general canvass database (NMFS, Southeast Fish- eries Science Center, Beaufort, NC, and Miami, FL). In addition. North Carolina commercial landing statistics by gear for red porgy were provided by NC Division of Marine Fisheries (NCDMF). Fishery-based length and weight sta- tistics, 1983-97, were obtained from the NMFS Trip Inter- view Program (TIP) database. Length frequencies from SC commercial landings, 1976-80, were taken from Vaughan etal. (1992). Commercial landings statistics, 1972-84, are recorded as "porgies," but include some other species. Following Vaughan et al. ( 1992), we multiplied such non-North-Caro- lina landings by 0.9 to approximate landings of red porgy only; North Carolina data were corrected by NCDMF before we received them. Commercial landings were available only in weight and were converted to counts by dividing catch in weight by mean weight per fish for the same gear and year To compute annual mean weights, fish lengths from the TIP database were converted to weight with estimated weight-length relationships, described below. Annual length-frequency distributions L were developed for commercial hook-and-line gear for each area (Carolinas vs. Florida), 1983-97. Commercial length frequencies for 1972-82 were taken from Vaughan et al. (1992). Because trap and trawl landings have been quite small since the mid-1980s and few fish have been sampled (1611 Itrapl; 1455 Itrawll, mostly between 1986 and 1988), two overall length-frequency distributions for each gear were used, one for 1972-89 and the other for 1990-97. Gear-specific commercial length-frequency distributions were weighted by catch in number by state to obtain catch at length by gear and state. Recreational fishery Recreational catch and effort esti- mates and data on length and weight composition, 1979-97, were obtained from the NMFS Marine Rec- reational Fisheries Statistics Survey (MRFSS) database (Gray et al., 1994; Marine Recreational Fisheries Statis- tics, 1999). Fishing modes used by MRFSS include shore- based, private-boat-based, and charter-boat based fishing. Within each mode, three catch types are defined: "A" catches were available to sampling personnel for identifi- cation and measurement; "Bl" catches were unavailable because they had been used for bait, filleted, discarded dead, etc.; and "B2" catches were unavailable because they had been released. Postrelease mortality of B2 catches was assumed to be 18% (Dixon and Huntsman"*). Following ■> Dixon, R. L. and G, R. Huntsman. 1993. Survival rates of released undersized fishes (abstract). Sixth Annual MARFIN Conference, Atlanta, GA, 12-13 October 1993. [Available from Beaufort Laboratory. 101 Fivers Island Road, Beaufort. NC, 28516.) Vaughan and Prager: Decline in abundance of Pagrus pagnis off the soutfneaslern United States 353 Table 1 Data sources for analysis of red porgy ofi'the sou theastern United States. Sample size n is nunilxT of fish lengths measured. Aging | data sources are given in Table 2. Symbol A? is n umber of fish. W, weight of fish caught. Data source Years Database Type n Commercial fishery 1972-97 NMFS, general canvass Landings iW) — 1972-97 NC Div. Mar Fish. NC landings (W — 1983-97 NMFS trip interview Lengths and weights 62,042 1976-80 SCDNR (Vaughan et al., Lengths 8855 Recreational fishery 1992) 1 private and charter boats) 1979-97 NMFS MRFSS Catch (N and W) — Lengths and weights 945 Headboat fisherv 1972-97 NMFS headboat survey Catch (N and W) — (NC.SC) Lengths and weights 44,131 1976-97 NMFS headboat survey Catch (TV and W) — (GA-N.E. FL) Lengths and weights 1454 1981-97 NMFS headboat survev Catch (TV and W) — (S.E. FL) Lengths and weights 1166 Fishery-independent 1979-97 MARMAP hook and line CPE (iV) and lengths 2085 1980-97 MARMAP traps (FL snapper and chevron) CPE (W) and lengths 10,784 Vaughan et aL (1992), we adjusted the data to avoid dupli- cation from inclusion of headboat landings in charter-boat landings reported by MRFSS, 1979-85. Mean recreational landings by mode were calculated for 1979-97 and used for 1972-78, a period of minimal ex- ploitation before MRFSS data were collected. Because rec- reational landings are a small fraction of the total, this procedure, while approximate, was better than assuming landings of zero in that period. Recreational length-frequency distributions from MRFSS, 1979-97, were weighted by catch (A-t-Bl) in number by mode, wave (2-month interval), and state. The minimal catch associated with shore-based fishing was pooled with the private-boat mode. Headboat length-frequency distri- butions from MRFSS ( 1979-85) were not used in develop- ment of catch matrices for the recreational fishery because more complete data were available from the headboat sampling program. Headboat fishery Estimates of headboat catch in num- bers and weight and fishing effort in angler-days (Table 1) were obtained from the NMFS (Beaufort, NC) head- boat sampling program (Huntsman, 1976; Huntsman et al.. 1978). Zero-intercept hnear regression analyses on NC and SC data were used to estimate landings from GA to northeast FL and landings from southeast FL. and fishing effort back to 1972. Catch per effort (CPE) in numbers was calculated by dividing annual catch in numbers by annual effort. Length samples from headboats were not available from GA or FL before 1976. nor from southeast FL before 1981. We used mean length-frequency distributions for 1976-80 to fill in for GA and FL, 1972-75, and length-frequency dis- tributions from GA to northeast FL to fill in for southeast FL. 1976-80. Annual headboat length-frequency distribu- tions from the headboat sampling program. 1972-97. were weighted by catch in numbers caught during that season (January-May. June-August, September-December), and geographic area (NC. SC. GA to northeast FL, southeast FL) to obtain catch in length by season and area. Development of catch-at-age matrices Estimates of catch at age m numbers were made by using an approach similar to that of Vaughan et al. (1992). Total catch in numbers (n. scalar) for each combination of time, gear, and area was multiplied by an age-length key (A, matrix of dimension a x 6); the product was multiplied by the corresponding length-frequency distribution (L, vector of length b) to obtain catch in numbers at age (N, vector of length a ): ^ = n -A-L. (1) Here, a is the number of ages ( for this study, ages 0 to 8* ) and b is the number of length intervals (for this study, 15 intervals of 25 mm each from 200 mm to 550+ mm). Age-length keys Separate age-length keys were devel- oped from fishery-dependent and fishery-independent data. Fishery-dependent keys for 1972-74 were those of Manooch and Huntsman ( 1977); for 1986. those of Vaughan et al. (1992); for 1996-97, those of Potts and Manooch (2002). Keys for 1975-85 and 1987-95 were approximated by linear interpolation from available keys (method of Vaughan et al., 1992). 354 Fishery Bulletin 100(2) We constructed fishery-independent keys from samples taken in 1979-94 (Harris and McGovern, 19971 by the Marine Resources Monitoring, Assessment, and Predic- tion (MARMAP) Program, a fishery-independent sampHng progi'ani of the SC Department of Natural Resources. In that study, 8660 red porgy were collected, primarily with hook and line and various traps, and preserved sagittae were used for age estimation. As a compromise between estimating annual keys based on smaller sample sizes and an overall key, which would disguise growth variability over time, we gi-ouped MARMAP aging data into 3-,yr pe- riods, with the last four years divided instead into two 2-yr periods: 1979-81, 1982-84, 1985-87, 1988-90, 1991-92, 1993-94. Potts and Manooch (2002) aged 111 red porgy collected during 1996-97 by the MARMAP program (pre- dominantly younger fish, maximum age of 6). We used an age-length key from those data for 1995-97. Where fewer than 10 fish occurred in a 25-mm total-length interval, we pooled the data over longer time periods. No aging data are available from fishery-independent sources before 1979. We constructed a key. used for 1972-74, from fishery samples taken in those years (Ma- nooch and Huntsman, 1977). For 1975-78, we interpo- lated linearly between that key and the earliest key de- rived from fishery-independent data (1979-81). Thus, the fishery-independent (primary) catch matrix described be- low reflects some fishery-dependent data in the earliest years. Catch-at-age matrices Equation 1 was applied to each fishery-gear combination and catch-at-age estimates were accumulated for each year to obtain estimates of annual catch in numbers at age (the catch matrix). 1972-97. This was done twice, first by using the fishery-independent age- length keys to obtain the primary catch matrix, and then by using fishery-derived age-length keys to obtain the alternate catch matrix. (Although the designation of "pri- mary" and "alternate" matrices is somewhat subjective, it was based on the more extensive and continuous qual- ity of fishery-independent data, as discussed later Despite the designation, most analyses were conducted twice, once with each matrix.) Final catch matrices each contained numbers of fish caught at ages 1 through 8-i- in fishing years 1972 through 1997; partial recruitment for age-0 red porgy was essentially zero. We judged coherence of catch matrices by examining pairwise correlations between ages as a cohort progressed through the fishery. For fully recruited ages and with F varying only moderately from year to year, we expected to be able to follow a strong or weak cohort through the catch matrices. Both matrices generally showed significant cor- relations (P<0.1) between lagged catches at adjacent ages. A few significant correlations were found among lagged catches for nonadjacent ages. Growth in length and weight Von Bertalanffy ( 1938) growth models were fitted from data on total length (mm) and age obtained from fishery-inde- pendent and fishery-dependent sources (Table 2). Disaggre- gated aging data could not be located for fishery-dependent data from 1972 to 1974 and from 1986; therefore observed midintei-val lengths at age from keys were used for those years. For fishery-dependent data for 1989-98 and fishery- independent data for 1996-97, back-calculated length at oldest age was used, a procedure that avoids potential bias by the use of multiple measurements per fish (Vaughan and Burton, 1994). For fishery-independent data, 1979-94, obsei-ved length at age was used, adjusted for month of col- lection. The particularly high standard error associated with the estimate of L . for fishery-independent data, 1996-97, was associated with a lack of fish older than 6 years of age in the sample (Table 2). We estimated weight W (in kg) from total length L (in mm) by using the relationship W = aV\ Parameter es- timates were as follows: for fishery-dependent modeling, we used values from Manooch and Huntsman (1977) (o = 2.524 X 10-''> and 6=2.894) for the 1970s and 1980s; and from Potts and Manooch (2002;a=8.85 x 10-6 ^^d 6=3.060) for the 1990s. For fishery-independent modeling, we used values for 1972-78 from Manooch and Huntsman (1977), and we estimated values for 1979-97 from MARMAP da- ta. 1979-94, (a=3.064 x lO-^ and 6=2.865). Abundance indices Fishery-independent data on length, weight, age, and CPE, 1979-97 (Table 1) were obtained through the MARMAP program (Collins and Sedberrv, 1991; Harris and McGov- ern, 1997). In computing fishery-independent indices of abundance from those data, we used CPE from hook-and- line and trap gears only. We extended data from chevron trap, 1988-97, back to 1980 with data on Florida snapper trap, using a conversion factor determined by MARMAP (Collins, 1990; Vaughan et al., 1998); we refer to the result- ing series as "extended" chevron trap. To obtain age-spe- cific indices of abundance for hook-and-line, 1979-97, and extended chevron trap, 1980-97, we applied Equation 1, using CPE in place of/;, and using annual MARMAP esti- mates of CPE, age-length keys, and length-frequency data for each gear Like the catch matrices, matrices of fishery-indepen- dent CPE estimates were examined for coherence through correlations among lagged CPE. The hook-and-line index showed greatest coherence among ages 1 through 4, where- as the extended chevron-trap index appeared quite coher- ent over a wider range of older ages (3 through 7 ). Mortality estimation Instantaneous total mortality rates (Z) were estimated from cohort-specific catch-cui-ve analyses (Ricker, 1975) of the two catch matrices by using only ages that appeared fully recruited (4 through 7). Empirical approaches were also used to estimate total mortality rate (Z). The more complex of two methods presented by Hoenig (1983) is based on age of the oldest fish obsei-ved, sample size, and age of recruitment to the sampling procedure. Assuming those quantities to be 18 yr, 10,000 fish, and 1 yr, we esti- mated Z = 0.58/yr, a likely upper bound on M. Vauglian and Prager Decline in abunidance of Pagrus pagrus off tfie soutfieastern United States 355 Table 2 Parameters of von Bertalanffy growth models for red porgy off southeastern United States. Maximum age in the sample is labeled /„ , ^, sample size is n. Parameters estimated from observed mid-interval length at age in fishery-dependent data for 1972-74 and 1986; from back -calculated length at oldest age in fishery-dependent data for 1989-1998 and fishery-independent data for 1996-97; and from obsei-ved length at age adjusted for month of collection in fishery-independent data for 1979-94. Asymptotic standard error is in parentheses below corresponding parameter estimate. Type of data and year n L. (mm) k ',. ^max lyr) Range (mm) Fishery-depend 1972-74 ^nt data 1685 575.8 (2.5) 0.16 (0.002) -1.88 (0.03) 15 185-640 1986 535 1252.9 (28.7) 0.40 (0.001) -1.05 (0.04) 12 213-615 1989-98 492 796.6 (43.9) 0.09 (0.009) -2.34 (0.23) 18 247-723 Fishery-independent (MARMAP) data 1979-94 8601 485.5 (5.3) 0.23 (0.01) -1.48 (0.06) 14 142-568 1979-81 1171 501.3 (5.4) 0.34 (0.01) -0.15 (0.06) 13 142-557 1982-84 2159 626.1 (24.6) 0.15 (0.01) -1.61 (0.14) 14 145-568 1985-87 2332 542.5 (20.4) 0.16 (0.011 -2.24 (0.16) 12 176-523 1988-90 1268 424.2 (8.6) 0.31 (0.02) -1.06 (0.141 12 176-487 1991-92 842 425.1 (10.1) 0.25 (0.02) -1.62 (0.20) 13 179-456 1993-94 829 375.7 (5.5) 0.43 (0.04) -0.78 (0.18) 12 183-501 1996-97' 111 754.7 (246.3) 0.10 (0.05) -1.27 (0.381 6 151-384 ' Otoliths from fis h collected during 1996-9' ■ by MARMAP but aged by C. Man ooch and J. Potts i NOAA Beaufort Laboratory). We estimated natural mortality rate empirically (Pau- ly. 1979) from parameters L_ and K of the von Berta- lanffy growth function and mean environmental tempera- ture. Estimates ranged frotn M = 0.27/yr to M = 0.57/yr, based on mean sea temperature of 22°C (Manooch et al., 1998) and the range of growth parameters estimated in our study (Table 2). A similar method, that of Ralston ( 1987) and based solely on growth parameter K. provided estimates from M = 0.22/yr to M = 0.64/vr. Given the un- certainty in M, we chose a base M = 0.28/yr (Vaughan et al., 1992). We also performed bracketing estimates of most quantities using M = 0.20/yr and M = 0.3.5/yr. Fishing mortality rate F was estimated by subtracting the assumed natural mortality rate from the total mortal- ity rate. These estimates of F were used only as terminal- year values in the separable VPA. Virtual population analyses Primary and alternate catch matrices were analyzed by using virtual population analysis (VPA) to obtain annual age-specific estimates of F and population size. First, a separable VPA (Doubleday, 1976; Pope and Shepherd, 1982), as implemented by Clay (1990), was used to esti- mate partial recruitment by age for later use in calibrated VPA. Data from 1992-97 were used for this purpose because a 12-inch minimum size limit was introduced in 1992 (amendment 4|SAFMC''| ), and using data from both before and after imposition of the size regulation would likely violate the separability assumption. The starting value of F for the separable VPA was the mean of three final year-class estimates of Z (~0.78/yr), less M.; i.e. F ~ 0.50/yr. A calibrated \TA (Pope and Shepherd, 1985; Gavaris, 1988) was then applied, as implemented in the FADAPT ' SAFMC (South Atlantic Fishery Management Council). 1991. Final, amendment 4, regulatory impact review, initial regula- tory flexibility analysis and environmental assessment for the fishery management plan for the snapper grouper fishery of the South Atlantic region. South Atlantic Fishery Management Council, Charleston, SC, 99 p. + appendices. 356 Fishery Bulletin 100(2) computer program,^ by using the two fishery-independent abundance indices (hook-and-line and extended chevron trap). Most calibrated VPA runs were made with both indices, but a few were made with individual indices to assess sensitivity. The hook-and-line index included esti- mates of zero in 1992 and 1996, to which a small value (0.0001) was added before log transformation, a technique often used in statistical AN OVA models (e.g. Snedecor and Cochran, 1980). Effects of that approach were explored by assuming missing values in place of zeroes in an addition- al calibrated VPA run. Several additional techniques were used to examine sensitivity to assumptions. Sensitivity of estimated F and recruitment to age 1 to uncertainty in M was investigat- ed by conducting the separable and calibrated VPAs with alternate values of M (0.20/yr and 0..35/vr). Retrospective analyses of the calibrated VPA were conducted to inves- tigate the possibility of systematic deviations ("retrospec- tive patterns") in estimates of F and recruitment in most recent years, in comparison to estimates obtained for the same years from later analyses. In the retrospective anal- yses, we varied the final year of data used from 1992 to 1997; the initial year was 1972 throughout. For presentation of results, we averaged estimated quantities within three time periods: 1972-78, repre- senting a lightly fished stock; 1982-86. representing the stock after increasing exploitation during the early 1980s; and 1992-96, representing conditions since the last ma- jor change in management (amendment 4|SAFMC'^|), but omitting the terminal year (1997) to reduce retrospective effects (discussed below). We refer to these as early, mid- dle, and recent periods. Recruitment was defined as the number offish at age 1 on January 1. Yield per recruit Equilibrium yield per recruit (YPR) was estimated by the method of Ricker (1975), which divides the exploited life span into phases of constant mortality and gi'owth rates; total YPR is obtained by summing yield across phases. Parameter estimates from YPR analysis were for both sexes and across the three time periods described above. It has long been recognized that fishing at F^^^.^^ (F that maxi- mizes YPR) can cause recruitment failure (Gulland and Boerema, 1973; Clarke, 1991; Caddy and Mahon, 1995), and the reference point F^ , was introduced as a more con- servative alternative (Gulland and Boerema, 1973). Using most recent growth and selectivity, we computed both F,,,.,^ and Fq j. Spawning potential ratio Static spawning potential ratio (static SPR) — also known as equilibrium SPR or maximum spawning potential (Gabriel et al., 1989) — is a measure of fishing mortality rate scaled to a species' biology. Static SPR for a given fishing mortality rate F* is computed as the ratio SPR = SS(F*) / SS(0), where SS(F*) is the spawning-stock size (lifetime cohort spawning contribution) expected under F ■ , and SS(0) is the corresponding spawning-stock size expected under F = 0. Other life history parameters are assumed constant at most recent values. Increases in F reduce static SPR. In species that do not change sex, spawning-stock bio- mass is usually computed as total mature female biomass or total egg production (Prager et al., 1987), quantities that are highly correlated. Because red porgy change sex, we used four representations of spawning-stock biomass: female mature biomass, male mature biomass, total (male + female) mature biomass, and total egg production, each assuming the sex ratios at age given above ( Roumillat and Waltz-). To compute total egg production, we used the relation- ship between fecundity i£, number of eggs) and total length (TL, mm) of Manooch (1976): In £ = -14.1325 + 4.3598 (In TL). (2) Sex ratio in the mature population can be changed by fishing. We estimated reduction in the proportion of males among mature fish from the proportion expected with no fishing and assuming that the rate of sex transformation is not affected by changes in population structure. Spawner-recruit relationships The relationship between spawning-stock biomass and resulting recruitment to age 1 was modeled by using the Beverton-Holt spawner-recruit function (Ricker, 1975): /? = SSB/(6„ SSB-^6,), (3) '■ FADAPT by V. R. Restrepo (International Commission for the Conservation of Atlantic Tunas, Calle Corazon de Maria, 8, Sixth Floor. 28002 Madrid, Spain). where R = recruitment; SSB = spawning-stock biomass, and 6,i, fe, = fitted parameters. To reduce the influence of retrospective patterns, SSB and R series (estimated from VPA) used in recruitment mod- eling were terminated at 1992. We used total mature bio- mass, rather than the more usual female mature biomass, to represent SSB in recruitment modeling. With the estimated stock-recruitment relationship, 500- year simulations of the stock were made to estimate equilibri- um yield and spawning-stock biomass as functions of fishing mortality rate. The simulations used mean population num- bers, 1992-96, as starting values. Age-specific selectivity for the same period and the most recent gi-owth patterns were used throughout. We thus estimated MSY, B^^^, and F^^a,-^ for each catch matrix. Surplus-production model A surplus-production model was fitted to obtain additional estimates of management benchmarks and stock status. We used the Prager (1994, 1995) formulation and imple- mentation of the continuous-time Graham-Schaefer (logis- Vaughan and Prager: Decline in abundance of Pagrus pagnK off \he southieastem United States 357 tic) model (Schaefer, 1954, 1957; Pella, 1967; Schnute, 1977; Prager, 1994). The approach is an ohservation- error estimator assuming proportional error variance and conditioned on observed landings. We attempted to fit the generalized production model (Fella and Tomlinson, 1969; Fletcher, 1978) by similar methods. Data were total landings in weight. 1972-97, and CFE indices derived separately from fishery-independent hook-and-line and trap data, 1979-97. Initial difficulty in obtaining estimates was traced to two years of zero CPE in the fishery-independent hook-and-line index. Because the fitting was conduct- ed in logarithmic transform, zeroes could not be ac- commodated; our initial, unsuccessful approach was to substitute very small values. We obtained estimates through the alternate procedure of considering those values missing, which is statistically equivalent to setting the observation weights to zero. Because some estimates were moderately sensitive to assumptions about the starting stock biomass, Bj in relation to carrying capacity. A', we obtained estimates at a range of values, namely B^IK = (0.55, 0.65, 0.75, 0.85, 0.951 , a range chosen to reflect the stocks lightly fished status before the first year (1972). Fixing the starting value of relative biomass can reduce variance of estimates considerably (Punt, 1990). Bootstrapping with 700 repetitions was used to estimate 80% bias- corrected confidence intervals (Efron and Tibshirani, 1986) on the central set of estimates (Bj/A'=0.75). Results S 750 -■ 250 B Headboat D Recreational B Commercial 1250 1000 S Headboat D Recreational 0 Commercial I Figure 1 Annual landings of red porgy off the southeastern United States by fishery (headboat, recreational, and commercial) in (A) weight, and (Bi numbers. In this section, all results are reported for the base value of M = 0.28/yr unless otherwise stated. Results reflect analysis of the primary catch matrix unless use of the alternate matrix is specifically mentioned. Summary of landings and CPE During the study period (1972-97), total landings in weight rose from about 325 t in 1972 to a maximum of 880 t in 1982 (Fig. lA). Landings have been at or below 300 t since 1992, with 234 t in 1997. The headboat fishery accounted for about 52^-80% of total landings through 1977; commercial landings have been most prevalent since (Fig. IB, Table 3). Maximum landings in numbers were about 1.13 million fish in 1982 (Table 4). Commercial landings in weight rose from 45 t in 1972 to 678 t in 1982 and declined to 200 t/yr or less since 1992 (Fig. lA); landings in numbers had a similar pat- tern (Fig. IB). Commercial trawl landings averaged about 14*^^ of commercial landings. 1972-84 but were quite small by 1985, and trawl gear was prohibited as of 12 January 1989 by amendment 1 to the Fishery Management Plan (amendment 12 [SAFMC^]). Hook-and-line landings have been about 95'7f of commercial landings by weight since 1985. Recreational landings in numbers have shown no con- sistent trend, averaging about 70,000 fish per year, with some outstanding years of more than 100,000 fish between 1985 and 1990 (F'ig. IB). Catch by mode of fishing has been highly variable, but recreational landings have main- ly been taken by charter boat (43*? by number during the 1990s) and private boat (55*7^). Although remaining land- ings (<3%) are attributed to fishing for red porgy from shore, this amount is unlikely to be correct. These records may result from errors in recorded fishing mode or from erroneous species identification, but in either case would have negligible effect on our analyses. Headboat landings have declined from 200,000 fish/yr during the 1970s to around 100,000-200,000 fish/yr in the 1980s and fewer than 100,000 fisli/yr since (Fig. IB). Most landings have come from NC and SC. Mean weight in the commercial hook-and-line and head- boat fisheries in the Carolinas has decreased, especially since about 1978 (Fig. 2). Mean weight in all fisheries com- bined has shown a general decline from 1972 to 1991 (Fig. 2) but has since increased somewhat. All data on catch per effort (CPE) exhibited long-term decline. Declining headboat CPE was apparent from both NC and SC (Fig. 3). The fishery-independent abundance indices, also based on CPE, showed corresponding declines (Fig. 3); the extended chevron-trap index decline was pre- cipitous; that in the hook-and-line index was similar, but somewhat less dramatic. (The CPE series derived from 358 Fishery Bulletin 100(2) _ +-> CO <^ II II r^ ■* lO CN ^ CO -tT a> r^ Ol OJ t^ ^ in ^ C^ LO o rt X CD Tt* Tt OJ o in oi 05 d in d CO CO CO d in ^ ^ tT ^ t-^ in d d t^ CM d 1> in d ^ d OJ 00 to 00 00 in 05 OJ OJ 05 00 O r- CO 05 [:- X t:^ 1— 1 o Tf CM CO u cq CO (M (N ^ ^ d d d r-i d CO •^ o c~^ d CM o d [^ d CO >T 00 r^ r^ t^ CO CO o o 00 o in ^ OJ in X CO CD i> X o CD in CD t>- CD 6 CO CO ■* t^ 00 00 CO in CO in •* in in t- CD CO OJ CM CM CM CM o c o CJ 'b o ^ -o 03 3 o cc CO to oq 05 t^ ■^ 00 00 ^ cr: I— 1 CD ^ CO CD CO ^ co .-H 00 OJ c^ q [^ ^ t: o6 uo CO 00 oi ^' c-^ c-^ d ^ r-^ CO in d d Ol d d t^ t~-^ ^ CO OJ ^ CO ^ CC C^l j—i Ol Tf cv) ■ — 1 ^H Ol CM j-^ 7] o _Q J l_ B u. o a + V- o o o o o o o o OJ •-H o '— < o X o o o o o o o 00 o ^ o O > < ro ^ -■ '" '-' '-' -i -■ -• CO -i" d d d --i d d d d d d d d ■^ d d in d a ^ O o a o p -Q c: o X c: 05 a> 05 05 05 05 05 00 LO ''^ CO 05 o 05 ^H ^. CO o 05 in q ^ in °° CO 00 -C > ci ai d 05 d d d -t — ;, d d d d d d d ^ CO ^ oi d ^ -o Q- '"' ^ X '"' C3 ra 15 "rt rt c ^o T3 ■^ ■^ CO CO 00 £75 ^ 00 c^ 05 q t:~ -f rH X o c:5 ■* in f-; ^H in t^ X CM X C C3 ^ 't ^ t~^ CO 00 d CO CO d in oi -t t^ d ^ t>^ t-^ t-^ oj ^ 00 [> d r-^ d g3 o c^ o 05 r^ t-- O) -T m CO LO Ol CD '^ LO LO CO 05 CM X r- CO CD CO ^ ■1; (>) c^ cq OJ OJ •-H f— ( 1— 1 X X 0/ t: o CO + S i; Oj ^ r- rH ^ -^ ^ — q -JO -- o r- tT 05 q ^ CTS ^ 00 05 CO CO CO o in 00 fi ^ t-^ t~^ [-^ t-^ t^ t^ f- oj t^ LO d d t- .-^ oi d d in CO Ol .-H CO t-^ d r--^ t-^ o 2 CA (M oa OJ OJ OJ OJ ' ' CD CO en ' ' '^ CD " ^^ t^ ' ' o c Zj *2 m t^ t^ t^ t~ c- l:~ t^ in C^ .—1 ^H CO c- ^ -*■ OJ X T)- I^ CO CO X c^ Tf CO q > ^ ^ ^ ^ ^ ^ ^ c-^ 00 x' LO 00 in iri CO '<:T' I~^ c-^ t~^ in d d ^ CO d CO 5 7" « CO CO oo CO CO CO CO -* ^ in CO 05 CM CM CO CJ £ ^H T3 rt 0) ^ 05 lO c- 05 CO -^ CO "* q in LO m •* ^ m CO q •^ t~ CO CO CTJ o X X C3 c in d t~^ in t--^ oi c6 d d 'i" d ^ 00 oj 00 d d ^ 00 CO oi LO CO CM d d J ^ to '^ 05 CO 05 CD ■* CO i> CO o m -r p LO ^ CD Ol r^ X CO in f- r- 05 |>^, _1 ~7. + CO T^ LO m ^ OJ O o o o o q Ol Ol c c 3 <^ LO LO CM i> CO •* o o u ^ d ^ d d CO d d d d d o d d d d ^ ^ d LO d 00 d i-i d d >, < H »-' O o o o 1 ^- lO o ^ CO OJ q c^ OJ ^^ ^^ LO in X CO CO OJ X CO o o CO •— ' TT q c Cj o OJ CO CO ':r CO d CO ^ oj d 00 r-i ,— i in 00 d d d ^ ^ LO d t^ 00 oi J !M CO CO LO , — 1 Tf in r- o CO X o CD X CO CM .—I t^ X r~ X t-- c OJ OJ CO -sr CO CO oi Ol CO CO tT CO Ol •-^ ^^ ^^ *— •-~ X M o c M X 05 00 CO m 05 05 OJ CO '^. ^ CD Tj' CD CD X ^ CO ^ m t:~ •T CTJ ^ t T ^ -5 + c^ c, bjj — o 5 c yj uo O OJ o OJ 05 q CO '^ CO O t^ O CM 05 OJ O o o o O o o o q a d CO d ^ d in d c-^ ,—1 d oi ^ d d 00 d 00 d d d o d d d d o ■a E- ^^ in 00 t^ t^ CD OJ o OJ Ol c^ oi u (M CO ^ in CD t^ CO 05 o , 1 O] ■-^ T LO CD t^ X 05 o , 1 CM CO ^ LO ■^ l:^ CO C- c^ t^ i> I> c^ c^ r^ 00 X 00 X X X X X X X 05 05 05 en CTJ OJ 05 (35 ^=1^ CT) m 05 05 05 05 05 05 c:5 05 CJ5 05 05 05 05 05 C75 05 C:5 05 05 05 CTJ 05 CJ5 cn > Vaughan and Prager Decline in abundance of Pagiw, pagni'i off the southeastern United States 359 Table 4 Primary and lish. is based year's modal alternate calcli-in-numbers on fishery-independent agt age is underlined. -at-age matrix for red porgy off the southeastern United States -length keys, and includes 187(. (by number) of released recrea Matrix is in thousands of tionally caught fish. Each Year Age (vi-) Total (lOOOl 1 2 3 4 5 6 7 8-f Primary catch-in-nunibo rs-at-age matrix 1972 13.2 60.7 56.2 55.9 87.3 44.1 23,5 19.0 359.9 1973 13.7 63.0 62.8 64.6 102.4 55.3 36,8 46.3 444.9 1974 13.7 64.8 63.1 58.8 76.9 35.5 21,7 26.0 360.5 1975 16.7 116.5 96.3 51.3 37.4 38.9 28.2 30.3 415.6 1976 13.8 88.3 87.5 60.4 56.4 46.1 32.5 36.7 421.7 1977 12.9 92.7 102.4 88.9 106.0 78.2 60,2 80.5 621.9 1978 15.3 101.0 106.2 88.2 115.6 69.1 46.9 58.0 600.3 1979 13.9 134.4 145.5 90.6 57.9 95.5 78.4 81.0 697.2 1980 14.6 153.9 212.2 128.6 76.6 121.6 100.2 130.9 938.6 1981 16.3 206.5 218.3 215.6 156.6 86.8 55.7 99.2 1055.1 1982 24.7 289.8 246.1 218.6 147.7 76.5 47.3 83,9 1134.8 1983 35.6 128.2 184.0 185.5 98,2 65.6 45.5 63,0 805.7 1984 31.1 137.7 202.5 186.6 85.5 50.8 28.3 37,5 760.0 1985 34.1 116.4 118.1 155.6 127.6 84.3 38.0 24.0 698.0 1986 29.0 143.4 134.9 160.6 118.6 71.3 32.5 22.3 712.7 1987 71.8 227.2 176.9 95.9 38.3 44.4 32.4 39.0 725.8 1988 87.9 288.7 219.0 109.0 42.2 46.8 30.5 38.2 862.4 1989 40.0 232.9 247.2 122.9 92.5 37.5 36.6 46.8 856.3 1990 66.2 325.1 322.5 146.1 105,2 40.2 40.3 49.9 1095,4 1991 46.7 160.4 162.9 181.1 97,5 54.9 29.6 49.9 783.0 1992 19.6 63.7 95.7 124.2 73,3 42.1 22,0 38.9 479,4 1993 3.9 38.9 53.2 95.7 94,7 51.0 24,0 23.2 384.5 1994 3.3 32.5 47.7 89.2 91.0 .50.4 24.7 23.5 362.2 1995 5.2 39.9 54.7 98.0 97.2 52.6 23,8 24.1 395.5 1996 4.9 44.9 57.9 99.9 95.6 50.6 21,5 21.1 396.3 1997 2.3 31.9 46.2 85.3 85,2 45,2 20,2 18,8 335.2 continued headboat catche.s are ,shown for comparison, but were not used in calibrating VPAs,) In the primary catch matrix, modal age (underlined in Table 4) varied between 2 and 5; age 4 has been most com- mon in recent years. Modal age in the alternate catch ma- tri.x (Table 4) was similar, but showed less annual varia- tion, probably because of interpolation across years of the age-length keys. A few strong year classes have moved through the population, most notably those that were age 2 in 1987-90. but no strong year classes have followed. The modal age has increased since 1992, the year in which a 12-inch minimum size limit was imposed (amendment 4 ISAFMC-'^]). Trends in mortality and recruitment In our calibrated VPAs, ages 4 through 7 were always con- sidered fully recruited, and estimates of F were typically averaged over those ages, weighted by population num- bers at age. Those weighted averages are referred to here as full F. In analysis of the primary catch matrix, estimates of F for all ages were lowest in the early years (Table 5). Fish- ing mortality on ages 1-3 increased between the early and middle periods and then declined, but full F (ages 4-1- ) in- creased and remained high. This pattern probably reflects the imposition of the 12-inch TL minimum size limit in 1992 with amendment 4 (SAFMC"'). Estimates of F from the alternate catch matrix in the recent period were gen- erally higher for ages 3-1- than estimates from the prima- ry catch matrix. Slightly lower estimates of F in the re- cent time period were obtained from the primary catch matrix when missing values, rather than small numbers, replaced zeroes in computing the hook-and-line index. Estimates of full F from both catch matrices started low, rose abruptly from the late 1970s to 1982, decreased some- 360 Fishery Bulletin 100(2) Table 4 (continued) Age yri Total Year 1 2 .3 4 5 6 7 8+ (1000 1 Alternate catch-in-numbers-at-age matrix 1972 13.2 60.7 56.2 55.9 87.3 44.1 23.5 19.0 359.9 1973 13.7 63.0 62.8 64.6 102.4 55.3 36.8 46.3 444.9 1974 13.7 64.8 63.1 58.8 76.9 35.5 21.7 26.0 360.5 1975 18.7 71.0 79.4 78.7 74.7 40.6 21.5 31.7 416.2 1976 15.2 57.6 69.1 79.7 85.8 49.2 26.6 39.0 422.2 1977 14.2 65.6 83.3 105.3 128.4 82.8 52.8 89.9 622.3 1978 16.6 78.5 99.0 102.4 121.2 75.2 43.9 63.7 600,5 1979 15.2 78.2 98.4 116.6 150.7 99.9 59.3 79.7 697,9 1980 15.8 86.0 129.2 171.7 213.4 116.6 70.1 136.6 939.3 1981 12.6 93.9 149.2 197.9 247.8 136.6 86.2 131.2 1055.1 1982 18.7 150.5 191.4 216.7 252.0 122.2 73.9 109.5 1134.8 1983 41.5 132.9 135.4 146.2 162.0 70.3 48.7 69.0 805.8 1984 36.4 145.9 154.0 151.0 153.5 54.5 27.9 36.8 760,1 1985 27.5 120.7 142.0 140.3 156.6 56.3 25.8 28.9 698,0 1986 25.2 120.5 159.5 169.0 131.4 52,3 23.0 31.7 712.7 1987 6.9 88.3 204.0 180.5 129.8 56.4 26.1 34.6 726.6 1988 15.4 111.7 262.3 224.5 141.3 53.8 23,8 31.3 864.0 1989 15.1 132.5 255.2 199.0 140.6 58.5 25.2 32.7 858.8 1990 32.5 216.4 330.7 247.5 158.7 57.7 23.0 32.9 1099.4 1991 22.5 165.6 238.3 180.9 104.5 34.4 14.7 22.6 783.4 1992 10.5 65.6 138.4 129.1 86.9 28.8 9.6 11.0 479.8 1993 1.9 29.8 104.6 125.7 80.5 24.6 8,4 9.1 384.5 1994 1.8 27.2 91.9 119.7 79.7 24.1 9.0 9.0 362.3 1995 3.6 37.7 108.4 128.4 81.1 21.6 6.8 7.9 395.5 1996 0.3 9.1 104.6 142.5 89.8 30.4 9,2 10.6 396.3 1997 0,1 3,3 76.9 125,8 83.9 28,4 8,1 8.8 335,2 •a I S' 0.8 0) g . " A * »■ X "X* A ' a"\ ■ ^ ^^!c:v<*»*, *A ' A A - ^ - Commercial Hook & Line A Headboat (Carotinas) — B— .All Fisheries ^S'ff'^^5\^a-^ Figure 2 Annual mean weight of red porgy off the southeastern United States in landings from commercial hook-and- line. headboats from the Carolinas, and from all fisheries. Minimum size limit was introduced in 1992. what, and fjradually increased through the present (Fig. 4Ai. Of the two matrices, the primary matrix provided lower estimates of full F in the recent period, particularly with zeroes replaced by missing values. Estimates of recruitment were not particularly sensi- tive to estimation procedure. A general decline in recruit- ment, 1973-97, was estimated from both catch matrices and with either treatment of zeroes (Fig. 4B). Sensitivity runs on the primary catch matrix, but with M = 0.20/yr or M = 0.35/yi", estimated that patterns in F and re- cruitment were similar to those with the base value of M = O.'28/yr (Fig. 5). Under the assumption M = 0.20/yr, esti- mated values of F were higher than the base assessment; whereas under the assumption M = 0.35/yr, estimated val- ues of F were lower (Fig. 5A). Effect of the value of M on recruitment increased with each year backwards in time. Regardless of the value of A/, the pattern of estimates was of initially high recruitment, followed by a long period of severe decline (Fig. 5B). For comparative purposes, VPA fits were also made by using only the hook-and-line index or only the extended chevron-trap index. Essentially no differences were noted in estimates of full F or recruitment to age 1. In our VPAs, we estimated statistical weights for the two calibration indices. Estimated weights varied only slightly Vaughan and Piager: Decline in abundance of Pagrus pagrus off the soutfieastern United States 361 MARMAP (Honk and l.inel 0.9 0.8 0.7 0.6 0.5 ■ 0.4- 0.3 0.2 0.1 • MARMP I Exended chevron) Figure 3 Catch per effort (standarciized to series maximum) from the headboat fishery in North and South Carolina (fish caught per angler day, 1972-97); and from MARMAP sampling by gear Ihook-and-line, 1979-97; and extended chevron trap. 1980-97), for red porgy off the southeastern United States. with different assumptions about M. Base analysis of the primary catch matrix estimated weights of 0.08 for hook- and-line CPE and 0.92 for extended chevron-trap CPE; analysis of the alternate catch matrix estimated weights of 0.05 and 0.95. When missing values were used in place of zeroes in the hook-and-line index, estimated weights were 0.50 for hook-and-line CPE and 0.50 for extended chevron trap CPE. demonstrating that treatment of zeroes is an important consideration. The main change in results from using missing values was decreased estimates of F in recent years (Fig. 4A). In the retrospective analysis, estimates of F generally converged in about .3 to 4 years; convergence in estimates of recruitment took a year or two longer (Fig. 6). A large positive retrospective pattern was evident in full F in the most recent year, and a corresponding negative pattern in estimates of recruitment (Fig. 6). The retrospective pat- tern was similar when using missing values for zeroes. Yield per recruit and related benchmarks Estimates of equilibrium YPR were somewhat sensitive to the value of M assumed, with larger assumed values of M corresponding to smaller estimates of YPR. The highest YPR was obtained during the middle time period (Table 6); however, high theoretical values of yield per recruit can result from fishing mortality rates that are not sustainable. The two biological reference points F,„,^ and Fg j (Bever- ton and Holt. 1957; Sissenwine and Shepherd, 1987) were estimated from the yield-per-recruit analysis. Assuming 362 Fishery Bulletin 100(2) partial recruitment based on average F at age, these refer- ence points were estimated for the recent time period from the primary and alternate catch matrices as F^^.^^ = 0.5/yr and 0.7/yr, respectively; and F,, j = 0.23/yr and 0.25/yr, re- spectively (Table 7). Spawning potential ratio, spawning-stock biomass, and stock-recruitment model Estimates of static SPR were calculated by catch matrix, time period, and assumed value of M and for four mea- sures of spawning biomass (total mature biomass, mature female biomass, mature male biomass, and egg produc- tion) (Table 6; Fig. 7). Static SPR based on mature female biomass was less affected by increases in F than static SPR based on mature male biomass because younger fish are predominantly female and older fish, male. For exam- ple, full F that produced 59% SPR in mature female bio- mass produced 49% SPR in egg production, 42% SPR in total mature biomass, and 27% SPR in mature male bio- mass (Table 6). Corresponding estimates were made of reduction in the proportion of males due to fishing in the three periods. For example, a value of 69% in the last column of Table 6 means that if the proportion of males in the mature un- fished population had been 50'/( , introduction of fishing mortality would reduce this proportion to 34.5% of the population (69% x 50%). These estimates were made on the basis of observed sex ratios at size and assuming that the rate of transformation from female to male does not increase as the population is fished down. Total spawning stock biomass (SSB) was estimated from the primary catch matrix to have reached its peak in about 1979 at 3530 t, dechning to 397 t in 1997 (Fig. 8A). When missing values were substituted for zeroes in computing the CPE index, estimates of SSB near the end of the se- ries were slightly higher (Fig. 8A). Estimates from the al- ternate matrix were similar, with the largest differences occurring before 1983 (Fig. 8A). Static SPR estimated from the primary catch matrix was high during the 1970s, reached 67% in 1975, and de- clined to a minimum of 33% in 1982, increased again to 55% in 1993, and declined to about 18% in 1997 (Fig. 8B). Estimates from the alternate catch matrix were similar, but markedly lower from 1983 to 1994 (Fig. 8B). Values of full F providing spe cified values of static SPR (based on female egg production and total spawning stock biomass) are summarized in Table 7, along with values of full F av- eraged by period, for comparative purposes. The recent management definition of overfishing is stat- ic SPR <35% (amendment 12 ISAFMC'*]). Our primary es- timates are of static SPR >35'7f from 1972 to 1994 (except for 1982), lower thereafter Estimates from the alternate catch matrix show static SPR >35% from 1972 to 1980, lower thereafter During the early to mid 1970s, a large spawning stock produced high recruitment. Despite a high spawning- stock size estimated through about 1983, SPR and re- cruitment declined, and the unsustainable landings of the late 1970s and 1980s reduced the spawning stock to levels Table S Mean estimates (from calibrated virtual population anal- yses 1 of age -specific instantaneous fishing mortality rate (F) on red porgy off southeastern U S. durin % three time periods. Estimates from primary catch matrix (see text) | are given for three levels of n atural mortality rate M, two treatments jf missing values for M= 0.28. Est imates fi'oni alternate catch matrix given for M = 0.28 on ly. Exploita- tion rate in final column is based on catch of ages 1-8 1 divided by estimated population number of same ages. Age (yr) ixploitation rate Analysis 1 2 3 4+ (ages 1-8) Primary (MARMAP) catch matrix (M=0.28) 1972-78 0.007 0.06 0.08 0.15 0.06 1982-86 0.018 0.12 0.19 0.35 0.12 1992-96 0.014 0.10 0.15 0.44 0.16 lM=0.28, zeroes in index treated as mi.ssing) 1972-78 0.007 0.06 0.08 0.15 0.06 1982-86 0.018 0.12 0.19 0.35 0.12 1992-96 0.011 0.08 0.13 0.40 0.14 (A/=0.20i 1972-78 0.010 0.08 0.11 0.20 0.08 1982-86 0.02.5 0.16 0,24 0.43 0.16 1992-96 0.017 0.12 0.17 0.49 0.19 (M=0.3.5) 1972-78 0.00.5 0.04 0.06 O.U 0.04 1982-86 0.01.3 0.09 0.14 0.28 0.09 1992-96 0.012 0.09 0.14 0.40 0.14 Alternate catch matrix iM=0.28) 1972-78 0.006 0.04 0.06 0.14 0.05 1982-86 0.017 0.10 0.18 0.42 0.13 1992-96 0.007 0.06 0.27 0.66 0.18 not expected (under the stock-recruitment model) to pro- vide good recruitment. Similar patterns were apparent in stock-recruitment cui-ves derived from both catch matri- ces (Fig. 9). Estimates of benchmarks from simulations Estimates of MSY, Bmsy s^^ F^igy were made from the stock-recruitment model and age-structured simulations (Table 8; Fig. 10). In addition, minimum stock size thresh- old (MSST), a benchmark recently introduced into U.S. Federal fishery management, was calculated from B-^^^y as suggested in Restrepo et al. (1998). namely as MSST = (1-M) S[^,t;Y' with M = 0.28/yr Although benchmarks are somewhat sensitive to assumptions examined, sensitivity was small compared with the difference between estimated benchmarks and current status of the stock and fishery (Table 8). For example, estimates of MSY obtained in this way (240-280 t/yr) were similar to recent landings (about VaLighan and Prager: Decline in abundance of Pngrtjs pagnis off the soutfieastern United States 363 2r Primary -ft— Primary (Miss) -**- Alternate 4000 3000 2000 Primary -ft— PrifiBry (Miss) -e— Alternate B Figure 4 (A) Annual estimated instantaneous fishing mortality rates (mean F on ages 4-8), and (B) recruitment to age 1 from calibrated VPA applied to red porgy off southeastern United States based on differ- ent catch matrices and treatment of zeroes (alternative treatment of zeroes as missing value is indicated as "Miss" after Primary! in MARMAP hook-and-line CPUE. 200-300 t/yr), but well below landings taken earlier (e.g. >800 t in 1981 an(i 1982). Estimates of fishing mortality rates in the recent period are about 2.7 to 3.6 times those that could produce MSY (Table 8). Because benchmark estimates are based on selectivity patterns estimated for 1992-96, their values are expected to change with the recent introduction of higher minimum size limits. How- ever, the general picture is unlikely to change. Surplus-production model It was not possible to estimate the shape parameter of the generalized production model from these data; therefore, we present results only for the logistic model. Fits of that model with various assumptions about B,//^ were statisti- cally equivalent (log likelihoods from -2.29 to -2.2.5 ), indi- cating that the data-model combination is not informative about relative biomass level at the start of the time series. We present results under a range of assumptions centered on B^/K = 0.75 (Table 9). The estimate of MSY was moder- ately sensitive to the assumption on B/A', varying about 18.5'^ around the central estimate (Table 9); it was also higher than our estimates from age-structured methods I Table 8). Nonetheless, the bias-corrected 80*^/ confidence interval on the central MSY estimate (Table 9 1 and the corresponding 50% interval (not shown) included the age- structured estimates of MSY; therefore there was some agreement between methods. Estimates of stock and fishery status expressed in rela- tive terms (Table 9) were similar to those from the age- structured analysis (Table 8) and like those estimates de- picted a heavily overexploited stock. Stock biomass at the start of 1998 was estimated as about 25% of the biomass that can support MSY (Bmhy'- ^ result that was remark- ably insensitive to the assumption on B^/K. Fishing mor- tality rate in 1997 was estimated as about twice the fish- ing mortality rate associated with MSY (Fj^jsy'- a result only slightly sensitive to the assumption on B^/K. Equilib- rium yield available in 1998 was estimated as about 44% of MSY; this reduction was due to the stock's reduction well below B^^^y- Estimates of relative stock and fishery status over time displayed a pattern of increasing fishing mortality rate and decreasing abundance over time (Fig. 11). It appears that the biomass moved below B^sy '" about 1982 and that the fishing mortality rate has been above F^j^y since about 1978. Estimated confidence intervals on F/F^gy are remarkable in encompassing a wide range that since 364 Fishery Bulletin 100(2) <2 2000 Figure 5 (Ai Annual estimated instantaneous (ishing mortality rates (mean F on ages 4-8), and (B> recruitment to age 1 from calibrated VPA applied to red porgy off the southeastern United States based on primary catch matrix varying instantaneous natural mortality rate (M). about 1980 has included only values greater than unity (Fig. IIB). Discussion This work includes numerous advances from previous assessments (Vaughan et al., 1992; Huntsman et al.^). The major improvements are availability of additional fish- ery-dependent and fishery-independent aging data and of fishery-independent indices for calibrating VPAs. Other changes include expanded geographic range, new esti- mates of sex ratios and maturity schedules, new growth parameters, application of a more modern algorithm for calibrated VPA, and application of a nonequilibrium sur- plus-production model. Our conclusions agree with both earlier assessments in finding a population in decline. Because unbiased sampling of wild fish stocks is ex- tremely difficult and fishery data sets are small (in num- ber of years of data), fish population models are not pre- cise approximations of reality. Accurately characterizing uncertainty in assessment results is difficult or impossi- ble: although statistical estimates of uncertainty can be obtained, it is not known how to balance them against fac- tors that tend to increase certainty, such as agreement among data sources or methods and among assessments conducted over time. Anv formal estimate of uncertainty hi the present assessment results would be large. None- theless, in light of our use of complementary models, sev- eral independent data sources, and two largely indepen- dent sets of age-length keys, and based on agreement with past assessments, we believe that the present results are unequivocal in their major finding of a severely depleted population. By using two treatments of observed zeroes in CPE se- ries, we illustrated that some results can be sensitive to the method used to treat zeroes. Although addition of a constant before logarithmic transformation is a widely used practice and has been a common statistical recom- mendation (e.g. Snedecor and Cochran, 1980; Berry, 1987), it is becoming evident that the practice can be problematic in estimation of abundance indices. Logarithmic transfor- mation is made under the assumption of lognormally dis- tributed data, and presence of zeroes in itself violates that assumption. Moreover, results can depend strongly on the additive constant chosen (Porch and Scott, 1994; Ortiz et al., 2000). The root of the problem is that transformed ze- roes are usually extreme values in the data set and thus highly influential. Several other statistical models have been proposed, of which the most promising in this appli- cation may be the delta-lognormal model (Lo et al., 1992; Ortiz et al., 2000). However, this is still an active area of methodological research, and the matter remains unre- solved. We thus used two treatments of zeroes and have Vaughan and Prager: Decline in abundance of Pagrus pagnis off the southeastern United States 365 Table 6 Static yield per recruit (\TR) and spawning potential ratio (SPR) of red porgy off southeastern United States. Estimates are based on mean age-specific fishing mortahty rates from calibrated VPA on primary or alternative catch matrices (see text), are made under several assumptions about natural mortality rate, M. and treatment of missing values in hook-and-line abundance index, and use selectivity for most recent time period, ( 1992-961. Spawning potential ratio Analysis YPR(gl Total Female Eggs Male Percentage of males in relation to no fishing' Primary (MARMAP) catch matrix (A/=0.28) 1972-78 1982-86 1992-96 (A/=0.28, zeroes in index treated as missing) 1972-78 1982-86 1992-96 (M=0.20) 1972-78 1982-86 1992-96 (A/=0..35) 1972-78 1982-86 1992-96 Alternate catch matrix (M=0.28) 1972-78 1982-86 1992-96 152.3 61 74 67 49 83 214.3 40 58 47 26 71 162.2 42 59 49 27 69 152.3 61 74 67 49 83 214.2 40 58 47 26 71 160.9 44 61 51 29 70 263.5 43 60 49 33 74 306.6 25 43 31 15 61 229.5 29 47 37 17 60 89.8 74 84 78 63 89 149.5 54 69 60 39 78 118.7 52 67 58 36 75 145.1 64 77 70 53 84 261.4 33 53 37 20 69 198.2 27 46 33 14 64 Percent relative reduction in numbers of mature males between fished and unfished conditions; i.e. proportion of males under fished conditions is .v'7 of proportion of males under unfished conditions. presented both sets of results, which are different, but not remarkably so. Several issues arise from choosing as "primary" the catch matrix developed from fishery-independent age data and as "alternate" the matrix from fishery-dependent da- ta. The choice was by necessity somewhat subjective, but was based on the long-term ( 1979-94). continuous nature of the fishery-independent data on age at size and its larg- er sample sizes, in contrast to the intermittent age sam- pling in the fishery, which required frequent interpolation of age-length keys to construct the alternate catch matrix. The different selectivity of the fishery-independent gear should not significantly bias the primary catch matrix, but fishery-independent samples at larger sizes were some- times very small, making it necessary to pool data across longer periods for the largest sizes. Parallel analyses were conducted with the two catch matrices to determine sensi- tivity of major results to the choice of matrices as primary and alternate. Size selectivity of the gear can have much greater effect in fitting growth models. Fishery-dependent sampling, by se- lecting larger fish, may overestimate mean size at younger ages, but estimate L^ more accurately, whereas fishery- independent sampling, by selecting smaller fish, may un- derestimate L^. In a simulation study, Goodyear (1995) noted: ". . . samples drawn from size-selective gears or fish- eries yield biased estimates of mean length at age. . . . Even slight changes in sampling protocol can result in mis- leading temporal shifts of estimates of size at age." This as- pect of the differing selectivity between gears seems impor- tant to us, and it can bias estimation of growth models. As a result, we cannot say whether estimated temporal differences in size at older ages result from changes in growth patterns in response to long-term overexploitation (Harris and McGov- em, 1997) or from a change in sampling-gear selecti\'ity in the fishery-independent sampling gear (Potts et al., 1998). Despite questions about size at age over time, patterns of population benchmarks and stock status estimated from 366 Fishery Bulletin 100(2) Table 7 Potential biological reference points developed from static jield per recruit i\TR) and spawning potential ratio (SPR) analyses for red porgy off southeastern United States estimated from output from calibrated VPA using selectivity for most recent time period | ( 1992-96), and based on primary catch matrix (three levels of M, and run with missing val je for 0.0 in parentheses) or alternate catch matrix (A/=0.28i. See text for derivation of catch matrices. Primary catch matrix Alternate catch matrix M=0.20 M=0.28 A/=0.28 (missing M=0.35 M=0,28 Yield per recruit (\TR) analysis f'oi 0.15 0.23 (0.23) 0.31 0.25 F max 0.25 0.53 (0.56) 0.84 0.72 Static spawning potential ratio analysis static SPR) Female (eggsi F30 0.25 0.49 (0.50) 0.79 0.77 ^35 0.20 0.37 (0.381 0.60 0.55 F,o 0.17 0.29 (0.30) 0.46 0.41 Total spawning-stock biomass ^30 0.25 0.42 (0.43) 0.69 0.51 F,, 0.20 0.33 (0.34) 0.52 0.38 ^4,, 0.17 0.26 (0.27) 0.40 0.30 Observed full F by period 1972-78 0.20 0.15 (0.15) 0.11 0.14 1982-86 0.43 0.35 (0.35) 0.28 0.42 1992-96 0.49 0.44 (0.40) 0.40 0.66 Table 8 Parameters and potential biological reference points developed for Beverton-Holt spawner-recruit model of red porgy off southeast- ern United States. Estimates are based on calibrated VPA by using selectivity for most recent time period (1992-96). and based on primary catch matrix (fishery-independent keys; includes separate run with missing value for zeroes in hook-and-line index) or alternate catch matrix (fishery-dependent keys). Parameter estimate or reference point Parameter estimates (standard error in parentheses) maxlfl,! = l/6u Reference points MSY '' Msy MSST Estimates of SSB (spawning stock biomass) by period 1972-78 1982-86 1992-96 Estimates of stock status i/n>sv '^92-96 "sL- 9i;'"MSY Primary Primary Alternate catch matrix CM (missing) catch matrix 1.83 X 10-^ 2.01 X 10-* 2.32 X 10-1 (7.0 X 10-5) (6.7 X 10-5) (5.3 X 10-5) 0.669 0.617 0.605 (0.185) (0.175) (0.157) 5.46 X 1015 4.98 X lOfi 4.31 X lO** 242 248 277 0.14 0.15 0.18 2564 2425 2387 1846 1746 1719 2884 2885 3367 2326 2330 2651 876 953 669 3.2 2.7 3.6 0.34 0.39 0.28 Vaughan and Prager- Decline in abundance of Pagrus pagrus off thie soutfieastem United States 367 2 t 1.6 A 2 1.2 o E g> 0.8 - jjjjjj ^ 0.4 ^ — „^^^^^^i^^^^^^^ 0 o ~- — ^ 1972 - 1974 - 976 - 978 - 980 ■^ 982 • 984 986 - 988 990 - 992 • 994 - 996 998 B — 4000- Recruits to age = ill "'^^V. r 1974 - 976 978 - 980 - 982 984 986 988 ■ 990 992 994 996 998 Figure 6 Retrospec tive patterns in (Ai annual estimated instantaneous fish- ing morta lity rates (mean F on ages 4-8). and (B) recruitment to age 1 from calibrated VPA applied to red porgy off southeastern | United S tates based on primary catch matrix with earlier final year compared to analysis with the most recent final year ( 1997). Table 9 Estimates of biological benchmarks and stock status from application of logistic surplus- production model to red porgy off the southeastern United States Symbols are B^ = biomass in first year K = canying capacit. ,■■ MSY = : maximum sustainable vield; B<|,/Bj[j^Y= biomass at start of 1998 (year following analysis 1 relative to biomass at MSY; F^ 7/^MSV = fishing mortalit\ • rate in 1997 compared with F at MSY; Y ?g^ = equilibrium yield available in 1998 . Because B, could not be estimated with anv precision, the model was fitted under a range of assumptions on its value in relation to that of A'. For cen tral case , 80*^ confidence inten'als are | shown in parentheses. Benchmark or Assumed value of S,/A' status indicator 0.55 0.65 0.75 0.85 0.95 MSY(t/>T) 506 482 459 (98-555) 437 416 ^gs/^MSY 0.247 0.247 0.248 (0.116-0.373) 0.250 0.252 ^97/^MSY 1.85 1,93 2.01 (1.36-5.08) 2.09 2.17 Keggd) 219 209 199 (69.0-304) 191 183 Ye^JMSY 0.434 0.433 0.434 (0.219-0.6071 0.437 0.440 368 Fisher/ Bulletin 100(2) Q. w 1970 1980 1985 1990 I Total -♦-Female -s-Eggs -Male 1(10 50 B 1970 1975 1980 1985 Year 1990 1995 Total - Female - Eggs - Male Figure 7 Static spawning potential ratio for total biomass, female biomass, egg production, and male biomass for red porgy off the southeast- ern United States from calibrated VPA based on (A) primary catch matrix, and (B) alternate catch matrix. the two matrices (and. indeed, in all sensitivity runs) were similar, depicting much higher population size and spawn- ing biomass during the 1970s and severe decline through the 1980s and 1990s. This pattern was independent of the large retrospective pattern in the terminal year. Declining trends in CPE from headboats and in fishery-independent sampling (Fig. 3), combined with the observed decline in recruitment (Fig. 4B), also depict an increasingly deplet- ed population. Additional evidence is provided by the sur- plus-production model, which estimated that the popula- tion was at high levels during the 1970s, but that since the mid 1980s it has been far below the level capable of pro- ducing MSY (Fig. 11). Implementation of a 12-inch mini- mum size limit in 1992 (amendment 4 ISAFMC''! ) appears to have reduced F only on ages 1-3, but F on fully recruit- ed ages (4-I-) has increased (Table 5). Static SPR (in total mature biomass) is estimated for the recent period at 42*^?^ and 27'? from the primary and alternate catch matrices, respectively (Table 6). Mean SPR estimated from the primary catch matrix for all three pe- riods is above SPR=35'7f . the value used by the South At- lantic Fishery Management Council to define overfishing for red porgy (amendment 12 |SAFMC-'| i. Estimated SPR from the alternate catch matrix is below SS"? in the recent period (Table 6). Coleman et al. (1999), in a discussion of reef fish man- agement, pointed out that SPR as usually estimated from female biomass or egg production has proven less effec- tive for protogynous reef fishes than for fishes that do not change sex. With the loss of the largest size classes, predominantly male, remaining males may not be able to increase harem size sufficiently to fertilize all available females. Thus, determining SPR solely on female repro- ductive contribution may not capture the true decline in reproduction caused by fishing. For that reason, Vaughan et al. (1992, 1995) recommended calculating static SPR from total mature biomass, broadening the concept of stat- ic SPR for protogynous species. Vaughan et al. (1995) cal- culated reduction in the proportion of males for black sea bass due to changes in fishing mortality and under the as- sumption that transition rate does not change with pop- ulation abundance. They estimated a reduction of about 40'/f due to recent fishing mortality, a reduction similar to the ~2>Q% we estimated for red porgy. Whether males are currently limiting, or the degree to which increased fishing mortality can cause them to be- come limiting, is unknown. Regardless, increased fishing mortality on all ages would be expected to reduce the pro- portion of males in the mature population. The rate at which the imposition of fishing mortality may alter the av- Vaughan and Prager Decline in abundance of Pagnis pagnis off the southeastern United States 369 o 3000 ■ S- 1000 Primary matrix -«- Primary (Miss) -a- Alternate matrix tr 9r so Primary matrix -*— Primary (Miss) -e- Alternate matrix B Figure 8 Annual estimates of (A) total spawning stock biomass, and IB) static spawning potential ratio (SPR) for red porgy off the south- eastern United States from calibrated VPA based on different catch matrices and treatment of zeroes (alternative treatment of zeroes as missing value is indicated as "Miss" after Primary) in MARMAP hook-and-line CPUE. erage age of transition to male is unknown, as is the effect of population density on transformation rate. Increased sex transformation due to reduced abundance of males, as reported in other protogynous reef fish (Sha- piro. 1979), would cause declines in mature female bio- mass and egg production beyond those expected from fish- ing alone. McGovern^ found a greater percentage of males in smaller size classes in 1991-99 than in 1979-81 and suggested that red porgy have compensated for the loss of large individuals by becoming male at smaller sizes. If so, a smaller change in sex ratio, but a greater reduction in egg production, would be expected than implied by in- creasing fishing mortality alone. New evidence suggests that red porgy are indetermi- nate, serial spawners (Daniel"). That is, the size and num- ber of batches of eggs are not fixed at the start of the spawning season, but older, larger females spawn in larger ^ McGovem, J. C. 2001. Personal commun. South Carolina Department of Natural Resources, P.O. Box 12559. Charleston, SC 29422. ^ Daniel, E. A. 2001. Personal commun. University of Charles- ton, Charleston, SC 29424. batches, more frequently, and possibly over a longer sea- son than younger, smaller females. If so. SPR based on egg production estimated from a fixed weight-fecundity rela- tionship would be unrealistic, and SPR based on total ma- ture biomass would underestimate the contribution of old- er fish, and thus underestimate the effects of fishing. Huntsman and Schaaf (1994) proposed accounting for sex-composition changes by multiplying total egg produc- tion by a factor representing the probability of successful fertilization. They calculated that probability (Q) for a giv- en fishing mortality rate F' from the relative change in male contribution of spawning products under fishing: ^jG^IGf)\F- ^GJGf)\F = 0 where G„, and G,= the gamete contributions of males and females, respectively. Huntsman and Schaaf (1994) computed G,„ as total male biomass and G. as total female egg production. When new information on red porgy batch spawning becomes avail- 370 Fishery Bulletin 100(2) 3000 2000 ° 1000 0 1000 2000 3000 Total Spawning Stock Biomass (nit) 4000 ■ B |» 1970s ■1980s ♦ 1990s •^ — t-* — 1 — 1 — 1 — CL 4000 -\ 3000 2000 1000 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 Total spawning stocl< biomass (t) Figure 9 Total spawning-stock biomass and rt'cruitniont. to age 1 for red porgy off the southeastern United States from calibrated VPA based on lA) primary catch matrix, and (B) alternate catch matrix. Solid lines represent recruitment predicted from Beverton-Holt models; dashed lines run from the origin with slope of median of recruits divided by the spawning-stock bio- mass for 1972-94 spawning years. able and allows calculation of total population fecundity at Z and M, we recommend exploring that approach. Popu- lation simulations for a protogynous grouper (Huntsman and Schaaf 19941 suggest greater vulnerability to fishing than for a comparable gonochoristic stock. It appears that the questions of what SPR measure is most appropriate and how it should be computed cannot be answered until the overall reproductive biology of the species is far better understood. As usual for fish populations, the spawner-recruit rela- tionship is only roughly defined by the data (Fig. 9), and the years in which the model does not fit raise interesting questions. In the fit based on the primary catch matrix (Fig. 9A), the two large residuals at the start might be taken as suggestive of a dome-shaped recruitment curve, they simply might reflect random error, or they might re- sult from the additional computations needed to extend the catch matrix back to 1972. The analysis based on the alternate catch matrix (Fig. 9B) shows a more pronounced pattern, in which data from the 1970s demonstrate higher recruitment for a given level of spawning-stock size than later data. This pattern might be taken as evidence of a regime shift (change in underlying ecological conditions, from natural or anthropogenic causes), or it might be an artifact of the smaller data sets of age-length data used in this approach. Other explanations are equally possible for residual patterns in Figure 9, A and B. These obsei-ved da- ta suggest "depensation"( reverse compensation;decreased, rather than increased, recruitment per spawner at lower population sizes) at very low spawning biomasses, but this suggestion is uncertain because of high retrospective error in estimating recruitment in the most recent years. Growth overfishing was not estimated to be severe. Mean full F in the early period was estimated well below both F,,,,,. and F,, ,. Mean full F in the middle period was estimated as about \'a2'7( of F,, j, and mean full F for the recent period was estimated as about VdVc of F^ j, both below F^,,,^ (Table 7). This stock appears to be one in which recruitment overfishing can be induced before growth overfishing, and for that reason reference points such as Fj,^ j^ and even F,, , would be insufficiently conser- vative for its management. The estimate of B^yg^ computed from the recruitment model and selectivity vector can be an appropriate target biomass for management and the corresponding MSST can be a threshold to define overfished status (Restrepo et al., 1998). Estimates of spawning-stock biomass were well above MSST, 1972-78; approached MSST. 1982-86; and were well Vaughan and Prager: Decline in abundance of Pagivs pagrus off the soutfieastern United States 371 Full(F) ■ Total SSB -Yield Figure 10 Sustainable yield and spawning stock biomass at varying levels of fishing, projected for red porgy off the southeastern United States from selectivity pattern estimated for 1992-96. Selectivity pattern estimated through calibrated VPA of (A) primary catch matrix, iBi alternate catch matrix. below MSST. 1992-96 (Table 8). Still lower estimates of spawning-stock biomass were obtained for the terminal year, although we view terminal-year estimates with scepticism because of retrospective patterns. The production model, not commonly subject to such retrospective patterns, also esti- mates current biomass as well below MSST (Table 9). Current estimates of full F (Table 5 for 1992-96) are considerably higher than estimates of F^c^y (Tables 8 and 9). In addition, estimates of F^jgy are considerably lower than F values corresponding to static SPR of SO'/; to 40^* . If static SPR is used as a reference point, a value higher than SPR = 40'> will be necessary to maintain the stock at reasonably productive levels. It seems inescapable that heavy fishing mortality has been a major cause of the decline in recruitment (Fig. 4B), particularly in the late 1970s and 1980s. Although fishing mortality was lower in the mid-1980s, it increased again in the 1990s for fully-recruited ages (Fig. 4A), and thus static SPR has also declined (Fig. 7). Patterns of increas- ing F and declining SPR are even more pronounced in es- timates from the alternate catch matrix (Figs. 4A, 7). Even though spawning-stock biomass remained high in the late 1970s, recruitment began to decline (Fig. 9). This finding suggests that the usual effects of heavy fishing pressure may have been compounded by other population-specific factors (e.g. those relating to sex structure) or extrinsic factors (e.g. unfavorable environmental conditions! affect- ing the reproductive success of the population. Management of hermaphroditic reef fishes is difficult, and it has had few successes (Coleman et al., 1999). As we have shown, interaction of a population's size and sex structures complicates the attempt to model reproductive capacity (e.g. SPR) as a function of fishing mortality rate. Size limits can have subtle side effects if they result in a deficit of males and if larger females in the population consequently change sex at a higher rate. Use of bag lim- its and size limits in management makes it likely that fish will be caught and released, imposing discard mortal- ity that is incompletely observable. Discard mortality has been found an important factor in collapse of at least one fish stock (Myers et al., 1997). Fish that aggregate become more catchable (increased catchability coefficient q = F/f) as abundance declines, another phenomenon associated with risk of collapse (Clark, 1974; Gulland, 1975). Ludwig ( 1998) has suggested on theoretical grounds that the best management strategy for stocks that may collapse is rapid adjustment of harvest size in response to changes in stock abundance. Ludwig noted: "It is noteworthy that the com- 372 Fishery Bulletin 100(2) A 1.5 - •>^v. V v.. V, • - ■ -. V •. >. •■•■.\ U) ••V m 1.0 - \. (S '\-: ■••^s:-. ■■■■♦'.;•. -.N. ■•■•s 0.5 - •'•■■, ■■. \ ■••• ■■"x ■•.•-•-. I 1 1 1 r 1975 1980 1985 1990 1995 B 10.0 - 5.0 - >, E /^\ LL .<•> <•'■■■■ '■'^-^'■■' / y/ 10 - y / ,•■' • A . / 0 5 " • . •. •-•-•.■ 1 1 1 1 1 1975 1980 1985 1990 1995 Year Figure 11 Time trajectories of (Al relative population biomass (B/iSn,^Y* '"^d (Bl relative fishing mortality rate IF/F^^y; log scale) estimated from non- equilibrium logistic surplus production model of red porgy off the south- eastern United States under assumption B,/A' = 0.75. Dotted lines are 80'!'r confidence intervals from bootstrapping. plications in stock dynamics all have the same effect up- on performance of |management| strategies: they all may cause drastic increases in the probability of extinction and decreases in the expected discounted total yield." It was in the face of such difficulties and uncertainties that the con- cept of precautionary management was developed (FAO, 1995). In conclusion, as of 1997 the stock of red porgy off the southeastern United States appears to be in poor condition and getting worse. Even after allowing for retrospective pat- terns and regardless of catch matrix, estimates are of long- term declines in spawning-stock biomass, recruitment, and catch per effort in the headboat fishery and in fishery-inde- pendent sui-veys. Such estimates strongly suggest overex- ploitation — overexploitation at an unsustainable rate. Stat- ic SPR since 1985 has generally been above the present criterion for overfishing (SPR > 35%), based on estimates from our primary catch matrix. (Estimates from the alter- nate catch matrix are lower, and estimates from that ma- trix are generally even more pessimistic about stock sta- tus. ) During that period, recruitment and spawning stock have continued to decline. Given the present low popula- tion level and poor recruitment, we believe that substantial reductions in fishing mortality rate will be necessary to re- Vaughan and Prager; Decline in abundance of Pagrus pagrus off tfie soutfieastern United States 373 build the stock, whether or not factors other than fishing have contributed to the dcchne. The situation may be analo- gous to that oC striped bass (Morone saxatilis) ofT the U.S. Atlantic coast in the 1980s, a population that was extremely depressed, with landings reduced by about 75'^ from their peak. Models of that decline suggested that sharp reduc- tions in fishing mortality would increase the population's growth rate, regardless of the relative importance of fishing and environmental conditions in the preceding decline. Un- der vigorous management (with sharp reductions in fishing mortality I the population has made a remarkable recovery (Richards and Rago, 1999). Since completion of this assessment, the South Atlantic Fishery Management Council has taken several actions to promote recovery. As noted in amendment 12 (SAFMCM, amendment 9 to the FMP increased the red porgy minimum size limit to 14-inch total length, established a bag limit of 5 fish per person per day, prohibited purchase or sale of red porgy during March and April, and restricted commercial hai-vest to the recreational bag limit during those months. Subsequently, a prohibition on landing red porgy (moratori- um) was enacted for the period 8 September 1999 through 28 August 2000. Amendment 12 (SAFMC^) to the FMP, which became effective on 29 August 2000, reduces the recreation- al bag limit to 1 fish per person per day, extends the closed season to 4 months (January-April), and allows a 50-pound trip limit for commercial landings during the remaining 8 months of each year. Furthermore, the maximum fishing mortality threshold (MFMT) was increased from SPR = 30% (amendment 4 [SAFMC""'] ) to SPR = .35% (amendment 12 (SAFMC'^I ). It remains to be seen whether these actions will be sufficient to achieve the recovery of red porgy. Acknowledgments We thank the following for their efforts in obtaining and processing the data upon which this study was based: Robert Dixon (headboat data); Laura Bishop, Guy Daven- port, and Linda Hardy (commercial data); Jack McGovern and Boxian Zhao (MARMAP data). New aging data from fishery-dependent collections were provided by Charles Manooch and Jennifer Potts. The authors also acknowl- edge the thoughtful reviews of Felicia Coleman, Charles Manooch, Jennifer Potts, and two anonymous reviewers. Literature cited Berry, D. A. 1987. Logarithmic transformations in ANOVA. Biometrics 43:439-456. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Re- printed 1993 by Chapman and Hall. New York, NY, 533 p. Caddy, J. F, and R. Mahon. 1995. Reference points for fisheries management. FAO Fish. Tech. Pap. 347. FAO, Rome, 83 p. Clark, C. W. 1974. Possible effects of schooling on the dynamics of exploited fish populations. J. Cons. Int. Explor. Mer 36:7-14. Clarke, W. G. 1991. Groundfish exploitation rates based on life history parameters. Can. J. Fish. Aquat. Sci. 48:734-750. Clay, D. 1990. TUNE: a series of fish stock a.ssessment computer programs written in FORTRAN for microcomputers (MS DOS). Int. Comm. Conscn'. Atl. Tunas, Coll. Vol. Sci. Pap. 32:44.3-460. Coleman, F. C, C. C. Koenig, A.-M. Eklund, and C. B. Grimes 1999. Management and conservation of temperate reef fishes in the grouper-snapper complex of the southeastern United States. Am. Fish. Soc. Symp. 23:233-242. Collins, M.R. 1990. A comparison of three fish trap designs. Fish. Res. 9:325-332. Collins, M. R., and G. R. Sedberry 1991. Status of vermilion snapper and red porgy stocks off South Carolina. Trans. Am. Fish. Soc. 120:116-120. Doubleday, W. G. 1976. A least squares approach to analyzing catch at age data. Res. Bull, Int. Comm. Northwest Atl. Fish. 12:69-81. Efron, B., and R. Tibshirani. 1986. Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy. Stat. Sci. 1:54-75. FAO (Food and Agriculture Organization of the United Nations). 1995. Precautionary approach to fisheries. Part 1: Guidehnes on the precautionaiy approach to capture fisheries and species introductions. FAO Fish. Tech. Rep. 350. FAO. Rome, 52 p. Fletcher, R. I. 1978. On the restructuring of the Pella-Tomlinson system. Fish. Bull. 76:515-521. Gabriel, W. L., M. P. Sissenwine, and W. J. Overholtz. 1989. Analysis of spawning stock biomass per recruit: an example for Georges Bank haddock. N. Am. J. Fish. Manag. 9:383-391. Gavaris, S. 1988. An adaptive framework for the estimation of popula- tion size. Canadian Atl. Fish. Sci. Adv. Comm. Res. Doc. 88/29, St. Andrews, New Brunswick, Canada, 12 p. Goodyear, C. R 1995. Mean size at age: an evaluation of sampling strate- gies with simulated red gi-ouper data. Trans. Am. Fish. Soc. 124:746-755. Gray, G. W.. L. L. Kline, M. F Osburn, R. J. Salz, D. A. Van Vorhees, and J. F. Witzig. 1994. MRFSS user's manual: a guide to the National Marine Fisheries Service marine recreational fisheries statistics sui-vey database. Special report 37, Atlantic States Marine Fisheries Commission. Washington. DC. Gulland, J. A. 1975. Thestabihty offish stocks. J. Cons. Int. E.\plor. Mer 37:199-204. Gulland, J. A., and L. K. Boerema. 1973. Scientific advice on catch levels. Fish. Bull. 71:325- 335. Harris, P. J., and J. C. McGovern. 1997. Changes in the life history of red porgy. Pagrus pagrus, from the southeastern United States, 1972-1994. Fish. Bull. 95:732-747. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mortality rates. Fish. Bull. 82:898-903. Huntsman, G. R. 1976. Offshore headboat fishing in North Carolina and South Carolina. Mar. Fish. Rev 38(3):13-23. 374 Fishery Bulletin 100(2) Huntsman. G R., D. R. Colby, and R. L. Dixon. 1978. Measuring catches in the Carolina hcadboat fishery. Trans. Am. Fish. Soc. 107:241-245. Huntsman, G. R., and W. E. Schaaf. 1994. Simulation of the impact of fishing on reproduction of a protogynous grouper, the gi'aysby. N. Am. J. Fish. Manag. 14;41-.52. 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. 49:2525-2526. Ludwig, D. 1998. Management of stocks that may collapse. Oikos 83: 397-402. Manooch, C. 8., III. 1976. Reproductive cycle, fecundity, and sex ratios of the red porgy, Pagrus pagrus (Pisces: Sparidae) in North Carolina. Fish. Bull. 74:775-781. Manooch, C. S., Ill, L. E. Abbas, and J. L. Ross. 1981. A biological and economic analysis of the North Caro- lina charter boat fishery. Mar. Fish. Rev. 43(8):1-11. Manooch, C. S., Ill, and W. W. Hassler 1978. Synopsis of biological data on the red porgy, Pagrus pagrtis (Linnaeus). U.S. Dep. Commer, NOAA Tech. Rep. NMFSCirc. 412, 19 p. Manooch, C. S., Ill, and G. R. Huntsman. 1977. Age, growth, and mortality of the red porgy, Pagrus pagriiti. Trans. Am. Fish. Soc. 106:26-33. Manooch. C. S., Ill, J. C. Potts, D. S. Vaughan, and M. L. Burton. 1998. Population assessment of the red snapper from the southeastern United States. Fish. Res. 38:19-32. Marine Recreational Fisheries Statistics. 1999. Marine recreational fisheries statistics survey data. NMFS/Office of Science & Technology/Fisheries Statistics & Economics Division, September 10, 1999. [Available from http://www.st.nmfs.gov/stl/recreational/database/index. html. I 14 August 2000. Myers, R. A., Hutchings, J. A., and N. J. Barrowman. 1997. Why do fish stocks collapse? The example of cod in Atlantic Canada. Ecol. Appl. 7:91-106. NRC (National Research Council). 1998. Improving fish stock assessments. National Acad- emy Press, Washington, D.C., 177 p. 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. Pauly, D. 1979. On the inter-relationships between natural mortality, growth parameters and mean environmental temperature in 175 fish stocks. J. Cons. Int. Explor Mer 39:175-192. Pella, J. J. 1967. A study of methods to estimate the Schaefer model parameters with special reference to the yellowfin tuna fishery in the eastern tropical Pacific ocean. Ph.D. diss., Univ Washington, Seattle, WA, 156 p. Pella, J. J., and P K. Tomlinson. 1969. A generalized stock production model. Bull. Inter-Am. Trop.Tuna Comm. 13:419-196. Pope, J. G., and J. G. Shepherd. 1982. A simple method for the consistent interpretation of catch-at-age data. J. Cons. int. Explor Mer 40:176-184. 1985. A comparison of the performance of various methods for tuning VPAs using effort data. J. Cons. Int. Explor Mer 42:129-151. Porch, C. E., and G. P Scott. 1994. A numerical evaluation of GLM methods for estimat- ing indices of abundance from west Atlantic bluefin tuna catch per trip data when a high proportion of the trips are unsuccessful. Int. Comm. Cons. Atl. Tunas (Madrid), Coll. Vol. Sci. Pap. 42( 1 ):240-245. Potts, J. C, and C. S. Manooch, III. 2002. Estimated ages of red porgy from fishery-dependent and fishery-independent data and comparison of growth parameters. Fish. Bull. 100(l):81-89. Potts, J. C, C. S. Manooch III, and D. S. Vaughan. 1998. Age and growth of vermilion snapper from the south- eastern United States. Trans. Am. Fish. Soc. 127:787-795. Prager, M. H., J. R O'Brien, and S. B. Saila. 1987. Using lifetime fecundity to compare management strategies: a case history for striped bass. N. Am. J. Fish. Manag. 7:403-409. Prager, M. H. 1994. A suite of extensions to a nonequilibrium surplus-pro- duction model. Fish. Bull. 92:374-389. 1995. User's manual for ASPIC: a stock-production model incorporating covariates, program version 3.6x. NMFS Southeast Fisheries Science Center, Miami Laboratory Docu- ment MIA-2/93-55, 4th ed., 29 p. [Available from author.] Punt. A. E. 1990. Is S| = K an appropriate assumption when applying an obsei-vation error production-model estimator to catch- effort data? S. Afr. J. Mar Sci. 9:249-259. Ralston, S. 1987. Mortality rates of snappers and groupers. //; Trop- ical snappers and groupers: biology and fisheries man- agement (J. J. Polovina and S. Ralston, eds.), p. 375-404. Westview Press, Boulder, CO. Restrepo, V. R., and ten coauthors. 1998. Technical guidance on the use of precautionary approaches to implementing national standard 1 of the Magnuson-Stevens Fishery Conservation and Management Act. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/ SPO-31. Richards, R. A., and P J. Rago. 1999. A case history of effective fishery management: Chesa- peake Bay striped bass. N. Am. J. Fish. Manag. 19:356- 375. Ricker, W. E. 1975. Computation and interpretation of biological sta- tistics of fish populations. Bull. Fish. Res. Board Can. 191:1-382. Schaefer, M. B. 1954. Some aspects of the dynamics of populations impor- tant to the management of the commercial marine fisher- ies. Bull. Inter-Am. Trop. Tuna Comm. l(2):27-56. 1957. A study of the dynamics of the fishery for yellowfin tuna in the eastern tropical Pacific Ocean. Bull. Inter-Am. Trop. Tuna Comm. 2:247-268. Schnute, J, 1977. Improved estimates from the Schaefer production model: theoretical considerations. J. Fish. Res. Board Can. 34:583-603. Shapiro, D. Y. 1979. Social behavior, group structure, and the control of sex reversal in hermaphroditic fish. Adv. Stud. Behav. 10: 43-102. Sissenwine, M. P., and J. G. Shepherd. 1987. An alternative perspective on recruitment overfish- ing and biological reference points. Can. J. Fish. Aquat. Sci. 44:913-918. VciLighan and Piager Decline in abundance of Pagi us pagnu off tfie southeastern United Slates 375 Sncdocor, G. W.. and W. G. Cochran 1980. Statistical methods, seventh edition. Iowa State Univ. Press, Ames, lA, 507 p. \'aughan, D. S., and M. L. Burton. 1994. P^stimation of von BerlalanlTy growth parameters in the presence of size-selective mortality: A simulated exam- ple with red gn"Ouper. Trans. Am. Fish. Soc. 123:1-8. X'aughan, D. S., M. R. Collins, and D. J. Schmidt. 1995. Population characteristics of the black sea bass Cen- tropristis striata from the southeastern U.S. Bull. Mar. Sci. 56:250-267. Vaughan, D. S., G. R. Huntsman, C. S. Manooch III, F. C. Rohde and G. F Ulrich. 1992. Population characteristics of the red porgy, Pagrus pagriia. stock off the Carolinas. Bull. Mar. Sci. 50:1-20. Vaughan, I). S.. B. Zhao. M. R. Collins, J. C. McGovern, and H. S. Meisler 1998. Evaluation of multiple survey indices in assessment of black .sea bass from the U.S. south Atlantic coast. In Fishery stock assessment models (F. 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. 121-136. Lowell Wake- field Fisheries Symposium, Univ. Alaska Sea Grant College Program, Report 98-01. von Bertalanffy. L. 1938. A quantitative theory of organic growth. Hum. Biol. 10:181-213. 376 A nearsurface, daytime occurrence of two mesopelagic fish species iStenobrachius leucopsarus and Leuroglossus schmidti) in a glacial fjord Alisa A. Abookire Kodiak Laboratory National Marine Fisheries Service 301 Research Court Kodiak, Alaska 99615 E-mail address alisa abookireiiJ'noaa gov John F. Piatt Alaska Biological Science Center U^ S^ Geological Survey 101 1 E. Tudor Road Anchorage, Alaska 99503 Suzann G. Speckman School of Aquatic and Fishery Sciences University of Washington 1122 NE Boat Street Seattle, Washington 98105 the day to the epipelagic zone (0-200 m) after sunset (Sobolevsky et al., 1996). Larval northern smoothtongue are abundant in the upper 50 m of Alaskan coastal waters during daylight (Haldorson et al., 1993; Norcross and Frandsen, 1996), but after the post- larval and early juvenile stages they are typically found at depths greater than 150 m (Taylor, 1968; Mason and Phillips, 1985). Previous accounts of adult northern smoothtongue at shal- low depths were those of collections made after sunset (Barraclough, 1967; references in Hart, 1973), and in some areas adults do not exhibit diel vertical migration but remain at depths below 240 m (Mason and Phillips, 1985). There has previously been no documen- tation of either juvenile or adult north- ern smoothtongue in the upper 50 m of the water-column prior to sunset. The purpose of our study was to doc- ument the nearsurface (<15 m), day- time occurrence of both juvenile and adult northern lampfish and northern smoothtongue in Glacier Bay, Alaska. The northern lampfish iStenobrachius leucopsarus, family Myctophidae) and northern smoothtongue (Leuroglossus schmidti, family Bathylagidae ) are mesopelagic fishes, defined by their ver- tical distribution in the mesopelagic zone (200-1000 m) during daylight hours. Northern lampfish range from the Bering Sea to southern California (Shimada, 1948). where their abun- dance is highest along the continental slope and decreases over the continen- tal shelf They are the most abundant species in the mesopelagic zone of the Bering Sea (Pearcy et al, 1977; Sobo- levsky et al., 1996), the Gulf of Alaska (Purcell, 1996), and the eastern North Pacific Ocean off Oregon (Pearcy, 1964; Pearcy et al, 1977). Northern smooth- tongue also concentrate in areas bor- dering the continental slope and are widely distributed from southern Brit- ish Columbia to the Bering Sea (Peden, 1981) and are very abundant in the Okhotsk Sea (Sobolevsky et al., 1996). Although lai-val myctophids spend day and night in nearsurface waters (Frost and McCrone, 1979), juvenile and adult northern lampfish typically inhabit depths of 300-600 m during the day (Paxton, 1967; Pearcy et al., 1977; Frost and McCrone, 1979; So- bolevsky et al., 1996; Watanabe et al., 1999). Off the Oregon coast north- ern lampfish exhibit seniimigrant be- havior: a large portion of the popula- tion migrates from a daytime depth of 300-600 m to about 50 m at night, and a nonmigratory portion of the popula- tion has both a day- and night-time vertical distribution at about 500 m depth (Pearcy et al, 1977). This semi- migrant behavior also occurs in the Gulf of Alaska (Frost and McCrone, 1979), in the western North Pacific off Japan (Watanabe et al., 19991, and off California in both the San Pedro Ba- sin (Paxton, 1967) and Santa Barbara Basin (Cailliet and Ebeling, 1990). Af- ter sunset, migratory northern lamp- fish have been collected at depths as shallow as 20 m in the Bering Sea (Na- gasawa et al., 1997). We are unaware of any studies in which adult northern lampfish were collected at depths of less than 100 m prior to sunset. Northern smoothtongue migrate from the mesopelagic zone (200-1000 m) in Materials and methods Glacier Bay is a fjord in Southeast Alaska that extends northward and bifurcates into a west arm and Muir Inlet (Fig. 1). We conducted 48 mid- water trawl tows at 34 stations from 10 to 23 June 1999. Fishes were lo- cated with a Biosonics DT4000 digital 120-kHz echo sounder, and significant targets were fished with a modified herring mid-water trawl with a mouth opening of 50 m-. Mesh sizes dimin- ished stepwise from 5 cm in the wings to 1 cm at the codend, which was lined with 3-mm mesh. A plastic col- lecting bucket with 1-mm mesh was attached to the end of the codend, and was detached and rinsed after each tow. A Furuno net-sounding system monitored the depth of the headrope during fishing operations. A temper- ature depth recorder (TDR, Wildlife computers model MK7) was mounted on the headrope to determine the exact depth of the net during fishing oper- Manuscript accepted 12 September 2001. Fish Bull. 100;376-380 (2002). NOTE Abookire et a\ NearsLirface, daytime occurrence of two mesopelagic fishes in a glacial fiord 377 136-W 59 N S8N Muir Glader .X - ■*/ !* f ■ . >}\ jf^ Alaska West, Arm Glacier 0» c^ . : ;' Bartiett ■^ *15 FTU from 3 to 16 m and >10 FTU to depths of 41 m; turbidity at site 1 was low and ranged from 7.8 to 8.7 FTU (Fig. 3). At site 2 the maximum chlorophyll concen- tration (16.1 pg/L) occurred at 5 m; whereas at site 1 the chlorophyll concentration peaked (24.7 vig/L) at 12 m and remained high (>10 pg/L) to 28 m. Discussion Prior to sunset, juvenile and adult mesopelagic fishes were found at depths of 10 to 15 m near Muir Glacier in Glacier Bay, Alaska. Diel migrations of northern lampfish may be a response to light intensity (Paxton, 1967; Pearcy et al., 1977) or be stimulated by hunger (Pearcy et al.. 1977; Cailliet and Ebeling, 1990). In the spring-summer period, when daylight hours increase in the Bering Sea, vertical migrations of northern lampfish and northern smooth- tongue had smaller amplitudes and were not as distinct iSobolevsky et al., 1996). Phytoplankton blooms and fog banks can reduce the depth to which light can penetrate, also causing northern lampfish to migrate higher in the water column (Barham. 1957). Turbidity in the upper water layer next to Muir Glacier was higher than anywhere else in this study (15 FTU), and this high turbidity corresponded with the shallowest occurrence of mesopelagic fishes. In contrast, in Icy Strait at site 1, where the turbidity was lower, northern lampfish were captured at 90 m depth during daylight hours (Table 1, Fig. 3). High turbidity from glacial silt at Muir Glacier may significantly reduce light penetration and account for the presence of northern lampfish and northern smooth- tongue at 10-15 m depth prior to actual sunset. Although these mesopelagic fishes are more commonly distributed along the continental slope, both northern lampfish (Shimada, 1948; Taylor. 1968) and northern smoothtongue (Mason and Phillips. 1985) also occur in deep, nearshore waters from the eastern Gulf of Alaska to Puget Sound. Washington. Northern lampfish were ob- served in Icy Strait and Lower Glacier Bay. Alaska, pre- viously.' but not as far inland as Muir Glacier. Northern smoothtongue have not been obsei-ved previously in the same area. Their occurrence along the inside waters and ijords of Southeast Alaska makes them available to a suite of predators that otherwise might feed upon them only off- shore. ' Wing. B. 2000. Personal commun. National Marine Fish- eries Sei-vice, Auke Bay Laboratory, 11305 Glacier Highway. Juneau, Alaska. 99801 . ' NOTE Abookiie et a\ Nedrsurface, daytime occurrence of two mesopelagic fishes in a glacial fjord 379 10 , Northern 8 6 lampfish (/;=«) 1 4 - ..L. 2 - D a 1 s a >^ o c 0 3 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90 0 20 n Northern 15 - 10 ^ smoothtongue (/;=133) In j 5 - h Ik 2 6 10 14 18 22 26 30 34 3S 42 46 50 54 5S 62 66 70 74 78 82 86 90 Fork length (mm) Figure 2 Length-frequency histograms for northern lampfish (Sfenobrachiiis leii- copsarus) and northern smoothtongue (Leuroglossus schinidti). Lengths were grouped in 2-mm intervals. The total number offish that were mea- sured is given in parentheses. Northern lampfish and northern smoothtongue serve as important Hnks in oceanic food webs (Cailliet and Ebel- ing, 1990). Myctophids provide a high-lipid energy source for a variety of predators (Van Pelt et al., 1997; Springer et al., 1999). For example, stomachs of marbled murrelets iBrachyramphiis marmoratus) collected at the entrance to Glacier Bay contained northern lampfish almost exclu- sively.'^ Black-legged kittiwakes {Rissa tridactyla), which are restricted to foraging on fish at the surface, must feed nocturnally to obtain northern lampfish at oceanic islands in the Aleutians (Springer et al., 1996). However, near Muir Glacier and other tide-water glaciers in Glacier Bay, kittiwakes may be able to regularly forage on myctophids during daylight hours. Further investigations of fish dis- tributions and predator diets are necessary to understand trophic interactions in Alaska's glacial fjords such as Gla- cier Bay. Acknowledgments We thank Mayumi Arimitsu, Janene Driscoll, Captain Mark Hottman, and the crew of the RV Pandalus for their assistance during the cruise. Becka Seymour helped to process fish samples, and Claire Armistead made Figure 1. Turbidity (FTU) 0 5 Id 15 20 m m » m 15 -♦- Site I • Site 2 Figure 3. Vertical profiles of turbidity (FTU, formazine tur- bidity units! at site 1 and site 2 from the surface to 1.5 m depth. Black diamonds symbolize data from site 1 and gray squares symbolize data from site 2. 2 Piatt, J. F. 1991. Unpubl. data. U.S. Geological Survey. 101 ! E. Tudor Road, Anchorage, Alaska, 99503. We thank Greg Cailliet, Michael Litzow, James Orr, and David Somerton for critical reviews of the manuscript. This project was supported by U.S. Geological Survey (USGS) Natural Resource Preservation Program (NRPP) 380 Fishery Bulletin 100(2) funds obtained for studies of forage fish in Glacier Bay National Park ( GBNP). We thank Jim Taggart ( USGS ) and Mary Beth Moss (GBNP) for encouragement and logistic support. Literature cited Barham. E. G. 1957. The ecology of sonic scattering layers in the Monterey Bay area. Hopkins Mar. Sta. Tech. Rep. 1:1-182. Barraclough, W. E. 1967. Number, size, and food of lai-val and juvenile fish caught with an Isaacs-Kidd trawl in the surface waters of the Strait of Georgia, April 25-29 1966. Fish. Res. Board Can. MS Rep. Ser 926, 79 p. Cailliet, G. M., and A. W. Ebeling. 1990. The vertical distribution and feeding habits of two common midwater fishes (Leuroglossus stilbius and Steno- brachius leucopsariis ) off Santa Barbara. CalCOFI Rep. 31:106-123, Frost, B. W., and L. E. McCrone. 1979. Vertical distribution, diel vertical migration, and abundance of somemesopelagic fishes in the eastern Sub- arctic Pacific Ocean in summer Fish. Bull. 76:751-770. Haldorson, L., M. Prichett, A. J. Paul, and D. Ziemann. 1993. Vertical distribution and migi-ation of fish larvae in a Northeast Pacific bay Mar Ecol. Prog. Ser. 101(1-2): 67-80. Hart, J. L. 1973. Pacific fishes of Canada. Fish. Res. Board Can. Bull 180, 740 p. Mason, J. C, and A. C. Phillips. 1985. Biology of the bathylagid fish, Leuroglossus schmidti. in the Strait of Georgia, British Columbia, Canada. Can. J. Fish. Aquat. Sci. 42:1144-1153. Nagasawa, K., A. Nishimura, T. Asanuma, and T. Marubayashi. 1997. Myctophids in the Bering Sea: Distribution, abun- dance, and significance as food for salmonids. In Forage fishes in marine ecosystems: proceedings of the interna- tional symposium on the role of forage fishes in marine eco- systems; 13-16 November 1996, Anchorage. Alaska (C. W. Mecklenburg, ed.), p. 337-350. Alaska Sea Grant College Program Report 97-01, Univ. Alaska, Fairbanks, AK. Norcross, B. L., and M. Frandsen. 1996. Distribution and abundance of lai-val fishes in Prince William Sound, Alaska, during 1989 after the Exxon Valdez oil spill. In Proceedings of the Exxon Valdez oil spill sym- posium; 2-5 February 1993, Anchorage, Alaska (S. D. Rice, R. B. Spies, D. A. Wolfe, and B. A. Wright, eds. I, p. 463-486. Am. Fish. Soc. Symp., vol. 18., Bethesda, MD. Paxton, J. R. 1967. A distributional analysis for the lantcrnfishes i family Myctophidae) of the San Pedro Basin. California. Copeia 1967(21:422-440. Pearcy W. G. 1964. Some distributional features of mesopelagic fishes off Oregon. J.Mar. Res. 22( 1):83-102. Pearcy, W. G., E. E. Krygier, R. Mesecar, and F Ramsey. 1977. Vertical distribution and migi-ation of oceanic micro- nekton ofr Oregon. Deep-Sea Res. 24(31:223-245. Peden, A. E. 1981. Recognition of Leuroglossus sclunidti and L. stilbius (Bathylagidae, Pisces) as distinct species in the North Pacific Ocean. Can. J. Zool. 59:239-2398. Purcell, J. 1996. Mesopelagics. /;i Mass-balance models of northeast- ern Pacific ecosystems (D. Pauly, V. Christensen, and N, Haggan, eds.), p. 23. Fish. Cent.Res. Rep. 4( 1). The Fisher- ies Centre, Univ., British Columbia. Shimada, B. M. 1948. Records oflantem fish in Puget Sound. Copeia 1948(3): 227. Smoker, W., and W. G. Pearcy 1970. Growth and reproduction of the lanternfish Sleno- brachius leucopsarus. J. Fish Res. Board Canada 27(7): 1265-1275. Sobolevsky, Y. I., and T G. Sokolovskaya. 1996. Development and distribution of the young of north- ern smoothtongue(Lt'»/-og/o,s'S(/s schmidti ) in the Northwest Pacific Ocean and Western Bering Sea. In Ecology of the Bering Sea: a review of Russian literature lO. A. Mathisen and K. O. Coyle, eds. ), p. 257-263. Alaska Sea Grant Col- lege Program Report 96-01, Univ. Alaska Fairbanks, AK. Sobolevsky, Y. I., T. G. Sokolovskaya, A. A. Balanov, and I. A. Senchenko. 1996. Distribution and trophic relationships of abundant mesopelagic fishes ofthe Bering Sea. In Ecology of the Bering Sea: a review of Russian literature (O. A. Mathisen and K. O. Coyle, eds.), p. 159-167. Alaska Sea Grant Col- lege Program Report 96-01, Univ. Alaska, Fairbanks, AK. Springer, A. M., J. F. Piatt, and G. Van Vliet. 1996. Sea birds as pro.xies of marine habitats and food webs in the western Aleutian arc. Fish. Oceanogr. 5( 1 ):45-55. Springer, A. M., J. F Piatt, V. P. Shunton, G. B. Van \r\iet. V. L. Vladimirov, A. E. Kuzin.and A. S. Perlov 1999. Marine birds and mammals of the Pacific subarctic gyres. Prog. Oceanogr. 43( 2-4 1:443-487. Taylor, F H. C. 1968. The relationship of midwater trawl catches to sound scattering layers off the coast of northern British Colum- bia. J. Fish. Res. Board Canada. 25(3):457-472. Van Pelt, T. I., J. F. Piatt, B. K. Lance, and D. D. Roby 1997. Proximate composition and energy density of some North Pacific foragefishes. Comp. Biochem. Physiol. 1 18A(2): 1393-1398. Watanabe, H., M. Moku, K. Kawaguchi, K. Ishimaru, and A. Ohno. 1999. Diel vertical migi'ation of myctophid fishes (family Myctophidae I in the transitional waters of the western North Pacific. Fish. Oceanogr. 8(2):115-127. 381 Opportunistic feeding of longhorn sculpin {Myoxocephalus octodecemspinosus) : Are scallop fishery discards an important food subsidy for scavengers on Georges Bank? Jason S. Link Frank P. Almeida National Marine Fisheries Service Northeast Fisheries Science Center 166 Water St Woods Hole, Massachusetts 02543 E mail address Jason LinkiQ'noaa gov There has been much recent interest in the effects of fishing on habitat and non-target species, as well as in protecting certain areas of the seabed from these effects (e.g. Jennings and Kaiser, 1998; Benaka, 1999; Langton and Auster, 1999; Kaiser and de Groot, 2000). As part of an effort to deter- mine the effectiveness of marine closed areas in promoting recovery of com- mercial species (e.g. haddock, Mela- nogrammus aegelfinus; sea scallops, Placopecten inagellanicus; yellowtail flounder, Limanda ferruginea\ cod, Gadus morhiia ). nontarget species, and habitat, a multidisciplinary research cruise was conducted by the Northeast Fisheries Science Center (NEFSC), National Marine Fisheries Sei'vice. The cruise was conducted in closed area II (CA-II) of the eastern portion of Georges Bank during 19-29 June 2000 (Fig. 1). The area has historically pro- duced high landings of scallops but was closed in 1994 principally for ground- fish recovery (Fogarty and Murawski, 1998). The southern portion of the area was reopened to scallop fishing from 15 June to 12 November 1999, and again from 15 June to 15 August 2000. While conducting our planned sam- pling, we observed scallop viscera (the noncalcareous remains from scallops that have been shucked by commercial fishermen at sea) in the stomachs of several fish species at some of these locations, namely little skate (Raja eri- nacea ), winter skate (R. ocellata ), red hake iUrophycis chuss), and longhorn sculpin [Myoxocephalus octodecemspi- nosus). We examined the stomach contents of a known scavenger, the longhorn sculpin, to evaluate and doc- ument the e.xtent of this phenomenon. Fishery discards provide food subsi- dies that help maintain fish popula- tions, but to what extent is unclear. There is some evidence that fishery discards allow fish populations to be more abundant than they would be with just ambient resources (e.g. Polls and Strong, 1996). Others would coun- ter that these discards may maintain a population but not necessarily lead to population growth (Fonds and Groe- newold, 2000). As an extension of our opportunistic field obsei"vations from the closed area (CA)-II 2000 cruise, we examined NEFSC historical bottom trawl survey and food-habits databas- es to ascertain the role of fishery dis- cards in the Georges Bank ecosystem, with particular respect to the longhorn sculpin population. Materials and methods Our sampling was conducted concur- rently with the scallop fishery, there- fore it was necessary to continuously monitor vessel activity in the area opened to fishing prior to our arrival and during our sampling. Prior to sampling, we collected biweekly sum- maries of scallop fishing vessel activ- ity during the 1999 open season, and day by day summaries of the 2000 season were obtained from the north- east region vessel monitoring system (VMS), National Marine Fisheries Ser- vice. With this system, individual vessel location.s are transmitted to NMFS every 30 minutes (McSherryM. The location of each fishing vessel was plot- ted (with Surfer 7.0; GSI, 1999) on a map overlaid with the sampling sites occupied during the 1999 study (June 1999). Stations from the 1999 study were chosen on the basis of fleet activ- ity during the 1999 open season and our sampling goals. Each day while the RW Albatross /Vwas in CA-II, cur- rent individual scallop-vessel activity data were transmitted to us by email, plotted, and sampling stations were selected after an examination of the data. As part of our sampling protocol, a 15-minute otter trawl haul was made at each station, towed at a speed of 6.5 km/li. A standardized NEFSC no. 36 Yankee otter trawl rigged with a rub- ber-disc-covered chain sweep, 11 floats, 5-m ground cables, and 450-kg poly- valent trawl doors (commonly referred to as a "flatfish net") was used. Once the trawl was on deck, we sorted fish and macro-invertebrates by species, weighed each species (0.1 kg), mea- sured lengths of all fish (cm), examined subsamples of the fish to determine sex, maturity, and stomach contents, and collected structures to determine ages. For further details of the survey and food habits methods used in our study see Azarovitz (1981) and Link and Almeida ( 2000 ), respectively. After initial observations of scallop viscera in fish stomachs at several pre- vious locations, we developed an ad hoc study that deliberately selected two stations located in areas of high scallop fishing effort and two stations in areas of little or no scallop fishing effort. On 21-22 June 2000 we sampled stations 8B, 7D, 7E, and 8E in the northeast portion of the region open to the scal- lop fleet (Fig. 1). At stations 8B and 7E, we observed a high frequency of seal- ' McSherry, M. 1998. NMFS Northeast vessel monitoring system — operations logic and geographical display interface. NMFS NE Law Enforcement Office Internal Docu- ment, 20 p. [Available from NMFS, NEFSC, 166 Water St., Woods Hole, MA 02543.] Manuscript accepted 18 September 2001. Fish. Bull. 100:381-385 (2002). 382 Fishery Bulletin 100(2) 71,0 70.5 70,0 69,5 69,0^„.--^8.5 68,0 67.5 67 0 66 5 \66,0 65,5 21 June 2000 22 June 2000 Figure 1 The northeast U.S. continental shelf region, with scallop fishing fleet activity i ■ I in closed area II of Georges Bank on 21-22 June 2000 overlaid with the sampling sites I ▲ I occupied by the NOAA FRV Albatross IV i Cruise AL 00-0.3 ). lop viscera in the diets of longhorn sculpin. among other species, and undertook extra samphng of sculpin stomachs to assess the magnitude of this phenomenon. We also un- dertook extra samphng of sculpin at the other two sta- tions. We present mean stomach volume and mean percent (by volume) diet composition for longhorn sculpin at these four stations. To determine the prevalence of this phenomenon and whether discarded scallop viscera are a regular food sub- sidy that significantly influences the sculpin population, we examined the stratified mean number of sculpin per tow on Georges Bank (Azarovitz, 1981). We calculated this index of abundance across the fall bottoni-trawl survey time series from 1963 to 1998. Additionally, we examined the mean stomach contents (cm'), maximal stomach con- tents, percent frequency of occurrence of bivalve, inollusk, or pectinid viscera, and an index of gorge feeding on an annual basis from the food habits time series (Link and Almeida, 2000). The index of gorge feeding (or more sim- ply, the gorge index) was calculated as the percent of stom- achs examined with total stomach contents greater than 10 cm'. We executed a simple linear correlation between the sculpin food-habit metrics and the index of abundance to determine if any significant relationship exists between gorge feeding and sculpin population abundance. Results On 21 and 22 June 2000, scallop fishing clearly occurred at stations 8B and 7E, whereas no scallop fishing occurred at 7D or 8E (Fig. 1 ). Rock crabs (both Cancer irroratus and C. boreal is) and small crustaceans typify the diet of long- horn sculpin (Fig. 2). However, at stations in areas with scallop fishing activity, the diet of sculpin was predomi- nately made up of scallop viscera. Crabs and small crusta- ceans were still a part of the diet at those stations, but did not comprise as large a percentage as was found at non- dredged stations. The volume of food in sculpin stomachs was also an or- der of magnitude higher at stations with intense scallop fishing activity (8B=38.7 cnr' ±6.9; 7E=19.4 cnr' ±12.7) than at those with no scallop fishing (8E=1.7 cm'^ ±1.4; 7D=0.3 cm' ±0.1). When scallop remains were present at NOTE Link and Almeida: Opportunistic feeding of Myoxocephnkis octodecemspinosus on Georges Bank 383 100 80 > 60 F, 40 20 1 Rock crab 1 c^ ■ 8B| □ 7e! D8E^ H7D i Gammarid Other crustaceans Hermit crab Sea mouse Polychaete Scallop viscera Figure 2 Diet composition {7i by volume t of longhorn sculpin at two dredged and two undredged loca- tions in closed area II. Bars are 959i confidence intervals. Sample sizes are 15, 8, 11,6 fish at stations 8B. 7E, 8E, and 7D respectively. a location, the amount of food consumed by sculpins in- creased significantly, suggesting that sculpins continued to feed on crabs and small crustaceans, but opportunis- tically gorged on the remains of scallops that had been shucked and thrown overboard from scallop fishing ves- sels operating in the region. Although anecdotal, the stom- achs of all sculpins at station 8B, the station most intense- ly dredged, were distended beyond normal proportions for 20-25 cm fish, often extending below the pectoral fins. The mean stomach contents of sculpin were effectively constant across the time series, averaging approximately 2 cm^ (Fig. 3). However, the maximal stomach contents, the percent frequency of bivalve viscera, and the index of gorge feeding showed distinct patterns across the time se- ries. All three of these metrics were generally coincident. Years with a higher gorge index corresponded to years with a high occurrence of bivalve viscera in the diet of sculpins (r=0.70. P<0.01). This finding implies that many, but not all. of the gorging events were on scallop viscera and simi- lar discards. The highest values of the gorge index were ob- served in 1987 and 1998. Longhorn sculpin abundance exhibited an initial peak and then a relatively steady period for the first 15 years of the survey, followed by a period of lower abundance dur- ing the mid-1980s and an increasing trend in the 1990s (Fig. 4). In most years sculpin abundance ranged from 10 to 20 fish per tow. The years with the highest index of scul- pin abundance were 1966 and 1998. In relation to the pre- ceding years, the index of sculpin abundance notably in- creased during 1966, 1987, and 1998. The latter two years correspond to years when the index of gorge feeding was highest. The correlation between the gorge index and scul- pin abundance was weak (r=0.21). but significant and pos- itive (P<0.01). Discussion Much of the information describing the impact of fishing gear on benthic communities and habitats has been decid- edly negative. It is unclear whether populations of scaven- gers such as crabs, flatfish, or other demersal fish benefit from indirect effects of fishing (Kaiser and Spencer, 1994; Greenstreet and Hall, 1996; Ramsay et al, 1996) if they are not caught by the fishing gear. It is known that in the short term, scavengers, especially those small enough to fit through the mesh, are attracted to trawl and dredge tracts (e.g Ramsay et al., 1998; Demestre et al.. 2000). Addi- tionally, some studies have actually shown that the abun- dance of scavenger populations increases in areas that have been trawled (reviewed in Greenstreet and Rogers, 2000). Longhorn sculpin on Georges Bank have main- tained their ubiquity and abundance, albeit at relatively low levels compared with other species, during a period of intense fishing pressure in this ecosystem (Fogarty and Murawski. 1998). In recent years the sculpin population on Georges Bank has begun to increase. Opportunistic feeding on scallop viscera appears to have a positive influence on sculpin populations. The relation- ship between gorge feeding and sculpin abundance is ad- 384 Fishery Bulletin 100(2) 1 1 1 cm^ 80 JX » 70 — 1 — Mean — « — Max / - 60 E o c S 40 ■ -o- - Gorge index (%) -.-X-. % Bivavie viscera ? ; \ f- 10% o CJ a ( \ ' / : ■© Stomach NJ CO o o r^ '■ ■ / r / 5% 10 0 ♦--. -*,^"^ ^Nodata-^ 1 6 TJ* -^ff-^ ^ * £— * * ^ ^ »^ « ^ ""^-"^ 5 - 0% CO CJ> in r^ cn T- CO 1^ h- 1^ CO 00 m 1^ CO CO G) at en CO ■<- CO in h- Oi cy> (ji Oi &} &i Oi Figure 3 The mea n OS'"; CIl and maximal (largest amount eaten) stomach contents (cm^) of long- | horn SCI ilpin from the NEFSC food habits datab ise Also, the index of gorge feeding (percent ige of stomachs with >10 cm'l ind percoi tage of stomachs containmg bivalve, 1 mollusk or pectinid viscera. mittedly weak, yet significant and positive. We rec- ognize that numerous other factors could influence sculpin abundance. However we find it intriguing that the two years with the most drastic increase in sculpin abundance (in the context of recent pre- ceding years: i.e. 1987 and 1996) occurred in years when sculpin gorge feeding was highest. The amount of food eaten and the diet of long- horn sculpin at stations without scallop fishing ac- tivity were similar to those reported for the species on the entire northeast continental shelf (Link and Almeida, 2000) and for Georges Bank in particular (Garrison and Link, 2000). The amount of food eat- en and diet at stations with scallop fishing activity showed clear evidence of gorge feeding by sculpin. Other studies have documented opportunistic feed- ing and changes in diet after groundfish trawling (e.g. gurnards and dabs; Kaiser and Spencer, 1994; Kaiser and Ramsay. 1997), however, this note is the first documentation of scavenging after scallop fish- ing. The long-term trend shows that, although the average amount of food sculpin consume is relatively con- stant, the composition of the diet and the frequency of gorging events varies notably. These variations may influ- ence sculpin abundance. In the 2000 open fishing season in CA-II, more than 760 metric tons (t) of scallop meats were landed (NERO. 2000). If one conservatively assumes that scallop viscera equals the meat weight, then over 760 t (>340,000 lb) of viscera were returned to the southern part of CA-Il dur- ing this two-month period. The discarded scallop viscera 30 25 20 E 15 ro 10 in I^ O) n in r-- Ol fO in r^ cn CO in r- i ^ ^ 2 O) CT) CT) a> en o^ O) Q) en Figure 4 Index of relative abundance of longhorn sculpin from the NEFSC fall bottom trawl sun.'ev. are a relatively small organic input, on an areal basis, to this closed area, which averages ca. 430,000 t in total an- nual benthic production (111 g/m', Cohen and Grosslein, 1987; approximately 3900 km- for the southern part of CA-11). How many tons of scallop viscera and similar dis- cards have been deposited across the entire Bank over the past several decades is unknown. Yet, at a local scale scal- lop viscera may be an important source of energy, akin to similar processes for fish in the deep ocean (Sedberry and Musick, 1978). The biomass of scallop viscera, although NOTE Link and Almeida: Opportunistic feeding of Myoxocephalus octodecemspinosus on Georges Bank 385 previously a part of the system, is returned in a condition much more readily available to benthic scavengers. Even weak links in terms of diet or energy can be important dy- namic links in terms of population regulation (reviewed in Polls and Strong, 1996). The reintroduction of scallop vis- cera may not be a large addition of energy to the Georges Bank ecosystem, but it may be an important one for scul- pin and other opportunist populations. There is evidence that energy derived from fishery dis- cards can cause scavengers to become more abundant than they would be w^ith ambient resources (Polls and Strong, 19961, or at least these discards can help scavengers main- tain population abundance under disturbed circumstanc- es (Fonds and Groenewold, 2000). Further studies may indicate that events which produce food subsidies such as those documented in this study may be more common than generally suspected. Given the ubiquity of naval, commercial, and private vessels on the ocean, it likely that organic waste from these ships, from fishing activities, and from similar processes serves as a significant "rain of food" for benthic scavengers. Acknowledgments We thank M. McSherry for access to the vessel monitoring system (VMS) data, L. Garrison for sending the data to us daily while we were on the vessel, the crew and scientific party of the RV Albat7-oss FV for assistance in collecting the samples, V. Guida for help in recording the logsheets, R. Reid and P. Valentine for stimulating discussions on habitat and gear effects issues, and R. Reid, H. Lai, W. Gabriel, M. Kaiser, and other reviewers for constructive comments on earlier versions of the manuscript. Literature cited Azarovitz. T. R. 1981. A brief historical review of the Woods Hole Labora- tory trawl survey time series. Can. Spec. Pub. Fish. Aq. Sci. 58:62-67. Benaka, L. 1999. Fish habitat; essential fish habitat and rehabilitation. Am. Fish. Soc. Publ., Bethesda, MD. 4.59 p. Cohen, E. B.. and M. D. Grosslein. 1987. Production on Georges Bank compared with other shelf ecosystems. In Georges Bank (R. H. Backus, ed), p. 381-391. MIT Press. Cambridge, MA. Demestre, M., and P. Sanchez, and M. J. Kaiser. 2000. The behavioural response of benthic scavengers to otter-trawling disturbance in the Mediterranean. In Effects of fishing on non-target species and habitats (M. J. Kaiser and S. J. de Groot, eds. ). p. 121-129. Biological, conser\-a- tion and socio-economic issues. Blackwell Science, Oxford. Fogarty, M. J., and S. A. Murawski. 1998. Large scale disturbance and the structure of marine systems, fishery impacts on Georges Bank. Ecol. Appl. 8 (suppl. 1):S6-S22. F'onds, M., and S. Groenewold. 2000. Food subsidies generated by beam-trawl fishery in the southern North Sea. In Effects of fishing on non-tar- get species and habitats (M. J. Kaiser and S. J. de Groot, eds.), p. 130-1.50. Biological, conservation and socio-eco- nomic issues. Blackwell Science, Oxford. Garrison, L. P., and J. S. Link. 2000. Fishing effects on spatial distribution and trophic guild structure of the fish community in the Georges Bank region. ICES J. Mar. Sci. 57:723-730. GSI (Golden Software Inc.). 1999. Surfer version 7.0. Surface Mapping System. Golden Software Inc., Golden, CO. Greentstreet, S. P. R.. and S. I. Rogers. 2000. Effects of fishing on non-target fish species. In Effects of fishing on non-target species and habitats iM. J. Kaiser and S. J. de Groot, eds.), p. 217-234. Biological, con- servation and socio-economic issues. Blackwell Science, Oxford. Greenstreet, S. P. R., and S. J. Hall. 1996. Fishing and groundfish assemblage structure in the northwestern North Sea: an analysis of long-term and spa- tial trends. J. Anim. Ecol. 65:577-598. Jennings, S., and M. J. Kaiser 1998. The effects of fishing on marine ecosystems. Adv. Mar. Biol. 34:201-352, Kaiser, M. J., and S. J. de Groot. 2000. Effects of fishing on non-target species and habitats. Biological, consen/ation and socio-economic issues. Black- well Science, Oxford, 399 p. Kaiser, M. J., and K. Ramsay. 1997. Opportunistic feeding by dabs within areas of trawl disturbance: Possible implications for increased survival. Mar. Ecol. Prog. Sen 152:307-310. Kaiser, M. J., and B. E. Spencer. 1994. Fish scavenging behaviour in recently trawled areas. Mar. Ecol. Prog. Sen 112:41-49. Langton, R. W,. and P. J. Auster. 1999. Marine fishery and habitat interactions: to what extent are fisheries and habitat interdependent? Fisher- ies 24:14-21. Link, J. S.. and F. P Almeida. 2000. An overview and history of the Food Web Dynamics Program of the Northeast Fisheries Science Center, Woods Hole, Massachusetts. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NE-159, 60 p. NERO (Northeast Regional Office). 2000. Sea scallop exemption program — weekly quota land- ing reports web page, http://www.nero.nmfs.gov/ro/fso/scal 2000.htm. Nov 2000. Polls, G. A., and D. R. Strong. 1996. Food web complexity and community dynamics. Am. Nat. 147:81.3-846. Ramsay, K., M. J. Kaiser, and R. N. Hughes. 1998. The responses of benthic scavengers to fishing distur- bance by towed gears in different habitats. J. Exper Mar. Biol. Ecol. 224:73-89. 1996. Change in hermit crab feeding patterns in response to trawling disturbance. Mar Ecol. Prog. Ser. 144:63-72. Sedberry, G. R., and J. R. Musick. 1978. Feeding strategies of some demersal fishes of the con- tinental slope and rise off the mid-Atlantic coast of the U.S.A. Man Biol. 44:337-375. 386 First record of a yellowfin tuna (Thunnus albacares) from the stomach of a longnose lancetfish (Alepisaurus ferox) Evgeny V. Romanov Southern Scientific Research Institute of Marine Fisheries and Oceanography (YugNIRO) 2, Sverdlov St 98300 Kerch Cnmea, Ul*' Figure 1 Sampling regions and station locations along the West Florida shelf (WFS). Thick black lines demarcate the three sampling regions: region 1, the Flor- ida Panhandle; region 2, Big Bend; and region 3, Southwest Florida. Koenig, C. C, and F. C. Coleman. 1998. Recruitment indices and seagrass habitat relationships of the early juvenile stages of gag, gray snapper, and other economically important reef fishes in the eastern Gulf of Mexico. Final report (MARFIN award no. NA.57FF0055) to Florida State University, 66 p. Depart- ment of Biological Sciences, The Florida State Univ., Tallahas- see, FL 32306-2043. and shoal-grass, Halodule wrightii (Zieman and Zieman, 1989). Station depth generally ranged from 1 to 3 meters. We collected juvenile fish for this study every other month in 1996 and 1997, beginning in late February of 1996 and early March of 1997. At each station we made five replicate tows of 150 meters (m) in approximately 5 minutes (1.8 km/hour) using two different types of trawls. To collect the smallest postsettlement juveniles, a 1-m by 0.40-m benthic scrape constructed of a stainless steel frame with a 2-mm nylon mesh tail bag was used. This trawl was used from the beginning of the year until the fall. We used a 5-m by 3-m otter trawl with a 3-mm nylon mesh tail bag to collect larger individuals year round. The contents of each trawl were sorted and all juvenile fish of commercial value were removed, bagged, and stored on ice. At the end of a sampling trip, all juvenile fish collected were frozen. In the laboratory juvenile gray snapper were removed from samples, thawed, measured (standard length [SL] to the nearest 0.1 mm), and weighed to the nearest 0.1 g. Otolith preparation and interpretation Both the sagittae and lapilli were extracted from each snapper under a dissecting microscope by the "open-the- hatch method" (Secor et al., 1992). One lapillus was chosen at random from each fish for measurement. Whole lapilli were viewed with a compound microscope at 50x magni- fication with a video camera and monitor and the image was digitized on a microcomputer with image analysis software (BioScan, 1990). A measurement (nearest pm) Allman and Grimes Spawning, settlement, and growth of Lutjanus griseus from tfie West Florida shielf 393 was made from the anterior edge to the posterior edge along the longest axis. We computed the hnear regression of standard length on lapillus length to determine if otolith growth and so- matic growth were proportional. To prepare lapilli for age determination, both lateral surfaces of the otolith were ground and polished on glass plates covered with a 3600- and 6000-grit polishing cloth. After being polished, the lapillus was mounted on a microscope slide with Pro-texx® mounting medium. Thirteen sagittae were also prepared to determine if sagittal section counts were consistent with lapilli counts. Because of the concave-convex shape of the sagittae, the technique used for examining the lapillus was not possible. Therefore, the more labor intensive pro- cess of embedding and sectioning the whole sagittae had to be used (Secor et al., 1992). Growth in mm/day was estimated as the slope of the lin- ear regression of standard length on the number of daily increments in the lapillus. Analysis of variance (ANOVA) and analysis of covariance (ANCOVA) were used to deter- mine if estimated growth rates varied significantly among sampling regions and years. Increment counts were made from sections of the la- pilli and sagittae by using a compound microscope at 400-lOOOx magnification. Oil immersion was used at the lOOOx magnification. Daily increments were distinguished from subdaily increments with the method of Campana (1992). Increments were counted twice by the same reader and counts that were different by no greater than 5% were averaged. Increment counts that differed by more than 5% were counted a third time, and if the third count differed by more than 57? of the previous counts, the otolith was rejected. An increment correction (3 days) was added to the total number counted based on the assumption that the first increment was not formed until after first feeding (i.e. approximately 3 days after fertilization) (Lindeman, 1997; Lindeman et al., 2000). This corrected increment count was subtracted from capture date to determine the fertilization date. We interpreted the settlement mark in the lapillus to be where the pattern in the increment widths changed markedly (Wellington and Victor, 1989). Increments were counted from the primordium to the settlement mark and a correction of three increments was added to estimate the number of days before settlement or the planktonic larval duration. Those lapilli that did not have an obvious settle- ment mark were excluded from the analysis. Settlement dates were calculated by subtracting the number of post- settlement increments from the date of capture. ANOVA was used to compare the age at settlement (i.e. planktonic larval duration) with region, year, and region-year inter- action. After converting lunar day of fertilization or settle- ment to a circular scale (Zar, 1984), we used a chi-square test to determine if fertilization and settlement dates were uniformly distributed across the lunar month. To validate the periodicity of increment formation in la- pilli we conducted an otolith marking experiment. Lapilli of live fish were marked with alizarin complexone, which has been shown to produce a well-defined mark in the otoliths of juvenile fish without producing high mortality (Tsuka- moto et al., 1989; Lang and Buxton, 1993). Approximately 30 juveniles were captured with a benthic trawl and held in an 800-gallon flow-through seawater system for several days so that they might acclimate to captivity. Juveniles were fed once daily ad libitum. Because of the photosensi- tivity of alizarin complexone, juveniles were immersed in 20 L of an aerated 200 mg/L solution of alizarin complex- one for 24 h in the dark. Juveniles were removed from the solution and returned to the flow-through seawater tank. Juveniles were sacrificed 7, 14, 21, or 28 days after marking and their lapilli were removed and prepared as previously described. Increments deposited after the alizarin complex- one mark were counted as previously described, except that transmitted ultraviolet light was used to read increments. Linear regression was used to compare number of incre- ments counted after the alizarin-complexone mark in the lapillus to the number of days the juveniles were held after marking. A /-test was used to determine if the slope of this regression was significantly different from 1.0. Gonad histology and analysis To obtain an estimate of the spawning time that was inde- pendent of otolith-based spawning dates, adult gray snap- per gonads were examined. All adult gray snapper were collected as part of an ongoing study by the National Marine Fisheries Service on reef fish reproduction. Fish were col- lected from commercial and recreational landings from Panama City, Florida, to Ft. Myers, Florida, from January to December in 1996 and from April to November in 1997. Each fish was weighed whole (i.e. ungutted [g|) and total length (TL) and fork length were measured (mm). All fish collected were at least 252 mm TL (minimum legal size) and were assumed to be sexually mature (Domeier et al., 1996). Gonads were removed and stored on ice and then pro- cessed by the methods of Collins et al. (1996). A small sample of each gonad was examined with a dissecting mi- croscope (250x) to measure the maximum oocyte diame- ter (nearest 0.1 mm). The developmental stage of ovarian sections was determined by using the methods of Wal- lace and Selman (1981). Ovarian stages were assigned on the basis of the most advanced ovarian stage or fol- licle stage present: 1 — primary growth (early oocytes); 2 — cortical alveolar (previtellogenic); 3 — vitellogenic; 4 — hydrated; and 5 — spent (i.e. presence of postovulatory fol- licles). Sections of testes were staged according to the methods of Moe (1969): 1 — spermatogonia; 2 — primary spermatocytes; 3 — secondary spermatocytes; 4 — sperma- tids; and 5 — spermatozoans. The timing and duration of spawning were determined by plotting oocyte diameter and histological development stage by sampling date. Results Sampling and collection Juvenile gray snapper were found in seagrass meadows along the west Florida shelf from June to November in the two most northern sampling regions, and from April 394 Fishery Bulletin 100(3) 12 — 10 — 8 — fi — 4 — June Southwest T I I I I I 1 I I j- 0 20 40 60 eO 100 120 140 160 ISO z — July Pantiandle I 1, , M U^ — 1 1 1 1 r 0 20 40 60 BO TOO 120 140 160 180 it ' July Southwest iJu T 1 I I 1 1 1 1 1 r 0 20 40 60 eO 100 120 140 160 180 Sept. Pan handle 1 1 1 1 r 0 20 40 80 eO ^°^ 120 140 180 180 12 - 10 - 8 6 4 - 2 0 — L Sept. Southwest T 1 ~ I " I I ■ ^ 1 1 1 r 0 20 40 60 80 100 120 140 160 180 Oct Panhandle T 1 1 I ( 1 1 1 1 r 0 20 40 80 80 100 120 140 ISO ISO Oct. Southwest 0 20 40 SO 80 100 120 140 180 180 10 — 8 — Nov Panhandle , , ,M , , — I 1 r 0 20 40 SO BO 100 120 140 1«0 1B0 Nov Big Bend B — e — Nov. Southwest T 1 1 I 1 1 1 1 1 r 0 20 40 60 BO 100 120 140 160 160 SL(mm) Figure 2 Length (SL. mm) frequency distribution of gray snapper juveniles by region and sampling month in 1996. to November in the most southern area. The first juvenile thic scrape and only one in the otter trawl along the entire was collected in the otter trawl in June 1996 from the WFS. The largest numbers were collected in September Southwest (Fig. 2). In July we collected 14 fish in the ben- and October from the Panhandle and Southwest. The fol- Allman and Grimes: Spawning, settlement, and growth of Lutianus gnscus from the West Florida shelf 395 5 - 4 — 3 - 2 - 1 - 0 —I April Soulhwest II I I "T 1 r 20 40 60 80 1 00 1 20 140 1 60 1 80 June Panhandle June Sou th west I I I July Panhandle I ■ I I July Sou th west Sept. Big Bend 1 1- 20 40 T 1 1 1 1 r 80 100 120 140 160 180 Sept Sou th west 20 40 60 aa 100 120 I'O 1 so la Oct. Panhandle ui Oct Southwest 2 0 4 0 SL(mm) Figure 3 Length (SL, mm) frequency distribution of gray snapper juveniles by region and sampling month in 1997. lowing year (19971, juveniles were initially collected in The catch rate decreased in June, then increased during April in the otter trawl in the most southerly part of the late summer, and peaked in September. Only two fish were Southwest sampling area (i.e. near Sanibel Island) (Fig. 3). collected in the benthic scrape, both in June. 396 Fishery Bulletin 100(3) There was evidence of multiple cohorts in the Southwest in October 1996 and in April 1997. Ju- veniles collected in 1997 (mean=88.3 mm SL) were, on average, about 50% larger than those collected in 1996 (mean=59.9 mm SL). Fish collected from the Southwest were also larger on average than those from the Panhandle both years. Age and growth Increments in the lapillus occurred at regular inter- vals and a consistent width and marked contrast was evident between each one. More closely spaced increments occurred at irregular intervals with less contrast between them. Usually we counted fewer increments in sagittae than in lapilli. Lapilli were thinner and flatter than sagittae and therefore required less grinding and could be examined whole, whereas sagittae had to be sectioned first. Incre- ments near the edge of the lapilli were also clearer and easier to distinguish compared with those from sagittal sections. Increment counts were success- fully assigned to 137 (71%) of the juveniles collected in 1996 and 97 (82%) of the 1997 collection. Uncor- rected increment counts ranged from 35-128 (mean=75) in 1996 and 50-226 (mean=125) in 1997. Several lines of evidence suggested that increments were deposited daily. The linear regression of SL (mm I on lapillus length (jam) indicated a highly significant re- lationship between body size and otolith size for fish col- lected in 1996 (P<0.000, r2=0.97) and 1997 fP<0.000, r- = 0.96). Juveniles selected for the otolith marking experi- ment ranged from 31.4 to 44.6 mm SL (mean=38.4 mm, standard error=0.63 mm). Only 21 out of 30 surviving alizarin-complexone-marked juveniles had otoliths with visible marks and no fish harvested on day 21 had read- able marks. Mean growth rate for survivors was estimat- ed at 0.33 mm/d. The slope (6=0.87) of the regression of the number of increments counted after the alizarin mark on the number of days juveniles were held after marking was not significantly different from 1.0 (Fig. 41 (^test, P<0.05); therefore we did not reject the hypoth- esis that increments were deposited daily. Instantaneous daily growth of juvenile snapper was 1.02 mm/d in 1996 compared with 0.60 mm/d in 1997 ( Fig. 5). AN OVA indicated a significant difference in slopes (i.e. growth rates) between years (P<0.001). However, ANCOVA indicated that there was no significant differ- ence in growth rate among regions. Because there was a significant difference in standard lengths between years and regions, only those size ranges that overlapped be- tween the two regions (i.e. Panhandle and Southwest, 1996: 29-83 mm SL, 1997: 39-112 mm SL) and between years (i.e. 32-110 mm SL) were included in the analysis. Fertilization-date distribution There were distinct patterns in the temporal and spatial distribution of fertilization dates. Our results however provided little support for lunar periodicity in spawn- 7 14 21 28 Days after mark Figure 4 Linear regression of increment count on number of days after alizarin marking of the lapillus. 140 120 100 ■g' 80 g W 60- 40 - 20 - 0 - Linea 1996 y= 10236X- 20.91 8, r = 0,88 1998 y = 0.6014X + 5.5573, r' = 0 79 ♦♦ i ♦♦ 01996 ♦ 1997 D 20 40 60 80 100 120 140 160 180 200 Age (d) Figure 5 - regression of SL (mml on age (d) for 1996 and 1997. ing. Back-calculated fertilization dates indicated that surviving juveniles in the Panhandle and Big Bend were mainly summer spawned fish, whereas Southwest juve- niles had winter and summer fertilization dates. In 1996, the earliest back-calculated birth date was 25 February from a fish from the Southwest, although a few (i.e. six fish) were recorded from late April and early May from the Panhandle and a few more by mid-May from the Southwest (Fig. 6). Most fish spawned beginning in mid- June and a peak in fertilization dates occurred during mid-July for both regions. Frequency of spawning then declined and the latest fertilization date of the year was recorded in mid-September from the Southwest. A chi- Allman and Grimes: Spawning, settlement, and growth of Lut/anus gnseus from tfie West Florida sfielf 397 M ■ Southwest 25 D Panhandle 20 - 15 - 10 1 5 0 - n 1996 2/20 3«1 4/20 5/20 6/19 7/19 8/18 9/17 30 25 20 15 10 5 n Panhandle B Big Bend ■ Southwest 1997 Jlii iJn, n 11/11 12/11 1/10 2/9 3/11 4/10 5/10 6/9 7/9 8/8 9/7 Fertilization date Figure 6 Fertilization-date distribution for 1996 and 1997. square test revealed only a marginal difference (P<0.10) from a uniform distribution in fertilization dates across lunar month. As in 1996, fertilization in 1997 began in mid-June, peaked in early August, and included birth dates from all three regions; however, the distribution for 1997 was bi- modal. Spawning began in November of 1996 and peaked in late January of 1997, then declined into April. With the exception of two individuals collected in late February from the Panhandle, all individuals from the first fertiliza- tion-date peak were from the Southwest. The last back- calculated fertilization date for the year was in early Sep- tember from the Panhandle. Fertilization dates for 1997 did not conform to a lunar cycle because a chi-square test indicated no significant difference from a uniform fertil- ization-date distribution across lunar month. Results from the reproductive condition of adult gonads were compared with results from back-calculated fertiliza- tion dates from otoliths to determine if survivingjuveniles were derived from a subset of the original propagules. We compared otolith data with oocyte diameters because large ova are indicative of ripe females (i.e. >0.59 mm maximum oocyte diameter) (Davis and West, 1993). In 1996, the maximum oocyte diameter distribution indicated that fe- males were in spawning condition from May to September and that a peak in spawning activity occurred in mid-July (Fig. 7). Spawning activity, signified by the presence of hy- drated oocytes, also occurred from May through mid-Sep- tember. Only one female with postovulatory follicles was found in 1996, collected in mid-July. Male spawning times (i.e. May-September) mirrored those of females; however, only five males contained ripening spermatids, and none contained spermatazoa in 1996. All gonads collected in 1996 were from the Panhandle. 40 -, 30 20- 10 ~ 0 - aM 1 1 1 \ 1 1 \ r 1/10 2/29 4/19 6/8 7/28 9/16 11/5 12/25 08- 0 7. 06- 0.5- 04- 0 3- 0,2- 0 1- 00- B — I 1 1 1 1 1 r 1/10 2/29 4/19 6/8 7/28 9/16 11/15 12/1 5 - 3 - 2- + F otvl ■M-l-M-MH-*»f HO 00«» ««C»4» «» O 0« 1— I #0 ++++++»+♦ T 1 1 1 1 1 1 T 1/10 2/29 4/19 6/8 7/28 9/16 11/5 12/25 Sampling date Figure 7 Comparison of spawning time as determined by back-calculation from otoliths (A), maximum oocyte diameter (B) and histological stage (C) for 1996. All fish were collected from the Panhandle. In 1997 gonad data were available from all three regions and the results of comparisons of gonads from the three regions may indicate the presence of juveniles that were not derived from local spawning. Gonad data for all regions indicated peak spawning during mid-July (Fig. 8). Gonad data from the Panhandle suggested spawn- ing times similar to those for the otolith data, i.e. a May through September spawning period and a summer peak. However, the back-calculated fertilization dates from late February and early March did not correspond to oocyte diameter or histological-stage data. Very few juveniles 398 Fishery Bulletin 100(3) Pdluile Bqtetl 01 1 — I — I — I — I — I — I — r V10 31 433 69 7® ST? 116 1355 1 I I I I I I r V10 31 4^99 /cggfi/iwiaa 0,9- 0,8- 0 7- 06- 05- 0,4- 03- 02- Q1- 00-. B a7- 06- as- 04- 03- 02- 1 — I — I — I — I — I — I — r 1/10 31 420 99 7/29 ff17 11/6 1226 01- 1 — I — I — I — r 1/10 3'1 «aO 59 7/29 9'17 11« 122E a) ^ 3- CO ' tt* m**** a s« t 00^ — I 1 1 1 1 1 T 1/10 3"! 4SD 69 7/29 9171 1,6 1336 — I 1 1 1 1 \ r 1/10 31 423 OT 7/29 9/17 11/6 13B6 SQifvest Di T — i — i — i — I — i — i — r IVI 1231 31 43062992810271326 10- 09- o,a 07- Q6- Q5_ Q4_ Q3- 02- 01- 11/1 1231 311 *X 529 928 10^1326 • » «*«aM * I I I I I I I r 1V1123131 430 629 92B 10271326 Sampling date Figure 8 Comparison of spawning time as determined by back-calculation from otoliths (A), maximum oocyte diameter (Bl and histological stage (Cl for 1997. or adult fish were collected from the Big Bend; however, these data also suggested a mid-summer spawning period. Oocyte diameter and developmental stage data from the Southwest indicated summer spawning (May to Septem- ber) but because no adult gonads were collected earlier in the year, a winter fertilization-date mode (as back-calcu- lated from otoliths) could not be determined. Larval duration Presumed settlement marks were distinguishable in some, but not all otoliths that were assigned ages. These marks were identified by a sharper contrast than preced- ing marks (Fig. 9). In addition, at this mark the increment pattern changed abruptly, postsettlement increment widths being consistently narrower than those laid down before settlement. Of the 135 otoliths collected during 1996 which were assigned ages, 80 (59%) had readily dis- tinguishable settlement marks, whereas in 1997, 62 (64%) had distinguishable marks. The estimated age at settle- ment (planktonic larval duration) for 1996 and 1997 fish ranged from 20 to 32 days (niean=25 days) and from 20 to 33 days (mean=24.7 days), respectively. ANOVA indicated no significant sampling region or year effect or region- and-year interaction on the age at settlement (planktonic larval duration). In 1996, settlement first began in mid May in the Pan- handle and continued throughout the summer, peaking during the first week of August in the Panhandle and Southwest, and ending finally on 1 October in the South- west (Fig. 10). Results for settlement date data showed no evidence for lunar periodicity in settlement, a chi-square Allman and Grimes: Spawning, settlement, and growth of Lutianus griseus from Ifie West Florida sfielf 399 test indicating that the distribution of hinar settlement dates was not sifjnificantly differ- ent from a uniform distribution. The back-cal- culated settlement date distribution for 1997 indicated two distinct settlement events. Set- tlement of the first mode began in December of 1996, peaked in late winter, and then declined into late spring. All juveniles from the first event were from the Southwest. The second settlement mode was similar to that seen in the 1996 settlement date distribution, begin- ning in late June, peaking in early August, and finally ending in the early fall. The evidence for lunar periodicity in settlement in 1997 was weak: a chi-square test indicated only a mar- ginally significant difference (P<0.10) from a uniform distribution across lunar month for 1997 settlement dates. Discussion Sampling and collection # isKo Sciilcnient mark Figure 9 Photomicrograph (400x) of settlement mark in a lapillus of gray snapper. The collection of juvenile gray snapper from June to November in 1996 was consistent with a May-September spawning period found from previous studies in which the reproductive condition of adults collected off South Florida was used (Starck, 1971; Domeier et al., 1996). However, in 1997 juvenile gray snapper were initially collected in April from southwest Florida, indicating a much earlier spawning time. Differences in the timing of the first appearance of juvenile snapper for 1997 could be related to the seasonal cycle of seagrass dieback and regeneration. Thalassia experi- ences leaf die off with temperatures below 15°C (Zimmerman and Livingston, 1976). In the northern most part of the sampling region, greater leaf die off and slower growth during the spring would be expected compared with more southerly areas. This would most cer- tainly limit the chances of survival of early settling larvae in the north because of greater predation risk and lower food availability. If gray snapper are truly recruitment-lim- ited and if postsettlement mortality is rela- tively constant (Shulman and Ogden, 1987) and individuals do not migrate out of the area during the monitoring period (Robertson et al., 1988), ju- venile abundance would be an effective predictor of adult abundance two to three years later when the adults are entering the fishery. Gag (Mycteroperca microlepis), a spe- cies with a similar early life history to that of the gray snapper, have a low rate of emigration from the seagrass nursery area during summer and a mortality rate of less than 1%/d (Koenig and Coleman, 1998). Johnson and Koenig (in press) found a strong correlation between ju- venile gag abundance in seagrass meadows and year-class 25 20 - 15 10 ■ Southwest n Panhandle -| 1996 r-i , nin .ri 1 -JU- 5/5 5/20 6/5 6/20 7/5 7/20 8/5 8/20 9/5 9/20 10/5 1997 20 15 10 ■ Southwest D Panhandle BBig Bend LiJjL I ^ n 12/5 1/5 2/5 3/5 4/5 5/5 6/5 7/5 8/5 9/5 10/5 Settlement date Figure 10 Settlement-date distribution for 1996 and 1997. strength in the fishery several years later. Juvenile indices have been used successfully to document recruitment for a number of temperate fish species and invertebrates. For example, juvenile indices for striped bass were found to ex- plain 83% of the variation in reported landings from 1963 to 1983 (Goodyear, 1985). These data for gray snapper could provide a fishery-independent method for forecast- ing size of year classes several years into the future. This would be a valuable addition to stock assessment which typically features age-based hind-casting methods such as 400 Fishery Bulletin 100(3) virtual population analysis from fishery-dependent data. However, before these data can be used reliably there must be a demonstrated relationship between juvenile abundance and year-class strength. Age and growth Because of the strong linear relationship between otolith size and fish size and the regression of increments depos- ited after alizarin marks on days after marking, we did not reject the hypothesis that one increment was depos- ited daily. In addition, Ahrenholz (2000) found no signifi- cant difference from 1 increment/d deposited from lapilli of alizarin-marked juvenile gray snapper Juvenile con- generic red snapper (Lutjanus campechaniis) have also been shown to deposit daily increments above a minimum growth rate (i.e. >0.3 mm/d FL) (Szedlmayer, 1998). The daily growth rate for juveniles varied between sam- pling years (1 mm/d in 1996 and 0.6 mm/d in 1997) and was similar to growth rates for juvenile red snapper (0.54— 0.86 mm/d, Szedlmayer and Conti, 1999). The difference in growth between years could be explained by the fact that most juveniles collected in 1997 were winter-spawned fish collected from SW Florida that may have experienced lower temperatures and that initially had a slower growth rate. Variability in the growth of larval and juvenile fish has been linked to water temperature in many species (Lang et al., 1994; Nixon and Jones, 1997). However, when the same size range of juveniles was compared between regions, there was no significant difference in growth rate. Similarly there was no difference in regional growth rate of co-occurring juvenile gag from the west Florida shelf (Fitzhugh-'). Fertilization-date distribution All indicators of the temporal distribution of spawning (i.e. juvenile otoliths, oocyte diameter, and histological stage of gonads) indicated peaks in spawning from May until August and a dominant peak in mid-July Very few gray snapper gonads were collected during the winter months in 1996 and none in 1997 because the majority of gray snapper are landed in July and August (U.S. Dep. Commer.'') when they aggregate on offshore reefs to spawn and are probably easier to catch (Starck, 1971; Domeier et al., 1996; Domeier and Colin, 1997). Spawning time has been related to both increasing water temperature and photoperiod for other lutjanids (Arnold et al., 1978; Grimes and Huntsman. 1980). No obvious temporal differences were observed in repro- ductive development between regions from adult gonads; however, we determined winter back-calculated fertiliza- ■' Fitzhugh, G. F. 2000. Unpubl. data. National Marine Fish- eries Service, 3.500 Delwood Beach Road, Panama City, Florida 32408. ■^ U.S. Department of Commerce. 1998. Fisheries of the United States, 1997. Current fishery statistics no. 9700, 1.56 p. Sta- tistics Division, Rm. 12340, 1315 East-West Highway, Silver Spring, MD 20910-3282. tion dates for the most southerly part of the sampling region in 1997. In contrast, Domeier et al. (1996) back- calculated spawning dates from June through September using the otoliths from juvenile gray snapper collected from south Florida. Gray snapper larvae have a relatively long pelagic stage of 25 days. The possibility exists that winter-spawned fish were produced from outside the Gulf of Mexico, perhaps from a nearby insular population, and transported from there to settle along the southwestern part of the Florida shelf The view that winter-spawned fish were Gulf of Mexico expatriates is supported by Grimes's (1987) review of lutjanid reproduction. He found two patterns: 1 ) continental populations, with extended summer spawning periods; and 2) insular populations, which spawned year-round with pulses in the spring and fall. However, Domeier et al. (1996) suggested after field observations that snapper spawning patterns are species specific, as opposed to habitat dependent. We collected winter-spawned juveniles in only one of the two years of sampling; therefore it is unknown how frequently expatri- ate settlement events occur and what their importance to gray snapper recruitment might be. We could not conclusively demonstrate that spawning occurs in association with lunar cycle peaks. Fertilization date hind-cast from otoliths in 1996 were only marginally associated with a lunar cycle and no relationship was evi- dent in 1997 data. Similarly gonad stage data from adult gray snapper did not reveal any lunar pattern. However, the association of spawning with the lunar cycle has been reported for congeneric species, as well as other conspecific populations. Domeier et al. ( 1996) used a GSI to determine that gray snapper spawning peaked around the time of the new and full moon. Upon examination of adult gonads, Starck (1971) speculated that gray snapper spawned dur- ing or near the full moon. The congeneric species Lutjanus vaigiensis spawned about the time of the full moon (Ran- dall and Brock, 1960) and Lutjanus kasmira spawned dur- ing both new and full moons in the laboratory (Suzuki and Hioki, 1979). Larval duration Presumed settlement marks were noted in many (60%) of the ageable otoliths, and they were similar in appearance to those observed for several other reef fish species ( Broth- ers and McFarland, 1981; Victor, 1982 and 1986; Sponau- gle and Cowen, 1994). Settlement marks are thought to be associated with morphological and ecophysiological changes which occur during the transition from a plank- tonic to a benthic life stage (Brothers and McFarland, 1981; Victor, 1982; Brothers, 1984). Several lines of evidence sug- gest that we correctly identified settlement marks. On average, planktonic duration, or length of the presettle- ment life history phase (i.e. difference between fertilization and settlement dates ) was 25 days for both sampling years. Similarly, Koenig and Domeier^ reported the planktonic 6 Koenig, C. C, and M. L. Domeier. 1993. Unpubl. data. De- partment of Biological Sciences, The Florida State Univ., Talla- hassee, FL 32306-2043. Allman and Grimes: Spawning, settlement, and growth of Lui/anus gnseus from \he West Florida sfielf 401 lar\'al duration to be 24 days (21 d + 3d correction) from otoliths of wild juvenile fish collected off the Florida Keys. Richards and Saksena ( 1980) recorded settlement for three laboratory-reared gray snapper that were 26, 26, and 36 days old respectively. Thus, gray snapper appear to have a planktonic duration similar to the majority of reef fish (i.e. from 20 to 30 days [Victor, 1991]). There was no obvious difference in the age at settlement among regions or years. For 1996 and 1997, the temporal distribution of settlement was about the same in the north- ern and southern parts of the west Florida shelf In 1997, birth dates and corresponding settlement dates were re- corded during the winter months for only the most south- ern sampling region. Settlement date was marginally asso- ciated with the new moon in 1997, possibly reflecting high transport into seagrass meadows at that time. Shenker et al. (1993) found a correlation between onshore transport of settlement-stage Nassau grouper (Epinephelus striatus) and the new moon. However. Smith ( 1995) found no signifi- cant association with recruitment of gray snapper through Sebastian Inlet, Florida, and lunar cycle. Conclusions This study established that there were differences in the timing of spawning and settlement of gray snapper across the West Florida shelf Juveniles collected from south- west Florida indicated earlier spawning times than previ- ously reported. Additional year-round sampling of juvenile gray snapper would help to identify the influence of early spawning events on recruitment. Further study of larval transport and genetic markers could aid in the identifica- tion of source populations. Acknowledgments We wish to thank Alan Collins for assistance in the col- lection and staging of gonads, Gary Fitzhugh for provid- ing guidance on otolith preparation and reading, William Fable, Todd Bevis, and Andrew David for helping to collect fish. This study was supported partially by a MARFIN grant (no. NA57FF0055) awarded to Christopher Koenig and Felicia Coleman of the Florida State University Insti- tute for Fisheries Resource Ecology. Joseph Travis, Doug- las DeVries, Kenyon Lindeman, and three anonymous reviewers provided constructive comments on previous versions of this manuscript. Literature cited Ahrenholz, D. W. 2000. Periodicity of growth increment formation in otoliths of juvenile gray snapper iLutjanus griseus) and lane snap- per (Z-u^anus synagns). J. Elisha Mitchell Sci. Soc.ll6(3): 251-259. Arnold, C. R., J. M. Wakeman. T. D. Williams, and G. D. Treece. 1978. Spawning of red snapper [Lutjanus campechanus) in captivity. Aquaculture 15:301-302. BioScan, Inc. 1990. BioScan Optimas 3.0. BioScan, Inc., Edmunds, WA. Brothers, E. B. 1984. Otolith studies. In Ontogeny and systematics of fishes (H. G. MosiT.W. J. Richards. D. M. Cohen, M. P. Fahay, A. W. Kendall Jr, and S. L. Richardson, ed.) p. 50-57. Am. See. Ichthyol. Herpetol. Spec. Publ. 1. Brothers, E. B., and W. N. McFarland. 1981 . Correlations between otolith microstructure,growth,and life history transitions in newly recruited French grunts [Hae- mulon flavolineatum (Desmarest), Haemulidae]. Rapp.P.-V. Reun. Cons. Perm. Int. Explor Mer 178:369-372. Campana, S. E. 1992. Measurement and interpretation of the microstruc- ture of fish otoliths. In Otolith microstructure examina- tion and analysis (D. K. Stevenson and S. E. Campana, eds.), p. 59-71. Can. Spec. Publ. Fish. Aquat. Sci. 117. Chester, A. J., and G. W. Thayer 1990. Distribution of spotted seatrout I Cynoscion nehulo- sus) and gray snapper ^Lutjanus gnseus) iuwemXes in sea- grass habitat of western Florida Bay. Bull. Mar Sci. 46(2): 345-357. Collins, L.A.. A. G. Johnson, and C. P. Keim. 1996. Spawning and annual fecundity of the red snapper iLut- Janus campechanus) from the northeastern Gulf of Mexico. In Biology, fisheries and culture of tropical groupers and snap- pers (F. Arreguin-Sanchez, J. L. Munro, M. C. Balgos, and D. Pauly, eds. ), p. 174-188. ICLARM Conf Proc. 48. Davis, T L. O., and G. J. West. 1993. Maturation, reproductive seasonality, fecundity and spawning frequency in Lu(/an!;s(u/?us(Quoy and Gaimard) from the North West shelf of Australia. Fish. Bull. 91: 224-236. Domeier, M. L., and P. L. Colin. 1997. Tropical reef fish spawning aggregations: defined and reviewed. Bull. Mar Sci. 60(3):698-726. Domeier, M. L., C. C. Koenig, and F. C. Coleman. 1996. Reproductive biology of the gray snapper iLutjanus griseus ), with notes on spawning for other Western Atlantic snappers (Lutjanidae). In Biology and culture of tropical groupers and snappers (F. Arreguin-Sanchez. J. L. Munro, M. C. Balgos. and D. Pauly, eds.), p. 189-201. ICLARM Conf Proc. 48. Frank, K. T, and W. C. Leggett. 1994. Fisheries ecology in the context of ecological and evo- lutionary theory. Annu. Rev. Ecol. Syst. 25:401-422. Goodyear, C. P. 1985. Relationship between reported commercial landings and abundance of young striped bass in Chesapeake Bay, Maryland. Trans. Am. Fish. Soc. 1 14:92-96. Grimes, C. B. 1987. Reproductive biology of the Lutjanidae: a review. In Tropical snappers and groupers: biology and fisheries man- agement (J. J. Polovina and S. Ralston, eds.), p. 239-294. Westview Press, Inc., Boulder, CO. Grimes, C. B., and G. R. Huntsman. 1980. Reproductive biology of the vermilion snapper, Rhom- hoplites aurorubens, from North Carolina and South Caro- lina. Fish. Bull. 78:137-146. Hoese, H. D., and R. H. Moore. 1977. Fishes of the Gulf of Mexico, Texas, Louisiana, and adjacent waters. Texas A&M Univ. Press, College Station, TX. Houde, E. D. 1987. Fish early life dynamics and recruitment variability. Am. Fish. Soc. Sym. 2:17-29. 402 Fishery Bulletin 100(3) Johnson, A. G., and C. C. Koenig. In press. Age and size structure of the fishery and juvenile abundance of gag, Mycteroperca microlepis, from the north- eastern Gulf of Mexico. Proceedings of the 47th Gulf and Caribbean Fisheries Institute. Koenig, C. C, and F. C. Coleman. 1998. Absolute abundance and survival of juvenile gags in sea grass beds of the northeastern Gulf of Mexico. Trans. Am. Fish. Soc. 127:44-55. Lang, J. B., and C. D. Buxton. 1993. Validation of age estimates in sparid fish using fluro- rochrome marking. S. Afr. J. Mar. Sci. 13:195-203. Lang, K. L., C. B. Grimes, and R .F Shaw. 1994. Variation in the age and growth of yellowfin tuna larvae, Thunnus albacares, collected about the Mississippi River plume. Environ. Biol. Fish. 39:259-270. Lindeman, K C. 1997. Development of grunts and snappers of southeast Flor- ida: cross-shelf distributions and effects of beach manage- ment alternatives. Ph.D. diss., Univ. Miami, Coral Gables, FL,419p. Lindeman, K.C., R. Pugliese, G. T. Waugh and J. S. Ault. 2000. Developmental patterns within a muhispecies reef fishery: management applications for essential fish habi- tats and protected areas. Bull. Mar Sci. 66(3):929-956. Lyons, W. G. 1979. Molluscan resources of the west Florida shelf Bull. Am. Malacol. Union Inc. 37-40. Lyons, W. G., and S. B. CoUard. 1974. Benthic invertebrate communities of the eastern Gulf of Mexico. In Proceedings on marine environments, impli- cations of offshore drilling in the eastern Gulf of Mexico (R. E. Smith, ed.), p. 157-165. Conference-Workshop, State University System, Florida Institute of Oceanography. St. Petersburg. FL. Moe, M. A., Jr 1969. Biology of the red grouper (Epinephelus morio Valen- ciennes) from the eastern Gulf of Mexico. Prof Pap. Ser., Fla. Mar Lab. 10, 95 p. Murah, R .S. 1982. Zero-energy coast. In The encyclopedia of beaches and coastal environments (M. L. Schwartz, ed.), p. 883. Hutchinson Ross Publ. Co., Narberth, PA. Nixon, S. W., and C. M. Jones. 1997. Age and growth of lan-al andjuvenile Atlantic croaker. Micropogonias undulatus, from the Middle Atlantic Bight and estuarine waters of Virginia. Fish. Bull. 95:773-784. Price, W. A. 1954. Shorelines and coasts of the Gulf of Mexico. U.S. Fish Wildl. Serv. Fish. Bull. 55:39-65. Randall, J. E., and V. E. Brock. 1960. Observations on the ecology of epinepheline and lut- janid fishes of the Society Islands, with emphasis on food habits. Trans. Am. Fish. Soc. 89:9-16. Richards, W. J., and V. P. Saksena. 1980. Description of larvae and early juveniles of laboratory reared gray snapper, Lutjanus griseus (Linnaeus) (Pisces, Lutjanidae). Bull. Mar Sci. 30:515-521. Robertson, D R., D. G. Green, and B. C. Victor 1988. Temporal coupling of reproduction and recruitment of larvae of a Caribbean reef fish. Ecology 69:370-381. Secor, D. H., J. M. Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstuctural examination. In Otolith microstructure examination and analysis (D. K. Stevenson and S. E. Campana, eds. ), p. 19-57. Can. Spec. Publ. Fish. Aquat. Sci. 117. Shenker, J. M., E. D. Maddox, E. Wihinski, A. Peari, S. R. Thorrold, and N. Smith. 1993. Onshore transport of settlement-stage Nassau grou- per Epinephelus striatus and other fishes in Exuma Sound Bahamas. Mar. Ecol. Prog. Ser 98:31-43. Shepard, J. G., and D. H. Gushing. 1990. Regulation in fish populations: myth or mirage? Phil. Trans. R. Soc. (Lond.) 330:151-164. Shulman, M. J., and J. C. Ogden. 1987. What controls tropical reef fish populations: recruit- ment or benthic mortality? An example in the Caribbean reef fish Haemulon flavolineatum , Mar Ecol. Prog. Ser 39:233-242. Smith, G. B. 1976. Ecology and distribution of eastern Gulf of Mexico reef fishes. Fla. Mar Res. Publ. 19, 78 p. Smith, S. L. 1995. Recruitment of larval snappers (family Lutjanidae) through Sebastian Inlet, Florida. M.S. thesis, Florida Institute of Technology, Melbourne, FL, 48 p. Sponaugle, S., and R. K. Cowen. 1994. Larval durations and recruitment pattens of two Caribbean gobies (Gobiidae): contrasting early life histories in demersal spawners. Mar Biol. 120:133-143. Starck, W. A., II. 1971. Biology of the gray snapper, Lutjanus griseus (Lin- naeus), in the Florida Keys. In Investigations on the gray snapper, Lutjanus griseus, studies in tropical ocean- ography, no. 10 (W. A. Starck II and R. E. Schroeder, eds.), p. 11-150. Univ. Miami Press, Coral Gables, FL. Suzuki, K., and S. Hioki. 1979. Spawning behavior, eggs and larvae of the lutjanid fish, Lutjanus kasmira, in an aquarium. Jpn. J. Ichthyol. 26:161-166. Szedlmayer, S. T. 1998. Comparison of growth rate and formation of otolith increments in age-0 red snapper J. Fish Biol. 53: 58-65. Szedlmayer, S. T, and J. Conti. 1999. Nursery habitats, growth rates, and seasonality of age-0 red snapper. Lutjanus campechanus, in the northeast Gulf of Mexico. Fish. Bull. 97:626-635. Tsukamoto, K., H. Kuwada, J. Hirokawa, M. Oya, S. Sekiya, H. Fujimoto, and K. Imaizumi. 1989. Size-dependent mortality of red sea bream, Pagrus major, juveniles released with fluorescent otolith-tags in News Bays, Japan. J. Fish Biol. 35:59-69. Victor, B. C. 1982. Daily otolith increments and recruitment in two coral- reef wrasses, Thalassoma bifasciatum and Halichoeres bivittatus. Mar Biol. 71:203-208. 1986. Duration of the planktonic larval stage on one hundred species of Pacific and Atlantic wrasses i family Labridae). Mar Biol. 90:317-326. 199 1 . Settlement strategies and biogeography of reef fishes. In The ecology of fishes on coral reefs ( P. F. Sale, ed. ), p. 231- 260. Academic Press, Inc., San Diego, CA. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am.Zool. 21:.325-343. Wellington, G. M., and B .C. Victor 1989. Planktonic larval duration of one hundred species of Pacific and Atlantic damselfishes (Pomacentridae). Mar Biol. 101:557-567. Wooster W. S., and K. M. Bailey 1989. Recruitment of marine fishes revisited. Effects of ocean Allman and Grimes: Spawning, settlement, and growth of Lut/anus gnseus from tfie West Florida shelf 403 variability on rocruitinent and an evaluation of param- eters used in stock assessment models. Canadian Special Publication of Fisheries and Aquatic Science 108:153- 159. Zar, J. H. 1984. Circular distributions: descriptive statistics, /n Bio- statistical analysis (J. H. Zar, cd.), p. 422-439. Prentice- Hall, Inc En{,'lewood Cliffs, NJ. Zieman, J. C, and R. T. Zieman. 1989. The ecology of the seagrass meadows of the west coast (if Florida: a community profile. U.S. Fish Wildl. Serv. Biol. Hep. 85(7.25), 155 p. Zimmerman, M. S., and R. J. Livingston. 1976. Seasonality and physico-chemical ranges of benthic macrophytes from a north Florida estuary (Apalachee Bay). Contrib. Mar Sci. 20:34-45. 404 Abstract— Light traps and channel nets are fixed-position devices that in- volve active and passive sampling, re- spectively, in the collection of settle- ment-stage larvae of coral-reef fishes. We compared the abundance, taxo- nomic composition, and size of such larvae caught by each device deployed simultaneously near two sites that dif- fered substantially in current velocity. Light traps were more selective taxo- nomically. and the two sampling de- vices differed significantly in the abun- dance but not size of taxa caught. Most importantly, light traps and channel nets differed greatly in their catch efficiency between sites: light traps were ineffective in collecting larvae at the relatively high-current site, and channel nets were less efficient in collecting larvae at the low-current site. Use of only one of these sampling methods would clearly result in biased and inaccurate estimates of the spatial variation in larval abundance among locations that differ in current velocity. When selecting a larval sampling de- vice, one must consider not only how well a particular taxon may be repre- sented, but also the environmental con- ditions under which the device will be deployed. Current velocity and catch efficiency in sampling settlement-stage larvae of coral-reef fishes Todd W. Anderson Department of Zoology Oregon State University Corvallis, Oregon 97331-2914 Present address: Department ol Biology San Diego State University San Diego, California 92182 E-mail address toddam'sunstroke sdsu edu Claudine T. Bartels Department of Biological Sciences Flonda Institute of Technology 150 West University Boulevard Melbourne. Florida 32901 Mark A. Hixon Department of Zoology Oregon State University Corvallis, Oregon 97331-2914 Erich Bartels Department of Biological Sciences Florida Institute of Technology 150 West University Boulevard Melbourne, Flonda 32901 Mark H. Carr Department of Ecology and Evolutionary Biology University of California Santa Cruz, California 95064 Jonathan M. Shenker Department of Biological Sciences Florida Institute of Tecfinology 150 West University Boulevard Melbourne, Flonda 32901 Manuscript accepted 23 January 2002. Fish. Bull. 100:404-413 (2002). " With few exceptions, benthic marine organisms exhibit a complex life cycle in which larvae are dispersed pelagi- cally and undergo a planktonic exis- tence before they settle to the sea floor. The vagaries of dispersal by currents and unpredictable mortality of larvae in the plankton contribute to tremen- dous variation observed in recruitment (here defined broadly as the input of young) to adult populations. For reef fishes, the relative importance of varia- tion in the supply of settlement-stage larvae versus postsettlement density- dependent mortality of recruits to the size and structure of local populations has been debated for some time, with evidence provided for both points of view (Doherty and Williams, 1988; Doherty, 1991; Hixon, 1991; Jones. 1991; Doherty and Fowler, 1994; Hixon and Carr, 1997; Schmitt and Holbrook, 1999). In order to determine the rela- tive contributions of presettlement and postsettlement processes to the popula- tion dynamics of reef fishes, knowledge of spatial and temporal variation in the delivery of settlement-stage larvae to local populations is essential. Various methods have been employed to assess the local abundance of pelagic larvae and juveniles of coral-reef fishes (Choat et al., 1993, and references there- in). Two common but relatively recent sampling devices developed to sample settlement-stage reef fish larvae are light traps (Doherty 1987; Thorrold and Milicich 1990; Milicich et al. 1992; Thorrold. 1992; Meekan et al., 1993; Thorrold, 1993; Milicich and Doherty 1994; Doherty et al., 1996; Sponaugle and Cowen, 1996a, 1996b; Thorrold and Williams, 1996; Doherty and Carleton, 1997; Sponaugle and Cowen, 1997; Leis et al., 1998; Munday et al. 1998; Meekan et al. 2000) and stationary nets, includ- ing channel nets (Shenker et al.,1993; Thorrold et al., 1994a, 1994b, 1994c) and crest nets (Dufour and Galzin, 1993; Dufour et al., 1996). Light traps are active sampling devices in that fish are attracted to, swim towards, and enter a transparent or semitransparent trap with a light source. By contrast, channel nets and crest nets passively catch larvae that are carried into them by currents or wave action, respectively. Unlike a previous study that compared light traps with towed nets (Choat et al., 1993), the goal of this study was to compare light traps and fixed-position nets, two commonly used methods for collecting settlement-stage fish larvae near coral reefs. Anderson et al : Current velocity and catch efficiency in sampling settlement-stage larvae of coral-reef fishes 405 76° 09' 76° 07' 76° 05' rA ^ (\ Bock T T 76° 03' 1 ^^. ^ » So. Bock ^ Adderly <., *>^ Co/ Norman's Pond Coy Sh«i( Edoe <9, . "^^ "<>, , '^-f 0 1000 2000 ■ Soolein mefere N t Lee Stocking Island V »SE. LSI Children's Bo/ Cay Rat .<^_ 23° 46' 23° 45' Figure 1 Location of South Bock Cay (So. Bock) and southeast Lee Stocking Island (S.E. LSI) study sites near the Caribbean Marine Research Center (CMRC), Lee Stocking Island, Bahamas. Light traps and channel nets may not only differ in cap- ture success for any particular species, but in the collection of larvae under varying environmental conditions such as current velocity and turbidity. Species (and the ontogenetic stages of these species) can differ in their sensitivity to light cues (Doherty, 1987; Choat et al., 1993), in swimming ability (Stobutzki and Bellwood, 1994; Leis et al., 1996; Stobutzki and Bellwood, 1997; Wolanski et al., 1997), and in the ability of larvae to interact with current velocity. For example, at relatively high current velocities, settlement- stage larvae that are photopositive but that have limited swimming speeds may not have the ability to respond to a light cue and swim into a light trap as they are carried past the sampling device. Conversely, larvae with strong swim- ming abilities may be able to avoid capture by channel nets at lower current velocities. Thus the relationship between the sensitivity to light cues and the ability to respond to such cues by larvae (determined by developmental stage, swimming ability, and current velocity) should determine the relative catch efficiency of these sampling devices. Here we compare and contrast the relative number, taxo- nomic composition, and size of settlement-stage fish larvae caught in light traps and in channel nets deployed at two reefs that differ substantially in current velocity. Because these devices collect larvae either actively or passively, we examined 1) whether light traps are more selective and catch fewer taxa (families) than channel nets, 2) whether light traps catch larger settlement-stage larvae than chan- nel nets, and 3) with higher current velocity, whether the relative effectiveness of light traps to channel nets de- creases, resulting in fewer taxa and a lower abundance of settlement-stage larvae in light traps (with lower current velocity the opposite is true). The relative abundance and taxonomic composition of larvae caught by these two sam- pling devices, and how they may be modified by current ve- locity, might result in different interpretations concerning both magnitude and variation in larval supply. Methods Study sites We conducted this study using light traps and channel nets deployed each day, from 30 July through 11 August 1997 at Lee Stocking Island (LSI), Bahamas, bracketing the new moon phase in the lunar cycle when settlement- stage fish larvae are more abundant (Thorrold et al., 1994b). Two fore-reef sites were selected a priori for study based upon our impression of marked differences in cur- rent velocity: South Bock Cay (So. Bock) northwest of LSI and southeast LSI (S.E. LSI), approximately 7 km south of So. Bock. So. Bock is near a channel between two cays and experiences moderate to strong tidally driven currents, whereas S.E. LSI is far from a channel where there is relatively low current velocity (Fig. 1). 406 Fishery Bulletin 100(3) Light traps We deployed three light traps at each site on each day, from 30 July to 11 August 1997. Each of the three traps at a site was a different modification from two basic designs, with each trap rotated among stations within a site to account for any trap- or location-specific biases. One trap design had a 65-cm diameter polycarbonate top and 1.5-mm nylon mesh enclosing a volume of 200 L, with four clear plastic funnels as fish entrance holes and a fluorescent light powered by nickel-cadmium rechargeable batteries (see Sponaugle and Cowen, 1996b). A second trap design was similar but had an 80-cm diameter wood top instead of a polycarbonate top, a larger volume of 360 L, twice the number of entrance holes, a fluorescent light powered by rechargeable sealed lead-acid batteries, and an automatic timer for turning on the light remotely. The third trap design was composed of a rectangular plexiglas trap (42 cm x 38 cm x 70 cm) with rigid plexiglas panels, a plastic tray, and four entrance slots, constructed to enclose a volume of 110 L based on the design of Munday et al. (1998). but having the same electronics design as that for traps made with polycarbon- ate tops. All light traps were placed 3^ m below the sea surface, suspended from moorings with subsurface buoys, and stationed approximately 12-15 m in front of each reef and 50-60 m from each other in a linear array along the offshore edge of the reef Traps were deployed between 1730 and 1830 h and retrieved the following morning between 0900-1030. All fishes (except for ubiquitous clupeids and atherinids) were collected and placed in vials of 70*^ or 95% ethanol, identified to family, and later measured with ver- nier calipers to the nearest 0.1 mm standard length (SL). Channel nets We deployed one surface (0-1 m depth) and one subsurface (2^ m depth) channel net at each site on each day (30 July to 11 August 1997) at a distance of approximately 50 m offshore of the center of the light-trap array. The channel nets were based on the design of Shenker et al. (1993), had a mesh size of 2 mm, and were positioned 30 m apart. The nets were suspended from surface buoys moored to concrete blocks or mooring anchors, allowing them to turn and fish both ebb and flood mixed-semidiurnal tides. The subsurface net mouth opening was 2 m wide x 2 m high. The mouth of each surface net was 2 m wide x 1 m high and was equipped with a General Oceanic Model 2030R2 flow meter and low-speed rotor blade suspended in the mouth opening. Flow-meter readings were recorded from the surface channel net to estimate relative current veloc- ity between the two sites. All nets were equipped with PVC (polyvinylchloride) rods along the length of the netting to prevent entanglement during slack tides, and the codends were constructed to sink and close the end of the net to contain fish larvae during times of very low current veloc- ity. Nets were deployed and retrieved at approximately the same time of day as the light traps. Channel nets were not sampled at dusk to distinguish catches during the day from the following night because previous studies indicated that daytime catches account for a very minor percentage of the total number of fish transported onto the Great Bahama Bank (Shenker et al., 1993; Thorrold et al., 1994c). At the laboratory, samples were rough-sorted to remove debris, fixed with 10% formaldehyde for 24—72 h, and then trans- ferred to 70% isopropyl alcohol for later identification. Fish were measured with vernier calipers to 0.1 mm SL. Analysis To compare the number of taxa collected by Ught traps and channel nets between sites and for both sites combined, we summed the total number of families caught by each sampling device to calculate family richness. We also used the Brillouin index of species diversity (Magurran, 1988) to compare the diversity of families of fish larvae between sampling methods and sites. This index is preferable to the Shannon-Weiner index because samples collected by light traps and channel nets are nonrandom; for example, light traps produce biased samples based on the sensitivity of species to a light cue. To compare the relative abundance of families caught by light traps and channel nets between sites and for both sites combined, we standardized catches in channel nets to the number of larvae per 1000 m-^, using flow meter read- ings recorded each day, and we standardized catches from light traps as the number of larvae per day We could not standarize catches to catch per unit of effort because the length of time that the lights were operational was vari- able and dependent on the type of light device (see heading "Light traps") and variance in battery life. Moreover, the volume over which light traps attract larvae is difficult to quantify, especially when external factors such as current velocity may largely affect catch rates (Thorrold, 1992; Meekan et al., 2000), and Meekan et al. (2000) suggested that it is useless to convert catch rates into densities. We used Spearman's rank correlation coefficient (Zar, 1984) to compare the relative abundance of taxa that repre- sented at least 1% of the catch (Cheat et al., 1993) for either sampling device. We also used this correlation coefficient to compare the relative abundance of taxa caught by surface and subsurface channel nets between sites for taxa that represented at least 1% of the catch for either net. In order to determine whether there were significant differences in the mean, median, and maximum length of families offish larvae between light traps and channel nets, we used a Wil- coxon paired-sample test (Zar, 1984) in which differences in length for each family caught by both sampling devices were ranked. Finally, we used a t-test to test for significant differ- ences in current velocity and the proportional abundance of total larvae caught by each sampling device between sites. Results Richness, diversity, relative abundance, and individual size According to our hypotheses in regard to active and pas- sive collection of larvae, both family richness and diversity would differ between light traps and channel nets. A total Anderson et al : Current velocity and catch efficiency in sampling settlement stage larvae of coral-reef fisties 407 Family Figure 2 Abundance of 25 families of larvae that constituted at least 1% of the catch by light traps (mean no. larvae per sampling date) or channel nets (mean no. larvae per 1000 m'') at (Al both study sites combined, (B) S.E. LSI (southeast Lee Stocking Island), and (C) So. Bock (South Bock Cay). Error bars are 1 SE. of 2111 larvae were collected from light traps (n=849; 78 samples) and channel nets (n=1262; 26 surface and 26 sub- surface samples), representing 20 and 33 famihes, respec- tively (Table 1), combining blenniids and labrisomids as blennioids, and not including clupeid and atherinid fishes. Synodontids were excluded from further analysis because of the presence of large postsettlement individuals in light traps. For families that constituted at least 1% of the total catch by at least one of these sampling methods, carap- ids, chaetodontids, gobiesocids, and holocentrids were not caught in channel nets. Conversely, carangids, chlo- psids, congrids, muraenids, ophichthids, ophidiids, and tetraodontids were not represented in light-trap samples. Channel nets had higher family richness and diversity at S.E. LSI, So. Bock, and at both sites combined (Table 2) than did light traps. There was no concordance in the rank order of abun- dance of families collected by light traps and channel nets at either S.E. LSI (r^=-0.358, n=25, P>0.05), So. Bock (Spearman's rank correlation coefficient: r, =-0.120, n=25, P>0.05), or at both sites combined (r^=-0.162, n=25, P>0.05), including all families that represented at least 1% of the catch of either sampling device (Fig. 2). Similarly, there was no correlation in rank abundance of 408 Fishery Bulletin 100(3) Table 1 Total number of fish larvae collected from light traps and channel nets at South Bock Cay and at southeast Lee Stocking Island from 30 July to 11 August 1997. The families Blenniidae and Labrisomidae are combined as Blennioidei. Collections from surface and subsurface channel nets each day have been standardized as no. /lOOO m-^. Percentage of total catch of larvae at each site for each family is denoted by (%) for light traps and for channel nets (surface and subsurface nets combined). Asterisks (*) denote families that represented at least 1% of the total catch by light traps or channel nets. Family South Bock Cay Southeast Lee Stockin g Island Light traps (%) Channel nets (%) Light traps (%) Channel nets {%) Surface Subsurface Surface Subsurface Acanthuridae* 10 (1.2) 0 4.0 (0.5) Albulidae 9.1 0 (0.9) 0 8.1 (0.9) Antennariidae 0 1.6 (0.2) 2.4 0 (0.3) Apogonidae* 14.1 10.7 (2.4) 148 (17.9) 10.6 3.6 (1.6) Belonidae 1.4 0 (0.1) Blenniioidei* 2 (9.5) 24.1 41.8 (6.5) 75 (9.1) 27.4 5.0 (3.7) Bothidae* 89.3 21.7 (11.0) 8 (1.0) 136.9 0 (15.6) Callionymidae 0.8 0 (O.ll Carangidae* 4.1 3.0 (0.7) 56.0 0 (6.4) Carapidae* 30 (3.6) Chaetodontidae* 9 (1.1) Chlopsidae* 24.4 0.9 (2.5) 52.8 0 (6.0) Congridae* 25.9 0.8 (2.6) 45.4 0 (5.2) Diodontidae 3.2 0 (0.4) Exocoetidae 0 0.8 lO.l) Gobiesocidae* 1 (4.8) 85 (10.3) Gobiidae 2 (9.5) 0 2.4 (0.2) Holocentridae* 4 (19,0) 14 (1.7) Labridae* 17.8 120.6 (13.7) 42 (4.9) 11.6 20.1 (3.6) Lutjanidae* 1 (4.8) 1.4 2.3 (0.4) 36 (4.4) 0 2.1 (0.2) Microdesmidae 1 (0.1) Monocanthidae* 118.4 24.0 (14.1) 124 (15.0) 146.1 5.4 (17.3) Moringuidae* 50.7 2.6 (5.3) 88.8 0 (10.2) Muraenidae* 33.0 1.6 (3.4) 18.6 0 (2.1) Ogcocephalidae 4.4 3.4 (0.8) 12.9 0 (1.5) Ophichthidae* 127.3 16.9 (14.3) 32.4 1.3 (3.8) Ophidiidae* 18.2 7.1 (2.5) 96.2 16.8 (12.9) Pomacanthidae* 14.6 1.7 (1.6) 5 (0.6) 3.2 0 (0.4) Pomacentridae* 35.9 8.9 (4.4) 41 (5.0) 11.0 0 (1.3) Priacanthidae 1.4 0 (0.1) Scaridae* 8 (38.1) 0 3.7 (0.4) 138 (16.7) Scorpaenidae* 4.6 4.8 (0.9) 3 (0.4) 4.0 0 (0.5) Serranidae* 0 11.5 (1.1) Sphyraenidae* 4.0 0.8 (0.5) 9 (1.1) Syngnathidae 4.5 0 (0.4) 1 (0.1) 18.6 0 (2.1) Synodontidae 2 (9.5) 1.4 0 (0.1) 48 (5.8) 4.8 0 (0.5) Tetraodontidae* 17.9 1.7 (1.9) 9.7 0 (1.1) Unidentified 1 (4.8) 7.3 28.9 (3.6) 12.3 0 (1.4) Total larvae 21 656.0 324.2 828 804.9 66.4 Anderson et al : Current velocity and catch efficiency in sampling settlement-stage larvae of coral reef fisfies 409 Table 2 Richness and diversity of families of fishes :aught by light traps and channel nets. South Southeast Sites Bock Lee Stocking combined Cay Island Richness (no. families) Light traps 19 6 18 Channel nets 32 30 23 Diversity (Brillouin index) Light traps 2.845 1.179 2.832 Channel nets 3.183 3.199 3.027 subsurface and surface channel nets, including only those families that represented I'Tc of the catch of either net for S.E. LSI (r,=0.227, n = \6. P>0.05; Fig. 3A) or for So. Bock (r,=0.088, n=l6, P>0.05; Fig. 3B). In addition, the sizes of larvae (Table 3) caught by light traps and channel nets (but excluding families with <3 individuals collected per sampling device [acanthurids, gobiids, and syngnathids] ) did not differ in the mean (sign test: n = ll, P>0.05), me- dian (« = 11, P>0.05l, or maximum (n = ll, P>0.05) length between sampling devices. For five families that differed in length between sites and that were represented by at least three individuals (Table 3), channel nets caught significantly larger larvae at S.E. LSI than at So. Bock in mean (n=5, P=0.03) and median {n=5. P=0.03) length, but not in maximum length (n=5, P>0.05). Current velocity and catch efficiency Over a 10-d period in which flow-meter readings were taken simultaneously at both sites, mean daily current velocity was three-times higher at So. Bock than at S.E. LSI (/test: ?=-7.92, df=18, P=0.0001; Fig 4A). Light traps deployed at So. Bock caught on average only 3.1'7f of the total number of fish collected by light traps at both sites, whereas the channel nets positioned at S.E. LSI caught an average of 43.8'7f of the total catch for nets positioned at both sites (Fig. 4B). The mean proportional abundance of all larvae caught by light traps was significantly lower at So. Bock than at S.E. LSI it =-50.3, df=24, P<0.0001) but there was no significant difference in proportional catch for channel nets (^=1.36, df=24, P=0.19). The difference in catch efficiency for each sampling device was also indicated by the opposite trends of abundance of taxa (apogonids, blennioids, labrids, pomacentrids, but not monocanthids) caught in at least nominally greater numbers by light traps at the low-current site (S.E. LSI) compared with the greater catch of these same taxa by channel nets at the high-current site (So. Bock). In addition, and consistent with our hypotheses concerning current velocity, light traps had higher family richness and diversity at S.E. LSI than at So. Bock, whereas the opposite was true for chan- nel nets (Table 2). Discussion Differential representation of taxa between sampling methods Light traps and channel nets differed in the taxonomic composition and the relative abundance of their catch. Expectedly, channel nets had appreciably higher family richness and diversity than did light traps, and this is consistent with our hypothesis that light traps are more selective (also see Choat et al., 1993) because not all larvae exhibit a photopositive response. One of the main differ- ences in the relative abundance of larvae between sam- pling devices in the central Bahamas is that channel nets collect a large proportion of labrids and leptocephalus larvae (Shenker et al., 1993; Thorrold et al., 1994a, 1994b, 1994c; Mojica et al., 1995; this study) but light traps do not (this study). Depth-dependent distributions of larvae are also likely to contribute to differences in the relative abundance of taxa because light traps catch larvae from an unknown depth range (over which larvae are attracted to the light) whereas surface and subsurface channel nets operate at discrete depths of 0-1 m and 2-4 m, respec- tively. Moreover, not only can surface and subsurface channel nets differ in their catch (Fig. 3), the relative abundance of particular taxa between surface and subsur- face nets can switch over time (Thorrold et al., 1994c). The similar sizes of larvae from families caught by both light traps and channel nets are somewhat inconsistent with the findings by Choat et al. (1993). who found that larvae caught by light traps were larger than those caught by towed nets and seines. They attributed this differerence in size to a stronger photopositive response by larger pe- lagic larvae. The smaller size of larvae that they caught in towed nets and seines may also indicate that larger larvae may better sense the presence of these nets and avoid cap- ture. Channel nets may decrease net avoidance by larvae because their stationary position may lessen water distur- bance and hence detection of the net by incoming larvae, or possibly because other larval behaviors are exhibited. Differential representation of taxa between sites The greatest difference between light traps and channel nets was in their relative catch between sites. Light traps were ineffective in collecting larvae at So. Bock, whereas channel nets collected a lower but not significantly differ- ent proportion of larvae at S.E. LSI. This result cannot be explained simply by differences in the abundance of taxa between sites; the same pattern was observed for taxa sufficiently represented by both sampling devices (apogonids, blennioids, labrids, pomacentrids). These sites did differ substantially in mean current velocity, with average flow rates at So. Bock three times higher than at S.E. LSI. Shenker et al. (1993) observed that larvae collected by channel nets were significantly more abun- dant at a sampling station that also had greater current velocity, and this finding is consistent with our results. Similarly, Thorrold (1992) observed that light traps that were allowed to drift with water masses collected signifi- 410 Fishery Bulletin 100(3) 16 12 A S.E. LSI ^^" Surface I I Subsurface i i ii xu B So. Bock j/- jj2f", j^*'. i^*- ,6*^- j^*- X,*- .-z!" v*- .*-_ j^ .->- jt!" j^- .-^^ .^>^ ^ s? 9P <-■?• '^^ # y>yt^^'$<«^^5^^^5/- ■6" .^" Family Figure 3 Abundance (mean no. larvae per 1000 m'l of 16 families of lar\'ae that constituted at least f^r of the catch by surface and subsurface channel nets at lA) S.E. LSI (southeast Lee Stocking Island.), and (B; So. Bock (South Bock Cay). Error bars are 1 SE. cantly more fish larvae than hght traps anchored to the sea floor, apparently independent of taxa. Thorrold noted that this pattern was unexpected because more water (and presumably more larvae) should pass by anchored traps, and he suggested that the ability of larvae to swim to and enter anchored traps may be difficult under high- current conditions. Although we conclude that between-site differences in relative catch by light traps and channel nets in our study were related to current velocity, the relationship between catch and average current velocity was not linear. The pro- portional abundance of larvae caught in light traps was over 31 times higher at S.E. LSI than at So. Bock, whereas the larval catch in channel nets was only 1.3 times higher at So. Bock than at S.E. LSI. The nonlinear relationship be- tween current velocity and larval abundance may represent a threshold response in which the efficiency of light traps and channel nets may change with different current veloc- ities. The mechanisms causing such relationships might be that fish larvae are able to orient to and swim into light traps more easily under lower current velocities, whereas fish larvae might more easily detect and avoid chan- nel nets because of greater hydrodynamic disturbances in front of nets during higher current velocities. Anderson et al : Current velocity and catch efficiency in sampling settlement-stage larvae of coral-reef fishies 411 Table 3 Mean (±1 SE), mt'dian, and range in size of lai-vae (mm standard length) coll ected in light trap s and ch innel nets at South Bock Cay and at southeast Lee Stocking Is and from 30 J uly to 1 1 August 1997. The families Blenniidae and La brisomidae are combined as Blennioidei. Family South Bock Cay Southeast Lee Stocking Island Mean(SE) Median Range n Mean(SE) Median Range n Light traps Acanthuridae 25.3(0.31) 25.2 (24.2-26.9) 9 Apogonidae 9.1(0.12) 8.6 (6.6-14.2) 150 Blennioidei 12.6(0.20) 12.6 (12.4-12.8) 2 11.8(0.18) 11.6 (7.8-18.7) 83 Bothidae 13.7(0.45) 13.5 (12.0-15.0) 7 Gobiidae 8.8(0.10) 8.8 (8.7-8.9) 2 Labridae 9.3(0.29) 8.6 (6.9-12.3) 32 Lutjanidae 11.7 11.7 (11.7-11.7) 1 15.0(0.39) 14.3 (12.6-20.8) 29 Monocanthidae 11.9(0.19) 11.4 (7.4-21.4) 128 Pomacanthidae 13.9 13.9 (13.9-13.91 1 9.0(0.24) 9.1 (8.3-9.4) 4 Pomacentridae 10.9(0.21) 10.8 (8.7-13.7) 40 Scaridae 9.1(0.24) 8.9 (8.7-10.3) 6 8.6(0.08) 8.7 (5.9-10.8) 120 Scorpaenidae 6.8(0.20) 6.8 (6.4-7.1) 3 SphvTaenidae 18.0(0.90) 18.4 (13.0-21.5) 9 Syngnathidae 40.5 40.5 (40.5-40.5) Channel nets Acanthuridae 24.7 24.7 (24.7-24.7) 1 Apogonidae 10.2(0.24) 9.8 (8.9-14.2) 26 9.8(0.43) 9.8 (9.0-10.5) 3 Blennioidei 12.7(0.12) 12.7 (10.8-17.8) 65 13.6(0.76) 12.8 (11.6-18.1) 8 Bothidae 14.8(0.14) 14.8 (7.0-17.8) 103 16.0(0.77) 15.1 (12.0-36.2) 29 Gobiidae 13.5(0.24) 13.6 (13.0-13.8) 3 Labridae 11.4(0.06) 11.4 (9.5-13.6) 154 13.7(1.3) 12.4 (10.9-28.7) 13 Lutjanidae 13.6 (0.38) 13.7 (12.0-14.7) 6 11.8 11.8 (11.8-11.8) 1 Monocanthidae 12.2(0.24) 12.0 (5.2-51.3) 204 12.8(0.29) 13.0 (7.7-18.6) 45 Pomacanthidae 8.9(0.14) 9.0 (7.5-9.4) 12 9.2 9.2 (9.2-9.2) 1 Pomacentridae 10.2(0.20) 9.8 (8.8-13.9) 32 12.6(0.95) 11.8 (11.5-14.5) 3 Scaridae 9.0(0.13) 9.0 (8.6-9.2) 4 Scorpaenidae 8.0(0.35) 8.1 (6.7-9.6) 9 8.1 8.1 (8.1-8.1) 1 Sphyraenidae 23.9 23.9 (23.9-23.9) 1 18.6(3.9) 19.9 (11.3-24.6) 3 Syngnathidae 32.9(16.9) 25.7 (7.8-65.1) 3 35.0(1.8) 35.0 (33.2-36.8) 2 Indeed, there may be important consequences in sam- pling reef fish larvae among multiple sites that vary in current velocity. Current-dependent catch efficiency could confound estimates of spatial variation in larval abun- dance if those sites vary substantially in current velocity. For example, deploying light traps to sample larval abun- dance at several locations may result in higher abundances at sites with low current. This result could indicate actual larval distributions, such as accumulation of larvae on the leeward side of an island, or it could simply reflect current- dependent catch efficiency. In the latter case, estimates of lan'al distribution and subsequent interpretations of such distributions (e.g. larval retention) would be flawed. As with all sampling devices, Hght traps and channel nets have both advantages and disadvantages for estimating lar\'al abundance. Light traps may collect a larger propor- tion of settlement-stage larvae for those fishes that exhibit a photopositive response, but they catch fewer taxa and sample an unknown volume of water (Cheat et al., 1993) in relation to channel nets. Our results indicate an important factor in selecting an appropriate device for estimating larval supply — namely, the hydrodynamic conditions under which the device will be deployed. Use of one sampling de- vice, either light traps or channel nets, would have resulted in biased and potentially inaccurate relative estimates of larval abundance between So. Bock and S.E. LSI. In more recent studies, several methods have been used to estimate the abundance of as many larval fishes as possible (e.g. Leis et al., 1998), and in a comparison of light traps and towed nets in sampling freshwater fishes, Gregory and Powles ( 1988) concluded that both sampling devices should be used to avoid bias in the collection of larvae. The use of both light traps and channel nets simultane- ously can provide less biased estimates of spatial variation in taxonomic composition and larval supply of coral-reef fishes among sites that vary in current velocity. This com- 412 Fishery Bulletin 100(3) 450 ' A 400 - X 350 - a> c E^ 300 - -a c o 250 - to Cl) ra > 200 - c o ■c 150 - Q. (5 r 100 - 50 - So. B oc ( S.E , LSI 1.0 0.8 0.6 - 0.4 0.2 0.0 B Light traps Channel nets T T So. Bock S.E. LSI Figure 4 (A) Mean current velocity averaged over 24 h (ni/hl, and (B) mean daily propor- tional abundance of larvae caught by light traps and channel nets at So. Bock (South Bock Cay) and at S.E. LSI (southeast Lee Stocking Island). Error bars are 1 SE. bination is problematic, however, because there is no "com- mon currency" between light traps and channel nets. The volume of water sampled by light traps is unknown, and capture rates of larvae are very low (Meekan et al., 2000), so that the catches of light traps and channel nets can- not be easily standardized. The value of these and other larval sampling devices is that they provide a measure of temporal and spatial variation in relation to larval supply. However, the nonlinear relationship between current ve- locity and larval abundances observed in our study could compromise such estimates and further complicate stan- dardization among sampling devices. Additional research is necessary to account for method-dependent differences in larval abundance among sites that differ in hydrody- namic or other environmental conditions. Acknowledgments We thank A. King, C. McPCinney-Richards, K. Overholtzer, and S. Whitcraft for field assistance, E. Maddox and B. Victor for verifying larval identifications from our col- lections, G. Almany and A. Summers for assistance with light-trap electronics, and the staff of the Caribbean Marine Research Center at Lee Stocking Island for logisti- cal support. This research was supported by NSF grants OCE-96-17483 and OCE-99-96053 (M. Hixon), NSF OCE- 99-96053 (M. Carr), and by NOAA National Undersea Research Program grant 97-3109 (M. Hixon). Literature cited Choat, J. H.. P. J. Doherty, B. A. Kerrigan, and J. M. Leis. 1993. A comparison of towed nets, purse seine, and light- aggregation devices for sampling larvae and pelagic juve- niles of coral reef fishes. Fish. Bull. 91:195-209. Doherty P. J. 1987. Light-traps: selective but useful devices for quantify- ing the distributions and abundances of larval fishes. Bull. Mar Sci. 41:423-431. 1991. Spatial and temporal patterns in recruitment, hi The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 261- 293. Academic Press, San Diego, CA. Doherty P. J., and J. H. Carleton. 1997. The distribution and abundance of pelagic juvenile fish near Grub Reef, central Great Barrier Reef Proc. 8"' International Coral Reef Symp. 2:1155-1160. Doherty, P. J., and A. J. Fowler. 1994. An empirical test of recruitment limitation in a coral reef fish. Science 263:93,5-939. Doherty, P. J., M. J. Kingsford, D. Booth, and J. Carleton. 1996. Habitat selection before settlement by Pomacentrus coelestis. Mar Freshwater Res. 47:391-399. Doherty P. J., and D. McB. Williams. 1988. The replenishment of coral reef fish populations. Oceanogr. Mar Biol. Annu. Rev. 26:487-551. Dufour, v., and R. Galzin. 1993. Colonization patterns of reef fish larvae to the lagoon at Moorea island, French Polynesia. Mar Ecol. Prog. Ser 102:143-152. Dufour, v.. E. Riclet, and A. Lo-Yat. 1996. Colonization of reef fishes at Moorea island, French Polynesia: temporal and spatial variation of the larval flux. Mar Freshwater Res. 47:413-422. Gregory, R. S., and P. M. Powles. 1988. Relative selectivities of Miller high-speed samplers and light traps for collecting ichthyoplankton. Can. J. Fish. Aquatic Sci. 45:993-998. Hixon, M. A. 1991. Predation as a process structuring coral reef fish communities. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 475-508. Academic Press, San Diego, CA. Anderson et aL: Current velocity and catch efficiency in sampling settlement-stage larvae of coral-reef fishes 413 Hixon, M. A., and M. H. Carr. 1997. Synorgistic predation, density dependence, and popu- lation regulation in marine fish. Science 227:946-949. Jones, G. P. 1991. Postrecruitment processes in the ecology of coral reef fish populations: a multifactorial perspective. In The ecol- ogj- of fishes on coral reefs ( P. F. Sale, ed.), p. 294-328. Aca- demic Press, San Diego. CA. Leis, J. M., H. P A. Sweatman, and S. E. Reader 1996. What the pelagic stages of coral reef fishes are doing out in blue water: daytime field observations of larval behav- ioural capabilities. Mar Freshwater Res. 47:401-411. Leis, J. M., T. Trnski, P J. Doherty, and V. Dufour 1998. Replenishment of fish populations in the enclosed lagoon of Taiaro Atoll: (Tuamotu Archipelago, French Poly- nesia) evidence from eggs and larvae. Coral Reefs 17:1-8. Magurran, A. E. 1988. Ecological diversity and its measurement. Princeton Univ. Press, Princeton, NJ. Meekan, M. G., P. J. Doherty and L. White Jr. 2000. Recapture experiments show the low sampling effi- ciency of light traps. Bull. Mar Sci. 67:875-885. Meekan, M. G., M. J. Milicich, and P. J. Doherty. 1993. Larval production drives temporal patterns of larval supply and recruitment of a coral reef damselfish. Mar Ecol. Prog. Ser 93:217-225. Milicich, M. J., and P. J. Doherty. 1994. Larval supply of coral reef fish populations: magni- tude and synchrony of replenishment to Lizard Island, Great Barrier Reef Mar. Ecol. Prog. Ser 110:121-1324. Milicich, M. J., M. G. Meekan, and P. J. Doherty. 1992. Larval supply: a good predictor of recruitment of three species of reef fish (Pomacentridae). Mar. Ecol. Prog. Ser 86:153-166. Mojica, R., J. M. Shenker, C. W. Harnden, and D. E. Wagner 1995. Recruitment of bonefish, Albula vulpes. around Lee Stocking Island, Bahamas. Fish. Bull. 93:666-674. Munday, P. L., G. P Jones, M. C. Oehman, and U. L. Kaly 1998. Enhancement of recruitment to coral reefs using light attractors. Bull. Mar Sci. 63:581-588. Schmitt, R. J., and S. J. Holbrook. 1999. Mortality of juvenile damselfish: implications for as- sessing processes that determine abundance. Ecology 80: 35-50. Shenker, J. M., E. D. Maddox, E. Wishinski, A. Peari, S. R. Thorrold, and N. Smith. 1993. Onshore transport of settlement-stage Nassau grou- per £p(>iep/ie/us striatus and other fishes in Exuma Sound, Bahamas. Mar Ecol. Prog. Ser 98:31-43. Sponaugle, S., and R. K. Cowen. 1996a. Larval supply and patterns of recruitment for two ("aribbean reef fishes, Stef^antes partitus and Acanthurus hahianus. Mar Freshwater Res. 47: 433-437. 1996b. Nearshore patterns of coral reef fish larval supply to Barbados, West Indies. Mar Ecol. Prog. Sen 133:13-28. 1997. Early life history traits and recruitment patterns of Caribbean wrasses (Labridae). Ecol. Monogr. 67:177-202. Stobutzki, I. C, and D. R. Bellwood. 1994. An analysis of the sustained swimming abilities of pre- and post-settlement coral reef fishes. J. Exp. Mar. Biol. Ecol. 175:275-286. 1997. Sustained swimming abilities of the late pelagic stages of coral reef fishes. Mar. Ecol. Prog. Ser. 149:35—41. Thorrold, S. R. 1992. Evaluating the performance oflight traps for sampling small fish and squid in open waters of the central Great Barrier Reef lagoon. Mar Ecol. Prog. Ser 89:77-285. Thorrold, S. R., and M. J. Milicich. 1990. Comparison of larval duration and pre- and post-settle- ment growth in two species of damselfish, Chromis atripec- toralis and Pomacentrus coelestis (Pisces: Pomacentridae), from the Great Barrier Reef Mar Biol. 105:375-384. Thorrold, S. R., J. M. Shenker, E. D. Maddox, R. Mojica, and E. Wishinski. 1994a. Larval supply of shorefishes to nursery habitats around Lee Stocking Island, Bahamas. II. Lunar and ocean- ographic influences. Mar Biol. 118:567-578. Thorrold, S. R., J. M. Shenker, R. Mojica, E. D. Maddox, and E. Wishinski. 1994b. Temporal patterns in the larval supply of summer- recruiting reef fishes to Lee Stocking Island, Bahamas. Man Ecol. Prog. Ser 112:75-86. Thorrold, S .R., J. M. Shenker, E. Wishinski, R. Mojica, and E. D. Maddox. 1994c. Larval supply of shorefishes to nursery habitats around Lee Stocking Island, Bahamas. I. Small-scale dis- tribution patterns. Mar Biol. 118:555-566. Thorrold, S. R., and D. McB. Williams. 1996. Meso-scale distribution patterns of larval and pelagic juvenile fishes in the central Great Barrier Reef lagoon. Man Ecol. Prog. Sen 145:17-31. Wolanski, E., P. Doherty, and J. Carleton. 1997. Directional swimming offish larvae determines con- nectivity of fish populations on the Great Barrier Reef Naturwissenschaften 84:262-268. Zan J. H. 1984. Biostatistical analysis. 2nd ed., 718 p. Prentice Hall, Englewood Cliff, NJ. 414 Abstract-A total of 7244 Greenland halibut (Reinhardtius hippoglossoides, Walbaum) were tagged in Greenland waters between 1986 and 1998 to in- crease information on stock delinea- tions, to clarify migration routes, and to describe the seasonal movements of fjord populations. At present 517 recaptured Greenland halibut have been recorded. For Greenland halibut released in Davis Strait, Baffin Bay, and the fjords of south- western and eastern Greenland, a sub- stantial portion of recovered fish dem- onstrated migratory behavior, up to 2500 km, primarily to Denmark Strait between Greenland and Iceland. The recaptured fish provided evidence of intermingling between the population in Denmark Strait and the populations in Davis Strait and the southwest Green- land fjords. These observations support those of other studies that indicate that Greenland halibut inhabiting Davis Strait and the fjords of southwestern and eastern Greenland originate in the spawning grounds west of Iceland. The high mobility of offshore Greenland hal- ibut within Baffin Bay and Davis Strait suggests that Greenland halibut migrate extensively between feeding and spawn- ing areas. Greenland halibut in the Qords of northwestern Greenland appear to be resident in behavior and do not inter- mingle with offshore or more southerly inshore populations. A seasonal pattern in the recovery of these fish indicates that Greenland halibut aggregate in the inner part of ^ords during the second half of the year (when inshore waters are not covered with ice). Intermingling and seasonal migrations of Greenland halibut {Reinhardtius liippogiossoides) populations determined from tagging studies Jesper Boje Department of Marine Fishenes Danish Institute for Fisheries Research Charlottenlund Slot DK-2920 Charlottenlund, Denmark Present address Greenland Institute of Natural Resources PO Box 570 3900 Nuuk, Greenland E mail address |esper(atnaturgl Manuscript accepted 3 October 2001. Fish. Bull. 100:414-^22 (2002). Greenland halibut, Reinhardtius hip- poglossoides (Walbaum), are widely dis- tributed in the Northwest Atlantic. Their range covers a geographical area from Smith Sound, between Greenland and Canada, southward throughout Baffin Bay and Davis Strait to the northeastern coast of the United States and eastward along eastern Greenland to Iceland (Smidt, 1969; Bowering and Brodie, 1995). The spawning grounds of Greenland halibut are believed to be located southwest of Iceland (Sigurds- son') and cover an extended area from Davis Strait, south of 67°N (Jensen, 1935: Smidt, 1969) to south of Flemish Pass off Newfoundland (Junquera and Zamarro, 1994) between 800 and 2000 m depths. The Greenland halibut popu- lations off the eastern coast of Canada, in Davis Strait, in the fjords of western Greenland, in the Denmark Strait, and in Icelandic waters are believed to be re- cruited from these spawning grounds (Templeman. 1973;Sigurdsson' ). Anum- ber of scientists have examined interac- tions among different local populations of Greenland halibut in the Northwest Atlantic. They have considered and, in some cases, attempted to quantify mer- istic characteristics (Templeman, 1970; Misra and Bowering, 1984; Riget et al., 1992; Rasmussen et al., 1999), genetic variability (Fairbairn, 1981; Riget et al., 1992; Vis et al., 1997), the occurrence of parasites as natural tags (Khan et al., 1982; Boje et al. , 1997 ), and tagging data (Smidt, 1969; Bowering, 1984; Riget and Boje, 1989; Boje-). The results of these investigations suggest that the entire Greenland halibut population between Greenland and Canada must be consid- ered a single stock unit. There is some evidence that the populations between Greenland and Iceland originate from a spawning stock situated along the conti- nental slopes west of Iceland. A number of Greenland halibut populations seem to maintain a degree of isolation with- out any prespawning migration to their original spawning areas; this is charac- teristic of the Greenland halibut popu- lation in the Gulf of St. Lawrence ( Khan et al., 1982) as well as for populations in the Qords of northwestern Greenland (Riget and Boje, 1989). Spawning has been observed in the Gulf of St. Law- rence (Bowering, 1980), but only a few ripe females have been sighted in the Qords of western Greenland (Riget and Boje, 1989). Previous tagging experiments in the fjords of western Greenland (Smidt, 1969; Riget and Boje, 1989; Boje^) suggest that Greenland halibut stocks in the southernmost fjords may be recruited from Icelandic spawning grounds. The Irminger Current and the East Greenland Polar Current carry eggs and larvae from these spawning ' Sigurdsson,A. 1979. The Greenland hal- ibut [Reinhardtius hippoglossoides (Wal- baum)) at Iceland. Hafrannsbknir. 16, Ma- rine Research Institute. Reykjavik, Iceland. - Boje, J. 1993. Migrations of Greenland halibut in the Northwest Atlantic from tag- ging experiments in West Greenland 1986- 89. ICES (International Council for the Exploration of the Sea) C. M. Doc. 1993/ G:65, 14 p. ICES, Pategade 2-4, DK-1261 Copenhagen K, Denmark. Boje: Intermingling and seasonal migrations of Reinhardtius htppoglossoides 415 grounds to eastern Greenland and probably as far as southwestern Greenland (Sigurdsson'; Boje^). In 1959, Smidt (1969) observed the re- capture west of Iceland of Greenland halibut that had been tagged in Licthenau Fjord in southwestern Greenland in 1954. Riget and Boje (1989) noted the recapture west of Iceland in 1980 of Greenland halibut that had been tagged in Godthaab Fjord in 1964. In the present study, tagging experiments in the ^ords of eastern and western Greenland from 1986 to 1998 are evaluated to determine the stock discreteness of Greenland halibut populations in the Northwest Atlantic and to describe seasonal migrations of Greenland halibut within inshore areas. Materials and methods N75- 70- 65- 60- 55- ■j*-' % pemuviK Greenland Uummonnnq TorsukkatUk llulissat / XlVa > OB -'«^xV IC x ^ 'Iceland." ^CxJih, J XI Vb Va IE 2G Greenland halibut from four areas around Greenland were tagged between 1986 and 1998. The tagging was carried out in the fjords of eastern Greenland at Ammassalik in September 1990, in the ^ords of southwestern Greenland from Cape Farewell to Godthaab Fjord in January 1987-88, in the Qords of northwestern Greenland from Disko Bay to Upernavik during July-August 1986-98, and off western Greenland from Davis Strait to Baffin Bay during May-August 1991-93 (Table 1, Fig. 1). Samples were collected with longlines (Mustad Autoline Systems) from research vessels fishing at depths between 400 and 900 m — a range that is standard for the commercial fishery (Boje^). Fishing time for each set was approximately six hours. A landing net was placed under each fish from the time it left the water until the fish was landed onboard the ship to avoid damage from the hook (i.e. the gravitational drag of the hook while the specimen was removed from the wa- ter). The hook was removed with care and only fish hooked in the mouth region were selected for tagging because these injuries were generally not fatal. Condition of the fish was judged visually, mainly by examining the color of the gills and by assessing internal hooking injuries. Length of the fish was measured to the nearest full cen- timeter (total length) and fish larger than 35 cm were se- lected to increase the possibility of immediate recapture in the fishery. Each fish was tagged in the musculature with ir '^L^CapO Farewell 2H xn 2J 55 -50 ■45 40 "T" -35 ■30 ■25 w Boje, J. 1997. Larval growth and spawning of Greenland halibut in West Greenland waters and the possible influences by hydrographic conditions. ICES (International Council for the Exploration of the Sea) international symposium on "Recruitment dynamics of exploited marine populations: physi- cal-biological interactions," Baltimore 22-24 Sept. 1997. ICES, Palaegade 2-4, DK-1261 Copenhagen K, Denmark. Boje, J. 1989. The fishery for Greenland halibut in Subarea 1. NAFO (Northwest Atlantic Fisheries Organization) SCR Doc. 89/27, sen N1603, 8 p. Figure 1 Map of release sites (shaded areas). NAFO (Northwest Atlantic Fisheries Organization) (0-2) and ICES (International Council for the Exploration of the Sea) (V-XIV) divisions are indicated. a yellow T-bar tag (Floy T-bar anchor tags, FD-68B yellow) just below the dorsal fin ray near the head. Immediately after having been tagged, the fish was released. For an evaluation of migratory routes, the numbers of recaptured fish were adjusted for fishing activities (an- nual landings) in recapture area (i.e. NAFO/ICES divi- sion) for the year of recapture and these numbers were expressed as the number of halibut per 100 metric tons of annual landings in the recapture division. Landing sta- tistics were obtained from Greenland Statistical Office, NAFO Statistical Bulletin 1986-93, NAFO STATLANT 21A data and ICES Cooperative Research Reports of the ACFM 1986-98. To analyze seasonal movements among "resident" Greenland halibut (i.e. fish caught less than 100 km from the release site) of northwestern Greenland, the §ords were stratified into five rectangles of equal area, and positioned in an east-west direction. A random-walk model (Skellam, 1951) was applied to the distribution of all releases (/i=4319) pooled for the entire period, 1986- 98, for all five areas, assuming equal probability that the fish would migrate either east or west or remain in the area (P=0.33). The model operated on a time unit of three months, thereby allowing each fish to move from only one rectangle to the next rectangle within a quarter of a year. 416 Fishery Bulletin 100(3) >> J2 T3 -a i. CD CD lO lO o r^ CJl CD lO CO CM '-' 00 ^H CD CM t=) r^ CT) ^H O CM in 00 CO ^ Q; (N t-^ lO t^ oi cri CD CO c-^ CO CD CD ^ C-^ d ^ in CM CM CO CD ^ ^ ci r- 3 t- , — , T-4 IM ^H »-l 1-H -tJ U a CO o t-, flj „ - CD a ca j3 CO m cc CO lO o c^ 00 CO ■^ tJ5 CM 00 I> ■^ lO CD CO CO C^ O ,-H f-i o t> ^ r^ OJ 5 a l>J Tf lO CM CM CM ^ CO t^ CO 5 ^ b lO c tC , , c CO 3 ^1 CO CO ;n "S T3 j3 (L T3 ilC 0; 00 1-H be t. CC 3 *J a. Oj I> ^ ^ ^ 1— 1 *H rt CD ■*t CO O] Tj- s CO 0) CD - CN CO o c o 15 c ^ in '-' •^ (N CO m ^ CM 00 o t— o CO CO -^ CJ5 ^^ !N CO I>] ^ CO CM 0) 1— » r-H Tf 03 o g u a CO o CO CO lO o CO -* 1- CM CO p c u i-H in -o l-H CO c Ol CO CN CO ^ -rr ^^ CM ■^ 00 Tf ,— 1 ■* •* CO cn CO Tf -!)• ^ ^H CM I-- CM in CD Tf o - CO 00 C^l CO Tf lO CO t^ C^ 00 00 ^ CO CD ^ CO 't CD C^ in CO Tt CD t^ 00 00 ^ -^ tuD 3 ^ 00 C^l CO CM Ol CO in ■^ CD CO CO f- CM s ' ti t^ ^ a. :C o o C^ CO 02 ,-H O fO c 1; _ ^ lO O CD 'OO IC o •M CM to CO 00 t^ Tf CO Oi o o o CO 00 o o CO rt ■& rr-- a> ^^ o> 0> CJ3 4 00 00 CT> t^ 05 05 00 00 00 t^ 00 CD 00 CD '- g 1 o 1 1 1 CO 1 1 1 1 00 1 00 1 CD 1 OS 4 1 CO ci 1 CO 1 1 CD 1 o 1 in 1 ID 1 CO 1 00 1 CM 1 CO 1 1 CO CD CO QJ O C '^^ 3 J ■* CO CO CO CO C^ CO CO CO xf CO ^ Tf CO CO ^ ^ CO Tf CO CO Tf CO Tf CO -3- CM O 0 1 CO X fc Q < < < < < < < < < < < < < < < < < < < < Q Q < -D ■c ;h 15 "q; o ^ ^ M Ji! j<: ^ cr a* a* cr cr a" -C c -^ s? CO CO CO CO 3 CO CO c CO CO c CO c CO c CO c ^ ^ ^ ^ *ra *co >. T3 C o CO Si .^ CO CO B « M CO c c c c c c u u CO 0) s c a; P '■+J CO 3 CO CO CO s [2 CO CO CO T3 o CO CO to CO CO to CO CO CO CO CO ]3 3 IS o 3 o 3 3 o 3 22 o CO E e 3 CO E e 3 to E g 3 CO E s 3 CO s S 3 CO E g 3 > CO c D. > CO c Oi Q. > CO c a; a > CO c tx "> CO to > CO c E CO o X ^ U o s s p p eS e2 e2 e2 D D D D tJ :d u Ij D D Q a CQ -a bjo c t- 3 ■Q CD C3 j= s b t: -*-> -*-> +-> -*-» -*-> *J ^ +-> -^ -*J -*j -*J ^j ^ ^j -^J ^ CO c o CO CO CO CO CO CO to CO CO CO to CO CO to to CO' to m CO c Cb a 3 C CO 3 c CO Ec 5c Ei M & & 5jD >•, a >> Ei Gc Ec Ei Ec >. a a >. >. a >. a V- O 3 3 :3 3 3 3 3 p 3 3 3 3 3 3 3 ^ 3 3 p p 3 s T3 CO C/2 •-J ■n < < < < < < < ■^ < ^ < < < < < ■-3 < < ■-5 -^ < < CO Q -a u o 00 c^ CD t^ CO rr to c^ CO rj- to t^ tr- CO ■^ in CD 00 Ol en ai oi (35 en (J> <7i (Jl cn (35 CD o> C71 i-H 03 CD (D (D CD (D CD CTl (D CD -.J o CO -a -a c ^ cn o CO CO s a t^ c c 3 .S "O a> 01 T3 c M — a C CO ^ § a O "c at O to CO CO CO 1jc CO W 3 O o o Z o CO 1 3= O "3 1 Bo|e; Intermingling and seasonal migrations of Reinhardtius hippoglossoides 417 The model run covered a total of eight years, which was the maximum timi' recorded for time at liberty of tagged (Greenland halibut at sea. The random walk distribution was compared with the true recapture distribution by means of a chi-square test. The true recapture distribution was corrected for fishing effort by using the frequency of recaptured halibut per unit landing in a statistical square (smallest statistical unit to assign landings), where landings were monthly averages for the period 1993-97. By pooling landings monthly for this 5-year period there were no zero val- ues in the data set. Effort data were not available for the inshore fishery; therefore landings data were used instead. Previously landings data for these inshore areas were found to correlate well with fishing effort (Anonymous, 1998). Migration distances are given in kilometers ( 1 km=0.540 nautical mile). For all analyses, SAS statis- tical (SAS, 1988) and Statistica (StatSoft, Inc., 2000) software were used. Results A total of 7244 Greenland halibut were tagged from 1986 to 1998. By the end of 1998, 517 recaptured Greenland halibut (7.1% of halibut released during the entire period) were recovered (Table 1). The majority of recaptured fish were reported within two years after tagging, but some were reported as late as eight years after tagging. In general, Greenland halibut tagged in the fjords of northwestern Greenland (Ilulissat, Torsuk- kattak, Uummannaq, and Upernavik) tended to remain resident, whereas some Greenland halibut from the three other tagging sites (East Greenland fjords. Southwest Greenland fjords, and West Greenland offshore) displayed migratory behavior. Prevailing migratory patterns An overview of the numbers of recaptured Greenland halibut (adjusted for fishing activity for each release site versus recapture area) is presented in Table 2. Most recaptured Greenland halibut were found at their release locality (especially at Ammassalik in East Greenland, in the fjords of Cape Farewell, in Godthaab Fjord, Uum- mannaq, and Upernavik in West Greenland), and 99% of the recaptured halibut were taken within the release area. For releases in the two neighboring fjords in north- western Greenland, Ilulissat, and Torsukkattak, 10-14% of recaptured Greenland halibut showed intermingling between the two fjords. Limited intermingling was also noted between Ilulissat and Uummannaq ( CO CO (J3 lO tN Ol O lO — ^ -*3 OJ [/; a 3 D ^ 03 ji m cC ^ *j CD -tJ ^ CO rt [^ <0 -^t -^ O CO ^ o E 3 a; ^ „ ^ ^ ^ (35 ^ ° 3 - — to — C en +-» C en a; t. CO a s 00 " I> o CO ^ CN a en 05 CO CD ^ V -^ en _3 u 0) ^ CO E s en -^ 3 a C3 ■> '5 Tf c5 [^ c- ■* OJ CO J3 CO <7i ^ r^ 1— ) 3 D w Q. D. a 3 T) — c •^ s « CO 4( u ^ 1 - 3 3 •^j-^s CO |2 5 £ S 0) a Z d u o £ ^ 'En c ^x. 05 S en a .2 ^ o CO OJ T— t D rt o 01 o I- at CO CO ^1 <31 5 S2 c^ c o c 'S te n >, c _Q .o -4J 3 j< c ^ "3 en o S CD Oi en O C3^ x: « CO O ^^ cd X 6 C»3 "3 -a IM H ■u S ffl "c j:: OJ -^ T3 0) I- j_, _ - _ _ „ O o OJ S 5 V ^ V - ^ ic C^ 3 a re 3 Q ■M a CO en cq ^ ^_l re o he £ c CO i2 Eb en V e C3 "^ t^ 03 C ic O e s a I-. 3? C as ji ^ -2, ^ ■^ £ -° 2 S |£ CO cfl p Tnous. 1998. Northwest Atlantic Fisheries Organization Scientific Council Reports, 1997, 274 p. NAFO, Dartmouth, Nova Scotia, Canada. Boje, J., F. Riget, and M. Koie. 1997. Helminth parasites as biological tags in population studies of Greenland halibut (Reinhardtius hippoglos- soides, (Walbaum)), in the north-west Atlantic. ICES (Intemation Council for the Exploration of the Sea) J. Mar Sci. 54:886-895. Bowering, W. R. 1980. Fecundity of Greenland halibut, Reinhardtius hippo- Nielsen, J. G., and J. Boje. 1995. Sexual maturity of Green- land halibut at West Greenland based on visual and histological observations. NAFO (Northwest Atlantic Fisheries Organiza- tion) SCR Doc. 95/18, ser no. N2525, 7 p. NAFO Secretariat, 2 Morris Drive, P. O. Box 638, P. O. Box 638, Dartmouth, Nova Scotia, Canada B2Y 3Y9. glossoides (Walbaum), from southern Labrador and south- eastern Gulf of St. Lawrence. J. Northwest Atl. Fish. Sci. 1:39-43. 1984. Migrations of Greenland halibut, Reinhardtius hip- poglossoides, in the Northwest Atlantic from tagging in the Labrador-Newfoundland region. J. Northwest Atl. Fish. Sci. 5:8.5-91. Bowering, W. R., and W. B. Brodie. 1995. Greenland halibut (Reinhardtius hippoglossoides): a review of the dynamics of its distribution and fisher- ies off eastern Canada and Greenland. In Deep-water fisheries of the North Atlantic slope (A. G. Hopper, ed.), p. 113-160. Kluwer Academic Publishers, Dordrecht, The Netherlands. Buch, E., S. Aa. Horsted, and H. Hovgaard. 1994. Fluctuations in the occurrence of cod in Greenland waters and their possible causes. ICES (Intemation Cou- ncil for the Exploration of the Sea) Mar Symp. 198:158- 174. Fairbairn, D. J. 1981. Biochemical genetic analysis of population differen- tiation in Greenland halibut (Reinhardtius hippoglossoi- des) from the Northwest Atlantic, Gulf of St. Lawrence, and Bering Strait. Can. J. Fish. Aquat. Sci. 38:669-677. Jensen, Ad. S. 1935. The Greenland halibut, (Reinhardtius hippoglossoi- des (Walb. )) its development and migrations. K. Danske Vidensk. Selsk. Skr, 9RK. 6(4):l-32. Junquera, S., and J. Zamarro. 1994. Sexual maturity and spawning of Greenland halibut (Reinhardtius hippoglossoides) from Flemish Pass area. NAFO (Northwest Atlantic Fisheries Organization) Sci. Coun. Studies 20:47-52. Jorgensen, O. A. 1997. Movement patterns of Greenland halibut,/fei/i/!arrf;ius hippoglossoides (Walbaum), at West Greenland, as inferred from trawl survey distribution and size data. J. Northwest Atl. Fish. Sci., 21:23-37. Khan, R. A.. M. Dawe, R. Bowering, and R. K. Misra. 1982. Blood protozoa as an aid for separating stocks of Green- land halibut (Reinhardtius hippoglossoides) in the North- west Atlantic. Can. J. Fish. Aquat. Sci. 39:1317-1322. Misra, R. K., and W. R. Bowering. 1984. Stock delineation of Greenland halibut in the North- west Atlantic using a recently developed multivariate sta- tistical analysis based on meristic character N. Am. J. Fish. Manag. 4:390-398. Rasmussen, E. B., M. -B. Salhauge, and J. Boje. 1999. The suitability of vertebral counts in population studies of Greenland halibut (Reinhardtius hippoglos- soides (Walbaum)) at West Greenland. ICES (Interna- tional Council for the Exploration of the Sea) J. Mar Sci. 56:75-83. Riget, F, and J. Boje. 1988. Distribution and abundance of young Greenland hal- ibut (Reinhardtius hippoglossoides) in West Greenland waters. NAFO (Northwest Atlantic Fisheries Organiza- tion) Sci. Coun. Studies, 12:7-12. 1989. Fishery and some biological aspects of Greenland halibut {Reinhardtius hippoglossoides) in West Greenland waters. NAFO (Northwest Atlantic Fisheries Organiza- tion) Sci. Coun. Studies, 13:41-52. Riget, F, J. Boje, and V. Simonsen 1992. Analysis of meristic characters and genetic differentia- tion in Greenland halibut (Reinhardtius hippoglossoides) in the Northwest Atlantic. J. Northwest Atl. Fish. Sci. 12:7-14. 422 Fishery Bulletin 100(3) SAS Institute Inc. 1988. SAS Statistics, release 6.03. SAS Inc. Gary. NC. Sigurdsson, A. 1981. Migrations of Greenland ha\ihut, Reinhardtius hippo- glossoides {V/a\h. ) from Iceland to Norway. Rit Fiskideildar 6:3-6. Skellam, J. G. 1951. Random dispersal in theoretical populations. Bio- metrika 38:96-218. Smidt, E. L. B. 1969. The Greenland halibut, Reinhardtius hippoglossoides (Walb. ), biology and exploitation in Greenland waters. Medd. Danm. Fisk.-Havunders. 6(4):79-148. StatSoft, Inc. 2000. STATISTICA for Windows. StatSoft, Inc., Tulsa, OK. Taning, A. V. 1937. Some features in the migration of cod. J. Cons. Int. Explor.Mer 12:1-35. Templeman,W. 1973. Vertebral and other meristic characters of Greenland ha\ihut. Reinhardtius hippoglossoides, from the Northwest Atlantic. J. Fish. Res. Board Can. 27:1549-1562. 1973. Distribution and abundance of Greenland halibut (Reinhardtius hippoglossoides (Walbaum)), in the North- west Atlantic. ICNAF (International Commission for the Northwest Atlantic Fisheries) Res. Bull. 10:83-98. Vis, M. L., S. M. Carr, W. R. Bowering, and W. S. Davidson. 1997. Greenland halibut iReinhardtius hippoglossoides) in the North Atlantic are genetically homogeneous. Can. J. Fish. Aquat. Sci. .54:1813-1821. 423 Abstract— Marine nianinuil (iict is typ- ically characterized by identifying fish otoHths and cephalopod beaks re- trieved from stomachs and fecal mate- rial (scats). The use and apphcabiHty of these techniques has been tlic matter of some debate given inherent biases associated with the method. Recent attempts to identify prey using skel- etal remains in addition to beaks and otoliths are an improvement; however, difficulties incorporating these data into quantitative analyses have limited results for descriptive analyses such as frequency of occurrence. We attempted to characterize harbor seal (P/iora vitulinn ) diet in an area where seals co-occur with several salmon species, some endangered and all managed by state or federal agencies, or both. Although diet was extremely variable within sampling date, season, year, and between years, the frequency and number of individual prey were at least two times greater for most taxa when prey structures in addition to otoliths were identified. Estimating prey mass in addition to frequency and number resulted in an extremely different rela- tive importance of prey in harbor seal diet. These data analyses are a neces- sary step in generating estimates of the size, total number, and annual biomass of a prey species eaten by pinnipeds for inclusion in fisheries management plans. Improving pinniped diet analyses through identification of multiple skeletal structures in fecal samples Patience Browne Jeffrey L. Laake Robert L. DeLong National Marine Mammal Laboratory Alaska Fisheries Science Center National Marine Fisheries Service. NOAA 7600 Sand Point Way NE Seattle, Washington 981 15 Present address (for P. Browne): Center for Health of the Environment University of California at Davis One Shields Ave. Davis, California 95616 E-mail address (for P Browne) pbrowne(a)ucdavisedu Manuscript accepted 12 February 2001. Fish. Bull. 100:423-433(2002). Increases in marine mammal popula- tions in Washington and Oregon have coincided with decreases in wild salmon populations in these and other western states. Recently, several salmon stocks in the western U.S. have been listed as threatened, endangered, or are under status review. These include coastal Oregon coho iOnchorhyncus kisutch), upper Columbia River steelhead (O. mykiss), and Snake River spring and fall Chinook (O. fshawytscha), steel- head, and sockeye salmon (O. nerka; NMFS, 1997). Salmon often co-occur with marine mammals and predation may substantially reduce fish popula- tions (Gearin et al., 1986). In these circumstances, an understanding of pinniped diet becomes necessary for the management of endangered fish populations. In 1994, the National Marine Mammal Laboratory began a project to quantify harbor seal ^Phoca vitulina) predation on salmon in the lower Columbia River and to incor- porate marine mammals in salmonid population models. Harbor seals are the most abundant pinniped in the lower river and annual maximum counts can exceed 2000 on the largest haul-out site, a tidal sandbar adjacent to Astoria, Oregon. Pinniped prey are typically identified from fish sagittae (otoliths) recovered from fecal material (scat) and stomach contents (Brown. 1980; Harvey, 1989; Peirce and Boyle, 1991; Ochoa-Acuna and Francis, 1995; Beach et al.'). These methods yield biased results because of partial or complete digestion of otoliths and because of greater probabilities of recovering otoliths from larger individu- als and species with robust otoliths and of identifying otoliths of species with distinctive morphological characteris- tics (Harvey 1989; Gates and Cheal, 1992; Cottrell et al., 1996; Tollit et al., 1997; Bowen, 2000). Estimates of harbor seal predation on adult salmonids are particularly poor due to extremely low recovery (because the otoliths are small and fragile) and because harbor seals may not completely ingest large prey and thus otoliths may not be ingested (Pitch- er, 1980; Harvey 1989; Boyle et al., 1990; Harvey and Antonelis, 1994; Cottrell et al, 1996; Riemer and Brown, 1997). We describe harbor seal diet on the lower Columbia River during spring, summer, and fall from 1) otoliths and 2) other skeletal elements (cranial bones, vertebrae, teeth, gill rakers, etc. ) to examine potential differences in the diet characterized by the two methods. In previous studies, identification of all Beach, R. J., A. C. Geiger, S. J. Jefferies, S. D. Treacy. and B. L. Troutman. 1985. Marine mammals and their interactions with fisheries of the Columbia River and adjacent waters, 1980-1982. NWAFC (Northwest Alaska Fisheries Science Cen- ter) processed rep. 85-03, 316 p. NWAFC, National Marine Fisheries Service, Seattle, WA. 424 Fishery Bulletin 100(3) skeletal elements resulted in at least two times greater frequency of occurrence of some prey taxa than frequen- cies exclusively derived from otoliths (Riemer and Brown, 1997; Boyle et al., 1990; Cottrell et al., 1996). For major prey taxa, we estimated frequency of occurrence and mini- mum number of individuals from fish otoliths and other skeletal remains recovered from scats and average mass from otoliths. Methods During 1995, 1996, and 1997, scats were collected from Desdemona Sands (river km 26, 123°52'W,46°13'N), the largest harbor seal haul-out site in the lower Columbia River (Huber'^). Scats were collected intermittently during 1995. From March through August 1996 and from March through October 1997, we attempted to collect 50 harbor seal scats every two weeks at extreme low tides. This sam- pling period coincided with Columbia River runs of spring, summer, and fall chinook salmon. Scats were collected from haul-outs, and upon arrival at the laboratory were rinsed in nested sieves (2-mm, 1-mm, and 0.5-mm mesh width). All skeletal elements were recovered, dried, and stored in vials. Cephalopod remains were stored in 70% isopropyl or ethyl alcohol. Other invertebrate remains were relatively rare (<2'7r frequency of occurrence) and their contribution to the diet was disregarded because of difficulties enumerating individuals and determining pri- mary from secondary (prey within large, ingested fishes) prey. Otoliths were identified to lowest possible taxon, their anatomical location recorded (left or right side), and enumerated (number for left side and number for right side). Lengths of intact left or right otoliths were mea- sured parallel to the sulcus to the nearest 0.1 ram with an ocular micrometer Micrometer measurements were verified with hand-held calipers. Other skeletal structures (such as teeth, vertebrae, and cranial bones) were identi- fied to lowest possible taxon by comparing prey remains to reference samples (NMMLM. Scat collections were divided into three seasons: spring (samples collected prior to 15 May), summer (samples col- lected from 15 May to 15 July), and fall (samples collect- ed after 15 July). These dates distinguish runs of spring, summer, and fall chinook salmon crossing the Bonneville Dam (river km 235), less two weeks estimated for travel from the lower Columbia River (Fryer, 1998). For each sea- son, harbor seal diet was described by frequency of occur- rence (FO), minimum number of individuals (MND, and average prey mass estimated from otoliths of all major prey taxa. Frequency of occurrence (FO) of prey taxon^' in season k was defined as £o,M FO, = - where O,^^, = a binary variate indicating presence (1) or absence (0) of taxon 7 in sample i in season k; and s^ = the total number of scats containing identifi- able prey remains in season k. Rare prey taxa were grouped with similar taxa for analy- ses. Unknown prey remains that were clearly distinct from known taxa were considered "unidentified taxa" in sam- ples containing "identified" hard parts. Scats containing skeletal remains considered "unidentifiable," i.e. extremely eroded bone or fragmented material, were excluded from analyses. The minimum number of individuals (MND was estimated from the greatest number of left or right otoliths and unique or paired bone structures and expressed within each season as total MNI or average MNI per scat (total MNL/number of scats collected). Presence of non-unique fish remains (non-unique vertebrae, gillrakers, teeth) con- stituted a single individual. For example, if a scat sample contained five left otoliths, three right otoliths, and six atlas or axis vertebrae of a prey taxon, the MNI was six. FO and MNI were calculated from otoliths and again from all prey remains. Prey masses were estimated for the three seasons from allometric relationships between otoliths and body size (Harvey et al., 2000; Table 1). If relationships were unavailable for a species, regressions generated for a similar species were used. Otolith lengths were multiplied by a species-specific correction factor when available or an average correction factor to account for reduction in length due to digestion (Harvey, 1989). All intact left or right oto- liths of a prey taxa were measured, and estimated masses were averaged for each season. Suitable morphometric regressions were not available for several salmon species or did not include juvenile fish; therefore, we generated regressions including subadult age classes specifically for our study. In addition to pub- lished regressions, relationships between salmon otoliths and fish mass used in this study were calculated from Na- tional Marine Mammal Laboratory reference samples, the private collections of Walker* and NRC^ (Table 1 ). Because of the large discrepancy in masses of adult and juvenile salmonids, otoliths were identified to species and classified as adult or juvenile according to species-specific lengths estimated from regression equations. "Adults" described all returning upriver migrants, including reproductively mature individuals and jacks. "Juveniles" were seaward migrants and may have included two-year-old fish of some species (Groot and Margolis, 1991). Onchorhyncus ^Huber, H.R. 1997. Unpubl.data. NationalMarine Mammal Laboratory, 7600 Sand Point Way NE, Seattle, WA 98115. ' NMML(NationalMarineMammalLaboratory). 1997. Marine Mammal Prey Osteological Reference Collection. National Marine Mammal Laboratorv, 7600 Sand Pomt Way NE, Seattle, WA 98115. •• Walker. W. 1998. Unpubl. data. National Marine Mammal Laboratory, 7600 Sand Pomt Way NE, Seattle, WA 98115. ■> NRC (National Resources Consultants). 1998. Unpubl.data. Natural Resources Consultants, Lie, 4055 2P' Ave. W, Suite 100, Seattle, WA 981 19. Browne et al.: Improving pinniped diet analyses 425 Table 1 Predicted standard length [L=a+bx: L=cstiinated st; weight Ig], /j=estimated standard length) of common indard length |cm|, .v=otolith length |mm|) and weight {W=cL''; lV=estimated harbor seal prey and sources of data. Calculations for groups offish are based on species (in parentheses). Sockeyc salmon calculations are based on regressior s for silver sa mon. Taxon Species Length Weight Source a h c d Ammodytidae Pacific sand lance 0.727 0.137 0.0529 3.46 NMML' Clupeidae Pacific herring -1.85 5.24 0.0044 3.398 Harvey etal., 2000. American shad -11.08 11.46 0.0135 3.046 Harvey et al., 2000. Cottidac Pacific staghorn sculpin -2.26 2.58 0.011 3.229 Harvey etal., 2000. Embiotocidae shiner surfperch -0.52 1.74 0.01 3.515 Harvey etal., 2000. Engraulidae northern anchovy 0.85 2.28 0.0485 2,413 Harvey etal., 2000. Gadidae Pacific hake 0.96 2.04 0.0081 2.966 Harvey etal., 2000. Gadidae Pacific tomcod -3.51 1.77 0.0064 3.191 Harvey etal., 2000. Hexagrammidae Hexagrammids (lingcod) -6.0.3 8 0.0023 3.567 Harvey etal, 2000. Osmeridae whitebait smelt 3.02 2.11 0.0063 3.233 Harvey etal., 2000. eulachon -2.7 4.71 0.0077 3.075 Harvey et al., 2000. Pleuroneetidae Dover sole 12.23 2.75 0.0094 3.092 Harvey etal., 2000. English sole -2.76 3.82 0.0163 2.939 Harvey et al., 2000. rex sole -2.5 4.8 0.0238 2.692 Harvey et al., 2000. slender sole 1.08 3.37 0.0058 3.293 Harvey et al., 2000. starry flounder 0.23 3.35 0.0107 3.268 Harvey et al., 2000. Salmonidae Chinook salmon -10.4 6.73 0.0043 3.207 length: Walker^; weight: NRC' cutthroat trout -91.2 89.3 0.0155 2.97 Walker^ silver salmon 3.29 9.33 0.0103 3.092 length: NRC^; weight: Harvey et al., 2000. sockeye salmon (silver salmon ) 3.29 9.33 0.0103 3.092 length: Walker^ steelhead salmon -32.43 14.77 0.0275 2.895 Harvey et al., 2000. Scorpaenidae rockfishes (black rockfish) 8.7 1.6 0.1225 2.499 Harvey et al., 2000. Stichaeidae and Pholididae gunnel and prickelback (wattled eelpout) 12.42 5.22 0.0007 3.483 Harvey etal., 2000. gunnel and prickelback (Pacific sand fish) -4.57 6.06 0.0171 2.953 Harvey etal., 2000. ' NMML (National Marine Mammal LaboratorvL 1997. Marine Mammal Prey Osteological Reference Collection. National Marine Mammal Laboratory, 7600 Sand Point Way NE, Seattle, WA, 98115. - Walker, W. 1998. Unpubl. data. National Marine Mammal Laboratory, 7600 Sand Point Way NE, Seattle, WA 98115. ■' NRC (National Resources Consultants, Inc. ). 1998. Unpubl. data. National Resources Consultants, Inc. 4055 21^' Ave. W, Suite 100, Seattle, WA, 98119. clarki, O. kisutch, O. nerka, and O. mykiss less than 30 cm in length and O. tshawytscha less than 35 cm in length were considered seaward migrating juveniles (Groot and Margolis, 1991). All distinguishable salmon otoliths were identified to species and all identifications were verified by W. Walker. Annual and seasonal variations in frequency of occur- rence (FO) were examined with generalized linear models (Venables and Ripley, 1994). We limited our analyses to prey taxon with FO >5% during one or more seasons. Fre- quency of occurrence on each sampling date was modeled as a binomial random variable and for each prey taxon, we fit- ted five models: constant, season (S), year (Y"), season-i-year tS-i-y), and season-i-year with interactions iSxY). To account for overdispersion, we scaled Akaike's information criterion (AIC) using 6=residual deviance/degrees of freedom from the SxY model (Venables and Ripley, 1994). The model with the smallest scaled AIC was considered the best descriptor of seasonal and annual variation in FO. Results Over 1500 scats were collected from March 1995 through October 1997. Sample sizes varied among years and within season (Table 2). Frequency and number of indi- 426 Fishery Bulletin 100(3) Table 2 Sample collection dates, harbor seal scats with some identifiable prey remains, without any identifiable remains, and without remains for samples collected from Desdemona Sands 1995 through 1997. Spring {<15 May), summer (15 May to 15 July), and fall | (>15Jul y) designate timing of chinook salmon runs on the Columbia River. Harbor seal scats With Without Season Collection date identifiable remains identifiable remains Without remains Spring 5 Mar 1995 13 1 0 5 Mar 1995 29 2 0 14 Mar 1996 29 1 1 21 Mar 1996 11 1 1 10 Apr 1996 42 4 0 2-8 May 1996 44 2 1 11 Mar 1997 16 1 0 26 Mar 1997 7 4 0 10-11 Apr 1997 29 7 1 15 Apr 1997 22 6 0 28 Apr-1 May 1997 28 4 11 9-10 May 1997 45 6 12 Subtotal 315 39 27 Summer 18-19 May 1995 53 1 4 14-16 Jun 1995 81 1 0 28-29 Jun 1995 78 1 0 14 Jul 1995 32 3 0 30-31 May 1996 53 2 1 18-19 Jun 1996 50 1 1 2 Jul 1996 52 3 0 27 May 1997 34 6 10 6 Jun 1997 24 8 0 23 Jun 1997 47 8 9 8 Jul 1997 74 2 1 Subtotal 578 36 26 Fall 15 Aug 1996 78 1 0 29 Aug 1996 59 2 0 22 Jul 1997 64 5 6 4 Aug 1997 102 1 0 19 Aug 1997 56 1 0 3 Sep 1997 51 5 0 16 Sep 1997 41 6 0 16-17 Oct 1997 41 6 0 Subtotal 492 27 6 Total 1385 102 59 viduals consumed by harbor seals on the Columbia River were extremely v'ariable, even among sample collections fewer than two weeks apart. Effort and sample sizes were unequal for season and years but we chose to include all data to better describe harbor seal diet. More than 45 prey tEixa were described in 1385 samples with identifiable prey remains; however, most of the diet by number and frequency was composed of about 17 prey taxa (Table 3). Seasonal effects were important for 15 of 17 harbor seal prey with FO >5% (Table 3). Annual effects also were important for FO of lamprey (Lampetra spp. ), Pacific hake iMerluccius productus), and northern anchovy (EngrauUs morda.x), and year-season interactions were included for three prey taxa (Table 3). All taxa had variances of FO greater than predicted by a binomial model (over-disper- sion values, b>l; Table 3). Browne et al.: Improving pinniped diet analyses 427 Table 3 Total minimum number of individuals (MNI). I'roqucncy ol occurrt'nce (FO), .significant difT't'renccs in frequency of occurrence of major prey taxa identified from all skeletal remains recovered from harbor seal scat iS indicates season, Y indicates year, SxY indicates interaction, and N indicates no effects), and an estimate of over-dispersion of the binomial model (h). Only taxa with F'O. >0.05 in at least one season were examined. Prey taxon MNI (all remains) FO (all remains) Effect b Spring Summer Fall Spring Summer Fall Pacific herring 168 511 141 0.36 0.57 0.22 S 8.4 Pacific staghorn sculpin 256 170 284 0.41 0.19 0.25 SxY 2,4 Smelt species 133 204 62.5 0.28 0.18 0.35 S 6.2 Pacific tomcod 66 73 251 0.18 0.09 0.39 S 7.2 Lamprey species 109 204 41 0.26 0.25 0.06 S+Y 2.0 Starry flounder 136 105 160 0.30 0.14 0.12 SxY 2.7 American shad 36 108 116 0.11 0.19 0.22 N 6.1 Other flatfish 102 129 132 0.20 0.12 0.18 S 2.2 Pacific hake 30 72 1.36 0.09 0.12 0.28 S+Y 4.0 Shiner surfperch 26 131 69 0.07 0.18 0.08 S 2.2 Gunnel and prickelback 47 55 30 0.15 0.08 0.06 S 3.0 Juvenile salmonids 92 71 30 0.19 0.05 0.05 S 3.2 Northern anchovy 3 63 290 0.01 0.06 0.19 S+Y 2.9 Adult salmonids 22 33 50 0.06 0.04 0.10 N 4.6 Peamouth chub 12 63 41 0.03 0.08 0.05 SxY 2.4 Pacific sand lance 37 18 11 0.10 0.03 0.02 S 1.5 Rockfish species 23 14 18 0.07 0.02 0.03 S 1.5 The inclusion of all skeletal elements recovered from scat increased the MNI and FO of all harbor seal prey taxa (Table 4). The FO more than doubled for most taxa and usually was more affected by including all prey elements than was the average MNI (Table 4). We compared the MNI of several common harbor seal prey estimated from all structures to an estimate based on the number of re- covered otoliths multiplied by a species-specific correction factor for recovery rate (accounting for complete digestion of the otolith; Harvey, 1989, Fig. 1 ). A value of 1.0 indicated that the same estimate was derived from both methods, whereas 0.5 indicated that the MNI estimated from all structures was twice the estimate from otoliths. Seasonal variation was also apparent in estimated prey mass (Table 5). In some instances, estimates were based on very few otolith measurements, values were taken from the literature, or mass was averaged from other seasons when no intact otoliths were recovered (Table 51. Because some species were difficult to discern or regression relationships were unavailable, species were grouped by phylogeny or size similarities (Table 5). Smelts (Osmerids) were pooled by family and mass was estimated from whitebait smelt (Allosmerus elongatus), the most abundant species by distinguishable otoliths, and eulachon (Thaleichthys pacificus), although longfin smelt (Spirinchus thaleichthys) and surf smelt (Hypome- sus pretiosits) were occasionally identified in harbor seal scats. Although smelt otoliths could be distinguished by species, smelt bone could not. Smelt mass estimates were based on eulachon (in their relative proportion) and white- bait smelt because although less common than the other three similar size species, eulachon were much larger Masses of juvenile and adult salmonids were estimated separately for five species identified in harbor seal scat, with the exception of sockeye salmon, which was repre- sented by one otolith from an adult fish. Mass for sockeye was based on regressions generated for silver salmon. Cutthroat (Onchorhyncus clarki) and steelhead salmon (O. mykiss) were not often distinguishable by otoliths, and mass estimates were based on steelhead because of their numerical dominance in the lower Columbia River. The sticheid-pholid group included fish from a variety of families: three spine stickleback {Gasterosteus aculeatus), snake prickleback (Lumpetius sagitta). high cockscomb (Anoplarchus purpurescens), wattled eelpout (Lycodes pa- learis). Pacific sandfish (Trichodon trichodon), and saddle- back gunnel iPholis ornata). Although not taxonomically related, these species were seldom represented by otoliths, individuals were very small, and with the exception of gunnels, were rare by number and frequency (Table 4). Little has been published about relationships between otolith length and fish length or fish length and mass for 428 Fishery Bulletin 100(3) Table 4 Average minimum number of individuals (MNI) per scat (total number of individuals identifiec in all scats collected in a season/ total number of scats collected in a season and frequency of occurrence (FO) for all skeletal remains and for otoliths exclusively for | major prey taxa of harbor seals in the Col umbia River during spring, summer. and fal . All = all skeletal remains Prey taxa spring Summer Fall Average MNI/scat FO Average MNI/scat FO Average MNI/scat FO All Otoliths All Otoliths All Otoliths All Otoliths All Otoliths All Otoliths American shad 0.11 0.01 0.11 0.01 0.19 0.01 0.19 0.01 0.24 0.03 0.22 0.02 Cephalopod 0.01 0.01 0.02 0.02 0 0 Elasmobranch 0.08 0.07 0.01 0.01 0.01 0.01 Gunnel and prickelback 0.15 0 0.15 0 0.10 0 0.08 0 0.06 0 0.06 Hexagrammid 0.02 0 0.02 0 0.01 0 0.01 0 0.03 0 0.03 Lamprey species 0.35 0.26 0.35 0.25 0.08 0.06 Northern anchovy 0.01 0 0.01 0 0.11 0.06 0.06 0.02 0.59 0.52 0.19 0.10 Other flatfish 0.41 0.25 0.20 0.07 0.16 0.03 0.12 0.03 0.28 0.12 0.18 0.06 Pacific hake 0.10 0.02 0.09 0.01 0.12 0.01 0.12 0.01 0.28 0.02 0.28 0.02 Pacific herring 0.53 0.24 0.36 0.09 0.88 0.52 0.57 0.21 0.29 0.16 0.22 0.09 Pacific mackerel 0.02 0 0.02 0 0 0 0 0 0.01 0 0.01 0 Pacific sand lance 0.12 0.06 0.10 0.04 0.03 0.01 0.03 0.01 0.02 0.02 0.02 0.01 Pacific staghorn sculpin 0.81 0.62 0.41 0.20 0.29 0.17 0.19 0.08 0.58 0.42 0.25 0.11 Pacific tomcod 0.21 0.09 0.18 0.06 0.13 0.05 0.09 0.02 0.51 0.24 0.39 0.13 Peamouth 0.04 0.02 0.03 0.01 0.11 0.06 0.08 0.03 0.08 0.05 0.05 0.02 Rockfish species 0.07 0.02 0.07 0.02 0.02 0.01 0.02 0 0.04 0.01 0.03 0.01 Salmon species — adult 0.07 0.04 0.06 0.03 0.06 0.04 0.04 0.02 0.10 0.02 0.10 0.01 Salmon species^uvenile 0.29 0.16 0.19 0.06 0.12 0.10 0.05 0.02 0.06 0.02 0.05 0.01 Shiner surfperch 0.08 0.04 0.07 0.02 0.23 0.15 0.18 0,05 0.14 0.10 0.08 0.04 Smelt species 0.42 0.31 0.28 0.12 0.35 0.30 0.18 0.11 1.27 1.21 0.35 0.21 Starry flounder 0.43 0.24 0.30 0.11 0.18 0.09 0.14 0.05 0.33 0.28 0.12 0.06 Unidentified flatfish 0.04 0.02 0.03 0 0.02 0 0.02 0 0.02 0 0.02 0 these families; rather than ignore their occurrence, these species were pooled and mass estimates were based on measurements of wattled eelpout and sandfish otoliths (Table 5). The family Scorpaenidae was composed mostly of juvenile fish (Sebastes and Sebastelobus spp. ) which can seldom be distinguished to species from bones and otoliths. Mass estimates were based on black rockfish (Se- bastes melanops; Table 5). Morphometeric relationships for peamouth (Mylocheilus caurinus) were unavailable in the literature. Peamouth are small, slender members of the minnow family, less than 36 cm in length, with a shape similar to several other small harbor seal prey. Mass was assumed to be less than 100 g. Hexagrammids included lingcod (Ophiodon elongatus) and greenlings (Hexagram- mos spp.) also were poorly represented by otoliths. Mass was estimated from lingcod otoliths (Table 5). Because bones and otoliths were often difficult to identify to spe- cies, flatfish other than starry flounder were pooled and mass was calculated from the average of estimated masses of identified otoliths (rex sole, Glyptocephalus zachirus; English sole, Pleuronectes vetulus; Dover sole, Microsto- mus pacificus; rock sole, Pleuronectes bilineatus; slender sole, Eopsetta exilis) for each season. In contrast, starry flounder remains were easily identified and much more abundant than any other flatfish species (Table 4). Several taxa were not represented by otoliths because they were completely digested or because the species lacked otoliths. No intact Pacific mackerel {Scomber ja- ponicus) otoliths were recovered from scat and their mass was assumed to be less than 700 g, the upper limit re- ported by Eschmeyer and Herald (1983). Lamprey species included river (Lampetra ayresii) and Pacific lamprey (L. tridentata). Mass was estimated from the upper limit of outgoing Pacific and river lamprey from Pacific Northwest river systems (Beamish, 1980). Little information was available for predicting the mass of elasmobranchs; how- ever, all elasmobranchs (spiny dogfish, Squalus acanthias; and skates, Rajidae) consumed by harbor seals appeared to be juveniles. Elasmobranch mass was extrapolated from a regression of vertebral centrum width on mass from another skate species (Zeiner and Wolf, 1993), yield- ing an upper estimate of 490 g. We assumed the mass of skates and spiny dogfish consumed by harbor seals to be of a similar size, and all less than 500 g. Cephalopod (Loligo Browne et al.: Improving pinniped diet analyses 429 Smelt Shiner surf Pacific herring Pacific Sebastes spp Pacific lomcod Salmonid spp Salmonid spp species perch staghorn (juvenile) (adull) scuipin Prey taxa Figure 1 For eight prey taxa. ratio of estimated mean number of individuals estimated from the number of otoliths recovered from scat multiplied by a species-specific correction factor for recovery rate (Harvey, 1989) to MNI estimated from all structures recovered from harbor seal scats collected from Desdemona Sands during spring, summer, and fall 199.5 through 1997. Values for smelt species are based on correction factors for euch- alon. Values for salmonid species are based on correction factors for steelhead salmon. opalescens and Octopus spp.) mass was estimated from regressions of beak measurements on mantle length and mass (Wolf. 1982). Discussion Diet of harbor seals in the Columbia River Identification of prey remains indicated that the diet of harbor seals in the Columbia River was temporally vari- able and seals appeared to exploit prey when species were abundant. Many of the dominant prey by number and frequency were small fish such as herring, smelts, northern anchovy, juvenile flatfish, and sculpins (Tables 3 and 4). Pacific herring. Pacific staghorn scuipin iLeptocot- tus armatus), and smelts were three of the top six prey taxa by number and frequency for all three seasons (Table 4), although estimated masses varied greatly between season, indicating that seals preyed on different size classes (Table 5). Interestingly, scats without remains were most common during late spring and summer (Table 2). Olesiuk et al. ( 1990) reported similar results from Brit- ish Columbia and suggested harbor seals were feeding on soft-bodied prey and roe. Occurrence of these scats in our study coincided with pupping on the lower Columbia River ^ Huber, H. R. 1997. Unpubl. data. National Marine Mammal Laboratory, 7600 Sand Point Way, NE, Bldg. 4, Seattle, WA 98115. (Huber*'), but an alternative explanation is that these may have been from nursing pups. Most common harbor seal prey were variable by season or year (or both). Seasonal effects in the diet indicated pe- riods when prey were reproducing, when young of the year were available, or perhaps an absence of highly abundant prey when seals relied more heavily on consistently avail- able species. Annual differences in frequency of occur- rence may have been largely the result of differences in sample timing or prey cohort strength (Moyle and Cech, 1982; Dark and Wilkins, 1994). Significant year and inter- active effects between season and year probably reflected differences in prey year class (Table 3). For example, FO of anchovy (known to have high variability in recruitment) collected during fall of 1996 was 63%, whereas during fall of 1997 it was only 2%, although samples were collected on similar dates (Table 1). Harbor seals are generalist feeders and differences in frequency and number of prey probably reflect the temporal availability of prey rather than predator selection. This hypothesis is supported by AIC over-dispersion constants (6) greater than 1.0 for all prey taxa (Table 3). A binomial model assumes constant probabilities of a prey taxon occurring in scats collected on any sampling date within a season or year. Ephemerally abundant prey will have highly variable probabilities of occurrence in harbor seal scats collected in each season. It is likely that the overdispersion constant underestimates deviations for taxonomic gi'oups, including more than one species such as salmonids and smelts, because temporal abundance of the different species in the group may be offset. 430 Fishery Bulletin 100(3) Table 5 Average mass (g) of harbor seal prey, standard deviation (SD), and number of otoliths measured (n) for spring, summer, and fall prey of harbor seals in the Columbia River. Boldface values are estimates calculated from other seasons (when no intact otohths were recovered within a season) or from literature sources (when no structures were available for measurement!. Family Prey taxa Spring Summer Fall Avg. mass SD n Avg. mass SD n Avg. mass SD n Clupeidae American shad 198 22 4 517 601 5 523 284 11 Pacific herring 93 35 50 96 64 238 97 73 54 Engraulidae Northern anchovy 9 1 12 2 30 14 5 192 Osmeridae Smelt species 15 23 84 6 2 147 7 3 331 Gadidae Pacific tomcod 128 115 24 180 106 25 228 132 107 Pacific hake 67 1 421 1 446 292 2 Pleuronectidae Starry flounder 70 117 64 114 243 52 89 168 101 Other flatfish 181 100 35 181 111 42 225 82 35 Cottidae Pacific staghorn sculpin 140 84 136 160 84 88 115 42 45 Salmonidae Chinook juvenile 206 150 22 41 93 21 0 Chinook adult 1385 8515 6854 2 6862 1757 4 Cutthroat juvenile 225 56 255 66 6 315 1 Cutthroat adult 509 51 426 52 2 0 Silver juvenile 277 0 88 103 2 Silver adult 1607 983 671 241 15 4317 3545 3 Steelhead juvenile 488 283 81 2 0 Steelhead adult 1637 897 1 0 Sockeye adult 2832 0 0 Embiotocidae Shiner surfperch 79 39 85 43 80 79 42 45 Stichaeidae and Pholididae Gunnels and prickelbacks 90 84 0 97 38 3 84 127 3 Hexagrammidae Hexagrammids 2090 1727 0 3410 1375 2 756 135 2 Scorpaenidae Rockfish species 187 87 6 132 7 2 114 1 Ammodytidae Pacific sand lance 72 63 3 62 58 8 151 1 Elasmobranchs Elasmobranchs 500 0 500 0 500 0 Scombridae Pacific mackerel 700 0 700 0 700 0 Petromyzontidae Lamprey species 50 0 50 0 50 0 Ptychocheilus Peamouth chub 100 0 100 0 100 0 Cephalopods Cephalopods 21 1 0 21 1 0 0 Salmon in the harbor seal diet Harbor seals consumed several species and sizes of salmon throughout our study, but frequency was greatest during spring. Size of otoliths recovered indicated that most of these fish were juvenile chinook and that adult salmon were consumed to a lesser extent, primarily during fall. Fryer ( 1998) reported no difference in mean fork lengths of adult spring-summer O. tshawytscha with scars from pin- nipeds and those without scars at the Bonneville Dam and observed a greater percentage offish with scars earlier in the year, although these findings do not necessarily con- tradict data from our study. Scarred fish represent failed predation and perhaps harbor seals attempted to capture fish beyond their ability when spring run-off results in greater water turbidity. In addition, part of the discrep- ancy may be due to the classification of "adult" salmon. For our purposes, we categorized all fish with estimated lengths greater than outgoing migrants (30 to 35 cm depending on the species) as "adults." The mean lengths of scarred spring-summer O. tshawytscha from 1994 to 1996 (75.9 to 79.3 cm standard length) were greater than the mean length of "adult" fish estimated from prey remains (73.4 cm mean standard length). Riemer and Brown (1997) also examined all skeletal structures; however, their results summarized data for four years and 154 samples collected on eight dates. Riemer and Brown (1997) reported salmon FO as great as 39% for a single sampling date but found no salmonid remains in scats collected during February and March. Browne et a\ Improving pinniped diet analyses 431 Table 6 Harbor seal prey taxa ran ked by mininiuin number of ir dividuals (MNI) estimated frr m all skeletal structures and otoliths frc- quency of occurrence (FO) estimat xl from i 11 skeletal structures and from otoliths. and average mass of prey estimated from allo- metric relationships between otoliths and f sh size. Prey are ranked for all seasons i.e smelt are the most numerically abundant prey over spring, summer, and fall Data (as opposed to ranks) are presented in Tables 4 and 5. Prey taxa MNI all MNI otoliths TO all FO otoliths Mass Smelt species 1 1 3 1 21 Pacific staghorn sculpin 2 2 1 2 10 Pacific herring 3 4 1 3 15 Starry flounder 4 6 5 5 16 Other flatfish 5 7 8 7 7 Lamprey species 5 5 19 Pacific tomcod 7 7 5 6 8 American shad 8 15 4 14 4 Shiner surfperch 9 7 10 8 18 Pacific hake 10 14 9 14 Salmon species— juvenile 11 10 12 10 6 Northern anchovy 12 3 15 4 20 Gunnel and prickelback 13 11 17 Cephalopod 14 13 13 Pacific mackerel 15 14 3 Salmon species — adult 16 12 15 12 1 Peamouth 16 11 17 11 12 Pacific sand lance 18 13 IS 13 14 Rockfish species 19 16 19 8 10 Unidentified flatfish 20 4 20 Elasmobranch 21 21 5 Hexagrammid 22 22 2 Monthly FOs were based on a single sampling collection. During our study, FO of juvenile salmon was 50% on a sampling date during March of 1997; however, no salmon remains were found in scats collected a week earlier dur- ing 1995. Additionally, we found salmonid remains in harbor seal scat during every month of our data collection and we found no significant differences in the seasonal oc- currence of adult salmonids. This finding, however, may have been due to our grouping species. Harbor seals eat Columbia River salmon; however, they feed mostly on juvenile fish during the spring, and otolith identifications have indicated that most of these are Chi- nook salmon. Currently, salmonid bone cannot be identi- fied to species; however, the National Marine Mammal Laboratory is investigating identification of salmonid spe- cies from skeletal remains by using genetic techniques. Comparison of identirication methods Identifying and enumerating prey from all skeletal struc- tures is more time consuming than relying exclusively on otoliths and the described diet may be substantially dif- ferent. MNI and FO both increase when all structures are used, particularly for taxa such as Pacific tomcod (Micro- gadus proximus), Pacific hake, American shad (Alosa sapi- dissima), salmon spp., hexagrammids, elasmobranchs, and lampreys that are vastly underestimated from oto- liths or are entirely lacking in otoliths (Table 4). Relative importance of prey in the diet also may be dramatically affected (Table 6). Prey of the greatest estimated masses are ranked among the least important prey by number and frequency with the use of all skeletal elements and these prey are often completely absent from the diet described from otoliths. If one were to rely solely on otolith identifications, these prey could be entirely overlooked. The extrapolation of estimated biomass of each prey taxa from average mass estimated by otolith length and MNI estimated from all skeletal remains has a variety of cave- ats— namely, the relative contribution of large, infrequent prey may be vastly underestimated by using otoliths. Although the identification of all structures increases the magnitude of both MNI and FO, estimates of the number of individual prey are likely to be less accurate. A description of pinniped diet from all prey remains is subject to many of the same biases inherent in otolith identifications. Identifi- cation of all skeletal structures assumes an equal probabil- 432 Fishery Bulletin 100(3) ity of detecting all prey species and of recovering remains after consumption; these assumptions were violated to some extent in our study because, like otoliths, passage and identification of prey structures are taxon-specific (Harvey, 1989; Cottrell et al., 1996; Marcus et al., 1998). In captive experiments, herring (Clupea pallasi) were identified by 11 structures recovered in scats (other than otoliths), whereas smelt were represented only by verte- brae (Cottrell et al., 1996). Smelt vertebrae cannot be used to enumerate individuals, whereas several herring bones commonly recovered in scats are unique or are structures with definite sides (prootics, atlas and axis vertebrae), and even highly eroded herring bone retain characteristics identifiable to species. In contrast, bones of some taxa such as pleuronectids erode rapidly, losing species traits and are identified only to family. Given these factors, the identification of all skeletal structures represents the rela- tive consumption of some prey more accurately than oth- ers. Further, MNI estimated from all skeletal structures is not corrected for complete digestion of structures useful for enumerating individuals. To date, there are no correction factors for complete digestion of bones and this lack, doubt- lessly, is a source of substantial bias. Behavior of both predator and prey also affects identifi- cation and enumeration of prey remains in scats. Small, schooling fish, such as smelts, are more likely to be con- sumed in greater numbers than larger, solitary fish such as hexagrammids. Smelts are most frequently identified from vertebrae; therefore MNI is more severely under- estimated because more than one individual is likely to be consumed. Captive feeding studies also have indicated that the activity of the pinniped, its meal size, size of prey, and the physical structure of the prey bone all affect pas- sage rate and the degree of erosion (Cottrell et al., 1996; Tollit et al., 1997; Marcus et al., 1998; Bowen, 2000). The estimation of MNI from all prey structures recov- ered in scats presents a variety of complications; however, the alternative — using otolith correction factors — also has problems. Otolith correction factors based on recov- ery rates from feeding experiments are highly variable between repeated trials of the same individuals, differ- ent individuals, and different pinniped species (Harvey, 1989; Cottrell et al, 1996; Tolht et al., 1997; Bowen, 2000). Although his results were inconclusive, Bowen (2000) sug- gested that differences in activity levels may account for much of the variability in digestion (and correction fac- tors). Activity levels among wild harbor seals are likely to be more variable than those between captive harbor seals with and without access to water, and thus otolith correc- tion factors may yield erroneous estimates of individuals consumed by free-ranging predators. Limiting analyses of diet to qualitative measures such as FO will reduce the bi- ases of including bone identification; however, the overall importance of frequent, small prey may be much less than indicated by their relative frequency. Pinnipeds are considered generalist feeders and may feed on large numbers of abundant, frequently encoun- tered prey; however, if mass is considered a measure of im- portance, they may be sustained by infrequent, large prey (Tables 4—6). A few prey species were both abundant and frequent (herring, smelts, sculpins, flatfish), yet their esti- mated masses were small. Large, infrequent species such as lingcod, hake, rockfish, and salmon may contribute more total mass to a hypothetical "meal." Unfortunately, these prey were also poorly represented by otoliths in our study and therefore our mass estimates may be inaccurate. All methods of examining marine mammal diets, such as fecal analyses, stomach lavage, or stomach content analysis, are inherently biased to some degree. Biases of fecal analyses have been discussed at length in the lit- erature; however, fecal analysis remains the least inva- sive and least expensive technique and allows for large sample sizes. Identification of all skeletal elements, rather than otoliths exclusively, is an improvement on other tech- niques. Although results are still subject to biases, prey taxa represented by hard parts in fecal material represent a minimum estimate of prey consumed. In addition, an ex- amination of skeletal elements other than otoliths is man- datory for assessing the impact of harbor seals on certain prey species — protected salmon stocks, for example. Acknowledgments The authors would like to thank M. Gosho, B. Hanson, H. Huber, K. Hughes, S. Melin, and L. Lehman for their assis- tance in the field and laboratory. They thank S. Reimer and W. Walker for assistance with prey identifications. This manuscript was improved by comments from S. Mizroch, D. Withrow, and the comments of three anonymous reviewers. Literature cited Beamish, R. J. 1980. Adult biology of the river lamprey (Lanipetra ayresi) and the Pacific lamprey (Lampetra tridentata) from the Pacific coast of Canada. Can. J. Fish. Aquat. Sci. 37:1906- 1923. Bowen, W.D. 2000. Reconstruction of pinniped diets: accounting for com- plete digestion of otoliths and cephalopod beaks. Can. J. Fish. Aquat. Sci. 57:898-905. Boyle. P R., G. J. Pierce, and J. S. W. Diack. 1990. Sources of evidence for salmon in the diet of seals. Fish. Res. 10:137-150. Brown, R. F 1980. Abundance, movements and feeding habits of the harbor seal, Phoca vitulina, at Netarts Bay, Oregon. M.S. thesis, Oregon State Univ., Corvallis, OR, 69 p. Cottrell, P E., A. W. Trites, and E. H. Miller 1996. Assessing the use of hard parts in faeces to identify harbour seal prey: results of captive-feeding trials. Can J. Zool. 74:875-880. 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, 73 p. U.S. Dep. Commer., NOAA Tech. Rept. NMFS-117. Eschmeyer, W. N., and O. W. Herald. 1983. A field guide to Pacific coast fishes of North America, 336 p. Houghton Mifflin Co., Boston, MA. Browne et a\: Improving pinniped diet analyses 433 Fryer, J. K. 1998. Frequency of pinniped-caused scars and wounds on adult spring-summer chinook and sockeye salmon returning to the Columbia Kiver N. Am. J. Fish Manag. 18:46-51. Gates, N. J., and A. J. Cheal. 1992. Estimating the diet composition of the Australian sea-lion (Neophoca cinerea) from scat analysis: an unreli- able technique. Wild. Res. 19:447-456. Gearin, P., B. Ffeifcr and S. Jefferies. 1986. Control of California sea lion predation of winter- run Steelhead at the Hiram M. Chittenden Locks, Seattle, December 1985-April 1986, with observations on sea lion abundance and distribution in Puget Sound. Wash. Dep. Game Fish. Manag. Rep. 86-20, 108 p. Groot, C, and L. Margolis. 1991. Pacific salmon life histories, 564 p. Univ. British Columbia Press, Vancouver, B.C. Harvey, J. T. 1989. Assessment of errors associated with harbour seal iPhoca vitulina) faecal sampling. J. Zool., Lond. 219:101- 111. Harvey, J. T., and G. A. Antonelis. 1994. Biases associated with non-lethal methods of deter- mining the diet of northern elephant seals. Mar. Mamm. Sci. 10:178-187. Harvey J. T., T. R. Loughlin, M. A. Perez, and D. S. Oxman. 2000. Relationship between fish size and otolith length for 63 species of fishes from the eastern north Pacific Ocean. U.S. Dep. Commer, NOAA Tech. Rep. 150, 36 p. Laake. J. L., P. Browne, R. L. DeLong, and H. R. Huber 2002. Pinniped diet composition: a comparison of estima- tion models. Fish. Bull. 100:434-447. Marcus, J., W. D. Bowen. J. D. Eddington. 1998. Effects of meal size on otolith recovery from fecal sam- ples of Gray and harbor seal pups. Mar Mamm. Sci. 14: 789-802. Moyle, R B., and J. J. Cech Jr. 1982. Fishes: an introduction to ichthyology, 592 p. Pren- tice-Hall Inc., Englewood Cliffs, NJ. NMFS (National Marine Fisheries Service). 1997. Investigation of scientific information on the impacts of California sea lions and Pacific harbor seals on salmo- nids and on the coastal ecosy,stem of Washington, Oregon, and California. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NWFSC-28, 172 p. Ochoa-Acuna, H., and J. M Francis. 1995. Spring and summer prey of the Juan Fernandez fur sea\,Arctocephalus philippii. Can. J. Zool. 73:1444-1452. Olesiuk, P F, M. A. Bigg, G. M. Ellis, S. J. Crockford, and R. J. Wigen. 1990. An assessment of the feeding habits of harbour seals iPhoca vitulina) in the Strait of Georgia, British Columbia, based on scat analysis. Can. Tech. Rep. Fish. Aquat. Sci. 1730,135 p. Pierce, G. J., and P. R. Boyle. 1991 . A review of methods for diet analyses in piscivorus ma- rine mammals. Oceanography and Marine Biology Ann- ual Review 29:409-486. Pitcher, K. W. 1980. Stomach contents and feces as indicators of harbour seal foods in the Gulf of Alaska. Fish. Bull. 78:544-549. Riemer, S. D., and R. F. Brown. 1997. Prey of pinnipeds at selected sites in Oregon identi- fied by scat (fecal) analysis, 1983-1996. ODFW (Oregon Dep. Fish and Wildlife) Wildlife Diversity Program Tech. Rep. 97-6-02, 34 p. Tollit, D. J., M. J. Steward, P. M. Thompson, G. J. Pierce, M. B. Santos, and S. Hughes. 1997. Species and size differences in the digestion of oto- liths and beaks: implications for estimates of pinniped diet composition. Can. J. Fish. Aquat. Sci. 54:105-119. Venables, W. N., and B. D. Ripley 1994. Modem applied statistics with S-Plus, 462 p. Springer- Verlag, New York, NY. Wolf G. A. 1982. A beak key for eight eastern tropical Pacific cephalo- pod species with relationships between their beak dimen- sions and size. Fish. Bull. 80:357-370. Zeiner. S. J.,andPWolf . 1993. Growth characteristics and estimates of age at matu- rity of two species of skates (Raja bmoculata and Raja rhina) from Monterey Bay, California. In Conservation biology of elasmobranchs, p. 87-99. U.S. Dep. Commer., NOAA Tech. Rept. NMFS 115. 434 Abstract— Along the west coast of the United States, the potential impact of increasing pinniped populations on declining salmonid (Oncorhynchus spp. ) stocks has become an issue of concern. Fisheries managers need species-spe- cific estimates of consumption by pin- nipeds to evaluate their impact on salmonid stocks. To estimate consump- tion, we developed a model that esti- mates diet composition by reconstruct- ing prey biomass from fecal samples. We applied the model to data collected from harbor seals (Phoca vitulina) that are present year-round in the lower Columbia River where endangered stocks of salmonids pass as return- ing adults and as seaward-migrat- ing smolts. Using the same data, we applied the split-sample frequency of occurrence model, which avoids recon- structing biomass by assuming that each fecal sample represents an equal volume of consumption and that within each sample each prey item represents an equal proportion of the volume. The two models for estimating diet compo- sition yielded size-specific differences in consumption estimates that were as large as tenfold for the smallest and largest prey. Conclusions about the im- pact of harbor seal predation on adult salmonids, some of their largest prey species, remain uncertain without some appropriate rationale or further infor- mation (e.g. empirical captive studies) to discrimmate between these models. Pinniped diet composition: a comparison of estimation models Jeffrey L. Laake Patience Browne Robert L. DeLong Harriet R. Huber National Marine Mammal Laboratory Alaska Fishenes Science Center National Marine Fishenes Service, NOAA 7600 Sand Point Way NE Seattle, Washington 98115 E-mail address (for J. L Laake): |eff laakeig'noaa gov Manuscript accepted 15 February 2002. Fish. Bull. 100:434-447 (2002). During the last three decades, harbor seal {Phoca vitit!i/ia) and California sea lion iZalophus californianux) popula- tions along the west coast of the United States have increased dramatically (Forney et al., 2000). During the same period, numerous salmonid (Oncorhyn- chus spp.) stocks that are consumed by these pinnipeds have declined and some of these stocks have been clas- sified as threatened or endangered (NMFS, 1997). To evaluate the impact of pinnipeds, fisheries managers need species-specific estimates of salmonid consumption by pinnipeds. In some limited situations, pinniped prey consumption can be determined from direct observation if the pinniped brings the prey to the surface and feed- ing occurs in a few predictable areas (Bigg et al., 1990). However, in most situations, consumption estimates have relied on a less direct approach that uses estimates of pinniped energetic requirements, prey energy density, and pinniped diet composition (Olesiuk, 199.3; Hammill et al., 1997; Stenson et al., 1997; Nilssen et al., 2000). In their simplest form, these models express biomass consumption of prey species / as j5, = s^^/e,. where ^ is the total en- ergy requirement of the pinnipeds, re, is the proportion of the energy in the diet derived from species i, and e^ is the energy density (kcal/g) of species ;. The total energy requirement of a popula- tion depends on the size of the popu- lation and requirements of each seal, which vary by sex, age, and status (i.e. whether it is molting, pregnant, lactat- ing). If prey energy density is constant or an average energy density is used, biomass consumption can be expressed as fi, = i^Tt,, where ^ is the total biomass requirement and 7t, is the proportion of the biomass derived from species i. Valid estimates of pinniped energetic requirements and prey energy density are important, as well as accurate esti- mates of pinniped diet composition. Diet composition can be determined from skeletal remains in scat (feces) and from stomach and intestinal con- tents, or stomach lavage. Each of these methods has some inherent bias (Bigg and Perez, 1985; Pierce et al., 1991). Examination of prey remains in scat is noninvasive and allows for the larg- est sample size. There are, however, numerous well-recognized problems in describing marine mammal diet from scats (Jobling, 1987; Harvey, 1989; Har- vey and Antonelis, 1994; Tollit et al., 1997b; Marcus et al, 1998). In particu- lar, nonrandom passage of hard parts, primarily otoliths, biases estimates of diet composition; however, the bias can be reduced by inclusion of all hard parts (e.g. bones) (Browne et al., 2002). In 1994, we began an investigation of harbor seal consumption of salmonids in the lower Columbia River. Initial attempts to survey the lower 110 km stretch of the river to estimate adult salmonid consumption by direct ob- sei^ation proved infeasible. The only remaining noninvasive alternative was to develop a consumption estimate Laake et al ; Pinniped diet composition 435 based on analysis of scat colloctions. During 1995-97, harbor seal scats were collected at the Dusdemona Sands haulout from 1 March to 15 October. By combining diet composition obtained from scat analysis and contempo- raneous surveys of seal abundance, we estimated the average consumption of salmonids and other prey in the Columbia River by harbor seals during spring, summer, and fall of 1995-97. In our study, we focussed on the method of estimating diet composition from a sample of scats. We describe an estima- tor for diet composition based on reconstruction of the prey biomass represented in the scat and show how it is related to an alternative estimator described by Olesiuk (1993). Using the data collected on harbor seals in the Columbia River, we demonstrate the sensitivity of the consumption estimates to the method for estimating diet composition. Materials and methods parts may be affected by factors that influence digestion and deposition (e.g. seal activity level). Also, it may not always be possible to collect an entire scat or even reason- ably define a scat as a discrete entity. Thus, even though consumption by seals varies to some degree, the biomass reconstructed from a scat may vary much more than the variation in consumption. Thus, one could argue that each scat should be treated as a "representative" variable-size sample of a nearly constant amount of biomass consumed during some feeding interval. With that conceptual sam- pling model, the most appropriate estimator would be a simple average of the proportions in each scat: I*^' (0 V X^.'- *=i X "'*"''* (2) Diet composition models If we could randomly select a sample of /( prey items con- sumed by pinnipeds, a ratio estimator (Cochran, 1977) would be appropriate to estimate the proportion of bio- mass (./r,) represented by the ;'*' prey species from w pos- sible prey species: where 6^^ = the biomass of species /; «,;(. = the number of species i consumed; and u\f, = the average mass of species / in the k*^ scat. Equation 2 is similar to the estimator of Olesiuk (1993), which he called split-sample frequency of occurrence (SSFO): Z^' S"'"' (1) where 6, = the total biomass of the n^ prey items that are species /; w^ = 6, /«, = the average mass for species i; and "=X"'- /,* s *=1 (0 1=1 (3) where /^^ = an indicator variable which equals 1 if one (or more) prey items of species ; is in scat k, and 0 otherwise. Prey hard-parts in a scat represent a filtered selection of the prey species that were consumed by a single animal over some unknown and variable amount of time. Captive feeding studies have shown that a scat does not represent a single meal or even a single fixed period of feeding time (Harvey, 1989). Moreover, pinnipeds are unlikely to con- sume the same amount of prey in each meal or in a speci- fied amount of time. Therefore, the biomass represented by the prey remains in a scat is unlikely to be constant because consumption varies. From a collection of s scats in which each scat represents a variable amount of biomass that is proportional to consumption, the ratio estimator (Eq. 1) is also appropriate, where 6, is the total biomass of species ; from the s scats. We will refer to (Eq. 1 ) as biomass reconstruction (BR), which is equivalent to the estimator used by Harvey (1988) and Hammond and Rothery (1996). Alternatively, one could argue that the biomass recon- structed from prey remains in a scat may vary for numer- ous reasons other than consumption. Variation in scat volume and production and the resulting amount of hard Equation 2 is equivalent to 3 when you make Olesiuk's (1993) assumption that an equal amount of biomass of each species in the scat was consumed. SSFO requires only a determination of the presence or absence of the prey in a scat, and thus, it is much easier to implement than either Equations 1 or 2, which require an enumera- tion of the individual prey in each scat and their mass. Enumeration of prey in a scat sample is straightforward with unique structures such as otoliths. However, by using nonunique hard parts (e.g. gillrakers, vertebrae) to reduce selection bias resulting from unequal digestibility of oto- liths, problems are introduced with enumeration. When prey are exclusively represented in scat by nonunique structures, it may be possible only to determine that a sin- gle individual was consumed or at best a minimum num- ber can be constructed by enumerating nonunique struc- tures and dividing by the average number of structures per fish (e.g. count of vertebrae divided by average number of vertebrae per fish). By including nonunique structures, enumeration of prey is replaced with an estimate of the 436 Fishery Bulletin 100(3) minimum number of prey («,) consumed. For some hard parts, it may not be possible to identify the species, but the hard part can be classified to a group of species, such as family. In these cases, the species can be grouped into a single unit for estimation (e.g. all hexagrammids) or the unidentified prey can be partitioned to species based on the sample of species-specific hard parts. For example, most salmonid bones are currently indistinguishable to species; therefore salmonids in scats represented by bones (u) can be apportioned into species from proportions ()^) obsei-ved from otoliths (o, ). If we denote f to be the set of all identified salmonids represented by otoliths, the num- ber of salmonid prey in species ; can be estimated as where -o, +u 7, for / e f . y, = =i^ for (■ 6 T (4) (5) Pinniped diet can be quite variable in response to prey availability (Pierce et al., 1991; Tolht et al., 1997a; Browne et al., 2002; Beach et al.'; Brown et al.^). Because scat collected at a single date may reflect what was lo- cally available at that time, collecting scats at different times throughout a season or year will provide a better representation of diet. From scats collected across several occasions, what is the best way to estimate average diet composition? If the amount of scat collected and resulting reconstructed biomass for an occasion is proportional to the amount of prey consumed, the data should be pooled for a single ratio estimate. However, in many cases the amount of scat collected will reflect a multitude of factors, such as tide height, storms, human disturbance, and the length of time that the seals were at the haulout prior to collection. Also, in many circumstances a fixed number of scats are collected rather than some fixed proportion of the scats available at the haulout. Therefore, we suggest that an average of the proportions (ratios) is appropriate. If there are T occasions, the average proportion is Prey mass can be determined from morphometric re- lationships between hard part dimensions (e.g. typically otolith length) and mass. Adjustments should be made to account for partial digestion of the hard part. From regres- sions between otolith length (corrected for degradation) and fish length and between fish length and mass, an esti- mate of the average mass of prey represented in scat can be constructed (Harvey et al., 2000; Browne et al., 2002). In some scats, a prey species may be represented only by hard parts (e.g. gillrakers) that do not have a quantifiable relationship to mass. We have to assume that the prey with unknown mass are represented by the average mass determined from the measurable hard parts (e.g. unbro- ken otoliths) of that species. Thus, the biomass cannot strictly be measured but must be estimated by using estimated average mass (w^) and estimated number of individual prey (r\): X'^" (6) b, = n, -^ where o* = the number of hard parts (typically unbroken otoliths); and ii',, = the estimated mass derived from the regres- sions relating otolith length to fish mass. To estimate number of prey consumed (P, ) rather than bio- mass, estimated biomass is divided by average mass: w, w, (7) where p, the number of prey species / consumed per unit of biomass consumed. (8) Often there are seasonal shifts in diet resulting from prey availability (Olesiuk et al., 1990; Tollit and Thompson, 1996; Browne et al., 2002). In such cases, the analysis should be stratified by season. If data are collected over several years, again a simple average of the seasonal proportions is appropriate. In the appendix, we provide variance estimators for diet composition and consumption estimates for data collected over several years stratified by season. Data collection and analysis For the data used in this paper, Browne et al. (2002) have described the scat collection and analytical methods and have provided descriptions of the food habits from these data. As in Browne et al. ( 2002 ), we stratified our data col- lection and analysis into three seasons, spring ( 1 March- 14 May), summer (15 May-15 July) and fall (16 July-15 October), based on the timing of chinook salmon runs at the Bonneville Dam, offset by two weeks to account for the travel of salmonids from the lower Columbia River to the Dam (at river km 235). We describe here additional meth- ' Beach, R. J., A. C. Geiger, S. J. Jefferies, S. D. Treacy, and B. L. Troutman. 1985. Marine mammals and their interactions with fisheries of the Columbia River and adjacent waters, 1980-1982. NWAFC Processed Rep. 85-03, 316 p. Northwest and Alaska Fisheries Sci. Cent., Nat. Mar. Fish. Serv., NOAA, 7600 Sand Point Way NE, Seattle, WA 98115. 2 Brown, R. F, S. D. Riemer, and S. J. Jeffries. 1995. Food of pinnipeds collected during the Columbia River Area Commer- cial Salmon Gillnet Observation Program, 1991-1994. Wildlife Diversity Program Tech. Rep. 95-6-01, 16 p. Oregon Depart- ment of Fish and Wildlife, 2501 SW 1st Ave., PO Box 59, Port- land, OR 92707. Laake et al : Pinniped diet composition 437 ods for development of estimates of prey consumption by harbor seals. Prey remains were usually identified to species, but in some cases could only be identified to genus, family, or larger taxon (e.g. flatfish). Our primary interest was salmonid consumption; therefore, where possible, we also classified salmonids as juvenile or adult. For non- salmonids, we were less interested in species-specific estimates of consumption: therefore we did not always identify nonotolith remains to species when identification was either time-consuming or uncertain. For example, we divided flatfish into starry flounder (Platichthys stellatus) and "other flatfish" because starry flounder were easily identified, but the remaining flatfish species were not eas- ily distinguishable. In some cases, prey remains could be identified to species, but they occurred infrequently; there- fore we grouped them by family (e.g. Hexagrammidae) or groups of families (e.g. Stichaeidae and Pholididae). We have used the term "prey group" to generically refer to our classification of prey remains into family (or more general taxon), species, or species and size. Both species and size of salmonids could be determined from unbroken otoliths. For most salmonids, even broken otoliths were classified as adult or juvenile because of a very apparent discontinuity in otolith size between adults and juveniles. However, for coho salmon (O. kisutch) and cutthroat trout (O. clarkii), the size difference between ju- veniles and adults was less obvious; therefore we did not classify broken otoliths. We partitioned coho salmon and cutthroat trout broken otoliths into adults and juveniles according to the observed seasonal proportion of unbroken otoliths for each species. Salmonids represented exclusive- ly by bone were not separated by species or size, but were apportioned according to seasonal average proportions measured from otoliths. The lengths of all unbroken otoliths were measured to compute an average mass for most prey. We corrected the measured length for an average amount of degrada- tion (Browne et al., 2002). For Pacific mackerel (Scomber japonicus). elasmobranchs (sharks and skates), lamprey (Petromyzontid spp.) and peamouth chub (Mylocheilus caurinus), size relationships were not available; therefore literature values of average mass were used (Browne et al., 2002). For many species, there were very few measurable otoliths for an individual collection date; therefore an aver- age weight was computed across all collection dates within that particular season and applied to each collection date during that season. If there were 10 or fewer hard parts measured in each season, we used the average and vari- ance of the predicted weights from the data pooled over the three seasons (i.e. we assumed no seasonal differences). The amount of prey biomass ( t) required to sustain the harbor seals in the Columbia River during a season is a function of seal abundance (A'^), the age and sex propor- tions (QJ and the sex- and age-specific daily biomass re- quirements of the seals (C^), and the length of the season (D): We confined our analysis to nonpup (>6 months) seals because weaned pups comprise a small proportion of the seals in the Columbia River and they primarily consume soft-bodied prey or crustaceans (Pitcher, 1980; Riemer and Brown') which could not be incorporated into our esti- mates of diet composition. The average number of seals was determined from aerial surveys that were flown over seal haulout sites. Pups and nonpups were counted from photographic slides or they were counted during flights over small haulout sites. Aerial surveys were flown on 10 occasions between March and July 1995, on 16 occa- sions between March and June 1996, and on 25 occasions between March and September 1997. In 1997, radio-tags and visual markers were attached to 26 seals (8 adult males, 10 adult females, and 8 subadults) to estimate the average proportion of nonpup seals that were hauled out during the surveys if) with the techniques described by Huber et al. (2001). The correction factor was used for all of the counts to construct average seasonal abundance estimates. Abundance of nonpup seals in each season was estimated by A^, f (10) where c = the average count of nonpup seals hauled out during season j. Age (other than unmolted pups) and sex of seals cannot be determined from aerial surveys; therefore, estimating the sex and age structure would require capturing seals at different times throughout the year. Instead, we relied on a predicted sex and age structure based on life-his- tory table data (Bigg, 1969; Pitcher and Calkins*), but rescaled the proportions to the nonpup portion of the population. We used the following sex and age structure for nonpup seals (d^): a=l, 23% subadult (1-4 yr); a=2, 35% adult males (>4 yr); a=3, 42% adult females (>4 yr). We assumed the following biomass requirements for the three groups: Ci= 1.89 kg/d , C,= 2.37 kg/d. and C3= 2.63 kg/d. We derived these values by averaging the age-spe- cific daily biomass requirements given by Olesiuk ( 1993). We did not include any variability in our estimates of bio- mass requirements, nor did we include any uncertainty in the estimates of the population structure (0^). Therefore, variability in ^ only included variation in the population estimate. ^ = iV^e„QD. (9) ^ Riemer, S. D.. and R. F. Brown. 1997. Prey of pinnipeds at selected sites in Oregon identified by scat (fecal) analysis, 1983-1996. Wildlife Diversity Program Tech. Rept. 97-6-02. 34 p. Oregon Department of Fish and Wildlife. 2501 SW 1st. Ave.. PO Box 59, Portland, OR 92707. 4 Pitcher, K. W, and D. G. Calkins. 1979. Biolog>' of the harbor seal, Phoca vitulina richardsi, on Tugidak Island, Gulf of Alaska. Final Rep. to OCSEAP (Outer Continental Shelf Environmental Assessemnt F*rogram), Dept. Interior, Bur. Land Manage., 72 p. 438 Fishery Bullelin 100(3) Table 1 Apportionment of unidentified salmonid remains basec on proportions of identified salmonid remains as species / in season/ (j;^). Species Age class Proportions ();,) Apportioned unidentified ( y.,",^ Spring Summer Fall Spring Summer Fall Chinook Adult 0.038 0.083 0.462 2.3 2.3 30.9 Juvenile 0.538 0.389 0.077 31.8 10.9 5.2 Coho Adult 0.051 0.292 0.231 3.0 8.2 15.5 Juvenile 0.026 0.000 0.154 1.5 0.0 10.3 Cutthroat Adult 0.144 0.049 0.000 8.5 1.4 0.0 Juvenile 0.144 0.146 0.077 8.5 4.1 5.2 Sockeye Adult 0.019 0.000 0.000 1.1 0.0 0.0 Steelhead Adult 0.019 0.014 0.000 1.1 0.4 0.0 Juvenile 0.019 0.028 0.000 1.1 0.8 0.0 Totald/,) 59 28 67 Table 2 Average and coefficient of var ation (in parentheses) of the count (c, of hau ed-out nonpup seals, estimated abundance (A'^ ) and prey biomass requirements ^, ) for each season. Season Avg. count Abundance Required biomass (?) Spring (1 Mar-14 May) 1012(0.11) 1659(0.12) 296.34(0.12) Summer ( 15 May-15 July) 823(0.07) 1349(0,10) 193.65(0.10) Fall(16July-15 0ct) 598(0.15) 980(0.16) 214.21(0,16) Results Scats were collected on 31 occasions during March-Octo- ber 1995-97 providing 1385 scats with identifiable prey remains (Browne et al., 2002). All seasons were sampled in each year except for the fall of 1995. Remains were identified from 5832 different prey which were assigned to (0=28 prey groups. We excluded the very minor unidenti- fied component. Salmonids of unknown species and size represented by bone (154) were partitioned based on the observed seasonal proportions (Eqs. 4 and 5, Table 1). During 16 of the flights in 1997, the proportion of the 26 radio-tagged seals hauled out was measured to construct a single average correction factor (\lf) of 1.64 (percent coefficient of variation (CV)=6.7%) to estimate seasonal abundance from the haulout counts (Table 2). Using the assumed age and sex structure and biomass requirements, we estimated that the seals in the Columbia River would consume 704 metric tons (t) of biomass during the 7.5 month period (Table 2). From the 1385 scats, we reconstructed 1.15 t of biomass which was only 0.16% of the required biomass consump- tion. Because the number of scats and seals varied be- tween seasons, the percentages varied from 0.077% in spring, 0.173% in summer, and 0.272% in fall. Using BR, we constructed diet composition estimates (Fig. 1) and seasonal consumption estimates (Fig. 2). Many of the estimates of salmonid consumption had exceedingly poor precision with the coefficient of variation exceeding 0.5 (Table 3). Using adult chinook salmon (O. tshawytscha) as an example, an examination of the variance components demonstrated that estimation of biomass (Table 4) was the predominant source of variance. Variance associated with biomass estimation includes variation in predicted weights and estimation of number of prey (n,) for salmo- nids. The latter accounted for one- to two-thirds of the total variance for salmonids depending on the species and size group. The use of genetics to obtain species identifica- tion of the unknown salmonids identified by bone would substantially improve the precision for salmonids. Using BR to estimate diet composition, adult chinook salmon was the only salmonid that was consistently in the five most important prey items for each season based on percent of biomass (Fig. 1). Their importance was derived from their average mass which was the largest of all the prey. Adult chinook salmon and the other salmonids ap- peared much less important if the ranking was based on the number consumed (Fig. 2). Smaller prey items such as herring iClupeid spp.), sculpin iLeptocottus armatiis), lamprey (Petromyzontid spp.), smelt iOsmeridae) and Laake et al : Pinniped diet composition 439 Figure 1 Average biomass proportion (/r^) of each prey group in the diet of harbor seals on the Columbia River between 1995 and 1997 for spring (A), summer (B), and fall (C). Table 3 Average seasonal estimates of the number of salmonids consumed by harbor seals during 1995- of variation (CV) based on the biomass reconstruction (BR) method for diet composition. -97(P,)(m 1000s) and its coefficient Species Size Spring Summer Fall P, CV P, CV P, CV Chinook (0. tshawytscha) Adult 3.9 0.65 3.1 0.44 15.6 0.30 Juvenile 57.1 0.39 18.0 0.32 2.4 1.11 Coho(0. kisutch) Adult 5.1 0.73 13.3 0.32 7.3 0.65 Juvenile 2.8 0.85 0.0 0.00 4.9 0.76 Cutthroat (0. clarkii) Adult 14.2 0.40 2.2 0.42 0.0 0.00 Juvenile 14.2 0.40 6.6 0.38 2.3 1.08 Sockeye (O. nerka) Adult 4.4 0.45 0.0 0.00 0.0 0.00 Steelhead (0. mykiss) Adult 2.0 0.99 0.6 0.64 0.0 0.00 Juvenile 2.0 1.00 1.3 0.57 0.0 0.00 anchovy (Engraulis mordax) were consumed in greater quantity but they were not always a large proportion of the reconstructed biomass. Using spHt-sample frequency of occurrence (SSFO) to estimate diet composition, the results were dramatically different for larger and smaller prey items (Fig. 3). For 440 Fishery Bulletin 100(3) B n,n ,n,n,n,n n , n I r-1 I Fi I pn ro i O 0 3 o E g 02 CD n ,n,n T3 — Figure 1 (continued) Laake et al : Pinniped diet composition 441 Figure 2 Estimates of the average consumption of each prey group (P ) by harbor seals on the Columbia River between 1995 and 1997 for spring (A), summer (B), and fall (C). prey items weighing 10-20 g, SSFO predicted consump- tion that was 10 times greater than BR. Likewise, for prey items with a 1 kg mass or greater, SSFO predicted con- sumption estimates that were less than one-tenth of the BR estimate. For prey items near the median mass of 173 g, both estimators produced similar results. Discussion The consumption estimates for the Columbia River harbor seals could be improved by incorporating differences in energy density across prey and by measuring the sex- and age-structure of the seal population through time rather than using a life-table which may not be appropriate. These are valid criticisms and they could be overcome by collecting additional data. However, we believe these are less important than the current inadequacies in measur- ing diet composition that will likely affect any study that attempts to estimate consumption based on scat analysis. It is well recognized that digestion does not act equally on all hard parts and is certainly species-specific for oto- liths (Harvey, 1989). We included other hard parts such Table 4 Proportion of variance of the adult chinook consumption estimate resulting from each estimation component. Diet composition Harbor seal Year Date Biomass Population Season (a,,) (a,) (CT^) size Spring 0.00 0.14 0.82 0.04 Summer 0.00 0.51 0.44 0.04 Fall 0.20 0.16 0.59 0.05 as bone to reduce bias due to differential otolith digestion; however, that inclusion may not remove all of the selectiv- ity bias and it certainly introduces several other problems discussed below. Also, because hard parts are used for diet composition, if seals are only eating the fleshy parts of large fish, a significant portion of their diet may be missed. 442 Fishery Bulletin 100(3) Laake et al : Pinniped diet composition 443 a: m 6 u. Median weight = 173g • ♦ 10 ♦ 1000 10000 Adult ctiinook Equivalent estimates Weight (g) on log 10 scale Figure 3 Comparison of spring consumption estimates for harbor seals on the Columbia River based on split-sample frequency of occurrence (SSFO) and biomass reconstruction (BR) in relation to prey weight. Similar estimates were obtained for prey near the median weight. SSFO increases estimates of smaller prey items (e.g. smelt) and decreases estimates of larger prey items (e.g. adult chinook salmon) in relation to BR. We have assumed an equal probability of recovering identifiable skeletal remains from all prey sizes and that masses predicted from otolith measurements represent prey identified from other structures. There are several situations when recovered otoliths may not correctly rep- resent the size of prey consumed. Small otoliths from small individuals of a species may be more likely to be completely digested. In that case, biomass would be over- estimated because the larger otoliths would be recovered from larger fish. Also, otoliths may have different passage rates due to changes in their structure as fish age result- ing in a size bias. Consumption estimates could thus be erroneously high or low because prey would be calculated from a single size group rather than the range of prey consumed. Unequal digestion may also create errors in estimated mass from the otoliths that are recovered. Our estimated mass may have been biased because we applied an average degradation factor to adjust otolith length. A better alternative would be to grade otolith condition and apply condition-specific degradation factors (Tollit et al., 1997b). Counting the number of individual prey items is not possible with nonunique bones. Instead, we estimated a minimum number of individuals (MNI) contained in the scat {7!,^=MNI). MNI from all skeletal elements is a mini- mum estimate because the presence of many nonunique structures are assigned to a single prey item when they could represent several different prey. This error would not bias diet composition if it did not vary over species and size. However, differential passage of unique structures and identifiability among species result in the greater probability of detecting some prey (Browne et al., 2002). To minimize interspecific biases, we could use an MNI of 1 for all bones, regardless of the enumeration of unique structures for some species. While this would reduce some problems with species differences, it would exacerbate differences associated with prey size because large prey would be accurately reflected by an MNI of 1 and small prey that are eaten in greater quantities would be se- verely underestimated. Bowen (2000) proposed estimating the number of prey consumed by correcting the otolith count with rates of oto- lith recovery from feeding trials. These correction factors vary widely between seals and studies and are influenced by a variety of factors, including size of individual prey, meal size, and activity of harbor seals (Harvey, 1989; Har- vey and Antonelis, 1994; Cottrell et al., 1996; Tollit et al., 1997b). Browne et al. (2002) examined the ratio of otolith- 444 Fishery Bulletin 100(3) corrected estimates of MNI for several species. The ratios were quite variable but the comparison did suggest that smaller prey such as smelt were more likely to be under- represented by using the minimum count. If diet composition is based on sagittal otoliths, a fish can be represented by, at most, two scats and because all fish have two sagittal otoliths, fish size should not influ- ence the probability that a particular fish is included, in a sample of scats, except through size-specific passage rates of otoliths. However, when all hard parts are included, the sampling may be size-biased if the size of the prey affects the number of scats in which the hard parts are deposited. Larger prey contain larger hard parts that may require longer passage times; therefore larger prey may be depos- ited in more scats than smaller prey Also, large prey may be shared among seals as a result of cooperative feeding behavior and could be deposited in several scats. If either situation occurs, larger prey would be more likely to be included in the sample and would be over-represented. Because the scat is the sampling unit, any prey-size or species-specific effects on scat deposition rate may also bias diet composition estimators. The effect of over-representing large prey depends on the estimator used for diet composition. The different outcomes with Equations 1-3 can be demonstrated with a simple example. Consider a sample of two scats in which one scat contains the remains of a 2-kg salmon and an- other scat contains the remains of ten 10-g anchovy and four 100-g herring. From Equation 1, the diet composition would be 80% (2000/2500) salmon, 4% anchovy and 16% herring based on proportions of total reconstructed bio- mass. From Equation 2, we would estimate that salmon represent 50% of the diet from the two samples that are 100% and 0% salmon, and likewise 10% anchovy and 40% herring. Finally from Equation 3, we would estimate that the diet was 50% salmon, 25% anchovy, and 25% herring. If the small prey were undercounted in relation to the large salmon, the influence of the error influences the composition within the scat for Equations 2 and 3, but for BR (Eq. 1) the error extends across all samples. As with the Columbia River harbor seal example (Fig. 3), the differences in the estimators are primarily the result of large prey in the weighted versus unweighted averages. Some difference would be expected in the results of Equa- tions 2 and 3 depending on the validity of the equal volume assumption. SSFO (Eq. 3) simplifies the analysis of diet composition to a measure of presence and absence by as- suming that prey within the same scat were consumed in equal volumes. The simplifying assumption removes the ne- cessity to enumerate prey and measure mass from morpho- metric relationships with prey remains. However, the equal volume assumption does not seem particularly reasonable and its implementation is arbitrary, depending on how the prey are classified unless all prey remains can be identified to species. Olesiuk ( 1993) showed that the diet composition percentages for the primary prey varied by a factor of two or three, depending on the assumed composition within each scat. We expected that these differences would depend on the diversity of the diet. How closely they represented the true diet would depend on the range in prey sizes. From our viewpoint, we do not see a clear choice between the estimators for diet composition. The use of consump- tion estimates from SSFO and BR to provide a range of es- timates may have limited application in cases where each approach would suggest a similar conclusion. However, for large prey, such as salmonids, a tenfold difference in esti- mates, compounded with the uncertainty from sampling and biomass estimation, may yield too little information to develop a reliable conclusion about the impact of pinniped predation on salmonid stocks. Acknowledgments The authors would like to thank M. Gosho, B. Hanson, K. Hughes, S. Melin, and L. Lehman for their assistance in the field and laboratory. The authors would also like to thank P. Olesiuk for his encouragement and for succintly demonstrating the problems with the minimum number computation in biomass reconstruction. This manuscript was greatly improved by comments from J. Jansen, P. Boveng, and four anonymous reviewers. Literature cited Bigg, M. A. 1969. The harbour seal in British Columbia. Fish. Res. Board Can. Bull. 172,31 p. Bigg. M. A., and M. A. Perez. 1985. Modified volume: a frequency-volume method to assess marine mammal food habits. In Marine mammals and fisheries (J. R. Beddington, R. J. H. Beverton, and D. M. Lavigne, eds.), p. 277-283. George Allen & Unwin, London. Bigg, M. A., G. Ellis. P. Cottrell, and L. Milette. 1990. Predation by harbour seals and sea lions on adult salmon in Comox Harbour and Cowichan Bay, British Columbia. Can. Tech. Rep. Fish. Aquat. Sci. 1769, 31 p. Bowen, W. D. 2000. Reconstruction of pinniped diets: accounting for com- plete digestion of otoliths and cephalopod beaks. Can. J. Fish. Aquat. Sci. 57:898-905. Browne, P.. J. L. Laake. and R. L. DeLong. 2002. Improving pinniped diet analyses through identifica- tion of multiple skeletal structures in fecal samples. Fish. Bull. 100:423-433. Cochran, W. G. 1977. Sampling techniques, 3'''' ed., 428 p. John Wiley and Sons, New York, NY. Cottrell. P E., A. W. Trites, and E. H. Miller 1996. Assessing the use of hard parts in faeces to identify harbour seal prey: results of captive-feeding trials. Can. J. Zool. 74:875-880. Forney, K. A., J. Barlow, M. M. Muto, M. Lowry, J. Baker, G. Cameron. J. Mobley, C. Stinchcomb, and J. V. Carretta. 2000. U.S. Pacific marine mammal stock assessments: 2000. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SWFSC- 300, 276 p. Hammill, M.O., C. Lydersen, K. M. Kovacs, and B. Sjare. 1997. Estimated fish consumption by hooded seals iCys- tophora cristata), in the Gulf of St. Lawrence. Northwest Atl. Fish. Sci. 22:249-257. Laake et al Pinniped diet composition 445 Hammond. P. S. and P. Rothcrv. 1996. Application ofcomputcr sampling in the estimation of seal diot. J. Appl. Stat. 23:525-533. Harvey, J. T. 1988. Population dynamics, annual food consumption, movements and dive behaviors of harbor seals Phoca vitu- Una richardsi, in Oregon. 177 p. Ph.D. diss., Oregon State Univ., Corvallis, OR. 1989. Assessment of errors associated with harbour seal (Phoca vitulina) faecal sampling. J. Zool. Lond. 219:101-111. Harvey, J. T.. and G. A. Antonelis. 1994. Biases associated with non-lethal methods of deter- mining the diet of northern elephant seals. Mar Mamm. Sei. 10: 178-187. Harvey, J. T., T. R. Loughlin. M. A. Perez, and D. S. Oxman. 2000. Relationship between fish size and otolith length for 63 species of fishes from the eastern North Pacific Ocean. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 150. 48 p. Huber H. R.. S. J. Jeffries. R. F. Brown. R. L. DeLong and G. VanBlaricom. 2001. Correcting aerial survey counts of harbor seals iPhoca vitulina richardsi) in Washington and Oregon. Mar. Mamm. Sci. 17(2):276-295. Jobling, M. 1987. Marine mammal faeces as indicators of prey impor- tance— a source of error in bioenergetics studies. Sarsia 72:255-260. Marcus. J. W. D. Bowen, and J. D. Eddington. 1998. Effects of meal size on otolith recovery from fecal samples of gray and harbor seal pups. Mar Mammal Sci. 14:789-802. Nilssen. K. T., O. Pedersen, L. Folkow, and T. Haug. 2000. Food consumption estimates of Barents Sea harp seals. In Minke whales, harp and hooded seals: major pred- ators in the North Atlantic ecosystem (G. A. Vikingsson and F. O. Kapel, eds. ), vol. 2, p. 9-27. NAMMCO < North Atlantic Marine Mammal Commission) Sci. Publ., Troms0, Norway. NMFS (National Marine Fisheries Ser%'ice). 1997. Investigation of scientific information on the impacts of California sea lions and Pacific harbor seals on salmo- nids and on the coastal ecosystem of Washington, Oregon, and California. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-NWFSC-28, 172 p. Olesiuk. R F 1993. Annual prey consumption by harbor seals {Phoca vitulina) in the Strait of Georgia, British Columbia. Fish. Bull. 91:491-515. Olesiuk, P F, M. A. Bigg, G. M. Ellis, S. J. Crockford, and R. J. Wigen. 1990. An assessment of the feeding habits of harbour seals (Phoca vitulina) in the Strait of Georgia, British Columbia, based on scat analysis. Can. Tech. Rep. Fish. Aquat. Sci. 1730, 135 p. Pierce, G. J., P R. Boyle, and J. S. W. Diack. 1991. Identification offish otoliths and bones in faeces and digestive tracts of seals. J. Zool. Lond. 224:320-328. Pitcher, K. W. 1980. Stomach contents and feces as indicators of harbour seal foods in the Gulf of Alaska. Fish. Bull. 78:544-549. Seber, G. A. F 1973. The estimation of animal abundance and related parameters, 506 p. Griffin, London. Stenson, G. B., M. O. Hammill, and J. W. Lawson. 1997. Predation by harp seals in Atlantic Canada: pre- liminary consumption estimates for Arctic cod, capelin and Atlantic cod. Northwest Atl. Fish. Sci. 22:137-154. Tollit, D. J., S. R R. Greenstreet, and P M. Thompson. 1997a. Prey selection by harbour seals, Phoca vitulina, in relation to variations in prey abundance. Can. J. Zool. 75: 1,508-1518. Tollit, D J.. M. J. Steward. P. M. Thompson, G. J. Pierce, M. B. Santos, and S. Hughes. 1997b. Species and size differences in the digestion of oto- liths and beaks: implications for estimates of pinniped diet composition. Can. J. Fish. Aquat. Sci. 54:105-119. Tollit, D. J., and P. M. Thomp.son. 1996. Seasonal and between-year variations in the diet of harbour seals in the Moray Firth, Scotland Can. J. Zool. 74:1110-1121. Appendix We have constructed variance estimators for diet com- position and consumption rates using finite population sampling methods (Cochran, 1977) and delta method approximations based on the Taylor series (Seber, 1973). Variance estimates and confidence intervals could also be constructed by using bootstrap techniques similar to the work of Hammond and Rothery (1996). To describe the variance estimators, we use the follow- ing subscripts: i for prey group, 7 for season, y for year, and t for collection occasion within year. A dropped subscript implies summation or averaging over that subscript (e.g. Sj^, is the total number of scats collected in season J of year y — summed over occasions). We define the following nota- tion which was not used in our article: Tj^ = number of scat collection occasions, Sjy = number of scats collected during the occasion, Y^ = number of years in which season _/ was sampled of a total of Y years. We have assumed that scat collected on each occasion is a varying proportion of a fixed and unspecified amount of prey biomass consumed. Thus, we computed unweighted averages of the proportions over occasions and then over years: ^ ''.,>' S^'y We have also assumed that scat samples are independent which would be a concern in the unlikely event that the same seal deposited multiple scats at one time. We have also assumed that the collection occasions which are limited to low tides represent a random sample of dates within a season. The variance of diet composition was based on mul- tistage sampling scheme stratified by season (Cochran, 1977) that was composed of the following stages: estima- tion of biomass proportion consumed for each occasion, sampling of occasions within season of a year, and sam- pling of years. We have limited inference to the Y years that were sampled. Thus, if a season was sampled in each 446 Fishery Bulletin 100(3) year there is no annual variance component. The variance of the diet composition for a prey group for a single season is the sum of three components corresponding to the three stages: estimation of biomass proportion W ^^rjy, r=l ^ ■p-^p. where /3 is a vector of partial-derivatives of ;r,,^,, with respect to 6^,^, for r=l, w and 1 is the w x w variance-cova- riance matrix of 6^„, for ;'=l,(o. Likewise, the variance of p was estimated by using Va7-(p„ ) + O, + o;. y< T,. SX^°'-.'p.v,) where ■2 v=l (=1 and Vaii{p,j^.,) = Var^ I^. t^L ^prp. The first term in the variance is the uncertainty re- sulting from the variability in the estimate of the mean weight. For salmonid species, the second term measures the uncertainty resulting from apportioning the number of unknown salmonids into prey groups (i.e. estimating '!,,,,,). For nonsalmonid species, all hard parts were classified into less specific prey groups; therefore a„ = 0. The group- specific estimates of biomass within season are uncorre- lated except for the estimates for salmonid species (iefO which are correlated because of the apportionment of the items which were identified as salmonid but could not be identified to species nor to the juvenile and adult sizes. For salmonid groups r and vff , the estimated covariance is Coi'(6,„,,6.,,,,, ) = Coi'(;v,,,,,n,j,„ )uv,u'>,- The total amount of biomass represented by the scats is the sum across all prey groups: ^v<=X^'^-VM and its variance was estimated as Varib jyt ' uarib,,^., ) +2 •jyi ' rsw v>rvEii/ COi'ffeavMV- The variance of n , was estimated assuming a binomial distribution: Varin,^ ,2 r„(i-x„^ and the covariance between salmonid groups was esti- mated as . . "(Vf O,, C0l'( «„„,/"!,, We assumed that the weights of the prey consumed were independent samples from a group- and season-specific distribution with mean ^,,. and variance a,-'. The param- eters ^,, and a^j were estimated with the predicted weights from the sample of o* measurable hard parts: ^^.J = '^'y -^ and a': = °;-i To incorporate uncertainty in regression parameters and prediction error for the measured hard parts and variabil- ity in mass for the unmeasured prey (n - o* ), the vari- ance of ^ should be estimated by Laake et al.: Pinniped diet composition 447 ^Var{^ Var(w^j) = 6l lation correction (fpc) factor on the second term. If we had used the fpc for salmonids, we would have had to replace the unknown «■ with their estimated values. For variance estimates of number of prey consumed (P- ) we also used a delta method approximation: where the variance of li' ,^ is the prediction variance lor the ''' otolith. When these data were analyzed, the available u repression equations in a draft version of Hai"vey et al. (2000) (lid not include standard errors for the parameters nor the residual variance that were needed to compute the prediction variance. Therefore, unfortunately we had to drop the first term from the variance. The necessary values are now available in Hai-vcy et al. (2000). The second-term measures variability between prey and is typically larger than the prediction error. However, to avoid underestimat- ing this component of variance we dropped the finite popu- Var{P,j ) = P,J[cu''{^j ) -I- cv'{p,j )J. We did not have variance estimates for energetic require- ments nor age structure; therefore the variation in estimated prey biomass requirement (£ ) reflected only variation in our estimates of population size: evil -cviN,} ,jcv^{Cj) + cv^{n, which was also approximated by the delta method. 448 Abstract— The vertical and horizontal movements of southern bluefin tuna (SBT). Thunnus maccoyii, in the Great Australian Bight were investigated by ultrasonic telemetry. Between 1992 and 1994, sixteen tuna were tracked for up to 49 h with depth or combined temperature-depth transmitting tags. The average swimming speeds (mea- sured over the ground) over entire tracks ranged from 0.5 to 1.4 m/s or 0.5 to 1.4 body lengths/s. The highest sustained swimming speed recorded was 2.5 m/s for 18 hours. Horizontal movements were often associated with topographical features such as lumps, reefs, islands and the shelf break. They spent long periods of time at the surface during the day (nearly 30%), which would facilitate abundance esti- mation by aerial survey. At night, they tended to remam just below the surface, but many remained in the upper 10 m throughout the night. SBT were often observed at the thermocline interface or at the surface while travelling. A characteristic feature of many tracks was sudden dives before dawn and after sunset during twilight, followed by a gradual return to their original depth. It is suggested that this is a behavior evolved to locate the scatter- ing layer and its associated prey when SBT are in waters of sufficient depth. SBT maintained a difference between stomach and ambient temperature of up to 9°C. Vertical and horizontal movements of southern bluefin tuna iThunnus maccoyii) in the Great Australian Bight observed with ultrasonic telemetry Tim L. O. Davis Clive A. Stanley CSIRO Division of Marine Research Castray Esplanade Hobart. Tasmania, Australia 7000 E mail address (for T L O Davis) tl davis®csiro au Manuscript accepted 17 May 2001. Fish. Bull. 100:448-465 (2002). Southern bluefin tuna, Thunnus mac- coyii, spawn in the northeast Indian Ocean south of the Sunda Islands from August to May (Farley and Davis, 1998). Young-of-the-year move down the west coast of Australia and first appear in the Great Australian Bight as 1-year olds. They aggregate in the Bight during the Austral summer, disperse to the east or west within latitudes of 30-40°S in autumn, and return to the Bight in spring. Juveniles between 1 and 4 years old return each summer to the Bight where they form extensive surface schools. To provide a fishery-independent in- dex of recruitment, line transect aerial surveys have been flown in the Bight each summer since 1990 to estimate the relative abundance of juveniles visible in the top 5 m (Chen et al.^). The largest source of variance in these estimates is thought to be environmen- tal factors that influence both surfac- ing behavior and aerial detection. To investigate these problems, we used ultrasonic telemetry to provide the first information on surfacing behavior and short-term horizontal and verti- cal movement patterns that might influence sightings from the air. More comprehensive information on tuna surfacing for the whole of the three- month survey period is expected from an archival tagging program begun in 1992 (Gunn et al.^). Because these data are dependent upon the recapture of archival-tagged fish, this information will not be available for some time. Because of Carey's pioneering work on Atlantic bluefin tuna (Carey and Lawson, 1973), ultrasonic telemetry has been used to study many tuna species, including Atlantic (Lutcav- age et al., 2000) and Pacific bluefin tuna (Marcinek et al., 2001), yellowfin tuna (Carey and Olsen, 1982; Cayre and Chabanne, 1986; Yonemori, 1982; Holland et al, 1990a; Cayre, 1991; Cayre and Marsac, 1993; Block et al., 1997; Brill et al., 1999; Marsac et al.^), skipjack tuna (Yuen, 1970; Dizon et al., 1978; Levenez, 1982; Cayre and Cha- banne, 1986; Cayre, 1991), bigeye tuna (Holland et al, 1990a, 1992; Holland and Sibert, 1994), and albacore (Laurs et al., 1977). A number of small (35-46 cm) southern bluefin tuna (SBT) have been tracked in the Indian Ocean off ' Chen, S., A. Cowling, and T Polacheck. 1995. Data analysis of the aerial surveys (1991-1995) for juvenile southern bluefin tuna in the Great Austrahan Bight. 1995 Southern Bluefin Tuna Recruitment Moni- toring Workshop Report RMWS/95/6, 57 p. CSIRO Marine Laboratories, Castray Es- planade, Hobart, Tasmania, Australia 7000. 2 Gunn, J., T. Davis, T, Polacheck, A. Betle- hem, and M. Sherlock. 1995. The appli- cation of archival tags to study SBT migra- tion, behaviour and physiology. Progress Report — 1994—95. Southern bluefin tuna recruitment monitoring and tagging pro- gram workshop, 7-10 August 1995, Ho- bart, Australia. Rep. RMWS/95/8, 17 p. CSIRO Marine Laboratories, Castray Espla- nade, Hobart, Tasmania, Australia 7000. 3 Marsac, F., P. Cayre, and F. Conand. 1995. Analysis of small scale movements of yel- lowfin tuna around FADs using sonic tag- ging. Expert Consultation on Indian Ocean tunas. 6th session. Colombo, Sri Lanka, 25-29 September 1995. Rep. TWS/95/2/ 10, 20 p. ORSTROM, Seychelles Fish- ing Authority, BP 570, Victoria, Mahe, Seychelles. Davis and Stanley: Movemenis of Thunnus maccoyli in the Great Australian Bight 449 Western Australia (Fishery Agency of Japan''"''). All SBT in the current study were released and tracked in shelf waters, except for one fish that moved from the shelf into slope waters. Methods Tracking experiments on SBT were carried out in the Great Australian Bight over three years. In 1992, the tracking system was tested on a short cruise late in the fishing season and close to Port Lincoln. In 1993, the strategy was to obtain extended tracks of up to 48 hours to investigate day and night patterns in depth dis- tribution and horizontal movement within the Great Australian Bight. In 1994, we planned short tracks to maximize the number of individ- ual fish observed during daytime, so that their behavior could be linked to sightings by aerial survey carried out at the same time. All SBT were caught by pole and line, except for tuna no. 7 (referred to simply as "tuna 7") which was caught on light game fishing tackle, and tuna 9, which swallowed the tag while free-swimming. The tuna were landed on a wet, plastic-covered, foam mat. The head was covered with a wet cloth, length of fish was measured, the tag attached or inserted, and the fish was released back into the school. V16 (16 mm diameter) sonic tags (VEMCO, Halifax, Nova Scotia, Canada) of various fre- quencies (50-69 kHz), equipped with pressure sensors, were deployed on two fish in April 1992 and four fish in January and March 1993 (Table 1 ). Five of these tags were attached externally, behind the second dorsal fin of the tuna, by us- ing the attachment technique described in Hol- land et al. (1985). The sixth was placed in the stomach with an oesophageal catheter. V22 (22 mm diameter) tags transmitting on 40 kHz were used on ten tuna tracked in January and February 1994. These tags, which have stronger signals and lower transmission frequencies, have a much larger detection range than V16 tags. They also have the capacity for * Fishery Agency of Japan. 1990. Report of 1988 research cruise of the RA' Shoyo-Marii. Distribution of juvenile southern bluefin tuna off west coast of Australia, December 1988-January 1989. Rep. Res. Dep., Fish. Agency Jpn. 63:1-125. Pelagic Fish Resource Division, 5-7-1 Orido. Shimizu, Shizuoka 424, Japan. 5 Fishery Agency of Japan. 1992. Report of 1989 research cruise of the R/V Shoyo-Maru. Distribu- tion of juvenile southern bluefin tuna off west coast of Australia, September-December 1989. Rep. Res. Dep., Fish. Agency Jpn. 64:1-166. Pelagic Fish Resource Division, 5-7-1 Orido, Shimizu, Shizuoka 424, Japan. I CM 05 C •c 3 T3 +-> J2 bo S 3 < O ."I c EC 3 3 -a c ■a c to « OJ CO 3 -a .2 > s ° s c J= 3 Q e2 S c 3 .6) s rt fc a a 5 " c c o .2 = ■" o -5 o s a > > .2 0) OO(M'^iC t-i a — .^ CualaobjD'^'^r.^X oocococoooojo C^COCD'Xl^C^lCDOOCOCOlO'^OCOfM cjocx^ocoiOTtcNo^oaiooooai^cccx) O0^ci0i-H^0'-H0000i-*00 T)>oiomoocMrfo-*a> ooaioDxait-c?5^c£)t> cjdddeboor-icjd J3 J3 J= X J2 j: j3 to to s s O O M o o o T3 c to f/i T3 C c*- t*.- crj >^ >, t« 1*- o ^ to K Od C CO m m ll< K « £ J3 £ 1 O ■o c _to In c _to In -a c _to c CO c c CO O m S c CO c a CO O "o S .a: 1X1 O w CO "5 g tn a CO .X! CO o CJ a; eg o H c CO C C to O c to c c CO o Cm o to -D to to to -o 1 o o O 3 z; S T— < e s s E CO CO s CN CM ro CO CO (^0 Tf i?l C75 >>>>>> a< Oi. 0. Oh Cu Cu cu >>>>>>>>> wcMco-^incDr^ooaio^cMco-^ioto 450 Fishery Bulletin 100(3) dual channels enabling sequential transmission of pres- sure and temperature. Seven pressure tags and three pressure-temperature tags were deployed. All, except one, were inserted into the stomach. To retard regurgitation, two small hooks were tied to each tag with VICRYL dis- solving sutures before being placed in the stomach. One tuna struck a tag with hooks during a calibration experi- ment, resulting in the tag attaching without the tuna leav- ing the water This fish was tracked for five hours until the tag detached. Standard tracking techniques were used (see Holland et al., 1985). The tracking equipment, including the re- mote-controlled hydrophone rotator, is described in Pep- perell and Davis ( 1999). We attempted to track the fish at a distance of about 400 m, which we judged from signal strength, using distance and signal-strength calibrations carried out before tracking. The decoded data from single- channel tags were logged together with global positioning system (GPS) latitude and longitude, nominally at 1-sec- ond intervals, on a computer with VEMCO VSCAN/GPS software (Vemco, 1992). Temperature and depth data from dual-channel transmitters were logged at about 3-second intervals. Conductivity-temperature-depth profiles were taken at the start and finish, and at convenient times during each track, with a digital data loggers international profiler Additional information collected during each track in- cluded bottom depth, air temperature, wind speed, cloud cover ( 1-8 sectors covered), brightness ( 1-4), and presence of surface and subsurface schools of SBT In 1994. we also recorded the depth distributions of SBT schools detected on the echosounder under the tracking vessel. The tracking data were processed using a filtering pro- gram to delete spurious data, interpolate missing data, and generate data files at nominated time intervals. The program compared each depth with the seven previous data points (which were nominally recorded at 1-second and 3-second intervals for single- and dual-channel tags, respectively) and then deleted spurious records based on expected maximum and minimum depths and maximum depth-change rates. Similarly, a temperature filter de- leted spurious values based on expected maximum and minimum temperatures and maximum temperature- change rates in data from the dual-channel transmitters. Complete vertical tracks were plotted against time with 20-second interval data. The position and sustained swim- ming speeds of tracked tuna were assumed to be the same as that of the tracking vessel. By using GPS data, speed was calculated over the ground from the straight-line distance travelled between points in 10 minutes. Verti- cal movements and horizontal movements in relation to the vessel were not considered. An attempt was made to match both the heading and speed of the tracked tuna so that the vessel reflected the movements of the fish. Depth distribution of tuna was examined from 20-sec- ond interval data aggregated in 5-m depth bins stratified by day and night. The number of observations in each depth bin was then expressed as a proportion of the total number of observations in each day-night stratum. The speed distribution of tuna was examined from 10-minute interval data aggregated in 0.1 m/s bins stratified by day and night. Results The tracking system was developed for use on a variety of commercial vessels. In April-May 1992, five fish were released and two were tracked successfully (Table 1). Modifications were then made to the system and different transmission frequencies were selected to reduce inter- ference. The improved system was used on two cruises in January and March 1993. Four fish were tracked for periods of up to 49 h (Table 1). In 1994, tracking was carried out farther west because fish schools were not sighted farther east (Fig. 1 ), probably due to warm water remaining in the northwest section of the study area (Fig. 2). The whole period (8 January-3 February) was characterized by windy conditions that made catching and tracking fish difficult. Ten SBT were released with tags (Table 1). The longest track was 31 h, but some tracks were short because the fish regurgitated their tags. Other tracks were terminated voluntarily after sunset to maximize the number of daytime observations on different fish. Tuna 1 was released SW of Liguana Island on 29 April 1992 at 08.55 h (Figs. 1 and 2A). It headed WSW towards the shelf edge until tracking was terminated after 12 hours. Acoustic interference from the tracking vessel affected calculation of depth at certain vessel speeds, re- sulting in some gaps in the vertical data (Fig. 3A). During the day the fish spent most of its time at the surface, but frequently dived to near the bottom. Tuna schools spotted at the surface appeared to break up and then re-appear sometime later, more or less at the same rate that the tracked SBT moved in and out of the surface waters. The fish dived shortly after sunset and then returned to the mixed layer and oscillated around 30 m until tracking was terminated at 2108 h. Tuna 2 was released SW of Liguana Island on 30 April 1992 at 1330 h (Fig. 1). The fish moved west throughout most of the track but slowed and changed direction many times from 0400 h. These direction changes occurred when the tuna reached cooler surface waters (Fig. 2A). It re- mained in the 17°C water, moving south along the bound- ary with the 16°C isotherm. Its vertical behavior can be seen in Figure 3B. Like tuna 1, this fish made a brief dive just after sunset. Tuna 3 was released at The Lump, 28 km west of Ward Island on 27 January 1993 at 15:22 h and was tracked for 49 h (Figs. 1 and 2B). It moved into waters near the shelf edge, a movement matched by a conventionally tagged tuna that was released at 33°15'S, 133°49'E on 26 Janu- ary 1993 and that was recovered five days later at 34°S, 132°17'E by a commercial fishing operation. The vertical behavior of tuna 3 is shown in Figure 4A. It spent con- siderable time near the surface in association with other tuna during the late afternoon on the first day It seldom surfaced on the second day; neither did the other tuna. A commercial pole-and-line boat fishing in the same area Davis and Stanley: Movements of Thunnus maccoyii in the Great Australian Bight 451 Figure 1 Geographic position of southern bluefin tuna tracks in the eastern Great Australian Bight. The number of each fish indicates the start of its track. Symbols along tracks represent hourly positions (open symbol=day, solid symbol=night). on the second day could not chum fish to the surface; the fishermen had also observed many tuna at the surface the previous day. Tuna 4 was released near Ward Island on 30 January at 16:40 h and was tracked for 48 hours. This fish remained within the general area of release until tracking was ter- minated (Figs. 1 and 2B). Vertical behavior is shown in Figure 4B. The fish was associated with surface schools until night on the first day. It was also associated with large numbers of tuna (observed on the echosounder) throughout the night. The surface activity of tuna schools was low throughout the next day. Tuna were observed on the echosounder at the same depth as the tracked tuna throughout the day and next night, but not in the same numbers as the day before. Tuna 5 was released on 13 March 1993 at 11:50 h at Rocky Island (Fig. 1). It was tracked for about 4 hours before it was lost. At 1500 h the tracked fish, together with other tuna, followed a pole-and-line vessel that was chumming at the time and that had crossed our path. The fish was temporarily lost but was relocated an hour later, again following the chum line of the pole-and-line vessel. Tuna 6 was released at Rocky Island on 14 March 1993 at 07:35 h and subsequently tracked for 26 h (Figs. 1 and 2C). At 08:15 h the tracked tuna was observed following the chum line of the pole-and-line vessel (Fig. 5B). It was associated with other tuna at or near the surface until about 11:00 h. At 12:00 h the fish moved rapidly offshore at about 2.5 m/s and was tracked to the shelf On reach- ing the shelf break, it remained within a region of warmer 452 Fishery Bulletin 100(3) NOAAll SST 28 Apr 1992 0647Z Copyright 2002 CSIRO 15 1.6 1/ la 1.9 26 2.1 2.2 134- 135 136 137 NOAAll SST 11 Mar 1993 Copyright £002 CSffiO NOAAU SST 21 Jan 1994 0602Z Copyright 2002 CSIRO 15 16 17 le 19 20 21 22 13+ 135 US 137 NOAAll SST 30 Jan 1993 0633Z Copyright 2002 CSIHO NOAAU SST i; Jan 1994 lb09Z Copyright 2002 CSIRO 1.4 15 16 1.7 1.8 134 135 NOAAll SST 02 Feb ly94 0715-1R42Z Copyright 2002 CSIRO Figure 2 NOAAll satellite sea-surface-temperature images of the study area during the periods of tracking with southern bluefin tuna tracks overlaid. (A) Tracks 1 and 2 on 29-30 April 1992. (B) Tracks 3 and 4 on 27-31 January 1993. (C) Track 6 on 14-15 March 1993. (D) Tracks 7. 8, and 10 on 10-16 January 1994. (E) Tracks 11-13 on 17-19 January 1994. (F) Tracks 14-16 on 31 January-2 February 1994. Davis and Stanley: Movements of Thunnus maccoyii in the Great Australian Bight 453 W -in w=4 w-in W=1fi 0 1 ^""^\|t f^ '^'^^\^J(fkf^ f^^VA^ \a 40 ' r 1/ ^vUvAf ^ 1 E 80 - ^ Q. Q 120 ' I^^H I^H 160 - ^1 H 200 H 1 T — ^ — 1 — \ — r ^^^^^^^^^^ ^ i 15 18 21 24 Local time (h) 03 06 Figure 3 (A) Track of southern bluefin tuna 1 on 29 April 1992. (B) Track of southern bluefin tuna 2 on 30 April 1992. Speed over the ground (upper graph) and depth of the tuna (lower graph I are plotted against time. The sea bottom is indicated by shading, the solid line represents the bottom of the mixed layer and the dotted lines represent isotherms (°C). The black bar represents nighttime, the unshaded bar, daytime and W = wind speed in km/h. surface water in association with a large subsurface school of tuna (Fig. 2C). While traveling to the shelf break, it re- mained near the bottom or just below the mixed layer. A deep sounding was made just after sunset and another just before dawn, the last reaching 180 m. Tuna 7 was released at The Lumps 7 km south of Can- nan Reef at 09:16 h on 10 January 1994 (Figs. 1 and 2D). This was the smallest fish tracked in this study. Unlike the others, it was caught on a trolled lure and landed after about a 5-minute struggle. Tracking was discontinued af- ter 30 h due to bad weather This fish was associated with large schools for the first five hours of tracking and again for three hours at first light the next day. It made postdusk and predawn dives (Fig. 6A). Tuna 8 was released at The Lumps 7 km south of Can- nan Reef at 10:58 h on 12 January 1994 (Figs. 1 and 2D). The fish was followed for 10 h until tracking was halted shortly after sunset. It remained in the upper 20 m for most of the track (Fig. 6B). At 12:50 h it followed the chum line of a pole-and-line vessel. From 14:00 to 16:00 h, it remained at the surface with a large school of tuna before slowing and sounding at a reef. 454 Fishery Bulletin 100(3) a a> CO 3 I C 3 O) H » 05 (s/Lu) psads (uj) L|)daa (s/oi) psads (Lu) mdaa Davis and Stanley; Movements of Thunnus maccoyn in the Great Australian Bight 455 12 13 14 15 16 Local time (hours) 17 Local time (tiours) Figure 5 (A) Track of southern bluefin tuna 5 on 13 March 1993. (B) Track of southern bluelin tuna 6 on 14—15 March 1993. The period when the tuna followed the chum line is shown. See Figure 3 for legend details. Tuna 9 struck at a tag during calibration experiments near Nuyts Reef at 09:05 on 15 January 1994 (Fig. 1). Conse- quently, it was tagged without leaving the water. The tag was either in the mouth or in the stomach. The tuna remained with tuna schools in the vicinity of tagging until 09:50 h and then moved away at considerable speed (2 m/s) for over an hour before sounding and slowing (Fig. 6C). It returned to the surface and remained there in association with a large surface school. The tag detached from the tuna at 1410 h. Tuna 10 was released 11 km southeast of St Francis Is- land at 11:40 h on 16 January 1994. It was tracked for 13 h before it regurgitated the tag (Figs. 1 and 2D). The vertical movements of this fish differed from other tracks in that they were of relatively high amplitude (30-1- m), traversed the water column, and were of moderate frequency (Fig. 7A). It spent little time at a particular depth, except for a short period at the bottom at the start of the track and again around 21:00 h. The fish was associated with tuna schools for most of the track. It made an extended dive just after sunset. This was the first SBT to provide stomach-temperature data. The stomach temperature dropped rapidly initially, then increased to 25.4°C, which was about 6°C above wa- ter temperature. Temperature then decreased gradually for the rest of the track, but remained at least 4°C above ambient temperature. Tuna 11 was released at The Lumps 41 km west of Sceale Bay at 13:57 h on 17 January 1994 with a temperature-pres- 456 Fishery Bulletin 100(3) 12 15 18 21 Local time (hours) 9 12 15 Local time (tiours) Figure 6 (A) Track of southern bluetin tuna 7 on 10-11 January 1994. (B) Track of southern blue- fin tuna 8 on 12 January 1994. (C) Track of southern bluefin tuna 9 on 15 January 1994. Shading in the water column represents depth range of tuna schools detected by echo- sounder. See Figure 3 for other legend details. sure tag in the stomach ( Figs. 1 and 2E ). This fish was associ- ated with tuna schools for most of the track (Fig. 7B). Most vertical activity was oriented towards the surface, but peri- odic dives were made to or through the thermocline. At night the fish moved slightly deeper but continued to dive at regu- lar intervals. It made an extended dive just after sunset. The stomach temperature gradually increased through the night and by 01:00 h was at 28°C, which was 9°C above ambient temperature. At about 01:30 h there was a sudden drop in stomach temperature towards ambient temperature and then a gradual recovery that was most likely caused by the swallowing of prey or water (or both). The tag was regurgitated two hours later. Tuna 12 was released at The Lumps 41 km west of Sceale Bay at 09:16 h on 18 January (Figs. 1 and 2E). It showed considerable vertical activity similar to tuna 10 and was associated with tuna schools throughout the track (Fig. 8A). It made an extended dive just after dusk. Davis and Stanley: Movements of Thunnus maccoyii in the Great Australian Bight 457 ra 20^ t " Water liiffrirrnTTTr^''^^ 1 1 1 r W=30 W=10 W=30 -1 1 1 r W=40 W==30 W=15 "I 1 r 80 100 n 1 1 1 r 12 15 18 Local time (hours) 21 24 16 18 20 22 Local time (hours) 24 02 04 Figure 7 (A) Track southern bluefin tuna 10 on 16 January 1994. (B) Track of southern bluefin tuna 11 on 17-18 January 1994. See Figure 6 for legend details of upper and lower graphs. In the middle graph, stomach temperature and water temperature (interpolated from the depth of the fish and data logger temperature profiles) are plotted against time. Tuna 13 was released 15 km west of Labatt Point at 1226 h on 19 January with a temperature-pressure tag in the stomach (Figs. 1 and 2E). This fish remained close to the surface during the day and deeper at the night, frequently staying just above the thermochne and some- times diving through it. It made an extended dive after dusk and another just before dawn (Fig. 8B). It was associ- ated with schools of tuna throughout the track. Although tuna schools did not appear on the echosounder during the afternoon, extensive bird activity and surface schools 458 Fishery Bulletin 100(3) 2- A VV^^AvxA /^VVV jvA^ u II 1 i 1 1 1 1 1 1 1 1 ' „ W=10 W=15 W=25 W=10 21 24 Local time (hours) 09 Figure 8 (A) Track of southern bluefin tuna 12 on 18 January 1994. iB) Track of southern bluefin tuna 13 on 19 January 1994. See Figure 7 for legend details. of tuna were observed during the whole period that the tracked tuna was near the surface. Stomach temperature showed a marked drop initially (presumably it swallowed water on its release) and then a gradual recovery. Temperature reached a peak of 23.8°C, about 4.5°C above ambient temperature, near midnight and then slowly declined throughout the remainder of the night. Tuna 14 was released 41 km west of Sceale Bay at 12: 26 h on 31 January 1994 (Figs. 1 and 2F). The tuna was lost for about 20 minutes early in the track (Fig. 9A). It had moved south before doubling back and returning to the reef where it had been tagged. At 16:40 h it then swam away with a school of tuna, remaining just above the thermocline and occasionally making excursions to the surface. It made an extended dive after dusk. Tuna 15 was released at The Lumps 7 km south of Can- nan Reef at 13:26 h on 1 February 1994 (Fig. 1). It joined a large school of tuna following the chum line of a pole-and- line vessel, but was lost one hour later Tuna 16 was released at The Lumps 7 km south of Can- nan Reef at 09:30 h on 2 February 1994 (Figs. 1 and 2F). It was associated with schools initially, but they dispersed at about 1400 h (Fig. 9B). The tag was regurgitated at 15: 20 h. Time at depth The proportion of time spent by tuna in 5-m interval depth strata was plotted by day and night for all tracks combined (Fig. 10). These data indicated that SET are surface-oriented by day in the Great Australian Bight, spending a significant Davis and Stanley: Movements of Thunnus maccoyii in the Great Australian Bight 459 Q 60 Local time (hours) Figure 9 (A) Track of southern bluefin tuna 14 on 31 January 1994. (B) Track of southern bluefin tuna 16 on 1 February 1994. See Figure 6 for legend details. proportion of their time (nearly 30"^) in the upper 5 m. At night they tend to remain somewhat deeper, although they are present at the surface for part of the night. Tunas 4 and 12 were not surface-oriented during the day, but ranged evenly within the mixed layer. Both these fish were associated with large schools throughout the track and they exhibited rapid vertical oscillations, which we assumed spanned the vertical boundaries of the school. Tuna 6 also spent little time at the surface. It was associated with a small school of tuna and spent most of its time at or below the thermocline while travelling at 2.5 m/s towards the shelf edge. Tuna 14 also spent much of its time at the ther- mocline boundary while traveling during the day. Swimming speed The average swimming speeds of SBT over the entire tracks were in the range of 0.5-1.4 m/s or 0.5-1.4 body lengths/s. These speeds are based on the movements of the tracking vessel and therefore are considered conservative because they do not incorporate the horizontal or vertical meandering of the tracked fish or water currents. There were small differences in the mean speeds between day (0.90 m/s) and night (0.94 m/s) for all fish combined. The frequency distribution of these speeds differed markedly between day and night (Fig. 11). However, the number of observations during the day (n=1119) far exceeded those 460 Fishery Bulletin 100(3) 10 0 10 20 Percent of observed deptfis 30 Figure 10 Distribution of observed depths (5-m inter- vals) by day (open bars) and night (shaded bars) for all tracked southern bluefin tuna. at night (n=607) and probably accounts for much of the difference in distributions. 10-| 8- 6- c/) -a 0) 4- Q. en •D 9 > "5 n rfl n ^ Day Mean=0 90 SEM=0 02 n n n=1ll9 o •S 10- C 0) Q o 8- ^ 6- 4- 1 1 ) i_ Night n- "I Mean=0 94 -| "I SEM=0 02 " - n=607 ... _ _. _ T_ 0 12 3 Speed (m/s) Figure 11 Distribution of observed speeds over the ground (0.1 m/s intervals) by day and night for all tracked south- ern bluefin tuna. Discussion Horizontal movements The average swimming speeds of SBT over entire tracks of 0.5-1.4 body lengths/s were similar to the range of mean speeds (0.6-1.0 body lengths/s) for Atlantic bluefin tuna (Lutcavage et al., 2000) and 1.02-1.34 body lengths/s for Pacific bluefin tuna (Marcinek et al., 2001). This is within the range of minimum speeds (0.5-2.0 body lengths/s) required to maintain hydrostatic equilibrium in scom- brids (Magnuson and Weininger, 1978). A much higher sustained swimming speed of 2.5 m/s was recorded for tuna 6, which traveled at this speed for 18 hours, virtually in a straight line, from Rocky Island to the shelf break. The vertical movements of this fish were minimized by it orienting itself with the thermocline interface or the surface. And horizontal movements were not apparent. A speed of 2.5 m/s or 2.6 body lengths/s is probably a real- istic sustained swimming speed for SBT. Although tunas are obviously capable of high bursts of speed, sustained swimming speeds seem to be very much lower. Yellowfin tuna have been tracked at speeds of 1.2-3.5 body lengths/ s, and bigeye tuna at 1.4 body lengths/s (Holland et al., 1990a, Block et al., 1997). The sustained swimming speeds of Atlantic bluefin tuna, determined from sequential aerial photographs, were 0.8-1.6 body lengths/s (NOAA^). Brill (1996) considered that the specialized anatomy, physiol- ogy, and biochemistry of tuna evolved, not for high burst and sustained swimming speeds as proposed by Dickson (1995), but for rapid growth, digestion, and recovery from exhausting activity. The horizontal movements and distribution of SBT suggest strong associations with regions of topographical variation or higher temperatures (or both). The SBT were in the warmer waters of the NW region of the study area in 1994, and the tracked fish also remained within areas of warmer water In 1992 and 1993, all the tuna that trav- eled some distance swam to waters of similar or higher temperatures. Lutcavage et al. (2000) found that tracked North Atlantic bluefin tuna also traveled on the warm side of surface fronts and that ofTshore movements were gener- ally associated with offshore warming of deep basins. The temperatures measured during tracking sometimes did not match up with satellite sea-surface temperatures, owing to the inability of sea surface temperature (SST) algorithms to correct for the variability in atmospheric water vapour, cloud, and aerosols, etc. However, the rela- tive temperature differences considered here appear to be robust. Within these temperature limitations, the tuna were initially located on lumps or near islands. Migration to new areas was often interrupted by brief stays at lumps on the way. The shelf break was clearly the destination of tuna 6. It was tracked near Rocky Island for three hours, before it left abruptly and moved in a straight line to '" NOAA (National Oceanic and Atmospheric Administration). 1975. A study of the applications of remote sensing techniques for detection and enumeration of giant bluefin tuna. Southeast Fish. Cent, contrib. no. 437 (MARMAP no. 108). 48 p. National Space Technology Laboratories, Bay Saint Louis, Mississippi 39520. Davis and Slanley: Movements of Thunmi-i maccoyii in Ihe Great Australian Bight 461 the shelf break, which it reached some 18 hours later It remained in that area until tracking ended. Those lump, reef, island and shelf-break associations matched the dis- tribution of sightings by aerial surs'ey (Cowling et al."). Surfacing behavior Most SBT spend a large part of the day in the upper 10 m (Fig. 10) in the Bight in summer unless they travel to other areas. At night, they tend to move to deeper water, but some tracked tuna remained in the upper 10 m through- out most of the night. The relatively long time spent at the surface during the day in the Bight (nearly 30%) would facilitate sightings by aerial survey, and surfacing behav- ior would significantly influence aerial detection. The surface behavior of southern bluefin tuna in the Bight contrasts with that of most other tunas studied. Yel- lowfin tuna near Hawaii have an average daytime depth of 71m during tracks offshore and away from fish aggregat- ing devices (FADs), with modes at 0-10 m and 50-60 m. When near FADs, they have an average daytime depth of 59 m. At night, yellowfin have modes at 0-10, 30-40, and 180-190 m (Holland et al, 1990a). Similar depth distribu- tions for yellowfin tuna were observed in the western Pa- cific (Yonemori, 1982). In the eastern Pacific Ocean, at the northern extent of their range, yellowfin tuna remained at somewhat shallower depths during the day than in previous studies (Block et al., 1997), presumably because of the shallowness of the mixed layer which limited their depth distribution. In the western Indian Ocean, yellowfin tuna tracked both on and off FADs spent most of the day between 60 and 110 m (Cayre and Chabanne, 1986; Cayre, 1991;Marsacet al.-'). Tracked skipjack tuna spent little time at the surface during the day in waters off Hawaii (Yuen, 1970; Dizon et al., 1978), and Tahiti (Cayre and Chabanne, 1986), although they were surface oriented at night. In the western Indian Ocean, two skipjack tuna tracks were found at somewhat shallower depths but the modal depths were still between 10 and 20 meters (Cayre, 1991). Bigeye tuna tracked off Hawaii had a modal depth of 220 m during the day and 80 m at night (Holland et al., 1990a). Atlantic bluefin tuna appear to spend some time near the surface in inshore wa- ters off the New England area (Carey and Lawson, 1973), although the modal depth of 0-15 m during the day (Lut- cavage et al., 2000) is somewhat deeper than that for SBT in the Great Australian Bight. This nearsurface orienta- tion of Atlantic bluefin tuna has prompted development of aerial surveys to assess the abundance and distribution of these tuna in these waters (Lutcavage and Kraus, 1997). The extensive surfacing behavior of SBT in the Bight does not appear to occur in the shelf waters of the Indian Ocean ' Cowling, A., T. Polacheck, and C. Millar. 1996. Data analysis of the aerial surveys (1991-1996) for juvenile southern bluefin tuna in the Great Australian Bight. 1996 Southern bluefin tuna recruitment monitoring workshop report. Rep. RlVrWS/96/4, 87 p. CSIRO Marine Laboratories, Castray Esplanade, Hobart, Tasmania, Australia 7000. off Western Australia. Of the six SBT tracked in the west, only one spent significant time at the surface during the day (Fishery Agency of Japan''-''). However, this fish oscil- lated rapidly between 0 and 40 m rather than remaining near the surface. The different surfacing behavior of these SBT in the west may have been due to their much smaller size (35-48 cm compared to 67-114 cm), and that they were in a migratory mode moving southwards, assisted by the Leeuwin Current (Shingu, 1967; Maxwell and Cress- well, 1981), rather than in summer residency. The conditions in which SBT come to the surface in the Great Australian Bight are quite varied. These tuna re- main within the surface 5 m for extended periods during the day under a range of weather and sea conditions, al- though surfacing is more pronounced on warm days with little wind. A clear example of this type of movement was evident with tuna 9, which remained below the surface in a school of tuna for the first hour of the track, during which winds were documented at over 35 km/h. After the winds dropped, the tracked tuna surfaced and remained there, apart from occasional dives, for the rest of the track. Surface schools of tuna were observed as far as the eye could see during these calm conditions. In contrast, exten- sive surfacing can occur in adverse weather, as observed in fish 7 between 07:00 and 09:00 h during rough seas and winds >35 km/h. However, in these conditions, no surface fish could be seen from the vessel. Protocols for aerial sur- veys of SBT in the Bight restrict surveys to wind speeds of <18 km/h because of a marked decrease in school sight- ings at higher wind speeds (Cowling et al.'). The highest number of sightings per unit of effort occurred when aerial surveys were flown in conditions of low wind speeds and little cloud cover The ultrasonic telemetry suggests that wind affects detection of schools from the air rather than their surfacing behavior. One of the objectives of the aerial survey was to reduce uncertainty in aerial survey estimates of surface abun- dance by incorporating environmental and behavioral data into the aerial survey analyses. Although the results from ultrasonic tracking provided preliminary informa- tion on vertical distribution that might affect sighting by aerial survey, the intent of our study was to use the more extensive data that would be obtained from archival tags to model responses in surfacing behavior to environ- mental conditions through space and time. Aerial survey estimates of abundance could then be adjusted by incor- porating the proportion of time that SBT would be visible during aerial surveys based on the environmental condi- tions that occurred at the time of survey flights. A distinct surfacing behavior regarded as a characteris- tic of SBT in the Bight during summer is called "rippling" (Hynd and Robins, 1967). This behavior occurs under very calm, sunny conditions. Tuna laze at the surface for extended periods, often breaking this inactive phase by rolling from side to side. Unfortunately we did not have suitable weather conditions to observe this behavior How- ever, data from archival tags suggest that these fish derive significant heating benefits through both insolation and the transfer of heat from the warm surface waters (Gunn etal.2). 462 Fishery Bulletin 100(3) Reference depths Predawn and postdusk dives Reference depths — the depths to which tuna return after regular excursions above or below them — are not well defined in SBT in the Bight. Bounce dives from the sur- face occur in tracks 1,3, and 11, and brief excursions above and below the bottom of the mixed layer could be seen in parts of tracks 4, 6, and 14. None of these showed the movements characteristic of bigeye tuna, which use the 15-16° isotherm as a reference point and make regular excursions to the surface (Holland et al., 1990a). These excursions are thought to be a way of regaining optimum body temperature after foraging in cooler deep water (Holland et al., 1990a; Holland et al., 1992). Bluefin tuna have a more advanced vascular anatomy for physiological thermoregulation than bigeye tuna (Holland et al., 1992); therefore one would expect SBT to display less pronounced behavioral thermoregulation than bigeye tuna under the same thermal regime. Unfortunately, the temperature range of the shallow waters of the Bight was also not large enough to induce such a response. The interface of the surface mixed layer and the ther- mocline is a reference point used by many tuna species. Yellowfin tuna often remain within this narrow depth band, especially when travelling in a straight line (Carey and Olsen, 1982; Holland et al., 1990a). Traveling fish have been observed to alternate abruptly between the thermocline interface and the surface, which might as- sist in straight-line orientation (Holland et al., 1990a). Tracks 6 and 14 provide the clearest examples of use of the thermocline interface during sustained straight-line migration. Tuna 6 remained at a depth of about 20 m while staying in the area. However, once it started swim- ming toward the shelf break, it maintained a relatively constant depth within or just above the thermocline. While this behavior is apparently a characteristic of yellowfin tuna travelling alone (Carey and Olsen, 1982), our tracks were for SBT travelling in schools. Tunas 7 and 8 remained at the surface during fastest straight- line swimming, suggesting that both depth zones might be used during straight-line swimming. Although it appears that the thermocline interface provides some advantage to sustained straight-line movement, it is not clear whether the advantage is in orientation: possibly this temperature range balances heat build-up from sus- tained swimming. In contrast to returning to a reference depth, some tracked fish oscillated rapidly within a depth band. This behavior was observed in parts of tracks 4, 7, 10, and 12. This "frenzied" activity might be ascribed to disturbed behavior caused by the tagging or tracking (or both). How- ever, this behavior might reflect individuals traversing the vertical boundaries of the school (Carey and Olsen, 1982). When this behavior was observed in our study, the tracked fish appeared to be part of a school. However, there were many times when the fish was in a school but rapid verti- cal oscillations were not apparent. Possibly the schools did not extend over sufficient depth range to make this behavior obvious. A feature of many of the tracks of SBT was the deep dive made just before dawn as the sky began to lighten, and just after sunset before all light was gone. In these dives, the tuna descended rapidly, stayed a short time at the bottom of the dive, and then usually rose gradually to its original depth. These signature dives were observed in tracks 1, 3, 4, 6, 7, 11, 12, 13, and 14 and possibly occurred in other cases obscured by repeated dives over the twilight period. Archival tag data showed clear signature dives in the Bight and more varied behavior off the shelf (Gunn et al.-). They also showed that in the central Indian Ocean, SBT dive at dawn, remain at great depth throughout the day feeding on squid, and return to the surface at night. The dawn dives we tracked might correspond to the first part of this movement, but the shelf waters of the Bight are too shallow to have a significant scattering layer, and therefore usually do not allow for the detection of prey. Perhaps this dive at dawn is a fixed behavior that has evolved to locate the scattering layer when SBT are in waters of sufficient depth. If tuna rely largely on sight to locate prey, twilight would be the first opportunity (dawn), and last opportu- nity (dusk), in which to find the scattering layer as it rises during the night and descends during the day. Dawn and dusk dives have been observed in eastern spotted dolphin in the eastern tropical Pacific Ocean (Holland*'). During the day, these dolphin remained in the shallow mixed lay- er (10-15 m). However, at late dusk they dived to depths of about 65 m. These were followed by further dives, each one a little shallower than the last, until diving was again largely confined to the mixed layer. Just before dawn the pattern was reversed. A line drawn along the bottom of the dives from dusk to dawn would describe a bell shape. The dolphins" behavior was thought to be in response to the vertical movement of the scattering layer. Dawn and dusk dives have since been noted in Atlantic bluefin tuna tracked in shelf waters off New England (Lutcavage et al., 2000) and are apparent in tracks of Pacific bluefin tuna in the eastern Pacific (Marcinek et al., 2001). Block et al. (1997) reported predawn dives in yellowfin tuna tracked in the California Bight. Their vertical track plots sug- gested that postdusk dives may also occur Similar verti- cal behavior has been apparent in the track of one skip- jack (Cayre and Chabanne, 1986) and one yellowfin tuna (Carey and Olsen, 1982). However, the absence of similar dives in other studies suggests that the timing of the dives was coincidental. Stomach temperature Muscle and stomach temperatures of Atlantic bluefin tuna were monitored in several tracking experiments (Carey and Lawson, 1973; Carey et al., 1984). Both were shown " Holland. K. N. 1996. Personal commun. University of Ha- waii, PO Box 1346, Coconut Island, Kaneohe, Hawaii 96744. Davis and Stanley Movements of Thunnus maccoyii in the Great Australian Bight 463 to be well above ambient water temperature. Stomach temperature increased after feeding, whicli was at Iribuled to the hydrolytic processes of digestion and an increase in metabolic rate. Possibly raised stomach temperature speeds digestion and enables a higher feeding frequency when food is abundant. The capacity to maintain higher stomach temperature is enhanced by thermal isolation through heat exchangers, the gas bladder, and the fatty body wall (Carey et al., 1984). Therefore, stomach tem- perature would reflect activities associated with feeding and digestion, rather than temperature changes occurring in the body. SBT, while very much smaller than the giant bluefin tuna tracked by Carey, also showed a marked dif- ferential between stomach and ambient temperatures. The largest differential of 9°C (track 11) was found in the largest SBT tracked, which may be relevant. Because tem- perature increased steadily during the track, it was prob- ably due to digestion of a meal. All pole-and-line-caught SBT tracked during these experiments were initially chummed with pilchards and could have eaten pilchards just before tracking. In some instances, tunas regurgitated pilchards while the tag was placed in their stomachs. A caecal temperature differential reaching 10°C has also been observed in a 103-cm SBT after being fed in cage experiments (Gunn et al.^). The smaller temperature dif- ferential observed in the other two tracked tuna (tuna 10 and 13) may have been a consequence of their having less food in their stomach to digest, or their smaller size. Tuna 11 showed the characteristic changes in stomach temperature associated with the swallowing of prey or water or both. Two swallowing events occurred, a minute apart, resulting in a drop in stomach temperature to ambi- ent temperature. Stomach temperature recovered rapidly, but not quite to the level before swallowing. Carey et al. (1984) managed to have Atlantic bluefin tuna retain tags in their stomachs for up to 13 days by feeding them to ensure that their stomachs never remained empty. Sev- eral hours before regurgitation of the tag, a series of cold pulses were observed in stomach temperature, presum- ably from swallowing water which lowered the stomach temperature briefly. If these cold pulses persisted, the tag would be spat out a few hours later. If the tuna was fed, the tag was usually retained. If regurgitation to purge indigestible items from the stomach normally occurs after digestion is complete and the stomach is relatively empty, as suggested by Carey et al. (1984), then it is likely that the swallowing events observed in tuna 11 involved little or no food. Because a very marked drop in stomach tem- perature was observed in tuna 11, a large amount of water must have been swallowed. =* Gunn, J., T. Polacheck, T. Davis. M. Sherlock, and A. Betle- hem. 1994. The development and use of archival tags for studying the migration, behaviour and physiology of southern bluefin tuna, with an assessment of the potential for transfer of the technology to groundfish research. Proc. ICES mini-sym- posium 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. The effects of chumming on tuna behavior Obsci-vations on tracked tuna being caught up in chum- ming by commercial pole-and-line fishing have provided us with insights into these operations. Tunas 5, 6, 8, and 15 were associated with chum lines for part of their tracks. They all followed the chum line at a depth of about 20 m for 15-30 minutes, and some came to the surface for brief periods. The tuna schools extended to at least 200 m behind the vessel that was chumming. Pole-and-line opera- tors usually steam forward at about 2 m/s, chumming with live and, sometimes, frozen pilchards. They progress from tuna school to tuna school hoping to combine schools. Most tuna appear to be below the surface and well behind the boat. Fish progress to the front of the school and come to the surface to take chum, replacing tuna that were there before them. Based on our tracking, it appears that tuna after spending a short time at the surface, break away from the chum line in groups or in schools. Effects of tagging and tracking on tuna behavior There is a concern that capture, tagging, and tracking procedures change the behavior of the fish being observed. A number of criteria are used to support the view that the fish behave normally: they remain within a school (Yuen, 1970; Cayre, 1991; Cayre and Marsac, 1993); show simi- larity in vertical and horizontal movements across tracks (Holland et al., 1990a; Cayre, 1991) or through experimen- tal procedures such as evaluating swimming performance when tags are attached externally (Arnold and Holford, 1979; Blaylock, 1990) or observing food consumption when tags are placed in the stomach (Lucas and Johnstone, 1990). There is however, evidence that SBT do react adversely to being tagged. Hampton (1986) found that lower than expected numbers of SBT tagged with dart tags were re- captured in the first 5 days after release. Also, recaptured fish were in significantly poorer condition than untagged fish — the effect being greatest in those at liberty for 5-20 days (Hampton, 1986) and 13-24 days (Hearn, 1986). Presumably, the adverse effects were related to reduced feeding. Fish generally undergo a phase of recovery from the trauma of capture and tagging. The resumption of "typi- cal" behavior is often taken as a sign that the fish is be- having normally (Holts and Bedford, 1990). In our study, five tuna initially dived to the bottom, five dived below the depth distribution of the tuna school, and six returned immediately to the school. All returned to schools, which were usually near the surface, within 20 minutes. Most significantly, four tracked tuna actually joined the chum lines of commercial pole-and-line vessels shortly after tag- ging. However, it could not be determined whether these four tuna were feeding because they did not have tempera- ture-pressure tags in their stomach. Stress associated with capture and tagging is more obvious in larger fish: blue and striped marlin remain at depth for many hours before returning to normal depths 464 Fishery Bulletin 100(3) (Holland et al., 1990b; Holts and Bedford, 1990; Block et al., 1992a). Black niarlin dive after release but usually return to normal depths soon after (Pepperell and Davis, 1999). High, sustained, swimming speeds for up to the first six hours of blue marlin tracks were attributed to increased ram ventilation to reduce anaerobic debt (Block et al., 1992b). Undoubtedly there is trauma associated with capture and tagging, but pole-and-line capture is fast, and SBT are not as tired as they would be if caught on a rod and line. In most of the SBT tracks, swimming speed was much slower in the half hour after release than for the remainder of the track. Possibly this slow swimming relates to a recovery phase. However, it more likely reflects the nature of pole-and-line fishing and the association of the chummed school with the boat. After catching and releasing the fish, chumming is stopped. However, the school usually remains with the boat for a short period. The strong schooling behavior of SBT ap- pears to quickly override trauma-induced behavior, such as remaining at depth, that might keep tagged fish away from the school. Association of the tracked fish with schools is probably the most encouraging sign to the field worker that the tuna is behaving normally, or at least that its movements are representative of the school as a whole. This associa- tion was marked in our study. However, on the occasions when the tracked fish appeared to be alone, it was not clear whether this isolation was due to the normal breakup and dispersal of the schools aggregated by the chumming op- eration or to the tagging and tracking procedure. We were concerned that "herding" (the vessel following the fish in- fluencing the direction of its movements) over a period of time might separate a fish from its school. However, it is also not clear whether it is possible to detect if the tracked fish is solitary or with a school. Depending on the skill and knowledge of the obsei-ver, schools can usually be detected in good weather if there is surface or subsurface activ- ity. At night or in rough seas, the usual way of detecting schools is by echo-sounder, but this works only if the school is under the tracking vessel. The school must therefore be large and extend well behind the tracked fish (which is nominally tracked at a distance of 400 m). Often "solitary" tracked tuna displayed the repeated and predictable be- haviors seen in other SBT tracks and generally accepted as criteria of "normal" behavior, the most striking of which are the predawn and postdusk dives. Three methods of tag attachment were used in our study — five tags were attached externally behind the sec- ond dorsal fin. ten were placed internally in the stomach, and one tag was accidentally struck by a free-swimming tuna and was attached by the force of the strike. The effects of attaching tags are not currently quantifiable because of the diversity of tuna behaviors. However, the tracks of SBT with stomach tags were more dynamic than the tracks of tuna with tags attached externally. Tunas with stomach tags seemed more responsive to their environment: their behavior included fast, sustained swimming, large translocations, and these tuna were frequently observed with other tuna attracted by chum- ming operations. Externally attached tags presumably created physical disturbances, such as drag, turbulence, and attachment wounds to which the tuna had to adjust. These are more likely to cause short-term disturbance in behavior than swallowed tags. However, stomach tags had unpredictable, and often short, retention times, which counteracted their advantages. Acknowledgments We thank Barry Bruce, Jeff Cordell, Jessica Farley, Ran- dall Gray, Lindsay MacDonald, and Matt Sherlock for helping with tracking operations. We very much appreci- ate the efforts of the captains of the tracking vessels, Stan Lukin, Rex Hall, Stuart Richey, and their crews in the field program. The filtering program was originally developed by Miroslav Ryba and subsequently modified for tempera- ture filtering by Andrew Betlehem. We thank John Gunn and Kim Holland for their reviews of the manuscript and Vivian Mawson for editing. This research was part of a joint Australian Commonwealth Scientific and Industrial Research Organisation/Japanese National Research Insti- tute for Far Seas Fisheries recruitment monitoring project funded by the Australian Fisheries Management Author- ity, Japan Marine Fishery Resources Research Center and Japan Fisheries Agency. Literature cited Arnold. G. P., and B. H. Holford. 1979. The physical effects of an acoustic tag on the swim- ming performance of plaice and cod. J. Cons. Explor. Mer 38(2): 189-200. Blaylock, R. A. 1990. Effects of external biotelemetry transmitters on behavior of the cownose ray Rhinoptera bonasut: (Mitchill 1815). J. Exp. Mar. Biol. Ecol. 141:213-220. Block. B. A., D. T. Booth, and F. G. Carey 1992a. Depth and temperature of the blue marlin. Makaira nigricans, observed by acoustic telemetry. Mar Biol. 114: 175-183. 1992b. Direct measurement of swimming speeds and depth of blue marlin. J. Exp. Biol. 166:267-284. 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 {Than- 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(1):3-15. Brill, R. W., B. A. Block. C. H. Boggs. K. A. Bigelow. E. V. Freund. and D J. Marcinek. 1999. Horizontal movements and depth distribution of large adult yellowfin tuna iThunnus albacares) near the Hawaiian Islands, recorded using ultrasonic telemetry: implications for the physiological ecology of pelagic fishes. Mar Biol. 133:395-408. Carey. F. G., J. W. Kanwisher. and E. D. Stevens. 1984. Bluefin tuna warm their viscera during digestion. J. Exp. Biol. 109:1-20. Davis and Stanley: Movements of Thunnus maccoyii in the Great Australian Bight 465 Carey, F. G., and K. D. Lawson. 1973. Temporaturo regulation in free-swimming bluefin tuna. Comp. Biochem. Physiol. 44(A):375-392. Carey, F. G., and R. J. Olsen. 1982. Sonic tracking experiments with tunas. Int. Comm. Conserv. Atl. Tunas, Coll. Vol. Sci. Pap. 17(2):458-466, Cayr6, R 1991. Behavior of ycUowfin tuna (Thunnus alhacares) and skipjack tuna {Katsuwiinus pelamis) around fish aggregat- ing devices (FADs) in the Comoros Islands as determined by ultrasonic tagging. Aquat. Living Resour 4:1-12. Cayrc, P., and J. Chabanne. 1986. Marquage acoustique et comportement de thons tropicaux (albacore: Thunnus albacares, et listao: Katsu- uionus pelamis) au voisinage d'un dispositif concentrateur de poissons. Oceanogr. Trop. 21 (2):167-183. Cayre, P., and F. Marsac. 1993. Modclhng the yellowfin tuna (Thunnus albacares) vertical distribution using sonic tagging results and local environmental parameters. Aquat. Living Resour 6:1-14. Dickson, K. A. 1995. Unique adaptions of the metabolic biochemistry of tunas and billfishes for life in the pelagic environment. Env. Biol. Fishes 42:65-97. Dizon, A. E., R. W. Brill, and H. S. H. Yuen. 1978. Correlations between environment, physiology, and activity and the effects on thermoregulation in skipjack tuna. In The physiological ecology of tunas (G. D. Sharp and A. Dizon, eds.), p. 233-259. Academic Press, New York, NY. Farley, J. H., and T. L. O. Davis. 1998. Reproductive dynamics of southern bluefin tuna, Thunnus maccoyii. in the Indian Ocean south of the Sunda Islands. Fish. Bull. 96:223-236. Hampton, J. 1986. Effect of tagging on the condition of the southern blue- fin tuna (Thunnus maccoyii). Aust. J. Mar Freshwater Res. 37(6):699-705. Heam.W. S. 1986. Mathematical methodsforevaluatingmarine fisheries. Ph.D. diss., 195 p. Univ. New South Wales, Australia. Holland, K. M., R. W. Brill, and R. K. C. Chang. 1990a. Horizontal and vertical movements of yellowfin and bigeye tuna associated with fish aggregating devices. Fish. Bull. 88:493-507. 1990b. Horizontal and vertical movements of Pacific blue marlin captured and released using sportfishing gear. Fish. Bull. 88:397-402. Holland, K. M., R. W. Brill, R. K. C. Chang, J. R. Sibert, D. A. Fomier 1992. Physiological and behavioral thermoregulation in bigeye tuna (Thunnus obesus). Nature 258:410-412. Holland, K. M., R. W. Brill, R. K. C. Chang, and R. Yost. 1985. A small vessel technique for tracking pelagic fish. Mar Fish. Rev 47(41:26-32. Holland, K. M., and J. R. Sibert. 1994. Physiological thermoregulation in bigeye tuna, Thun- nus obesus. Env Biol. Fishes 40:319-327. Holts, D, and D.Bedford. 1990. Activity patterns of striped marlin in the Southern California Bight. In Proceedings of the second interna- tional billfish symposium, Kailua-Kona, Hawaii, August 1-5, 1988, part 2: contributed papers, p. 81-93. National Coalition for Marine Conservation, Inc., Savannah, GA. Hynd,J. S.,and J. P Robins. 1967. Tasmanian tuna survey report of first operational period. CSIRO Div. Fish. Ocean. Tech. Pap. 22, 55 p. Laurs, R. M., H. S. H. Yuen, and J. H. Johnson. 1977. Small-scale movements of albacore, Thunnus ala- lunga, in relation to oceanographic features as indicated by ultrasonic tracking and oceanographic sampling. Fish. Bull. 75:347-355. Levenez, J. J. 1982. Note preliminaire sur I'operation sengelaise de track- ing de listao. Coll. Vol. Sci. Pap. Int. Comm. Cons. Atl. Tunas 17:189-194. Lucas, M. C, and A. D. F. Johnstone. 1990. Observations on the retention of intragastric trans- mitters, and their effects on food consumption, in cod, Gadus morhua L. J. Fish Biol. 37:647-649. Lutcavage, M. E., R. W. Brill, G. B. Skomal, B. C. Chase, and J. Tutein. 2000. Tracking adult North Atlantic bluefin tuna (Thunnus thynnus) in the northwestern Atlantic using ultrasonic telemetry. Mar. Biol. 137:347-358. Lutcavage, M., and S. Kraus. 1997. The feasibility of direct photographic assessment of giant bluefin tuna in New England waters. Fish. Bull. 93: 495-503. Magnuson, J. J., and D. Weininger. 1978. Estimation of minimum sustained speed and associ- ated body drag of scombrids. In The physiological ecology of tunas (G. D. Sharp and A. E. Dizon, eds.), p. 293-312. Academic Press, New York, NY. 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. Maxwell, J. G. H., and G. R. Cresswell. 1981. Dispersal of tropical marine fauna to the Great Australian Bight by the Leeuwin Current. Aust. J. Mar. Freshwater Res. 32:493-500. Pepperell. J. G., and T L. O. Davis. 1999. Post-release behavior of black marlin, Makaira in- dica, caught off the Great Barrier Reef with sportfishing gear. Mar Biol. 135:360-380. Shingu, C. 1967. Distribution and migration of the southern bluefin tuna. Nankai Reg. Fish. Res. Lab. Rep. 25:19-36. Vemco. 1992. VSCAN software, version 3.02. VEMCO Limited, Shad Bay, Nova Scotia, Canada. Yonemori, T 1982. Study of tuna behavior, particularly their swimming depths, by the use of sonic tags. Far Seas Fish. Res. Lab. (Shimizu) Newsletter 44:1-5. (Engl. Transl. No. 70 by T Otsu, 1982, 7 p. [Available from Honolulu Laboratory, Southwest Fish. Cent., Nat. Mar. Fish. Serv. NOAA, Hono- lulu, HI 96822-2396.1 Yuen, H. S. H. 1970. Behavior of skipjack tuna, Katsuwonus pelamis, as determined by tracking with ultrasonic devices. J. Fish. Res. Board Can. 27:2071-2079. 466 Abstract— All five species of sea tur- tles in continental U.S. waters are protected under the Endangered Spe- cies Act of 1973 and the population sizes of all species remain well below historic levels. Shrimp trawling was determined to be the largest source of anthropogenic mortality of many of the species. As a mechanism to reduce the incidental catch of turtles in trawl nets, turtle excluder devices have been required intermittently in the shrimp fishery since 1987. and at all times since 1994. The expanded turtle excluder device (TED) regulations, implemented in 1994, were expected to reduce shrimp trawl capture of sea turtles by 97%. Recent evidence has indicated that the sizes of turtles stranding were not representative of the animals sub- jected to being captured by the shrimp trawlers. The purpose of our study was to compare the sizes of stranded sea turtles with the size of the TED openings. We compared the sizes of stranded loggerhead tCaretta carctta), green iChelonta mydas), and Kemps ridley (Lepidochelys kempii ) sea turtles, the three species most commonly found stranded, to the minimum widths and heights of TED openings. We found that annually a large proportion of stranded loggerhead turtles (33^7%) and a small proportion of stranded green tur- tles ( 1-7% ) are too large to fit through the required minimum-size TED open- ings. The continued high mortality of sea turtles caused by bottom trawling is reason for concern, especially for the northern subpopulation of loggerhead turtles, which cun-ently is not projected to achieve the federal recovery goal of reaching and maintaining prelisting levels of nesting. Turtle excluder devices- Are the escape openings large enough?* Sheryan P. Epperly Wendy G. Teas Southeast Fisheries Science Center National Marine Fishenes Service 75 Virginia Beach Drive Miami, Florida 33149 E-mail address (for S P Epperly): sheryan, epperlyrainoaa gov Manuscript accepted 12 February 2002. Fish. Bull. 100:466-474 (2002). All five species of sea turtles in continen- tal U.S. waters are protected under the Endangered Species Act of 1973 (ESA, PL93-205). Elasticity models of turtle populations have indicated that the life stages with the highest elasticity are juveniles (i.e. a reduction in mortality in these stages would result in the great- est annual population multiplication rate) (Crouse et al., 1987; Crowder et al., 1994; Heppell, 1998a, 1998b; Epperly et al., 2001; Heppell et al., in press). Size data indicate that the sea turtles most often found dead on ocean beaches are immature (Crouse et al., 1987; STSSNM and shrimp trawling is thought to account for the majority of these deaths (Magnuson et al.. 1990; Caillouet et al, 1991, 1996; Crowder et al., 1995). Strandings, however, likely represent only a small proportion of the animals that die offshore (TEWG, 1998). Beginning in the fall of 1987, the National Marine Fisheries Service (NMFS) seasonally required turtle excluder devices (TEDs) in shrimp trawl nets on most vessels operating in ocean waters off the southeastern U.S. as a mechanism to reduce the inciden- tal catch of turtles in general and the catch of the large immature turtles in particular (Federal Register, 1987); ves- sels operating off Cape Canaveral and off southwest Florida were required to use TEDs all year. Boats working in inshore waters were allowed to use tow time limits in lieu of TEDs. The dif- ference between offshore and inshore regulations was due, in part, to the lack of information on the distribution and abundance of sea turtles in inshore waters and to the lack of documenta- tion of incidental captures by shrimp trawlers working in these inshore wa- ters (Federal Register 1992a). For the first few years implementation of the regulations was delayed by challenges in the courts and in Congress. The regulations were implemented fully in Spring 1990. Evidence of the importance of in- shore areas to sea turtles, along with evidence that shrimp trawlers catch sea turtles in inshore waters (Epperly et al., 199.5; NMFS- ) provided sufficient justification for NMFS to expand re- quirements for turtle excluder devices in the shrimp fishery to all areas at all times, including inshore waters; full implementation of these requirements was achieved by December 1994 (Fed- eral Register, 1992a, 1992c). The ex- panded TED regulations were expected to reduce shrimp trawling capture of sea turtles by 97% (Henwood et al., 1992). Since 1992, TEDs also have been required in the winter trawl fishery for summer flounder operating as far north as Cape Charles, Virginia (Fed- eral Register, 1992b). * Contribution PRD-99/00-07 of the South- east Fisheries Science Center. Miami, Florida 33149. ' STSSN (Sea Turtle Stranding and Salvage Network). 1998. Unpubl. data. The Sea Turtle Stranding and Salvage Net- work is a cooperative endeavor between NMFS, other federal agencies, the states, many academic and private entities, and innumerable volunteers. Data are archived at the National Marine Fisher- ies Service Southeast Fisheries Science Center, 75 Virginia Beach Dr, Miami. FL 33149. - NMFS (National Marine Fisheries Service). 1990. Unpubl. data. NMFS Galveston Laboratorv, 4700 Avenue U, Galveston, TX 77.551." Epperly and Teas; Escape openings in turtle excluder devices 467 Hi-i^ht: 1(1 in (25.40 cm) CJiill Ot Mexico 12 in (30.48 cm) Allunlic Width: 32 in (SI. 28 cm) Gulf of Mexico 3.'i in (88.90 cm) Atlantic Figure 1 Minimum dimensions (height and width) specified m current U.S. federal regulations (Federal Register, 1992c) for e.scape openings in single-grid hard turtle excluder devices. The copyrighted figure of the turtle was kindly provided by Jamie Serino of Flying Turtle Productions. TEDs are equipped with a trap door that allows sea turtles to escape from trawl nets (Seidel and McVea, 1982) and may be either rigid or soft in design ( Federal Register, 1992c). To be certified by NMFS, a TED design must be 97% effective, in comparison with a control net, in exclud- ing sea turtles (Federal Register 1987, 1992c). Since 1990, turtles used for TED trials have been small loggerhead turtles, mostly 2->T-olds that have averaged 34.4 cm SCL (SD=4.1, 7! = 1730, NMFS-^). Regardless of design, certain parameters of the TED architecture are regulated. Most important to this discussion are the requirements of the height and width dimensions of the opening in the net through which turtles escape. Along the Atlantic Coast these requirements are width >35 in (88.90 cm ) and height >12 in (30.48 cm) (Federal Register, 1992c). In the Gulf of Mexico these measurements are >32 inch (81.28 cm) and >10 inch (25.40 cm), respectively. Height is measured si- multaneously with width and is measured at the midpoint of the straight-line distance of width (i.e. the width and height of a taut triangle is measured. Fig. 1). The purpose of our study was to compare the sizes of stranded sea turtles with the size of the TED openings. This evaluation was prompted by the need, identified by the NMFS Turtle Expert Working Group (TEWG), to reduce the strandings of mature loggerhead sea turtles [Caretta caretta) from the northern subpopula- tion (TEWG, 1998). We compared the sizes of stranded loggerhead, green (Chelonia mydas), and Kemp's ridley iLepidochelys kempii) sea turtles, the three species most commonly found stranded, to the minimum widths and heights of TED openings. 3 NMFS (National Marine Fisheries Service). 2001. Unpubl. data. Galveston Laboratory, National Marine Fisheries Ser- vice, 4700 Ave. U, Galveston, TX 77551. Materials and methods To compare the sizes of stranded turtles with the min- imum size of TED openings we first constructed a pre- dictor of carapace width and body depth. Thus, a morpho- metric analysis for each species was conducted, generally with data from reared, live captured, or nesting turtles and not with data from strandings. The predictive regres- sion for carapace width was then applied to the strand- ings data when this measurement was not recorded for a given turtle; the predictive regression for body depth was applied to all turtles in the database because body depth was rarely measured at stranding sites. We then analyzed the entire strandings database to compare turtle sizes with the minimum size of TED openings. Morphometric analyses The species-specific relationship between both body depth and carapace width with carapace length was explored through regression analysis and predictive regression equations were developed. Regressions of untransformed data were compared with regressions of log^-transformed data by comparing goodness-of-fit values. Morphometric data (straight line carapace length, notch- to-tip ISCL], straight line carapace width [SOW], and body depth [BD] ) were recorded by a number of researchers throughout the southeast United States and at the Cay- man Turtle Farm, Cayman Islands. Data for loggerhead turtles were concentrated in the 20-30 cm and 30-40 cm SCL size classes and were censored (randomized selection of ^=37 in each of the two size classes) to create a more uni- form distribution for the analysis (Table 1). Green turtle data were more uniformly distributed across size classes and were not censored (Table 1). Data for Kemp's ridley turtles were concentrated in the 1-10 cm and 10-20 cm 468 Fishery Bulletin 100(3) Table 1 Dis tribution of sizes of sea turtles used in the morphometric analysis. Straight carapace length (cm) Loggerhead Green Kemp's ridley Frequency Censored frequency Censored Frequency frequency Frequency 1.01-10 1 1 17 3032 105 10.01-20 1 1 5 3778 105 20.01-30 123 37 37 155 155 30.01-40 629 37 49 58 58 40.01-50 24 24 22 137 137 50.01-60 48 48 21 71 71 60.01-70 26 26 16 70.01-80 10 10 1 80.01-90 32 32 3 90.01-100 27 27 2 100.01-120 7 7 3 SCL size classes and were censored (randomized selection of n=105) in each of the two size classes (Table 1). Strandings analyses The Sea Turtle Stranding and Salvage Network (STSSN) documents dead or injured sea turtles along the coasts of the eastern United States and the U.S. Caribbean (Schroeder, 1989). The STSSN relies on a trained group of volunteers, including state and federal employees and private individuals, to collect basic biological data on each stranded turtle. Each animal is identified to species, the condition or state of decomposition is determined, standard carapace measurements are taken, and any obvious wounds, injuries, or abnormalities are noted and described. Volunteers who have received additional train- ing may also perform necropsies, or internal exams, on a carcass to determine the general state of health of the animal prior to death, to determine sex, and to locate any obvious internal abnormalities. Data are recorded on stan- dardized report forms that are submitted first to a state coordinator and then to the national STSSN coordinator at the National Marine Fisheries Service, Southeast Fish- eries Science Center, Miami, Florida. The species-specific predictive regression equations from the morphometric analyses were used to estimate the carapace width for each turtle in the STSSN database for which this measurement had not been taken and to estimate the body depth for each turtle. For turtles with curved measurements only, straight line carapace lengths were estimated from curved carapace lengths (CCL) be- fore estimating body depth and carapace width by apply- ing equations reported by Teas ( 1993). Within each region (Fig. 2) carapace widths were com- pared with the currently required minimum widths of TED openings and body depths were compared with the currently required minimum heights of TED openings. Stranded turtles that were reared in captivity, cold- stunned, or known to have been captured incidentally were censored. Results Morphometric analyses Loggerhead sea turtles The relationships between cara- pace width and carapace length and between body depth and carapace length were linear. Coefficient of determina- tion (/■-) values of regressions with log-transformed data were slightly ( <0.002 ) higher than values based on untrans- formed data. Regression of each of the morphometric values on carapace length was highly significant (P<0.0001) and resulted in the following predictive equations: In sew = -0.0225 + (0.9507 x In SCL) [n=250, /-■-=0.9891, In BD = -0.5682 -t- (0.9100 x In SCL) \n='250, r2=0.9661. Straight line carapace lengths corresponding to turtles with carapace widths of 81.28 cm (32 inch; the minimum width ofTED openings in the Gulf of Mexico) and 88.90 cm (35 inch; the minimum width of TED openings in the At- lantic) were 104.5 cm and 114.9 cm, respectively. Straight line carapace lengths corresponding to turtles with body depths of 25.40 cm (10 inch is the minimum height of TED openings in the Gulf of Mexico) and 30.48 cm (12 inch is the minimum height of TED openings in the Atlantic) were 65.3 cm and 79.8 cm, respectively. Green sea turtles The relationships between carapace width and carapace length and between body depth and Epperly and Teas: Escape openings in turtle excluder devices 469 ^V 50° --^ ^ ^_ ^ -v-^^^ 40° l^ NE U.S. 30° \A WGulf EGulf ( SEU.S. N t 100° 90° 80° 70' Figure 2 Regions of reported sea turtle strandings along the coasts of the eastern United States and the Gulf of Mexico. carapace length were linear. Coefficient of determination (r^) values of regressions with log^-transformed data were sHghtly (<0.015) higher than values based on untrans- formed data. Regression of each of the morphometries on carapace length was highly significant (P<0.0001) and resulted in the following predictive equations: In SCN = -0.1608 + (0.9812 x In SCL), [« = 176, /■2=0.995], In BD = -1.0115 + ( 1.0023 x In SCL). [n = 176, r-'=0.9771. Straight line carapace lengths corresponding to turtles with carapace widths of 81.28 cm (32 inch) and 88.90 cm (35 inch) were 104.2 cm and 114.1 cm, respectively. Straight line carapace lengths corresponding to turtles with body depths of 25.40 cm (10 inch) and 30.48 cm (12 inch) were 69.2 cm and 83.0 cm, respectively. Kemp's ridley sea turtles The relationships between cara- pace width and carapace length and between body depth and carapace length were linear. Coefficient of determi- nation (r^) values of regressions with logj,-transformed data were slightly (<0.006) higher than values based on untransformed data. Regression of each of the morphomet- ries on carapace length was highly significant (P<0.0001) and resulted in the following predictive equations: In sew = -0.2039 -t- (1.0437 x In SCL) [n=631, r2=0.998], In BD = -0.6283 + (0.9075 x In SCL). ln=631, r2=0.989]. Straight line carapace lengths corresponding to turtles with carapace widths of 81.28 cm (32 inch) and 88.90 cm (35 inch) were 82.2 cm and 89.6 cm, respectively. Straight line carapace lengths corresponding to turtles with body depths of 25.40 cm (10 inch) and 30.48 cm (12 inch) were 70.6 cm and 86.3 cm, respectively. Strandings analyses Straight carapace length and width were not measured for a number of stranded sea turtles; body depth almost never was recorded. The total number of records, by species, for which the predictive regressions were applied to estimate straight carapace length or straight carapace width are given in Table 2. Note that the length of a turtle, straight line or curved, must have been measured for the turtle to be included in the analyses because the predictive mea- sures were based on length. It should also be noted that the conclusions from the strandings analyses were not altered by the choice of linear or log-transformed data in the morphometric analyses above. Loggerhead sea turtles Carapace width Strandings of loggerhead turtles with carapace widths greater than the currently required minimum widths of TED openings have not exceeded 1% 470 Fishery Bulletin 100(3) of the total measured strandings in any year since 1986 (Table 3). The majority of the stranded large (wide) turtles occur in the eastern Gulf of Mexico and the southeast U.S. Atlantic regions, areas where significant nesting occurs. Body depth Strandings of loggerhead turtles with body depths greater than the currently required minimum heights of TED openings have ranged between 33% and 47% of the total stranded turtles measured every year since 1986 (Table 4]). From 1995 to 1997 nearly 1300 stranded loggerhead turtles were deeper bodied that the currently required minimum TED height opening. The highec!, proportion of turtles that were too deep bodied to pass through TEDs was found to be in the Gulf of Mexico where TED openings are smaller The greatest numbers of large turtle strandings occurred on nesting beaches of the eastern Gulf of Mexico and the southeast U.S. Atlantic. Table 2 The total number of records, by species, for which the pre- dictive regressions were applied to estimate straight line carapace length or straight line carapace width for logger- head, green, and Kemp's ridley sea turtles. Kemp's Missing measurement Loggerhead Green ridley Straight line carapace 8340 1034 1209 length Straight line carapace 8555 1089 1261 width Green sea turtles Carapace width Strandings of green turtles with cara- pace widths greater than the currently required minimum width of TED openings have not exceeded two turtles or 2%' of the total stranded turtles measured in any year since 1986 (Table 5). Body depth Strandings of green turtles with body depths greater than the currently required minimum height of TED openings have ranged between 1% and 7% of the total stranded turtles measured since 1986 (Table 6). The large turtles were found sranded in the eastern Gulf of Mexico and the southeast U.S. Atlantic regions, the latter an area of nesting activity. Kemp's ridley sea turtles None of the nearly 3000 measured Kemp's ridley turtles that stranded during 1986-97 (total stranded=3476) had carapace widths or body depths greater than the currently required minimum widths and heights of TED openings. Discussion All ESA-listed species of sea turtles remain below their historic levels of abundance. The status of Kemp's ridley and loggerhead sea turtles was recently evaluated by the NMFS Turtle Expert Working Group ( 1998; 2000). Kemp's ridley turtles constitute a single management unit and the nesting population is increasing. If the population contin- ues to grow exponentially, the recovery goal of downlisting Table 3 Number of stranded loggerhead sea minimum width of openings current turtles and percentage of those measured with y required in turtle excluder devices. (predicted) carapace widths gi-eater than the Year Region of stranding Total number measured Total number stranded Western Gulf Eastern Gulf SE U.S. Atlantic NE U.S. Atlantic All regions n '■^ n ^'f n c; n ' i n 0'- 1986 0 0 2 2 5 1 1 5 8 1 959 1209 1987 1 1 3 2 5 1 0 0 9 1 1318 1728 1988 0 0 5 3 1 0 1 1 8 1 1105 1373 1989 0 0 3 1 1 4 0 0 7 1 1088 1425 1990 0 0 3 3 4 0 0 0 7 1 1258 1592 1991 0 0 2 2 1 0 0 0 3 0 777 975 1992 1 2 0 0 1 0 0 0 2 0 798 1101 1993 0 0 1 1 0 0 1 1 2 0 693 972 1994 0 0 5 7 2 0 0 0 7 1 1044 1342 1995 0 0 1 1 0 0 1 1 4 0 973 1424 1996 0 0 1 1 1 0 2 1 4 0 1461 1883 1997 0 0 6 4 0 0 2 1 8 1 1289 1643 Epperly and Teas: Escape openings in turtle excluder devices 471 the species from endangered to threatened status (10,000 females nesting in a season) will be met early in this cen- tury (TEW(!, 2000). There are five known subpopulations of loggerhead turtles in the Western North Atlantic (Enca- lada et al., 1998; Francisco Pearco, 2001; Francisco et al.. in press) but the status of only two could be addressed by the TEWCl. Nesting levels of the South Florida sub- population appear to be increasing, meeting one of the recovery goals set for Florida loggerhead sea turtles. The northern subpopulation, which nests from northern Table 4 Number of stranded loggerhead sea turtles and percentage of those measured with predicted body depths mum height of openings currently required in turtle excluder devices. greater than the mini- Year Region of stranding Total number measured Western Gulf Eastern Gulf SE U.S. Atla ntic NE U.S. Atlantic All regions n % n % n '^< n % n ^f 1986 68 44 72 89 188 27 4 24 332 35 954 1987 75 54 123 96 225 23 11 25 434 33 1309 1988 62 56 134 96 187 25 6 17 389 38 1027 1989 41 44 209 92 179 26 4 19 433 42 1042 1990 48 36 91 88 250 27 6 22 395 33 1188 1991 37 54 73 83 162 31 18 30 290 40 734 1992 35 58 66 85 198 34 13 28 312 41 763 1993 26 47 76 89 182 40 12 28 296 47 636 1994 97 57 64 88 237 35 15 15 413 41 1010 1995 56 52 66 90 207 30 18 19 347 36 956 1996 99 52 127 88 253 27 33 20 512 36 1436 1997 97 66 131 87 171 23 29 14 428 34 1266 Table 5 Number of stranded green sea turtles and percentage of those measured mum width of openings currently required in turtle excluder devices. with (predicted) carapace widths greater thar the mini- Year Region of stranding Total number measured Total number stranded Western Gulf Eastern Gulf SE U.S. Atlantic NE U.S. Atlantic All regions n 9c '! 'i n '7f n 9c n % 1986 0 0 2 22 0 0 0 0 2 2 101 125 1987 0 0 0 0 0 0 0 0 0 0 122 142 1988 0 0 0 0 0 0 0 0 0 0 178 193 1989 0 0 0 0 0 0 0 0 0 0 223 255 1990 0 0 0 0 0 0 0 0 0 0 280 308 1991 0 0 0 0 0 0 0 0 0 0 200 221 1992 0 0 1 4 0 0 0 0 1 1 184 208 1993 0 0 0 0 0 0 0 0 0 0 180 200 1994 0 0 0 0 0 0 0 0 0 0 268 320 1995 0 0 0 0 0 0 0 0 0 0 312 389 1996 0 0 0 0 1 0 0 0 1 0 508 584 1997 0 0 0 0 0 0 0 0 0 0 286 352 472 Fishery Bulletin 100(3) Table 6 Number of of openings stranded green turtles currently required in and percentage of those measured w turtle excluder devices. th predicted body depths greater than the minimum height Year Region of stranding Total number measured Western Gulf Eastern Gulf SE U.S. Atlantic NE U.S. Atlantic All regions n % II '7r 11 % n '^■'f II % 1986 1 17 3 33 3 4 0 0 7 7 100 1987 2 13 3 12 3 4 0 0 8 7 122 1988 1 11 2 8 2 1 0 0 5 3 177 1989 3 21 3 9 6 3 0 0 12 5 219 1990 0 0 1 2 9 4 0 0 10 4 277 1991 0 0 3 8 3 2 0 0 6 3 196 1992 0 0 2 8 8 6 0 0 10 6 180 1993 0 0 4 11 4 3 1 33 9 5 175 1994 1 2 4 12 9 5 0 0 14 5 258 1995 0 0 1 2 3 1 0 0 4 1 301 1996 1 2 10 9 10 3 0 0 21 4 500 1997 1 3 4 10 7 3 0 0 12 4 282 Florida through North Carolina may now be stable but is currently well below goals set for its recovery (return to prelisting nesting levels). Mortality on at least the north- ern subpopulation of loggerhead sea turtles needs to be reduced throughout its range to ensure recovery. Although subpopulations of loggerhead turtles can be easily distinguished by the geographic location of their nesting beaches, the subpopulations comingle on the for- aging grounds (Sears, 1994; Norrgard, 1995; Sears et al., 1995; Rankin-Baransky et al., 2001; Witzell et al., 2002; Bass et al., in press). Genetic studies of foraging and stranded animals indicate that the immature benthic ani- mals of the northern subpopulation are distributed along the Atlantic seaboard (Sears, 1994; Norrgard, 1995; Sears et al., 1995; Rankin-Baransky et al, 2001; Witzell et al, 2002; Bass et al., in press), in Florida Bay (Bass et al.^), and in the Gulf of Mexico (Bass et al.^). Non-nesting adult females from the northern subpopulation appear to occur exclusively along the east coast of the United States with rare exception, and none have been reported from inter- national waters (Bell and Richardson, 1978; Williams and Frick, 2001;CMTTP6). * Bass, A. L., M. Clinton, and B. W. Bowen. 1998, Loggerhead turtles iCaretta caretta i in Florida Bay: an assessment of origin based on genetic markers. Unpubl. report to Florida Depart- ment of Environmental Protection, 5 p. Department of Fish- eries and Aquatic Sciences, University of Florida, Gainesville, 7922 NW 71»' St., Gainesville, FL 32653. 5 Bass. A. L., S.-M. Chow, and B. W. Bowen. 1999. Final report for project titled; genetic identities of loggerhead turtles stranded in the Southeast United States. Unpubl. report to National Marine Fisheries Service, order number 40AANF 809090, lip. Department of Fisheries and Aquatic Sciences, University of Florida, 7922 NW 71*' St., Gainesville, FL 32653. Eight nesting subpopulations were identified for green turtles in the Atlantic Ocean (Encalada et al., 1996), but lat- er were reduced to five regional population units (Bass and Witzell, 2000). Like loggerheads, the subpopulations com- ingle on the foraging grounds (Bass et al, 1998; Lahanas et al., 1998; Bass and Witzell, 2000). The status of all these subpopulations has not been evaluated, but it appears that nesting levels are increasing on the east coast of Florida (Meylan et al., 1995; Florida Fish and Wildlife Conservation Commission" ) as well as at Tortuguero, Costa Rica, the larg- est Western Atlantic rookery (Bjorndal et al., 1999). A large proportion of stranded loggerhead turtles and a small proportion of stranded green turtles are too large to pass through the required minimum-sizes for TED openings. This finding is corroborated by analyses that suggest that the size distribution of stranded loggerheads is different (larger) than the size distribution of turtles in the nearshore waters (TEWG, 2000). The relatively large proportion of stranded loggerhead turtles with dimen- sions greater than the minimum height required for TED openings is cause for concern in light of the need to reduce mortality on the northern subpopulation of loggerhead sea ' CMTTP (Cooperative Marine Turtle Tagging Program). 2001. Unpubl. data. The Cooperative Marine Turtle Tagging Pro- gram was established by NMFS in 1980 to centralize the tagging programs among sea turtle researchers, distribute tags, manage tagging data, and facilitate exchange of tag information. Since 1999 the CMTTP has been managed by the Archie Carr Center for Sea Turtle Research at the Universitv of Florida, PO Box 118525, Gainesville, FL 32511. Florida Fish and Wildlife Conservation Commission. 2001. Unpubl. data. 100 Eighth Avenue S.E., St. Petersburg, FL 33701 Epperly and Teas: Escape openings in turtle excluder devices 473 turtles (TEWG, 1998: 2000) and indicates that, for logger- head sea turtles, TEDs have not achieved a 9T'/r reduction in captures by shrimp trawlers. Loggerhead turtles are ex- ceeding in size the required minimum height of openings in TED before reaching maturity, especially in the Gulf of Mexico where the allowed opening is smaller than that in the Atlantic. This anomaly had not been noted previously because, since 1990, turtles used for TED trials have been small (<21 cm in body depth, n=1415; NMFS-'). Based on stage elasticities, a proportional reduction in mortality in the smallest size classes not fitting through the TED open- ings (large immature turtles) would result in the greatest increased annual population multiplication rate (Grouse et al., 1987; Heppell. 1998a: Epperly et al., 2001). A reduc- tion in subadult and adult mortality from drowning in trawls would benefit all species and subpopulations of sea turtles (Heppell, 1998b). To decrease the mortality on large turtles caused by trawling, the opening dimensions of TEDs need to be larger than the current minimum requirements and need to be the same in the Gulf of Mexico and the Atlantic. Possible management options include the following: 1) in- crease the dimensions to accommodate some desired pro- portion of adults or the total population and 2) adopt the "leatherback" modification (Federal Register, 1993, 1994, 1995) for all areas and all times, which would allow the exclusion of turtles of all sizes, including leatherbacks which are the largest of the sea turtles. In response to our findings, an advance notice of a proposed rulemaking, to effect a change in TED requirements, was issued by NMFS (Federal Register, 2000). After consideration of public com- ments, NMFS advertised a proposed rule to change the TED requirements (Federal Register, 2001). NMFS should also consider extending the TED regulations to other bot- tom trawl fisheries of the Gulf of Mexico and along the Atlantic seaboard, including in the Northeast United States, whenever turtles and bottom trawling activity may co-occur. Acknowledgments We thank the following people for providing morphometric measurements: Allen Foley, Tim Fontaine, Jerris Foote, Jonathan Gorham, Ben Higgins, John Hammond, Joanne Braun McNeill, Joe Parsons, Mike Salmon, Jeffrey Schmid, and Jeanette Wyneken. We thank the state coordinators of the Sea Turtle Stranding and Salvage Network and its innumerable volunteers, past and present, for providing the numbers and size distributions of stranded sea turtles. We also thank Joseph Powers for his constructive review of the manuscript and Jamie Serino for allowing us to use his copyrighted drawing of a turtle. Literature cited Bass, A. L., S. P. Epperly, J. Braun-McNeill, and A. Francisco. In press. Variation in the stock composition of a loggerhead turtle (Caretta caretta) foraging habitat. Conserv. (jenetics. Bass, A .L., C. J. Lagueux, and B. W. Bowen. 1998. Origin of grcon turtles, Chetonia mydas, at "Sleeping Rocks" off the northeast coast of Nicaragua. Copeia 1998: 10(54-1069. Bass, A. L., and W. N. Witzell. 2000. Demographic composition of immature green turtles (Chelonia mydas) from the East Central Florida Coast: evi- dence from mtDNA markers. Herpelologica 56:357-367. F?ell, R., and J. I. Richardson. 1978. An analysis of tag recoveries from loggerhead .sea turtles (Caretta caretta) nesting on Little Cumberland Island, Georgia. Fla. Mar Res. Publ. 33:20-24. Bjorndal, K. A., J. A. Weatherall, A. B. Bolten, and J. A. Mortimer 1999. Twenty-six years of green turtle nesting at Tortu- guero, Costa Rica: an encouraging trend. Conserv. Biol. 13:126-134. Caillouet, C. W., Jr, M. J. Duronslet, A. M. Landry Jr, D. B. Revera, D. J. Shaver, K. M. Stanley, R. W. Heinly, and E. K. Stabenau. 1991. Sea turtle strandings and shrimp fishing effort in the Northwestern Gulf of Mexico, 1986-1989. Fish. Bull. 89: 712-718. Caillouet, C. W., Jr, D. J. Shaver W. G. Teas, J. M. Nance, D. B. Revera, and A. C. Cannon. 1996. Relationship between sea turtle stranding rates and shrimp fishing intensities in the northwestern Gulf of Mexico: 1986-1989 versus 1990-1993. Fish. Bull. 94:237- 249. Crouse, D. T., L. B. Crowder, and H. Caswell. 1987. A stage-based population model for loggerhead sea turtles and implications for conservation. Ecology 68: 1412-1423. Crowder, L. B., D. T. Crouse, S. S. Heppell, and T.H. Martin. 1994. Predicting the impact of turtle excluder devices on log- gerhead sea turtle populations. Ecol. Applic. 4:437^45. Crowder, L. B., S. R. Hopkins-Murphy, and J. A. Royle. 1995. Effects of turtle excluder devices (TEDs) on logger- head sea turtle strandings with implications for con- servation. Copeia 1995:773-779. Encalada, S. E., P. N. Lahanas, K. A. Bjorndal, A. B. Bolten, N. M. Miyamoto, and B. W. Bowen. 1996. Phylogeography and population structure of the green turtle iChelonia mydas) in the Atlantic Ocean and Mediterranean Sea: a mitochondrial DNA control region assessment. Mol. Ecol. 5:473-484. Encalada. S. E., K. A. Bjorndal, A. B. Bolten, J. C. Zurita, B. Schroeder, E. Possardt, C. J. Sears, and B. W. Bowen. 1998. Population structure of loggerhead turtle {Caretta caretta) nesting colonies in the Atlantic and Mediterra- nean as inferred from mitochondrial DNA control region sequences. Mar Biol. 130:567-575, Epperly, S. P., J. Braun, and A. Veishlow. 1995. Sea turtles in North Carolina waters. Conserv. Biol. 9:384-394. 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. U.S. Dep. Commer NOAA Tech. Memo. NMFS-SEFSC-455:3-66. Federal Register 1987. 52 (124):24244-24262. June 29, 1987. Government Printing Office, Washington D.C. 1992a. 57 (84):18446-18461. April 30, 1992. Government Printing Office, Washington D.C. 474 Fishery Bulletin 100(3) 1992b. 57 (219):53603-53606. November 12, 1992. Gov- ernment Printing Office, Washington D.C 1992c. 57 (234):57348-57358. December 4, 1992. Govern- ment Printing Office, Washington D.C. 1993. 58 (93):28795-28798. May 17, 1993. Government Printing Office, Washington D.C. 1994. 59 (95);25827-25831. May 18, 1994. Government Printing Office, Washington DC. 1995. 60 (178):47713-47715. September 14, 1995. Govern- ment Printing Office, Washington DC. 2000. 65 (66):17852-17854. April 5, 2000. Government Printing Office, Washington D.C. 2001. 66(191):50148-50159. October 2, 2001. Government Printing Office, Washington D.C. Francisco Pearce, A. 2001. Contrasting population structure of the loggerhead turtle iCaretta caretta ) using mitochondrial and nuclear DNA markers. M.S. thesis, Univ. Florida, Gainesville, FL, 71 p. Francisco, A. M., K. A. Bjorndal, A. B. Bolten, R. Reardon, M. Lamont, Y. Anderson, J. Foote, and B. W. Bowen. In press. Stock structure and nesting site fidelity in Florida loggerhead turtles f Caretta caretta ) resolved with mtDNA sequences. Mar. Biol. Henwood, T, W. E. Stuntz, and N. Thompson. 1992. Evaluations of U.S. turtle protective measures under existing TED regulations, including estimates of shrimp trawler related turtle mortality in the wider Caribbean. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC- 303:1-14. Heppell. S. S. 1998a. Application of life-history theory and population model analysis to turtle conservation. Copeia 1998:367- 375. 1998b. Eigenvalue elasticity analysis of species life histories for conservation and management: methods and applica- tions. Ph.D. diss., 162 p. Duke Univ, Durham, NC. Heppell, S. S., L. B. Crowder, D. T. Crouse, S. P Epperly, and N. B. Frazen In press. Population models for Atlantic loggerheads: past, present and future. In Biology and conservation of logger- head sea turtles (A. B. Bolten and B. Witherington, eds.). Smithsonian Institution Press. Lahanas, P N., K. A. Bjorndal, A. B. Bolten, S. E. Encalada, M. M. Miyamoto, R. A. Valverde, and B. W. Bowen. 1998. Genetic composition of a green turtle t Chelonia mydas) feeding ground population: evidence for multiple origins. Mar Biol. 1.30:345-352. Magnuson, J. J., K. A. Bjorndal, W. D. Dupaul, G .L. Graham, F. W. Owens, C .H. Peterson, P C. H. Pritchard, J. I. Richardson, G. E. Saul, and C. W. West. 1990. Decline of the sea turtles: causes and prevention, 259 p. National Academy Press, Washington, D.C. Meylan, A., B. Schroeder, and A. Mosier 1995. Sea turtle nesting activity in the state of Florida. Fla. Mar Res. Publ. 52:1-51. Norrgard, J. 1995. Determination of stock composition and natal origin of a juvenile loggerhead turtle population {Caretta caretta) in Chesapeake Bay using mitochondrial DNA analysis. M.A. thesis. College of William and Mary, Williamsburg, VA, 47 p. Rankin-Baransky, K., C. J. Williams, A. L. Bass, B. W. Bowen, and J. R. Spotila. 2001. Origin of loggerhead turtles stranded in the North- eastern United States as determined by mitochondrial DNA analysis. J. Herpetol. 35:638-646. Schroeder, B. A. 1989. Marine turtle database management: National Ma- rine Fisheries Service-Miami Laboratory. In Proceedings of the first international symposium on Kemp's ridley sea turtle biology, conservation and management (C. W. Cail- louet and A. M. Landry Jr, eds. ), p.15.3-156. Texas A&M Univ Sea Grant Publ. TAMU-SG-89-105. Sears, C. J. 1994. Preliminary genetic analysis of the population struc- ture of Georgia loggerhead sea turtles. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-351:135-139. Sears, C. J., B. W Bowen, R. W. Chapman, S. B. Galloway, S. R. Hopkins-Murphy, and C. M. Woodley. 1995. Demographic composition of the feeding population of juvenile loggerhead sea turtles [Caretta caretta) off Charleston, South Carolina: evidence from mitochondrial DNA markers. Mar Biol. 123:869-874. Seidel, W. R., and C. McVea Jr 1982. Development of a sea turtle excluder shrimp trawl for the southeast U.S. penaeid shrimp fishery. In Biology and conservation of sea turtles ( K. A. Bjorndal, ed. ), p. 497-502. Smithsonian Institution Press, Washington, D.C. Teas, W. G. 1993. Species composition and size class distribution of marine turtle strandings on the Gulf of Mexico and South- east LInited States coasts, 1985-1991. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-315:l-43. TEWG (Turtle Expert Working Group). 1998. An assessment of the Kemp's ridley (Lepidochelys kempii) and loggerhead [Caretta caretta) sea turtle popula- tions in the Western North Atlantic. U.S Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC-409:l-96, 2000. Assessment update for the Kemp's ridley and logger- head sea turtle populations in the Western North Atlantic. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-SEFSC- 444:1-115. Williams, K. L., and M. G. Frick. 2001. Results from the long-term monitoring of nesting log- gerhead sea turtles (Care/tacare50%) and subsidiary (>20%) substratum type (see descriptions in legend to Table 1), and then all sedentary biological features were recorded (as present or absent). A total of 381 samples was analyzed, most of which were in the northern area (Table 1). All nekton were identi- fied and counted. Counts were subsequently converted to densities (m^) with a knowledge of both the horizontal dis- tance traversed by the Jago (calculated from the dive track with Autocad (Autodesk Inc., 1988]), and the width of the video- frame (calculated by using the distance between the two lasers, approx. 1-1.5 m). Statistical analysis Associations with environmental features The associa- tion between nekton and features of the physical and bio- logical environment was examined by following Felley et al. ( 1989) and Felley and Vecchione (1995). In each sample, the primary substratum was given a score of 0.75, and the subsidiary substratum was allocated a score of 0.25: sed- entary biological features were scored as either 1 (present) or 0 (absent). The mean environment was then calculated for each species, after weighting values of each of the envi- ronmental variables in each sample by density of individu- als in that sample. Samples where no individuals were seen, made no contribution (since they had a weight of 0), whereas samples where the species was abundant made a marked contribution to the species mean. The mean for a variable reflected the relative state of that variable in samples where the species was most likely to be found, and can be thought of as its preference for that variable. A correlation matrix was then generated from all means for the measured environmental variables. Patterns of habitat use by species were reflected in patterns of inter- relations among variables. Principal components analysis (PCA) was then performed on the correlation matrix. PCA resolves patterns of interrelationships among variables into a smaller set of composite variables (PCs) to which 478 Fishery Bulletin 100(3) observed variables (species abundances) were correlated. Each PC represents a particular trend in habitat use, and the axis differentiates among sets of species that are likely to be found in contrasting conditions for variables that define the PC. Because the PCs define axes, it is possible to plot species on these axes (using the PC scores) and thus identify those with contrasting or similar patterns of habi- tat use. Species with intermediate scores might be char- acteristic of intermediate environments, or they might be found over the entire range of environments reflected by a PC (because a species' score is the weighted mean of scores of samples where it was found). Habitat associations by demersal nekton were investigated separately from sites in the north and south because previous observations have suggested that areas to the north and south of the Orange River are distinct (Field et al., 1996). Habitat selection by individual species can be deter- mined by comparing species' variances on each PC with variance of the environment. Environmental variance on a PC was determined by calculating PC scores for each sample, following standardization with the appropriate mean of means, and standard deviation of means (see Fel- ley and Vecchione, 1995). Then the scoring function was applied to each sample and the variance determined as the variance of sample scores. The score of each sample was assigned to each individual of all species seen in that sample. A species' variance was then calculated for each species as the variance of these scores. Levene's test was used to compare the species' variance with environmental variance. The null hypothesis invoked for these tests was "species" variance is not significantly different from en- vironmental variance with respect to each factor." Active habitat selection by a species was inferred when a species variance was significantly smaller than the observed en- vironmental variance ( 1-tailed test). This implies that the species was actively selecting a subset of the available en- vironment with respect to that PC, i.e. species distribution in the study area was not random (see Felley et al., 1989). Comparisons of species variances with sample variances on each axis were performed only on the most common species, in each area because the sample sizes for most species were too small to allow statistically meaningful tests. Assemblages Principal component analysis of the data revealed how species were distributed along composite environmental axes. Although species that are distributed in a similar way on a PC axis can be interpreted to share common responses to that axis, they do not necessarily occur together in the same assemblage. Alternative meth- ods of analysis for examining species associations directly can be used to determine assemblage composition; as an alternative method, the software package PRIMER (Clarke and Warwick, 1994) was used in our study. With this software, descriptive, multivariate statistics allowed an examination of the relationships among the samples (based on similarities in specific composition of nekton) to determine how they were distributed with respect to the physical environment. Abundance of only those species that occurred in greater than 5% of the samples were root-root transformed, and a similarity matrix was constructed between the samples containing two or more species by using the Bray-Curtis index (Field et al., 1982). These matrices were used to plot classification diagrams of percentage similarity between samples by means of group-average sorting (fuller details of this method are provided by Field et al., 1996). Owing to the small area filmed in individual videotape samples (mean 26.20 m^, SD 11.24 m-), and the generally low density of most fish species, the samples were pooled (as in Dennis and Bright, 1988) by the nature of the substratum within a dive (hard, soft, mixed), so that replicates were of dives and not of individual videotape samples. Activities In an effort to understand the behavior of demersal fishes in their natural environment, the activity of each demersal fish (on first sighting) was assigned to one of the following behavior patterns: 1 hovering off the substratum; 2 positioned on the substratum; 3 swimming in the water column; 4 positioned in a crevice or under an overhang; 5 occupying a shelter hole; 6 buried (fully or partially) in the substratum. Observations of the substratum over which the fish was seen were also recorded, in the hope that it would be possible to correlate behaviors with the environment. The activity of each of the dominant species of demersal fishes was analyzed by using percent occurrence. This value was calculated by dividing the sum of all individuals observed in each activity (per major substratum type) by the total number of individuals of the species for which activities were recorded. Results Associations with environmental features A total of 22 different taxa of demersal nekton were identi- fied from the samples (Table 2). Eight of these were seen only once or twice, and only hakes occurred in more than 50'S of the samples. The hakes were assumed to be Mer- luccius capensis on the basis of their inshore distribution (Roel, 1987). It was not always possible to clearly separate gobies, Sufflogobius bibarbatus, from dragonets, Paracal- lionymus costatus, and all cases of ambiguity were elimi- nated (i.e. only those fish that could be identified to species were included in the analyses). Each species tended to be associated with slightly differ- ent features of the physical or biotic environment (or both). North Eleven species were seen among these samples, and their mean environments are given in Table 3. PCA of the correlation matrix, generated from the mean envi- ronments of the six dominant species of nekton, produced three PCs with eigenvalues greater than one (Table 4). PCI alone explained more than 60% of the variance in the Gibbons et al : Habitat use by demersal nekton on the continental shelf in the Benguela ecosystem 479 Table 2 Full list of all demersal nekton observed on the vidoetapes; * = observed once only. Common name Scientific name Common name Scientific name Octopus Octopus vulgaris Kingklip Genypterus capensis Cuttlefish Sepia spp. Pelagic goby Sufflogobius bibarhatus Cliokka squid Loligo vulgaris reynaudii Cape gurnard Chelidonichthys capensis Sixgill shark Hexanchus griseus* West coast sole Austroglossus microlepis Smooth hound Mustelus mustelus* Rough scale grenadier Caelorinchus simorhynchus Izak catshark Holohalaelurus regani Ladder dragonct Paracallionymus costatus St Joseph's shark Callorhinchus capensis* Cape dory Zeus capensis* Spearnose skate Raja alba* Monk Lophius vomerinus* Shallow-water Cape hake Merluccius capensis Cape conger eel Conger wilsoni* False jacopever Sebastes capensis Snoek Thyrsites atun* Jacopever Helicolenus dactylopterus Maasbanker Trachurus trachurus capensis data and contrasted species found over soft bio- active substrata (sole, Austroglossus microlepis, and to a lesser extent hakes) with those over more mixed and hard substrata (false jacopever, Sebastes capensis). The other species were located midway along this axis (Fig. 2). PC2 accounted for almost 25% of the variance in the data set (Table 4). The significant load- ings on the environmental variables essentially contrasted broken expanses of hard substrata (cobbles, and the biota associated with more open areas) with unbroken expanses of "lifeless" bedrock (Table 4). Those species that were previ- ously contrasted along the PCI axis were grouped together (false jacopever and sole) and set against those that were positioned midway between them (kingklip, Genypterus capensis, and goby) (Fig. 2). PCS was responsible for almost 10% of the variance in the data set (Table 4). It can be interpreted in terms of biological richness and contrasted gobies and soles with kingklip and hake (Table 4). South A total of 10 species were seen in these samples, and their mean environments are given in Table 5. PCA of the correlation matrix gener- ated from the dominant species' means produced four PCs with eigenvalues greater than one (Table 6). The first two PCs explained almost 75% of the variance in the data, although PCI extracted only -60%. A clear trend that differen- tiated the species was determined by PCI, which contrasted dragonets and soles with kingklip and false jacopever (Fig. 2). The loadings on the variables of PCI (Table 6) opposed hard substrata and their associated epifauna at one extreme with sand and its associated organisms and structures at the other. PC2 contrasted gobies, kingklip, and hake with false jacopever and, to a lesser extent, jacopever, Helicolenus dactylopterus (Fig. 2). In- terpretation of this PC is difficult because few 1 0.5 ^A 0-- § -0.5 + CO LL -1 -1.5 -I- -2 Sole False Jacopever Cuttlefish Hake Kingklip Goby -" 1 1 1 1 1 1 (— •1.5 -1 -0.5 0 0.5 Sand Factor 1 B H 1 1 1 1 1 1.5 2 Rocks 1.5 -r 1 - - False Jacopever 0.5 -- 0 -- -0.5 -- -1 -- -1.5 Jacopever Grenadier Dragonet Sole Kingklip Goby Cuttlefish Hake -+- ■+- ■+- ■+- -t -2 -1.5 Rocks -1 -0.5 0 Factor 1 0.5 1 1.5 Sand Figure 2 Distribution of species scores along the first and second axes repre- senting habitat use by nekton species in the sites to the north (A) and south (Bl of the Orange River mouth. PCI (north and south) repre- sented habitat use according to hard and sofl substrata. PC2 (north) corresponded to the broken or continuous nature of the substratum, and PC2 (south) represented biological richness (see text). 480 Fishery Bulletin 100(3) Table 3 Species means for environmental variables in the undisturbed sites in the northern sampling area. Also given are the sample means total number of individuals of each species seen in samples (n). Data in bold typeface indicate those species used in the principal components Species Samples Area n Sand Cobbles Debris Boulders Bedrock Mounds Hake 177 5569.47 561 0.8266 0.0602 0.0971 0.0058 0.0102 0.9430 False jacopever 34 931.3 40 0.3750 0.1188 0.3750 0.0750 0.0563 0.7000 Jacopever 1 18.93 1 0.2500 0 0 0 0.7500 1 Kingklip 21 601.555 23 0.6087 0.0326 0.3261 0 0.0326 0.9565 Goby 21 604.98 25 0.5000 0.0300 0.3500 0 0.1200 0.8000 Gurnard 4 88.2 4 0.8125 0 0.1875 0 0 1 Sole 11 484.99 11 0.7727 0.1591 0.0682 0 0 1 Grenadier 1 18.41 1 0 1 0 0 0 0 Dory 1 20.04 1 0.7500 0 0.2500 0 0 1 Octopus 4 94.86 4 0.6250 0.0625 0.3125 0 0 0.7500 Cuttlefish 114 3721.355 183 0.7281 0.1011 0.1380 0.0041 0.0287 0.8852 Environment 251 7291.03 251 0.6434 0.1116 0.1922 0.0149 0.0378 0.8207 Table 4 Eigenvalues generated from the principal components analysis (PCA) of the species means in the northern study site, using the dominant species only (Table 3). PC load- ings of environmental means and PC scores of the domi- nant species are also s lown The representations below | are of the unrotated princip e components. PC loadings greater than 0.50 are in bold typeface FI, F2, F3 refer to | factors 1-3, respectively, generated by the PCA. Fl F2 F3 Eigenvalue 9.14 3.481 1.40 % Total variance 60.95 23.21 9.33 Cumulative eigenvalue 9.14 12.62 14.02 Cumulative % variance 60.95 84.15 93.48 Sand -0.98 0.18 0.11 Cobbles 0.34 0.89 -0.18 Rock Debris 0.92 -0.37 0.02 Boulders 0.83 0.49 -0.05 Bedrock 0.53 -0.77 -0.32 Mounds -0.92 -0.10 0.21 Holes -0.83 -0.47 0.07 Tracks -0.55 -0.33 -0.71 Bryozoans 0.93 -0.25 0.20 Ascidians 0.81 -0.21 -0.51 Sponges 0.94 -0.10 -0.14 Cerianthids 0.54 0.62 0.15 Corals 0.82 0.55 -0.03 Asteroids -0.80 0.54 -0.25 Hake -0.79 0.26 0.92 False Jacopever 1.73 0.94 -0.20 Kingklip 0.07 -0.78 1.26 Goby 0.28 -1.63 -1.04 Sole -1.12 0.74 -1.18 Cuttlefish -0.18 0.47 0.23 of the environmental variables had significant loadings (Table 6); however, it would appear to reflect some level of biological richness. PCS accounted for 13% of the variance in the data set (Table 6), whereas PC4 was responsible for only 7%. The significant loadings on the environmental variables of the former contrasted the biota on hard substrata of high relief (boulder and reefs) with the biota of low relief areas (Table 6). PC4 contrasted (much of) the biota with the sub- strata, although few of the environmental variables had significant loadings (Table 6). Selection of environmental features The observed associations between nekton and the bio- physical environment were often due to active selection. In the north only hake selected subsets of the available environment based on major substratum type (and its associated biota) (Table 7). This species avoided the rocky extremes, but otherwise showed a pattern of distribution that resembled that of the samples (Fig. 3A). Although false jacopever and hake selected opposite environments based on the broken or unbroken nature of the rock debris (PC2), it would appear that hakes again tended to avoid the extremes (Fig. 3, B and C). The other environmental PCs were not significantly selected for by any species. Although interpretation of the results generated from the northern data set is confounded by the (mixed) nature of the samples (see Table 1), the same problems did not hamper the data from the south. In the south, six species of demersal nekton selected subsets of the available envi- ronment based on the identified axes (Table 7). Hake, goby, sole, and dragonet all appeared to select their environments on the basis of general physical features (PCI). As in the northern area, hake did not show a strong selection for axis extremes (e.g. Fig. 4A), which in itself could mean that hake did not favor very rocky or sandy areas. Dragonets and sole Gibbons et al : Habitat use by demersal nel^ 30 -- .-4 QJ . . o CD -7-5-3-11357911 13 15 17 1 Sand Factor 3 Rocks Figure 3 Distribution of hake (stippled), and samples ( solid i, on the principal components 1 (PCI) axis which rep- resented hard and soft substrata, in the sites to the north of the Orange River mouth (A). Distribution of hake (B), and false jacopever (C) (both stippled) and samples (solid) on the PC2 axis which represented biotic richness, in the sites to the north of the Orange River mouth. The variance of all species scores was sig- nificantly less than that of sample scores. a rich infauna (Table 4), which probably reflects the strictly benthic nature of their diets (Macpherson and Roel, 1987). An apparent lack of strong habitat preference has been observed by longfin hakes (Urophycis chesteri) off North Carolina (Felley and Vecchione, 1995). Although similar re- sults were noted here for Merluccius capensis (e.g. Tables 4 and 6), this species tended to avoid the rocky extremes (as in Gibbons et al., 2000). Interestingly, the species was not associated with substrata that were conspicuously rich in either epifauna or infauna. However, because small hake Gibbons et al : Habitat use by demersal nel- CJ § 40 - cr Q> LlI 20 - ■ H -hake o o o o CN ■>- 1 1 0 -1 -rlW J^ -0 -4-2 0 2 4 6 8 10 12 14 16 18 2C High rel>ef Factor 3 lqw relief nn " « 80 - Q. 1 60 - § 40 - CD ^ 20- E - Frequency-false jacopeve 3 CO •<»■ CM I.I.I 0 - ^1 ■ -... _ _ fl 4-2 0 2 4 6 8 10 12 14 16 18 20 High relief Factor 3 Low relief Figure 4 (continued) the white lights of the Jago, these juvenile hake were pink and blotchy in color and appearance and were not strongly different from kingklip, and it is possible that such an ap- pearance may serve to camouflage individuals. Cuttlefish are well known to use camouflage, and this may account for the fact that they were not strongly as- sociated with soft or hard substrata (Tables 4 and 6). Cuttlefish are the commonest cephalopod caught in de- mersal trawls (Japp, 1997), implying that they are known to occur over sandy and mixed substrata. Cuttlefish feed predominantly on fishes and crustaceans (Lipinski, 1992) and therefore the lack of any strong association with ophiuroids is to be expected. Grenadiers, on the other hand, tended to be associated with brittlestars (Table 6). Unlike cuttlefish, however, grenadiers are known to feed quite extensively on ophiu- roids ^ and small epifauna (Meyer and Smale, 1991) and did not seem to show any strong association with extremes of substrata and were observed in both soft and hard habi- tats (Table 5). The smaller species of macruorids (such as C. simorhynchus) form a small but conspicuous part of the Adriaans,W. 2000. Unpubl.data. Zoology Department, Uni- versity of the Western Cape, Private Bag X17, Bellville 7535, South Africa. 486 Fishery Bulletin 100(3) North -^ -e HARD HARD HARD HARD MIXED HARD SOFT SOFT MIXED SOFT MIXED MIXED MIXED HARD MIXED HARD SOFT HARD HARD HARD HARD — SOFT — SOFT — HARD — MIXED — MIXED -HARD — MIXED — MIXED — MIXED — SOFT — MIXED — SOFT — SOFT — SOFT — SOFT — MIXED — SOFT — SOFT J- SOFT P-SOFT H r MIXED LrSOFT '-SOFT rSOFT '-SOFT SOFT MIXED MIXED SOFT MIXED 30. 40. 50. 60. 70. 80. 90. 100. 3rav-Curti^ siniilantv South SOFT 40. 50. 60. 70. 80. Bray-Curtis similarily 90. 100. Figure 5 Cluster analysis of the pooled (by major substratum per dive I samples of nekton collected over the areas to the north and south of the Orange River mouth. bycatch from the hake-directed trawl fishery and are thus assumed to occupy a wide range of habitats. The gobies seen in our study tended to show neither a strong association with any substratum type (and their associated biota ), nor were they linked to areas with a rich ophiuroid fauna (Tables 4 and 6). Sufflogobius bibarbatus is also known as the pelagic goby because it spends much of its time at night in the water column,- where it forms a conspicuous part of the fish community'^ and where it feeds largely on zooplankton.- If the gobies obsei-ved in our study were S. bibarbatus (as postulated), then the data suggest that their use of the demersal habitat by day was primarily linked to shelter. Othei-wide, it might be expected that they would be associated with areas rich in - Pillar, S. C. 2000. Personal commun. Marine and Coastal Management, Cape Town, South Africa. •^ OToole, M. 2000. Personal commun. Ministry of Fisheries and Marine Resources, Windhoek, Namibia. benthic food, such as ophiuroids, which are known to form part of the diet of other gobies (Gibson, 1982). Communities The overall diversity of the fish community obsei-ved by the Jago is less than that determined from the same region with trawl nets (Roel, 1987), principally because the area sampled was significantly smaller. The diversity of the fish and cephalopod fauna may also seem low by comparison with submersible surveys elsewhere (e.g, Felley et al., 1989). But, this region is not known to sup- port a high diversity of demersal species (Mas-Riera et al., 1990) because of the perceived harshness of the prevailing environment (such as frequent intrusions of low-oxygen bottom water (Bailey and Rogers, 1997)). The demersal nekton fauna, both as individual species and as a whole, was strongly influenced by substratum type (as in Parker et al., 1994). Despite the large research effort Gibbons et a\ Habitat use by demersal nekton on the continental shelf in the Benguela ecosystem 487 North Sole Kingklip Goby False Jacopever Hake South -H 1 1 10. 20. 30. 40. 50 60 70. 80. 90. 100. Bray-Curlis similarity Cuttlefish Goby Grenadier Jacopever Sole Dragonet False Jacopever Hake 20 30 40. 50. 60 70. 80. 90. 100. Bray-Curtis similarity -H Cuttlefish Figure 6 Inverse classification diagram showing species' associations among all the samples (pooled by major substratum and dive) collected in the sites to the north and south of the Orange River mouth. that has focusseci on the commercially important demersal fisheries along the west coast of southern Africa, there is a dearth of published material that can be usefully compared to the current observations. Roel ( 1987 ) recognized that the species composition of the demersal fauna in the southern Benguela ecosystem was mainly influenced by depth, and she identified two distinct communities of demersal nekton. One of these was dominated by Merluccius paradoxus and occuired at depths greater than 380 m, and the other was dominated by M. capensis and was confined to the continen- tal shelf In the northern Benguela ecosystem, Macpherson and Roel (1987) recognized five distinct demersal fish communities. Two of these were distributed over the shelf (and were subdivided by latitude), and the balance was es- sentially depth related and ran parallel to the coastline, offshore. Similar observations on the structure of the demersal fish assemblages were made by Macpherson and Gordoa ( 1992) and Mas-Riera et al. ( 1990), who noted that latitudinal variations in community composition were generally associated with the state of upwelling. The latter authors postulated that the especially low level of diversity of the demersal fish community in the northern Beguela ecosystem (the region of our study) was due to the presence of low-oxygen bottom water The "indicator" species for the assemblage in their analysis included Merlucxius capensis and Sufflogobius biharbatus, which were also among the common species observed in our study. Smale et al. (1993) observed three demersal communi- ties over the Agulhas Bank (south coast of South Africa) which were related to depth; inshore (<100 m), shelf (90- 190 m), and slope (>200 m). Although the species observed in the our surveys were all conspicuous components of the shelf communities identified by the these authors, they were relatively rare (or absent) from the samples collected by Smale et al. (1993). Unfortunately, the nature of these authors' data (30-60 minute trawls) is such that they do not provide any fine-scale information on species associa- tions, and this effectively limits further comparison. There appears to be a difference in the community of fishes from the northern and southern study sites (e.g. the presence of dragonets in the south but not in the north). Differences in the infaunal communities from undis- turbed sites to the north and south of the Orange River were noted by Field et al. ( 1996). Gibbons and Sulaiman'' Gibbons, M. J., and A. Sulaiman. 1998. A video-description of the mid-shelf benthic environment off the west coast of south- em Africa, with a comment on the habitat association of demer- sal nekton. A report to De Beers Marine (Pty) Ltd., Cape Town 8000, Republic of South Africa, 62 p. 488 Fishery Bulletin 100(3) 100 80 60 40 20 0 70 60 t 50 ^ 40 OJ = 30 0) iT 20 10 0 Hake >. 60 •■ 20 0 Grenadier 1 1 False Jacopever 100 80 c 20 0 Dragonet 3 4 5 Behavior Kingklip J lU L 12 3 4 5 6 Behavior Figure 7 Histograms showing the frequency (as percent) at which behavior patterns 1-6 were observed (x-axis) in species videotaped during survey dives between 100-140 m in the Orange River delta area, southern Africa. All data were pooled. Behavior patterns: 1 = hovering off the substratum; 2 = positioned on the substratum; 3 = swimming in the water column; 4 = positioned in a crevice or under an overhang; 5 = occupying a shelter hole; 6 = buried (fully or partiallyt in the substratum. Solid bars indicate behaviors over soft substrata, stippled bars indicate behaviors over hard substrata. also commented on some discrepancies in the structure of the physical and biological environment in regions to the north and south of the Orange River, which can be partly explained by the latitudinal influence of the River and by biogeography (Emanuel et al., 1992; Gibbons and Hutch- ings, 1996). Acknowledgments We are very grateful to James Felley (Office of Informa- tion Technology, Smithsonian Institution, Washington DC) for his assistance with some of the statistical analyses of the data, and for his comments on an earlier draft of the manuscript. We would like to thank the three anonymous referees for their valuable comments, which served to focus the text. The staff at the drawing office of De Beers Marine (Pty.) Ltd. are thanked for their patient delivery of maps, images, and yet more maps, and for putting up with the persistent glare of the television screen. We would like to thank Hans Fricke, Jurgen Schauer, and Karen Hiss- man for their safe delivery of personnel (and videotapes) from the bottom of the sea: rarely have we felt in such capable hands. We are also grateful to the personnel of the MV Zealous for their hospitality at sea, and to the helicop- ter pilots for ensuring speedy transport from ship to shore. De Beers Marine (Pty.) Ltd. are gratefully acknowledged for their financial support of the project and for being prepared to allow us to publish the data contained in our article. Gibbons et al : Habitat use by demersal nekton on the continental shelf in the Benguela ecosystem 489 Literature cited Autodesk Inc. 1988. Auto CAD LT 98. Autodesk Inc., Cupertino, CA. Badenhorst, A. 1988. Aspects of the South African longline fishery for king- kHp Genypterus capensis and the cape hakes Mercluvcius capensis and M. paradoxus. S. Afr J. Mar. Sci. 6:33-42 Bailey, G. C, and J. Rogers. 1997. Chemical oceanography and marine geoscience off southern Africa: past discoveries in the post-Gilchrist era, and future prospects. Trans. Roy. Soc, S. Afr. 52:51-79. Brown P. C, K. L. Cochrane, and S. J. Paiting. 1991. Estimates of phytoplankton and bacterial biomass and production in the northern and southern Benguela ecosystems. S. Afr. J. Mar. Sci., 11:537-564. Clarke, K. R., and R. M. Warwick. 1994. Change in marine communities: an approach to statistical analysis and interpretation, 144 p. Plymouth Marine Laboratory, UK. Gushing, D. H. 1971. Upwelling and the production of fish. Adv. Mar. Biol. 9:255-334. Dennis, G. D., and T. J. Bright. 1988. Reef fish assemblages on hard banks in the North- eastern Gulf of Mexico. Bull. Mar. Sci. 43: 280-307. Emanuel, B. P., R. H. Bustamente, G. M. Branch, S. Eekhout, and F. J. Odendaal. 1992. A zoogeographic and functional approach to the selection of marine reserves on the west coast of South Africa. S. Afr. J. Mar Sci. 12:341-354. Felley, J. D., and M. Vecchione. 1995. Assessing habitat use by nekton on the continental slope using archived videotapes from submersibles. Fish. Bull. 93:262-273. Felley, J. D., M. Vecchione, G. R. Gaston, and S. M. Felley. 1989. Habitat selection by demersal nekton: analysis of videotape data. NE Gulf Sci. 10:69-84 Field, J. G., K. R. Clarke, and R. M. Warwick. 1982. A practical strategy for analysing multispecies distri- bution patterns. Mar. Ecol. Prog. Ser. 8: 37-52. Field, J. G., C. A. Parkins, H. Winckler, C. Savage, and K. van der Merwe. 1996. Impact on benthic communities. In Impacts of deep sea diamond mining, in the Atlantic 1 mining licence area in Namibia, on the natural systems of the marine environment. EEU (Environmental Evaluation Unit) Report 11/96/158, 370 p. University of Cape Town, Cape Town, South Africa. Gibbons, M. J., and L. Hutchings. 1996. Zooplankton diversity and community structure around .southern Africa, with special attention to the Ben- guela upwelling system. S. Afr. J. Sci., 92:63-76. Gibbons, M. J., A. Sulaiman, K. Hissmann, J. Schauer, and P. A. Wickens. 2000. Video observations on the habitat association of demersal nekton in the mid-shelf benthic environment off the Orange River mouth. S. Afr. J. Mar. Sci., 22:1-7. Gibson, R. N. 1982. Recent studies on the biology of intertidal fishes. Oceanogr. Mar Biol. Ann. Rev. 20:36.3^14. Giguere, M., and S. Brulotte. 1994. Comparison of sampling techniques, video and dredge, in estimating sea scallop (Placopecten magellani- cus, Gmelin) populations. J. Shellfish Res. 13:25-30. Hutchings, L. 1992. Fish harvesting in a variable productive environ- ment— searching for rules or searching for exceptions? S. Afr. J. Mar Sci. 12: 297-318. Japp, D. W. 1990. A new study on age and growth of kingklip Genypterus capensis off the South and West coasts of South Africa, with comments on its use for stock identification. S. Afr J. Mar Sci. 9:223-237. 1997. Discarding practices and bycatches for fisheries in the Southeast Atlantic region (area 47). FAO Fish. Rep. 547 (suppl.):235-256. Lipinski, M. R. 1992. Cephalopods and the Benguela ecosystem: trophic relationships and impact. S. Afr. J. Mar Sci., 12:791-802. Macpherson, E. 1983. Feeding of the kingklip (Genypterus capensis) and its effect on the hake {Merluccius capensis) resource off the coast of Namibia. Mar Biol., 78:105-112 Macpherson, E., and A. Gordoa. 1992. Trends in the demersal fish community off Namibia from 1983-1990. S. Afr. J. Mar Sci., 12: 635-649. Macpherson, E., and B. A. Roel. 1987. Trophic relationships in the demersal fish community off Namibia. S. Afr. J. Mar Sci. 5: 585-596. Mas-Riera, J., A. Lombarte, A. Gordoa, and E. Macpherson. 1990. Influence of Benguela upwelling on the structure of demersal fish populations off Namibia. Mar Biol. 104: 175-182. Meyer, M., and M. J. Smale. 1991. Predation patterns of demersal teleosts from the Cape South and West coasts of South Africa. 2. Benthic and epibenthic predators. S. Afr. J. Mar Sci., 11:409-442. Murie, D. J., D. C. Parkyn, B. G. Clapp, and G. C. Krause. 1994. Observations on the distribution and activities of rockfish, Sebastes spp., in Saanich Inlet, British Columbia, from the Pisces IV submersible. Fish. Bull. 92:313-323. Parker, R. O., and S. W. Ross. 1986. Observing reef fishes from submersibles off North Carohna. NE Gulf Sci. 8:31-49. Parker, R. O., A. J. Chester, and R. S. Nelson. 1994. A video transect method for estimating reef fish abundance, composition, and habitat utilization at Gray's Reef National Marine Sanctuary, Georgia. Fish. Bull. 92: 787-799. Payne, A. I. L. 1985. The sole fishery off the Orange River, southern Africa. In International symposium on the most important upwell- ing areas off western Africa (Cape Blanco and Benguela) (C. Bas, R. Margaleff, and P Rubies, eds.), p. 1063-1079. Institute de Investigaciones Pesqueras, Barcelona. Payne, A. I. L., K. H. Brink, K. H. Mann, and R. Hilbom. 1992. Benguela trophic functioning. S. Afr. J. Mar Sci., 12: 1-1108. Payne, A. I. L., J. A. Gulland, and K. H. Brink. 1987. The Benguela and comparable ecosystems. S. Afr. J. Mar Sci. 5:1-957. Pillar, S. C, and M. Barange 1995. Diel feeding periodicity, daily ration and vertical migration of juvenile Cape hake off the west coast of South Africa. J. Fish Biol. 47:753-768. Richards, L. J. 1986. Depth and habitat distribution of three species of rockfish (Set>astes) in British Columbia: observations from the submersible PISCES IV. Env. Biol. Fish. 17:13-21. Roel, B. A. 1987. Demersal communities off the west coast of South Africa. S. Afr. J. Mar. Sci. 5:575-584. 490 Fishery Bulletin 100(3) Shannon, L. V. 1985. The Benguela ecosystem. I. Evolution of the Bengu- ela, physical features and processes. Oceanogr. Mar. Biol. Ann. Rev. 23:105-182. Sibuet, M.. S. K. Juniper, and G. Pautot. 1988. Cold-seep benthic communities in the Japan subduc- tion zones: geological control of community development. J. Mar Res. 46:333-348. Smale, M. J., B. A. Roel, A. Badenhorst, and J. G. Field. 1993. Analysis of the demersal community of fish and cephalopods on the Agulhas Bank, South Africa. J. Fish Biol. 43(suppl.A):169-191. 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. 491 Abstract-A total of IOO6 king mack- erel iScarnheromorus cavalla) repre- senting 20 discrete samples collected be- tween 1996 and 1998 along the east (Atlantic) and west (Gulf) coasts of Flor- ida and the Florida Keys were assayed for allelic variation at seven nuclear- encoded microsatcllites. No significant deviations from Hardy-Weinberg equi- librium expectations were found for six of the microsatellites, and genotypes at all microsatellites were independent. Allele distributions at each microsatel- lite were independent of sex and age of individuals. Homogeneity te.sts of spa- tial distributions of alleles at the micro- satellites revealed two weakly divergent "genetic" subpopulations or stocks of king mackerel in Florida waters — one along the Atlantic coast and one along the Gulf coast. Homogeneity tests of allele distributions when samples were pooled along seasonal (temporal) boundaries, consistent with the tem- poral boundaries used currently for stock assessment and allocation of the king mackerel resource, were nonsig- nificant. The degree of genetic diver- gence between the two "genetic" stocks was small: on average, only 0.19% of the total genetic variance across all samples assayed occurred between the two regions. Cluster analysis, assign- ment tests, and spatial autocorrelation analysis did not generate patterns that were consistent with either geographic or spatial-temporal boundaries. King mackerel sampled from the Florida Keys could not be assigned unequivo- cally to either "genetic" stock. The gen- etic data were not consistent with cur- rent spatial-temporal boundaries em- ployed in stock assessment and allo- cation of the king mackerel resource. The genetic differences between king mackerel in the Atlantic versus those in the Gulf most likely stem from reduced gene flow (migration) between the Atlantic and Gulf in relation to gene flow (migration) along the Atlantic and Gulf coasts of peninsular Florida. This difference is consistent with findings for other marine fishes where data indi- cate that the southern Florida penin- sula serves (or has served) as a biogeo- graphic boundary. Population structure of king mackerel (Scomberomorus cavalla) around peninsular Florida^ as revealed by microsatellite DNA* John R. Gold Elena Pak Center for Biosystematics and Biodiversity Department of Wildlife and Fisfneries Sciences Texas A&M University College Station, Texas 77843-2258 E mail address (for J R Gold) goldfishiaitamu edu Doug A. DeVrles National Manne Fisheries Service Southeast Fisheries Science Center 3500 Delwood Beach Road Panama City, Flonda 32408 Manuscript accepted 4 March 2002. Fish. Bull. 100:491-.509 (2002). The king mackerel (Scornberomorufi ca- valla) is a coastal pelagic fish distrib- utee] in the western Atlantic Ocean from Massachusetts (USA) to Rio de Janeiro (Brazil) and in the Gulf of Mexico and Caribbean Sea (Rivas, 1951; Collette and Nauen, 1983). Recreational and commercial catches of king mackerel in U.S. waters are substantial (Manooch, 1979; MSAPi; Legault et al.^) and the species is critical to the southeastern Atlantic coast (hereafter "Atlantic") and northern Gulf of Mexico (hereafter "Gulf) charter-boat industries. Man- agement of the king mackerel resource in U.S. waters is under the jurisdiction of two regional fishery management councils, and current management planning (GMFMC-^) is based on a two migratory unit (stock) hypothesis: one stock (Atlantic migratory unit) occurs in the U.S. southeastern Atlantic, the other (Gulf of Mexico migratory unit) occurs in the Gulf Separation of the two stocks has been based primarily on mark-and-recapture studies carried out prior to 1984 and on differences in growth rate (MSAP'; Williams and God- charles^). Based on the mark-and-recap- ture studies, it was hypothesized that the two stocks mixed extensively during * This is paper 34 in the series "Genetic Studies in Marine Fishes" and is contribu- tion 98 of the Center for Biosystematics and Biodiversity, Texas A&M University, College Station, Texas 7784.3-2258. ' MSAP ( Mackerel Stock Assessment Panel ). 1994. Contribution report of the mackerel stock assessment panel. Contribution re- port MlA-93/94-42, 27 p. Southeast Fish- ery Science Center, National Marine Fish- eries Service, 75 Virginia Beach Dr., Miami, FL 33149. - Legault, C. M., M. Ortiz, G. Scott, N. Cum- mings, and P. Phares. 2000. Stock as- sessment analyses on Gulf of Mexico king mackerel. Contribution report SFD-99/ 00-83, 48 p. Southeast Fishery Science Center, National Marine Fisheries Service, 75 Virginia Beach Dr., Miami, FL 33149. * GMFMC (Gulf of Mexico Fishery Manage- ment Council). 1996. Amendment 8 to the fishery management plan for coastal migratory pelagic resources in the Gulf of Mexico and South Atlantic includes envi- ronmental assessment, regulatory impact review, and initial regulatory flexibility analysis. Gulf of Mexico Fishery Manage- ment Council, 3018 U.S. Hwy 301 North, Suite 1000, Tampa, FL 33619-2266. ^ Williams,R.O.,andM.FGodcharles. 1984. Completion report, king mackerel tagging and stock assessment. Project rep. 2- 341-R, plus figures and tables., 45 p. FL Dept. Nat. Res., Florida Marine Research Institute, 100 8"^ Ave. SE, St. Petersburg, FL 33701. 492 Fishery Bulletin 100(3) the winter months along the southeast coast of Florida. For purposes of stock assessment and resource allocation (Sutter et al, 1991; GMFMC^), the boundary between the two stocks was specified as the Volusia/Flagler county line (east coast of Florida) in winter (1 November-31 March) and the Monroe/Collier county line (west coast of Florida) in summer (1 April-31 October). Pragmatically, this means that king mackerel caught south of the Volusia/ Flagler county line (including the Florida Keys) between 1 November and 31 March belong to Gulf stock, whereas fish caught south of the Monroe/Collier county line between 1 April and 31 October belong to Atlantic stock. Data from additional mark-capture (Fable et al., 1987; Sutter et al., 1991; Schaefer and Fable, 1994; Fable«), growth rate (DeVries and Grimes, 1997), otolith shape (DeVries et al., 2002), and temporal-geographic sampling studies (Collins and Stender, 1987; Trent et al., 1983; Grimes et al., 1990) are consistent with the hypothesis that king mackerel in the Atlantic differ from those in the Gulf In addition, on the basis of allozyme evidence (Johnson et al., 1993) and studies of early life history (Grimes et al., 1990; Grimes et al."), DeVries and Grimes ( 1997) suggested that there might be two distinct stocks of king mackerel in the northern Gulf Johnson et al. (1993) found a high frequency of the PEPA-2a allele of the nuclear-encoded dipeptidase (PEPA-2) locus among king mackerel from the western and northwestern Gulf whereas a high frequency of the PEPA-2b allele occurred among king mackerel from the Atlantic and northeastern Gulf Johnson et al. (1993) hypothesized that the two (putative) Gulf stocks mixed to varying degrees in the northern Gulf Considering all the data acquired to date, DeVries and Grimes (1997) sug- gested there may be three stocks of king mackerel in U.S. waters: one in the Atlantic, one in the eastern Gulf and one in the western Gulf The allozyme data of Johnson et al. (1993) did not distin- guish king mackerel in the eastern Gulf from those in the Atlantic, and to that extent, argued against the hypothesis that king mackerel in the Atlantic and Gulf represented two distinct stocks. Gold et al. ( 1997), however, assayed variation in restriction sites of mitochondrial (mt)DNA among king mackerel collected from 13 localities along the U.S. Atlantic ■5 GMFMC (Gulf of Mexico Fishery Management Council). 1984. Final amendment 1, fishery management plan and environ- mental impact statement for coastal migratory pelagic re- sources (mackerels I in the Gulf of Mexico and South Atlantic region. Gulf of Mexico Fishery Management Council, 3018 U.S. Hwy, 301 North, Suite 1000, Tampa, FL 33619-2266. '' Fable, W. A., Jr 1988. Stock identification of king mackerel based on mark-recapture. Unpubl. manuscript from meeting on stock identification of king mackerel in the Gulf of Mexico. Southeast Fish. Sci. Center Contribution Report 2-18-88, 24 p. Southeast Fisheries Science Center, 3500 Delwood Beach Road, Panama City, FL 32408. ' Grimes, C. B, J. H. Finucane, and L. A. Collins. 1988. Distri- bution and occurrence of young king mackerel, Scomberomorus cavalla, in the Gulf of Mexico. Unpubl. manuscript from meet- ing on stock identification of king mackerel in the Gulf of Mexico. Panama City Lab. Contribution Report 2-18-88, 31 p. NMFS SE Fish. Cntr, 3.500 Delwood Beach Road Panama City, FL 32408. coast and northern Gulf and found significant (but weak) heterogeneity only in comparisons of pooled mtDNA hap- lotypes from Atlantic localities with pooled haplotypes from Gulf localities. Thus, the mtDNA data did not support the hypothesis that two genetically identifiable stocks of king mackerel occur in the northern Gulf but rather were consis- tent with the hypothesis that separate stocks of king mack- erel may exist in the Atlantic and in the Gulf Estimates of Fgj,, a measure of population subdivision, between king mackerel in the Atlantic and Gulf were small, indicating that mixing between Atlantic and Gulf king mackerel occurs. Gold et al. (1997) also examined spatial variation in fre- quencies of the two alleles at PEPA-2. Results were essen- tially the same as those reported by Johnson et al. ( 1993): high frequency of the PEPA-2a allele among king mackerel in the western Gulf and high frequency of the PEPA-2b allele in the eastern Gulf and Atlantic. Tests of indepen- dence of PEPA-2 genotypes with age and sex of individual fish, however, revealed significant nonrandom associations among Gulf fish of PEPA-2a homozygous genotypes with males and of PEPA-2h homozygous genotypes with females. Moreover, among fish sampled from the Atlantic, there was a highly significant decrease in the frequency of PEPA-2b alleles with increasing fish age. The same trend was found among fish sampled from the Gulf but to a lesser extent. Tests of independence of sex versus age, and of mtDNA variation versus sex or age, were nonsignificant. These find- ings strongly indicated that the use of PEPA-2 genotypes to distinguish stocks of king mackerel is compromised and that the hypothesis of eastern and western stocks of king mackerel in the Gulf needs to be re-evaluated. Finally, Broughton et al. (2002) surveyed allelic variation at five nuclear-encoded microsatellites among a subset of the samples of king mackerel studied by Gold et al. ( 1997). Tests of homogeneity in allele distribution at the five micro- satellites indicated that samples from Port Aransas, Texas (western Gulf), and Gulfport, Mississippi (central Gulf), dif- fered from each other and from the remaining samples (in- cluding two samples from the Atlantic, one from the Florida Keys, one from the eastern Gulf one from the western Gulf and one from Veracruz, Mexico). No significant differences in allele frequencies at any microsatellite were found be- tween samples representing geographic extremes, and no significant geographic patterns were found when samples were combined into regional groupings reflecting current hypotheses of king mackerel stock structure in U.S. waters. Of concern to management of the king mackerel resource in U.S. waters is the degree of mixing between the pre- sumed stocks in the Atlantic and Gulf Analysis of mark- and-recapture data collected from 1985 to 1993 (MSAPM indicated that roughly 3.0% of fish tagged in the Atlantic were recaptured in the Gulf whereas 6.4% tagged in the Gulf were recovered in the Atlantic. More liberal estimates (SFC**) of recaptures (generated when utilizing summer ^ SFC(SoutheasternFisheriesCenter). 1992. Preliminary anal- ysis of southeastern US. king mackerel mark-recapture data: 1985-1993. Contribution report MlA-93/94-36, 19 p. South- east Fisheries Science Center, National Marine Fisheries Ser- vice, 75 Virginia Beach Dr, Miami, FL 33149. Gold et al.: Genetic studies of Scomberomorus cavalla in Florida 493 and winter seasons in the "mixing" zone) suggested that 2.6-30.9% of recaptured tagged-fish in the Atlantic were returned as Gulf fish and 1.5-13.6% of recaptures tagged in the Gulf were returned as Atlantic fish. These mixing rates, however, were questioned (Jones et al.**) because virtual population analysis (VPA) based estimates of fishing mor- tality for the directed king mackerel fisheries in the Gulf and Atlantic corresponded to annual exploitation rates of 0.30 and 0.11, respectively, whereas exploitation rates cal- culated from the 1985-93 (uncorrected) tag returns ranged from 0.027 to 0.033 (Gulf) and from 0.036 to 0.045 (Atlan- tic). The difference between the two estimates of exploita- tion rates implied either that the true exploitation rate was overestimated by VPA or underestimated by uncorrected tag-return data, leading to the conclusion (Jones et al.^) that little confidence should be placed in reported mixing rates based on mark-and-recapture data. The goals of this project were to use nuclear-encoded microsatellites to define more rigorously the spatial-tem- poral limits of the two stocks (if separate stocks exist) and to estimate the proportions of both stocks in the mixing zone. The issues of spatial-temporal limits and mixing of the two (presumed) stocks are important in relation to assessing and allocating the king mackerel resource, particularly during the winter season. For example, mark-recapture data (MSAPM indicated that -20% offish tagged in the mixing zone in southeastern Florida moved into the Gulf If this means that only -20% of winter catches from the east coast of Florida are Gulf stock, as opposed to 100% under the current management plan, the allowable biological catch (ABC) for the Gulf stock would decrease significantly (MSAPM. Because the Gulf stock of king mackerel currently is considered overfished (Legault et al.^) reductions in ABC of the Gulf stock could have sig- nificant economic impact. The choice to employ microsatellites for the project was straightforward. Briefly, microsatellites are rapidly evolving, short stretches of DNA composed of di-, tri-, and tetranucleotide arrays that are abundant, highly polymor- phic, and inherited in a codominant fashion (Weber, 1990; Wright, 1993; Wright and Bentzen, 1994). Because allele frequencies at microsatellites are generally consistent with equilibrium expectations of diploid, Mendelian loci and because identification of individual microsatellites is by polymerase-chain-reaction (PCR) amplification, both of which remove most of the problems associated with ho- mology of alleles, microsatellites have proven to be useful genetic markers of population structure in numerous taxa, including fishes (Angers and Bernatchez, 1998; Ruzzante et al., 1996; O'Connell et al, 1998; Nielsen et al, 1999). In addition, new alleles at microsatellite loci appear to arise rapidly (Schug et al., 1998), generating high allelic diversity important for statistical power in exact tests and ^ Jones, C. M., M. E. Chittenden, and J. R. Gold. 1994. Report to the mackerel stock identification working group. Unpubl. document of meeting held 8 Sep 94 to 9 Sep 94 at Panama City Lab., SE Fish. Sci. Cent. 7 p. Southeast Fisheries Science Center, National Marine Fisheries Service, 3500 Delwood Beach Road, Panama City, FL 32408. other tests of allclo-distribution homogeneity (Estoup et al., 1998; Ross et al., 1999). Materials and methods A total of 20 samples of king mackerel were procured be- tween 1996 and 1998 from 11 different offshore sites (Table 1, Fig. 1). The sample from Panama City was obtained from charter boat catches, and the samples from Sarasota and Jacksonville were obtained from tour- naments. The remaining samples were obtained from commercial catches. Tissue samples (heart and muscle) were removed from each fish, frozen in liquid nitrogen, transported to College Station, and stored at -80°C. Sex of individuals was recorded for all samples, except for the March 1997 sample from the Florida Keys. Approximate ages of individuals from all samples except for the July 1998 sample from Jacksonville, the March 1997 sample from the Florida Keys, and the April 1997 sample from Sarasota, were determined by otolith-increment analysis by following methods outlined in DeVries and Grimes (1997). Initially, we planned to deploy the five microsatellites developed in a prior study (Broughton et al., 2002). Two of these (Sca-8 and Sca-47), however, had proven difficult to amplify consistently in the prior study and therefore were omitted from our study. A third microsatellite, Sca- 30. developed by Broughton et al. (2002), also was omit- ted because of difficulties with consistent amplification and because allele distributions at Sca-30 were highly leptokurtic (Broughton et al., 2002). A total of five new microsatellites was then developed from the microsatel- lite-enriched genomic library generated by Broughton et al. (2002). Candidate microsatellites were sequenced from either or both ends by using standard M13 sequencing primers and an Applied Biosystems (Perkin Elmer) 377 automated DNA sequencer. The OLIGO software pack- age (National Biosciences, Inc., 1992) was used to identify primers from regions flanking microsatellites. Primers were designed according to preset criteria that included product length, internal stability, proportion GC content, and primer Tm difference. PCR amplifications were per- formed under a variety of experimental conditions to optimize procedures that produced high yields of target sequence and minimized additional fragments ("stutter" bands). Experimental tractability (reproducibility, consis- tency, range in allele size, frequency of "stutter" bands [if present], and microsatellite polymorphism) of PCR-ampli- fied microsatellites were evaluated by screening a panel of king mackerel samples available in the laboratory. PCR primer sequences, the length (in base pairs) of the cloned allele, and the annealing temperature in PCR amplifica- tion for the seven microsatellites used in the project are given in Appendix Table 1. Two of these, Sca-37 and Sca- 44, were developed previously by Broughton et al. (2002). For assay of individual fish, genomic DNA was isolated from frozen tissues as described in Gold and Richardson (1991). Genotypes at the seven microsatellites were deter- mined by PCR amplification and gel electrophoresis. Prior 494 Fishery Bulletin 100(3) Table 1 Localities, acronyms, dates of collection, and number of indivic from the east and west coasts of Florida and the Florida Keys. uals (by sex) of king mackerel iScomberomorus cavalla) sampled Sample locality Acronym Date of capture Number of individuals Migratory group' Female Male Total Atlantic Ocean (east coast) Jacksonville, FL JCKi Jul 1996 48 0 48 Atlantic Jacksonville, FL JCK^ Jul 1998 28 3 31 Atlantic New Smryna Beach, FL NSB Jul 1996 41 9 50 Atlantic Cape Canaveral, FL CCN Dec 1998 24 26 50 Gulf Sebastian, FL SEB' Mar 1997 29 21 50 Gulf Sebastian, FL SEB2 Mar 1998 24 26 50 Gulf Sebastian, FL SEB' Dec 1998 35 15 50 Gulf Ft. Pierce, FL FTP Apr 1996 31 25 56 Atlantic West Palm Beach, FL WPB May 1998 29 25 54 Atlantic Florida Keys Key West, FL KEY' Mar 1996 41 10 51 Gulf Key West, FL KEY2 Mar 1997 — — 29 Gulf Key West, FL KEY3 Jan 1999 29 19 48 Gulf Gulf of Mexico (west coast) Marco Island, FL MCI Apr 1996 29 26 55 Atlantic Boca Grande, FL BCG Apr 1996 31 4 35 Gulf Sarasota, FL SARI Apr 1996 39 5 44 Gulf Sarasota, FL SAR- Nov 1996 60 0 60 Gulf Sarasota, FL SAR' Apr 1997 55 2 57 Gulf Sarasota, FL SAR^ Apr 1998 68 2 70 Gulf Sarasota, FL SAR-' Nov 1998 62 6 68 Gulf Panama City FL PCY Oct 1996 25 25 50 Gulf ' Classification as Atlantic or Gulf migratory group based on the further details. two migratory uni L hypothesis where boundaries change seasonally. See text for to amplification, one of the primers was kinase-labeled with y^-P-ATP by T4 polynucleoti(ie kinase (30 min, 37°C). PCR reactions contained approximately 5 ng of genomic DNA, 0.1 units of Tag DNA polymerase, 0.5 ^M of each primer, 800 ^M dNTPs, 1-2 niM MgCl.^, IX Tag buffer at pH 9.0 (Promega, Corp., Madison, WI), and sterile deion- ized water in a total volume of 10 ^L. Thermal cycling was carried out in 96-well plates as follows: denaturation (4.5 sec, 95°C), annealing (30 sec, temperature as in Appendix Table 1), and polymerization (30 sec, 72°C) for 30 cycles. Aliquots (3 fih) of each PCR reaction were electrophoresed in 6% denaturing polyacrylamide ("sequencing") gels. Gels were dried and exposed to x-ray film. Alleles at individual microsatellites were scored as number of repeats by com- parison to the cloned (and sequenced) allele. Genotypes at each microsatellite for each individual were scored and entered into a database. Initial statistical analysis involved generation of allele frequencies and (direct-count) heterozygosity values, and significance testing of genotypic proportions in relation to those expected under conditions of Hardy-Weinberg equilibrium. Significance testing of Hardy-Weinberg equilibrium proportions involved exact tests performed using Markov-chain randomization (Guo and Thompson, 1992); probability (P) values for tests at each microsatel- lite within each sample were estimated by permutation (bootstrapping) with 1000 resamplings (Manly, 1991). Significance levels for simultaneous tests were adjusted with the sequential Bonferroni approach (Rice, 1989). Tests of genotypic equilibrium at pairs of microsatellites were carried out as a surrogate to assess whether any microsatellites were genetically linked. Probability values for (exact) tests of genotypic equilibrium were generated by 1000 resamplings, and significance levels for simulta- neous tests were adjusted with the sequential Bonferroni approach. Allele frequencies and heterozygosity values were obtained by using biosys-1.7 (Swofford and Selander, 1981), and tests of Hardy-Weinberg and genotypic equilib- rium employed the package genepop (Raymond and Rous- set, 1995). Exact tests also were used to test independence Gold et a\ Genetic studies of Scomberomorus cavalla in Flonda 495 I (W 12-4 -19 W u \m JCK 310406 N c^^ ^^ '''•w- r^ \ ATLANTIC \ \ V«NSB OCEAN GULF OF / % \ • CCN V* SEB \» KIP N A MEXICO BCG •W /• WPB Ma •> / ^^ ^ ••'• KEY Figure 1 Sampling localities for king mackerel examined in the present study. Acronyms for sample localities are defined in Table 1. of the distribution of genotypes at each microsatellite with the sex and age of individuals. Initial tests involved each of the 20 samples separately. We then pooled individuals sampled at Atlantic localities (nine samples), in the Flori- da Keys (three samples), at Gulf localities (eight samples), and over all localities (20 samples) in order to increase cell sizes in individual tests. Probability (P) values for these tests of independence were estimated by permutation (1000 resamplings) and significance levels for simultane- ous tests were adjusted with the sequential Bonferroni approach. Tests of genetic homogeneity among samples included exact tests, as implemented in genepop, the Monte Carlo procedure of Roff and Bentzen (1989), as implemented in the restriction enzyme analysis package of McElroy et al. (1992), and the analysis of molecular variance (amova) of Excoffier et al. (1992). Significance of tests of genetic ho- mogeneity employed permutation with 1000 resamplings per individual comparison, and significance levels for si- multaneous tests were adjusted by using the sequential Bonferroni approach. Tests of genetic homogeneity were carried out separately for each of the seven microsatel- lites. Individual tests were carried out 1) among all 20 samples, 2) among samples (nine) from Atlantic localities, 3) among samples (three) from the Florida Keys, and 4) among samples (eight) from Gulf localities. Analysis of molecular variance (amova) was employed to generate es- timates of (genetic) variance components and statistics for the same comparisons. (t> statistics are a set of hierar- chical F-statistic analogs that consider evolutionary dis- tance among alleles. Significance of

0). Probability values for Sca-14 and Sca-23 were marginal in relation to the (ini- tial) Bonferroni adjusted a of 0.007, whereas the probabil- ity that 0f.j. > zero at Sca-44 was nonsignificant after Bon- ferroni correction (Table 5). For the comparison of Atlantic, Florida Keys, and Gulf samples, significant heterogeneity was found at Sca-14 (all three statistical approaches) and Sca23 (exact test only) before but not after Bonfer- roni correction; heterogeneity at Sca-37 and Sca-44 in the same comparison was significant both before and after Bonferroni correction in at least one of the three statisti- cal approaches (Table 5). Frequency differences at Sca-14, Sca-23. and Sca-44 among the three regional groupings are shown in Table 6 and indicate that small differences in frequency of several alleles at each microsatellite appear to account for observed heterogeneity among the pooled sample comparisons. Gold et al ; Genetic studies of Scomberomorus cavalla in Florida 499 Table 6 Allele frequencies at Sea 14 and Sea 23 for king mackerel (Scombcniniortis carallal from the Atlantic, Florida Keys, and Gulf Allele numbers represent the size in base pairs of the fragment amplified. Microsatellite (allele) Atlantic Florida Keys Gulf Sea- 14 91 0.017 0.004 0.006 93 0.053 0.058 0.034 95 0.670 0,667 0.735 97 0.233 0.240 0.200 99 0.029 0.031 0.027 Sea-23 138 0.063 0.0.54 0.052 140 0.028 0.050 0.028 142 0.221 0.250 0.241 144 0.012 0.008 0.005 146 0.131 0.092 0.121 148 0.026 0.008 0.015 150 0.146 0.108 0.132 152 0.185 0.177 0.149 154 0.008 0.011 0.015 156 0.100 0.111 0.109 158 O.OU 0.011 0.011 160 0.009 0.011 0.012 162 0.001 0.000 0.001 164 0.015 0.027 0.016 166 0.002 0.004 0.009 168 0.002 0.004 0.000 no 0.009 0.019 0.018 172 0.017 0.019 0.039 174 0.009 0.015 0.010 176 0.002 0.008 0.009 178 0.002 0.008 0.001 180 0.001 0.004 0.002 182 0.000 0.000 0.002 184 0.002 0.000 0.000 Sca44 153 0.004 0.015 0.009 157 0.084 0.046 0.090 161 0.030 0.019 0.041 165 0.366 0.308 0.295 169 0..398 0.465 0.438 173 0.104 0.131 0.118 177 0.012 0.011 0.010 181 0.000 0.004 0 000 In general, results of the three approaches to homo- geneity testing were fairly consistent, with one notable exception. At Sca-37. probability values from the exact test and the Roff-Bentzen procedure were 0.009 and 0.001 in the comparison of Atlantic, Florida Keys, and Gulf samples, respectively, whereas the probability that 0[.j. differed from zero was 0.794 (Table 5). We examined this discrepancy further by carrying out "V" tests of ho- mogeneity (DeSalle et al. 1987) for each allele at Sca-37. Significant heterogeneity (P<0.05) was found only at Sca-37*12: this allele was found only in the March 1997 sample from the Florida Keys (KEY''), where it occurred at a frequency of 6.9'7f (Appendix Tables 2 and 3). Because there were far fewer alleles at Sca-37 sampled from the Florida Keys (258) than from either the Atlantic (884) or Gulf (872), the disproportionate frequency of this allele within the Florida Keys likely accounts for the signifi- cance encountered in the exact test and the Roff-Bentzen procedure. Given the absence of this allele in two of the three samples from the Florida Keys, we do not believe the significant heterogeneity detected at Sca-37 is mean- ingful biologically. Although homogeneity testing of pooled samples in- dicated that samples from the Atlantic differed from samples from the Gulf at Sca-14 and Sca-23, and that samples from the Florida Keys differed from the other two at Sca-44. the allele-frequency differences were small and accounted for only a fraction of the overall genetic vari- ance. Results of AMOVA for the comparison of Atlantic with Gulf samples revealed that on average 99.74% of the total genetic variance at the seven microsatellites occurred within samples, as compared to only 0.19% between re- gions. For the comparison of Atlantic, Florida Keys, and Gulf samples, 99.78'7( of the genetic variance on average occurred within samples, whereas only 0.11% occurred among regions. For both comparisons, the proportion of the variation among samples within regions accounted for the remainder of the genetic variance, and for both comparisons, this proportion was small and statistically nonsignificant. Finally, homogeneity tests were used to examine the temporal stock boundaries currently used in management of the king mackerel resource by classifying each of the 20 samples as either Atlantic or Gulf stock (Table 1). No sig- nificant heterogeneity at any of the seven microsatellites was found, providing no genetic evidence for existence of temporal boundaries dividing Atlantic and Gulf migratory units (stocks). Neighbor joining of Cavalli-Sforza's chord distances between pairs of samples yielded little evidence of geo- gi-aphic structure among the 20 samples. With few excep- tions, samples from the same or geographically proximate localities did not cluster together, and most nodes in the topology (available from the first author) were supported by well less than 50% of bootstrap proportions. Spatial au- tocorrelation (SAAP) analysis also indicated the absence of a relationship between allele frequency and geographic distance. Initially, SAAP analysis employed both equal geographic distances between each of five distance classes and equal numbers of pairwise comparisons in each dis- tance class. Analysis involving equal geographic distances between distance classes generated an uneven number of pairwise comparisons among distance classes, i.e. 18, 14, 15, 5, and 3 pairwise comparisons in distance classes 1-5, respectively, resulting in a high variance in Moran's I val- ues among alleles in distance classes 4 and 5. Accordingly, the analysis was restricted to equal numbers of pairwise 500 Fishery Bulletin 100(3) comparisons (eleven) in each distance class. A total of 50 alleles (five at Sca-14, sixteen at Sca-23, four at Sca-37, six at Sca-44, five at Sca-49, three at Sca-61, and eleven at Sca-65) was tested, resulting in 250 Moran's I values. Only 10 significant (P<0.05) Moran's I values were gener- ated: one at Sca-14 (positive in the third distance class); seven at Sca-23 (two positive in the second distance class, four positive in the third distance class, and one negative in the fifth distance class); two at Sca-37 (one negative in the fourth distance class and one positive in the fifth distance class), and one at Sca-65 (negative in the fifth distance class). No significant Moran's 1 values were found at Sca-44, Sca-49, and Sca-61. Only one of the "significant" Moran's I values (a positive value for Sca-23*22 in the third distance class) remained significant after Bonferroni correction. The general absence of spatial autocorrelation indicates that gene flow in king mackerel is consistent with expectations of an island model (sensu Wright, 1943) of population structure, meaning that there is roughly an equal probability of gene flow between any of the 20 sample localities. Assignment tests generally were concordant with other analyses of king mackerel microsatellites in that fish from the Atlantic appeared to be weakly divergent genetically from fish in the Gulf Each of the 20 samples included high proportions of both Atlantic and Gulf fish, and there ap- peared to be no strong geographic pattern in proportion of fish assigned to either Atlantic or Gulf groups (Table 7). On average, samples from the Atlantic contained more "Atlantic" fish (54%), whereas samples from the Gulf con- tained more "Gulf" fish (51.2%). The three samples from the Florida Keys, on average, contained more "Atlantic" fish (53.3%), but this finding is misleading because the proportion of "Atlantic" fish in the three samples from the Florida Keys ranged from 37.2% to 64.0% (Table 7). In addition, the estimated proportions of Atlantic versus Gulf fish in samples from the Florida Keys were not con- sistent with what might be predicted based on the time of sampling and the spatial-temporal boundaries used currently in king mackerel stock assessment. The KEY' and KEY" samples were obtained in March, a time when both would be considered Gulf stock, yet close to the tem- poral boundary (1 April) when they would be considered Atlantic stock. The estimated proportion of Gulf fish in these two samples was 62.8% (KEYM and 41.4% (KEY'-'). Alternatively, the KEY'^ sample was obtained in January when king mackerel in the Florida Keys are considered Gulf stock. The estimated proportion of Gulf fish in the KEY3 sample was 36.09i . Synopsis and conclusions Genetic data obtained in our study are compatible with the hypothesis that two, weakly differentiated "genetic" subpopulations of king mackerel exist in waters off Florida and that considerable, perhaps extensive, mixing occurs between them. King mackerel sampled from the Florida Keys cannot be assigned unequivocally to either "genetic" stock; all collections tested appeared to be mix- Table 7 Assignment (as percentage) of individuals from 20 samples of king mackerel iScomberomorus cavalla) from the east and west coast of Florida and the Florida Keys to either "Atlantic" or "Gulf group. Sample Assigned group Atlantic Gulf Atlantic Ocean JCK' 51.8 48.2 JCK^ 62.5 37.5 NSB 58.0 42.0 CCN 46.0 54.0 SEB' 58.0 42.0 SEB^ 62.0 38.0 SEB3 52.0 48.0 FTP 53.6 46.4 WPB 41.8 58.2 Avg. 54.0 46.0 Florida Keys KEY! 37.2 62.8 KEY2 58.6 41.4 KEY-' 64.0 36.0 Avg. 53.3 46.7 Gulf of Mexico MCI 49.1 50.9 BCG 45.7 54.3 SARI 46.8 53.2 SAR- 53.3 46.7 SAR3 53.6 46.4 SAR^ 41.1 58.6 SAR5 54.3 45.7 PCY 46.0 54.0 Avg. 48.8 51.2 tures, with approximately equal proportions of fish from the two "genetic" stocks. These results are not consistent with the current spatial-temporal boundaries employed in stock assessment and allocation of the king mackerel resource. Results are consistent with the hypothesis that considerable gene flow occurs among all of the localities sampled, and that differences in gene flow likely do not arise as a function of geographic distance. Similar find- ings were obtained by Gold et al. (1997) in their study of variation in king mackerel mitochondrial DNA. The genetic differences between king mackerel in the Atlantic versus those in the Gulf most likely stem from reduced gene flow (migration) between the Atlantic and Gulf in relation to gene flow (migration) along the Atlantic and Gulf coasts of peninsular Florida. This is consistent with the notion based on studies of other marine fishes ( Avise et al., 1992; Gold and Richardson, 1998) that the southern Florida peninsula serves (or has served) as a biogeo- graphic boundary. Gold et a\ Genetic studies of Scomberomoms cavalla in Florida 501 Acknowledgments We thank the following for direct and indirect assistance in obtaining specimens: R. Broughton, G. Davenport, D. Fable, C. (Jrimes, S. (Jrimes, T. Herbert, E. Heist, P. Kirwin, M. Murphy. L. Richardson, D. Roberts, R. Roman, T. Turner, and Q. Wlnite. We are especially indebted to C. Denis, E. Little, and C, Schaefer (National Marine Fisher- ies Service) and J. O'Hop (Florida Department of Environ- mental Protection) for their help in all aspects of specimen collection. We also thank L. Richardson for technical assis- tance, C. Burridge for carrying out the assignment tests, C. Bradfield and R. Ross for assistance with Figure 1, and C. Burridge and R. Broughton for comments on a draft of the manuscript. Work was supported primarily by the Marfin Program of the U.S. Department of Commerce (Grant NA57-FF-0295) and by the Texas Agricultural Experiment Station (Project H-6703). Part of this work was carried out in the Center for Biosystematics and Biodiversity at Texas A&M University, a facility funded, in part, by the National Science Foundation (award DlR-8907006). Literature cited Angers, B., and L. Bernatchez. 1998. Combined use of SMM and non-SMM methods to infer fine structure and evolutionary history of closely related brook charr {Salvelinus fontinalis, Salmonidae) populations from microsatellites. Mol. Biol. Evol. 15:143-159. Avise, J. C. 1992. Molecular population structure and the biogeographic history of a regional fauna: a case history with lessons for conservation biology. Oikos 6:62-76. Broughton, R. B., L. B. Stewart, and J. R. Gold. 2002. Microsatellite loci reveal substantial gene flow between king mackerel [Scomberomorus cavalla) in the western Atlantic Ocean and Gulf of Mexico. Fish. Res. 54:305-3 16. CavaUi-Sforza, L. L., and A. W. Edwards. 1967. Phylogenetic analysis: models and estimation procedures. Am. J. Hum. Genet. 19:233. Collette, B. B., and C. E. Nauen. 1983. Scombrids of the world: an annotated catalogue of tunas, mackerels, bonitos and related species known to date. FAO Fish Syn. 125, vol. 2, 137 p. FAO, Rome. Collins, L. A., and B. W. Stender. 1987. Larval king mackerel (Scomberomorus cavalla). Spanish mackerel (S. maculatus), and bluefish (Pomato- mus saltatrix) off the southeast coast of the United States, 1973-1980. Bull. Mar Sci. 41:822-834. DeSalle, R., A. Templeton, I. Mori. S. Pletscher, and J. S. Johnston. 1987. Temporal and spatial heterogeneity of mtDNA poly- morphisms in natural populations of Drosophila merca- torum. Genetics 116:215-233. DeVries, D. A., and C. B. Grimes. 1997. Spatial and temporal variation in age and growth of king mackerel from the southeastern U.S. and Mexico. Fish Bull. 95:694-708. DeVries, D. A., C. B. Grimes, and M. H. Prager. 2002. Using otolith shape analysis to distinguish eastern Gulf of Mexico and Atlantic Ocean stocks of king mackerel. Fish. Res. 57(1): 1-62. Do Woody, J. A., and J. C. Avise. 2000. Microsatellite variation in marine, freshwater and ••inadromous fishes as compared with oth(!r animals. J. Fi.sh Biol. .56:461-473 Estoup A., F. Rousset, Y. Michalakis, J.-M. Cornuet, M. Adriamanga, and R. Guyomard. 1998. Comparative analysis of microsatellite and allozyme markers: a case study investigating microgeographic dif- ferentiation in brown trout (Salmu trutta). Mol. Ecol. 7: 339-353. Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance inferred from metric dis- tances among DNA haplotypes: application to human mito- chondrial DNA restriction data. Genetics 131:479-491. Fable, W. A., Jr, L. Trent, G. W. Bane, and S. W. Ellsworth. 1987. Movements of king mackerel based on mark-recap- ture. Movements of king mackerel, Scomfteromoruscaua/Za, tagged in southeast Louisiana, 1983-1985. Mar. Fish. Rev. 49:98-101. Felsenstein, J. 1992. Phylip (Phylogeny Inference Package) version 3.4 manual. Cladistics 5:164— 166. Gold, J. R., and L. R. Richardson. 1991. Genetic studies in marine fishes. FV. An analysis of population structure in the red drum (Sciaenops occllatus) using mitochondrial DNA. Fish. Res. 12:213-241. 1998. Mitochondrial DNA diversification and population structure in fishes from the Gulf of Mexico and western Atlantic. J. Heredity 89:404-414. Gold, J. R., A. Y. Kristmundsdottir, and L. R. Richardson. 1997. Mitochondrial DNA variation in king mackerel {Scomberomorus cavalla) from the western Atlantic Ocean and Gulf of Mexico. Mar Biol. 129:221-223. Gold, J. R., E. Pak, and L. R. Richardson. 2001. Microsatellite variation among red snapper (Lutja- nus campechanus) from the Gulf of Mexico. Mar Biotech- nol. 3:293-304. Grimes, C. B., J. H. Finucane, L. A. Collins, and D. A. DeVries. 1990. Young king mackerel, Scomberomorus cavalla, in the Gulf of Mexico, a summary of the distribution and occur- rence of larvae and juveniles, and spawning dates for Mexi- can juveniles. Bull. Mar Sci. 46:640-654. Guo, S., and E. Thompson. 1992. Performing the exact test of Hardy-Weinberg propor- tion from multiple alleles. Biometrics 48:361-372. Johnson, A. G., W. A. Fable Jr, C. B. Grimes. L. Trent, and J. V. Perez. 1993. Evidence for distinct stocks of king mackerel, Scomb- eromorus cavalla, in the Gulf of Mexico. Fish. Bull. 92: 91-101. Manly, B. J. F. 1991. Randomization and Monte Carlo methods in biology, 281 p. Chapman and Hall, New York, NY. Manooch, C. S., III. 1979. Recreational and commercial fisheries for king mack- erel, Scomberomorus cavalla, in the South Atlantic Bight and Gulf of Mexico, U.S.A. In Colloquium on the Spanish and king mackerel resources of the Gulf of Mexico (E. L. Nakamura and H. R. Bullis Jr, eds.), p. 33-41. Gulf States Mar Fish. Comm., publ. 4. McElroy, D., E. Bermingham, and I. Kornfield. 1992. REAP - the restriction enzyme analysis package. J. Hered. 83:157-158. National Biosciences Inc. 1992. OLIGO, version 4.0. National Biosciences, Inc., Ply- mouth, MN. 502 Fishery Bulletin 100(3) Nielsen, E.E., M. M. Hansen, and V. Loeschcke. 1999. Genetic vanation in time and space: microsatellite an- alysis of extinct and extant populations of Atlantic sal- mon. Evolution 53:261-268. O'Connell, M., M. C. Dillon, J. M. Wright, P. Bentzen, S. Merkouris, and J. Seeb. 1998. Genetic structuring among Alaskan Pacific herring populations identified using microsatellite variation. J. Fish Biol. .53:150-163. Paetkau, D., W. Calvert, I. Stiriing, and C. Strobeck. 1995. Microsatellite analysis of population structure in Canadian polar bears. Mol. Ecol. 4: 347-354. Paetkau, D., L. P. Waits, P. L. Clarkson, L. Craighead, and C. Strobeck. 1997. An empirical evaluation of genetic distance statistics using microsatellite data from bear (Ursidae) populations. Genetics 147:1943-1957. Raymond, M., and F. Rousset. 1995. GKNEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J. Hered. 86:248-249. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223-225. Rivas, L. R. 1951. A preliminary review of the western North Atlantic fishes of the family Scombridae. Bull. Mar Sci. Gulf Carib. 1:209-230. Roflf, D. A., and P. Bentzen. 1989. The statistical analysis of mitochondrial DNA poly- morphisms: x^ and the problem of small samples. Mol. Biol. Evol. 6:539-545. Ross K.G., D. D. Shoemaker, M. J. B Krieger, C. J. DeHeer, and L. Keller 1999. Assessing genetic structure with multiple classes of molecular markers: a case study involving the introduced fire ant Solenopsis invicta. Mol. Biol. Evol. 16:525-543. 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. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for recon- structing phylogenetic trees. Mol. Biol. Evol. 4:406-425. Schaefer, H. C, and W. A. Fable Jr 1994. King mackerel, Scomberomorus cavalla, mark-recap- ture studies off Florida's east coast. Mar Fish. Rev. 56: 13-23. Schug, M. D., C. M. Hutter, K. A. Wetterstrand, M. S. Gaudette, T. F C. Mackay and C. F Aquadro. 1998. The mutation rates of di-, tri-, and tetranucleotide repeats in Drusophila melanogaster. Mol. Biol. Evol. 15: 1751-1760. Sokal, R. R., and N. L. Oden. 1978a. Spatial autocorrelation in biology. 1. Methodology. Biol. J. Linn. Soc. 10:199-228. 1978b. Spatial autocorrelation in biology. 2. Some biological implications and four applications of evolutionary and eco- logical interest. Biol. J. Linn. Soc. 10:229-249. Sutter, F C, III, R. O. Williams, and M. R Godcharles. 1991. Movement patterns of stock affinities of king mack- erel in the southeastern United States. Fish. Bull.: 89: 315-324. Swofford, D. L., and R. B. Selander 1981. BiosYs-1: a Fortran program for the comprehensive analysis of electrophoretic data in population genetics and systematics. J. Hered. 72:281-283. Trent, L., R. O. Williams, R. G. Taylor, C. H. Saloman, and C. S Manooch III. 1983. Size, sex ratio, and recruitment in various fisheries of king mackerel. Scorn beronwrus cavalla, in the southeast- ern United State.s. Fish. Bull: 81:709-721. Turner, T. F, L. R. Richardson, and J. R. Gold. 1998. Polymorphic microsatellite DNA markers in red drum iSciaenops ocellatiis). Mol. Ecol. 7:1771-1772. Wartenberg, D. 1989. SAAP: A spatial autocorrelation analysis program. Department of Environmentmal and Community Medi- cine, Robert Wood Johnson Med. School, Piscataway, NJ. Waser, P. M., and C. Strobeck. 1998. Genetic signatures of interpopulation dispersal. TREE (Trends in Ecology and Evolution): 13:43-44. Weber. J. L. 1990. Informativeness of human (dC dA)^ (dG dT)^ poly- morphisms. Genomics 7:524—530. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370. Wright, J. M. 1993. DNA fingerprinting in fishes. In Biochemistry and molecular biology of fishes, volume 2:57-91. Elsevier Press, Amsterdam. Wright, J. M., and P. Bentzen. 1994. Microsatellites: genetic markers for the future. Rev. Fish Biol. Fish. 4:384-388. Wright. S. 1943. Isolation by distance. Genetics 28:114-138. Gold et al : Genetic studies of Scomberomorus cavalla in Florida 503 Appendix Table 1 Microsatellik'S employed for kinj; mackerel (.Scon beronioriiK cavalla). Primer sequence ( 5 ' -> 3 ') Length Annealing Microsatellite (forward and reverse, respectively) (base pairs) temperature (°C) Sea- 14 ATT CCC CAA ACA ATA CAC AC AGT GGA CGA CCC ATT CTA C 93 56 Sca-23 AGO CCT CTT ACA ATC TGC TAC CC AAA CCT TTA AGG CCT CAA GTA AAG 146 58 Sca-37 GCG CCG TGA CTT TTT ATT GCT C CAA CAA TTA GTC GCA GCC CTA G 154 58 Sfa-44 ATG GCC AAA TGG CAC ATA ATC A GGG CAG CTC CAT GGG TCT GAG T 169 58 Sca-49 AGA TGT GAC AAC AGT GGG ATG GCA GCA GTA ATA AAG 157 56 Sra-61 GGT ACT GTC GGG AGA ATG AGA T TGA ATT TTA TAT GGA GGG TCT G 228 56 Sca-65 AGC TGC TGC CAT GAT TTG TT TCC TCC ACT GCC CCT TTC TT 129 52 504 Fishery Bulletin 100(3) Appendix Table 2 Allele frequencies at mici osatellites in king mackerel iScomheroinorus cave lla). Legend to samples: JKV' = Jacksonville, Florida 1 iJuly 19961; JKV- = Jack ionville. Florida (July, 1998); NSB = New Smryna Beach. Florida (July 1996); CCN = Cape Canaveral, | Florida (Decembe r 1998); SEE' = Sebastian, Florida (March 1997); SEE- = Sebastian, Florida ( March 1998); SEB3 = Sebastian, Florida (December 1998); FTP = Ft. Pierce, Florida (April 1996) WPB = West Palm Beach, Florida (May 1998) KEYi = Key West, Florida (March 19961. Microsatellite Sample (allele)' JKVi JKV2 NSE CCN SEE' SEE2 SEB' FTP WPB KEY' Sea- 14 91 0.000 0.031 0.040 0.030 0.010 0.030 0.000 0.009 0.009 0.010 93 0.046 0.063 0.070 0.040 0.041 0.070 0.050 0.063 0.037 0.030 95 0.741 0.719 0.650 0.630 0.735 0.590 0.670 0.643 0.667 0.680 97 0.185 0.188 0.220 0.250 0.184 0.280 0.250 0.259 0.2.59 0.260 99 0.028 0.000 0.020 0.050 0.031 0.030 0.0.30 0.027 0.028 0.020 Sea -23 138 0.046 0.016 0.060 0.080 0.090 0.030 0.060 0.091 0.073 0.029 140 0.019 0.032 0.030 0.050 0.020 0.000 0.060 0.027 0.018 0.039 142 0.231 0.210 0.120 0.250 0.230 0.230 0.210 0.2.36 0.264 0.265 144 0.009 0.016 0.000 0.020 0.020 0.000 0.010 0.018 0.018 0.000 146 0.102 0.129 0.140 0.100 0.140 0.170 0.140 0.127 0.136 0.147 148 0.019 0.048 0.030 0.000 0.060 0.010 0.010 0.036 0.027 0.020 150 0.176 0.161 0.140 0.180 0.120 0.190 0,100 0.155 0.100 0.088 152 0.241 0.177 0.180 0.120 0.180 0.280 0.150 0.145 0.191 0.098 154 0.009 0.016 0.000 0.010 0.010 0.000 0.000 0.018 0.009 0.020 156 0.111 0.113 0.130 0.110 0.060 0.050 0.140 0.064 0.091 0.118 158 0.019 0.016 0.020 0.020 0.000 0.000 0.010 0.000 0.018 0.020 160 0.000 0.016 0.020 0.000 0.010 0.010 0.030 0.000 0.000 0.000 162 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0.000 164 0.009 0.016 0.030 0.010 0.000 0.030 0.020 0.009 0.009 0.039 166 0.000 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0.009 0.000 168 0.000 0.000 0.000 0.000 0.000 0.000 0.010 0.009 0.000 0.010 170 0.009 0.016 0.010 0.020 0.000 0.000 0.010 0.018 0,000 0.0,39 172 0.000 0.016 0.060 0.010 0.030 0.000 0,010 0.018 0,009 0.029 174 0.000 0.000 0.020 0.000 0,010 0.000 0,010 0.018 0,018 0.010 176 0.000 0.000 0.000 0.010 0.010 0.000 0,000 0.000 0,000 0.010 178 0.000 0.000 0.000 0.000 0.000 0.000 0,020 0.000 0.000 0.010 180 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.009 0.000 0.010 182 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.000 0.000 0.000 184 0.000 0.000 0.000 0.010 0.010 0.000 0,000 0.000 0.000 0.000 Sca-37 138 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.000 0.000 0.000 142 0.000 0.031 0.030 0.000 0.000 0.010 0.010 0.000 0,000 0,020 144 0.667 0.688 0.600 0.694 0.720 0.630 0.540 0.623 0,630 0,637 146 0.009 0.016 0.040 0.010 0.030 0.010 0.030 0.019 0.019 0,000 148 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,000 150 0.000 0.000 0.010 0.000 0.000 0.000 0.000 0,000 0.000 0.000 152 0.009 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.000 0.000 154 0.296 0.266 0.320 0.286 0.240 0.330 0.400 0,330 0.286 0.333 156 0.019 0.000 0.000 0.010 0,010 0.020 0.020 0,028 0.010 0.010 Sea44 153 0.000 0.000 0.000 0.010 0.000 0.000 0.010 0,018 0.000 0.020 157 0.066 0.078 0.110 0.120 0.060 0.080 0.090 0,098 0.056 0.069 161 0.038 0.047 0.030 0.020 0.030 0.040 0.010 0,027 0.037 0.020 165 0.387 0.469 0.390 0.300 0.390 0.310 0.290 0.420 0.370 0.275 169 0.396 0.281 0.380 0.420 0.410 0.470 0.500 0..357 0.333 0.471 173 0.104 0.109 0.090 0.110 0.100 0.090 0.070 0.071 0.194 0.147 177 0.009 0.016 0.000 0.020 0.010 0.010 0,030 0.009 0.009 0.000 181 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.000 0.000 0.000 coTitintied Gold et a\: Genetic studies of Scomberomorus cavalla in Florida 505 Appendix Table 2 (continued) Microsatellitc Sample (allele)' JKV' JKV2 NSB CCN SEB' SEB2 SEB3 FTP WPB KEY' Sca-49 139 0.009 0.000 0.000 0.000 0.000 0.000 0.010 0.009 0.000 0.000 141 0.046 0.109 0.060 0.000 0.043 0.020 0.020 0.036 0.037 0.067 143 0.009 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.011 145 0.546 0.547 0.490 0.590 0.435 0.690 0.580 0.518 0.481 0.533 147 0.167 0.203 0.150 0.130 0.207 0.100 0.170 0.188 0.167 0.178 149 0.019 0.031 0.060 0.030 0.033 0.030 0.030 0.018 0.046 0.000 151 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 153 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 155 0.009 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0.000 0.000 157 0.046 0.031 0.090 0.070 0.076 0.060 0.080 0.063 0.093 0.089 159 0.130 0.063 0.130 0.170 0.196 0.090 0.090 0.161 0.148 0.111 161 0.009 0.016 0.000 0.010 0.011 0.010 0.010 0.009 0.009 0.011 163 0.000 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0.009 0.000 165 0.009 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0.000 167 0.000 0.000 0.000 0.000 0.000 0.000 0.010 0.000 0.000 0.000 Sea 61 208 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 210 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.010 224 0.083 0.078 0.100 0.041 0.071 0.102 0.130 0.074 0.046 0.088 226 0.139 0.063 0.050 0.112 0.061 0.071 0.090 0.065 0.083 0,137 228 0.769 0.859 0.850 0.847 0.857 0.827 0.780 0.861 0.861 0,765 230 0.009 0.000 0.000 0.000 0.010 0.000 0.000 0.000 0.009 0,000 Sea 65 117 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,010 123 0.380 0.422 0.360 0.430 0.440 0.350 0.360 0.384 0.464 0,438 125 0.028 0.047 0.040 0.050 0.030 0.080 0,040 0.063 0.036 0,063 127 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0.000 0,010 129 0.065 0.016 0.130 0.060 0.080 0.090 0.130 0.036 0.036 0,073 131 0.037 0.094 0.030 0.030 0.020 0.000 0.030 0.018 0.000 0,042 133 0.000 0.000 0.000 0.000 0.020 0.000 0.000 0.009 0.009 0,000 135 0.241 0.219 0.200 0.170 0.150 0.280 0.200 0.223 0.209 0,219 137 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0,000 139 0.000 0.000 0.000 0.010 0.010 0.010 0.000 0.018 0.018 0,000 141 0.000 0.000 0.000 0.000 0.000 0.010 0.000 0.000 0.000 0,000 143 0.009 0.000 0.000 0.010 0.020 0.010 0.000 0,009 0.018 0,010 147 0.093 0.047 0.150 0.080 0.080 0.070 0.150 0,080 0.109 0,063 149 0.028 0.016 0.010 0.020 0.020 0.010 0.010 0.018 0.009 0,021 151 0.009 0.016 0.020 0.040 0.030 0.030 0.020 0.045 0.018 0,000 153 0.046 0.016 0.020 0.000 0.040 0.020 0.020 0.045 0.027 0,031 155 0.037 0.031 0.020 0.050 0.010 0.020 0.020 0.036 0.027 0,010 157 0.019 0.016 0.010 0,030 0.020 0.020 0.010 0.009 0.000 0,000 159 0.000 0.016 0.000 0.020 0.010 0.000 0.000 0.000 0.000 0,000 161 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,000 163 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,010 165 0.000 0.016 0.000 0.000 0.010 0.000 0.000 0.000 0.000 0,000 167 0.009 0.031 0.010 0.000 0.010 0.000 0.010 0.000 0.000 0,000 171 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0,000 ' Allele number represents size in base pairs of the fragnien amplified. 506 Fishery Bulletin 100(3) Appendix Table 3 Allele frequencies at microsatellites in king mackerel tScomhci-omoriis cavalla). Legend to sanr pies: KEY = Key West, Florida (March 1997); KEY-' = Key West, Florida (January 1999); MCI = Marco Island, Florida April 1996); BCG = Boca Grande, Florida (April 1996); SAR ' = Treasure Island Florida (April 1996); SAB ' = Treasu re Island, F onda (November 1996); SAR-' = Treasure 1 Island, Florida (April 1997 ); SAR-" = Treasure Island, Florida (April 1998); SAR^ = Treasure Island Florida (November 1998); PCY= | Panama City, Florida (Octobe ■ 1996). Microsatellite Sample (allele)' KET-^ KEY* MCI BCG SARI SAR^ SAR-i SAR^ SAR'' PCY Sea- 14 91 0.000 0.000 0.009 0.014 0.000 0.009 0.009 0.000 0.007 0.000 93 0.069 0.080 0.036 0.057 0.033 0.018 0.063 0.021 0.036 0.020 95 0.724 0.620 0.745 0.771 0.685 0.764 0.741 0.771 0.714 0.680 97 0.190 0.250 0.173 0.143 0.250 0.200 0.170 0.186 0.221 0.2,30 99 0.017 0.050 0.036 0.014 0.033 0.009 0.018 0.021 0.021 0.070 Sea 23 138 0.086 0.060 0-018 0.057 0.096 0.033 0.027 0.036 0.096 0.060 140 0.017 0.080 0.027 0.029 0.032 0.017 0.027 0.043 0.029 0.020 142 0.224 0.250 0.309 0.257 0.266 0.217 0.200 0.2.39 0.2.50 0.200 144 0.017 0.010 0.000 0.000 0.000 0.000 0.018 0.000 0.000 0.020 146 0.069 0.050 0.145 0.171 0.064 0.183 0.118 0.109 0.103 0.080 148 0.000 0.000 0.018 0.000 0.032 0.008 0.009 0.029 0.007 0.010 150 0.121 0.120 0.155 0.043 0.160 0.108 0.173 0.130 0.1.32 0.1.30 152 0.207 0.240 0.136 0.157 0.117 0.142 0.164 0.174 0.176 0.110 154 0.000 0.010 0.000 0.029 0.011 0.017 0.009 0.007 0.022 0.030 156 0.086 0.120 0.045 0.157 0.117 0.117 0.127 0.109 0.096 0.130 158 0.017 0.000 0.000 0.000 0.021 0.008 0.027 0.007 0.000 0.030 160 0.052 0.000 0.027 0.014 0.000 0.033 0.000 0.014 0.000 0.010 162 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.007 0.000 0.000 164 0.034 0.010 0.000 0.029 0.000 0.033 0.018 0.014 0.022 0.010 166 0.017 0.000 0.009 0.014 0.000 0.000 0.009 0.007 0.000 0.040 168 0.000 0.000 0.000 0.000 0.000 0,000 0.000 0.000 0.000 0.000 170 0.000 0.010 0.027 0.014 0.011 0.025 0.009 0.014 0.015 0.030 172 0.017 0.010 0.036 0.014 0.064 0.025 0.036 0.036 0.037 0.060 174 0.017 0.020 0.018 0.000 0.000 0.017 0.009 0.007 0.015 0.010 176 0.000 0.010 0.027 0.014 0.000 0.000 0.009 0.007 0.000 0.020 178 0.017 0.000 0.000 0.000 0.011 0.000 0.000 0.000 0.000 0.000 180 0.000 0.000 0.000 0.000 0.000 0.017 0.000 0.000 0.000 0.000 182 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0.007 0.000 0.000 184 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Sca-37 138 0.069 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 142 0.034 0.000 0.009 0.000 0.000 0.017 0.009 0.007 0.000 0.023 144 0.603 0.653 0.627 0.700 0.628 0.642 0.679 0.600 0.579 0.593 146 0.000 0.010 0.018 0.043 0.021 0.025 0.036 0.007 0.007 0.000 148 0.000 0.000 0.000 0.000 0.000 0.009 0.000 0.000 0.000 0.000 150 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 152 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 154 0.276 0.327 0..327 0.257 0.319 0.292 0.250 0.343 0.379 0.349 156 0.017 0.010 0.018 0.000 0.032 0.017 0.027 0.043 0.036 0.035 Sca-44 153 0.034 0.000 0.009 0.014 0.000 0017 0.000 0.029 0.000 0.000 157 0.069 0.010 0.056 0.100 0.096 0.133 0.089 0.114 0.072 0.050 161 0.052 0.000 0.037 0.014 0.032 0.008 0.098 0.036 0.051 0.040 165 0.276 0.360 0.259 0.314 0.255 0.300 0.268 0.293 0.355 0.300 169 0.431 0.480 0.546 0.443 0.436 0.408 0.455 0.393 0.413 0.430 173 0.121 0.120 0.093 0.100 0.170 0.133 0.080 0.129 0.087 0.160 177 0.017 0.020 0.000 0.014 0.011 0.000 0.009 0.007 0.022 0020 181 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0-000 0.000 CO 0.000 ntiniicil Gold et al.: Genetic studies of Scomberomorus cavalla in Florida 507 Appendix Table 3 (continued) MicrosMtcllito Sample (alleU'l' KEY' KEV MCI BCG SARI SAR2 SAR3 SAR-" SAR'^ PCY Sca-49 139 0.000 0.000 0.009 0.000 0.011 0.010 0.009 0.000 0.000 0.000 141 0.069 0.070 0.055 0.043 0.054 0.031 0.054 0.043 0.036 0.020 143 0.000 0.000 0.000 0.000 0.022 0.000 0.000 0.007 0.000 0.000 145 0.621 0.580 0.464 0.457 0.576 0.594 0.455 0.614 0.529 0.560 147 0.121 0.100 0.218 0.200 0.076 0.156 0.152 0.114 0.200 0.180 149 0.017 0.040 0.009 0.029 0.011 0.010 0.027 0.014 0.014 0.060 151 0.000 0.000 0.009 0.014 0.000 0.000 0.000 0.007 0.000 0.000 153 0.000 0.000 0.000 0.000 0.011 0.000 0.000 0.000 0.000 0.000 155 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 157 0.052 0.080 0.100 0.129 0.076 0.063 0.089 0.021 0.071 0.050 159 0.086 0.130 0.118 0.129 0.141 0.135 0.196 0.171 0.143 0.130 161 0.034 0.000 0.018 0.000 0.000 0.000 0.009 0.007 0.000 0.000 163 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 165 0.000 0.000 0.000 0.000 0.022 0.000 0.009 0,000 0.000 0.000 167 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.007 0.000 Sea- 61 208 0.000 0.000 0.000 0.000 0.000 0.009 0.000 0.000 0.000 0.000 210 0.000 0.000 0.000 0.000 0.011 0.000 0.000 0.000 0.000 0.000 224 0.052 0.130 0.073 0.086 0.054 0.138 0.127 0.107 0.100 0.070 226 0.017 0.100 0.055 0.143 0.098 0.078 0.100 0.114 0.121 0.070 228 0.931 0.760 0.873 0.757 0.837 0.776 0.755 0.779 0.771 0.860 230 0.999 0.010 0.000 0.014 0.000 0.000 0.018 0.000 0.007 0.000 Sea -65 117 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.010 123 0.483 0.460 0.364 0.357 0.436 0.446 0.375 0.357 0.386 0.340 125 0.017 0.030 0.036 0.100 0.064 0.045 0.071 0.086 0.043 0.060 127 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 129 0.052 0.080 0.055 0.057 0.032 0.054 0.063 0.057 0.021 0.060 131 0.000 0.020 0.027 0.000 0.064 0.036 0.018 0.071 0.029 0.020 133 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 135 0.207 0.160 0.273 0.300 0.245 0.179 0.205 0.193 0.243 0.190 137 0.000 0.000 0,000 0.000 0.000 0.009 0.000 0.000 0.000 0.010 139 0.000 0.010 0.000 0.000 0.000 0.009 0.009 0.007 0.007 0.000 141 0.000 0.000 0.018 0.000 0.000 0.000 0.000 0.000 0.000 0.010 143 0.017 0.010 0.000 0.014 0.000 0.036 0.009 0.000 0.007 0.000 147 0.103 0.110 0.073 0.029 0.053 0.080 0.080 0.100 0.114 0.090 149 0.034 0.020 0.036 0.057 0.021 0.000 0.018 0.014 0.029 0.020 151 0.017 0.040 0.055 0.000 0.000 0.009 0.018 0.014 0.021 0.040 153 0.034 0.030 0.018 0.029 0.021 0.036 0.054 0.029 0.029 0.090 155 0.017 0.020 0.018 0.014 0.032 0.027 0.045 0.021 0.029 0.050 157 0.017 0.010 0.018 0.014 0.011 0.018 0.027 0.014 0.029 0.010 159 0.000 0.000 0.000 0.014 0.000 0.000 0.009 0.014 0.000 0.000 161 0.000 0.000 0.009 0.000 0.000 0.000 0.000 0.007 0.000 0.000 163 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 165 0.000 0.000 0.000 0.000 0.000 0.018 0.000 0.000 0.000 0.000 167 0.000 0.000 0.000 0.014 0.021 0.000 0.000 0.014 0.007 0.000 171 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.007 0.000 ' Allele number represents size in base pairs of the fragme nt amplified 508 Fishery Bulletin 100(3) Append ix Table 4 Summary statistics for each of seven microsatellites among samples of king mackerel iScomheromorus cave lla): n = number of individuals assayed; //q,. = direct-count heterozygosity; and P/„y = probability that genotypes cor form to expectations of Hardy- Weinberg equilibri jm. Legend 5 to sample localities are given in Appendix Table 2. Microsatellite JKV' JKV^ NSB CCN SEB' SEB2 SEB3 FTP WPB KEY> Sea- 14 n 54 32 50 50 49 50 50 56 54 50 ^DC 0.481 0.375 0.460 0.500 0.408 0.600 0.420 0.571 0.537 0.520 ^HW 0.710 0.297 0.141 0.354 0.333 0.431 0.440 0.203 0.070 0.103 Sca-23 n 54 31 50 50 50 50 50 55 55 51 ^DC 0.833 0.806 0.780 0.840 0.820 0.600 0.800 0.855 0.800 0.765 P 0.571 0.094 0.121 0.618 0.654 0.000 0.141 0.085 0.483 0,000 Sca-37 n 54 32 50 49 50 50 50 53 54 51 ^DC 0.500 0.531 0.540 0.449 0.380 0.520 0.560 0.585 0.463 0.608 ^HW 0.039 0.306 0.903 0.193 0.607 1.000 0.784 0.291 0.293 0.069 Sca-44 n 53 32 50 50 50 50 50 56 54 51 ^DC 0.604 0.719 0.720 0.680 0.720 0.620 0.660 0.643 0.722 0.569 p 0.584 0.965 0.911 0.518 0.583 0.618 0.719 0.875 0.095 0.095 Sca-49 n 54 32 50 49 46 50 50 56 54 45 ^DC 0.611 0.656 0.820 0.680 0.739 0.540 0.620 0.696 0.704 0.600 ^HW 0.722 0.543 0.871 0.509 0.290 0.753 0.313 0.958 0.943 0.571 Sca-61 n 54 32 50 49 49 49 50 54 54 51 ^DC 0.407 0.281 0.280 0.306 0.224 0.306 0.360 0.204 0.278 0.392 ^HW 1.000 1.000 0.709 1.000 0.152 0.306 0.703 0.093 1.000 0.851 Sea -65 n 54 32 50 50 50 50 50 56 55 48 ^DC 0.870 0.813 0.900 0.900 0.820 0.800 0.780 0.839 0.727 0.729 p 0.975 0.388 0,647 0.985 0.777 0.242 0.917 0.978 0.479 0.838 Gold et al : Genetic studies of Scomberomorus cavalla in Florida 509 Appendix Table 5 Summary statistics for each of seven microsatellites among samples of king mackerel (.Scomberomorus cavalla): n = number of individua s assayed; //,„. = direct-count heterozygosity; and P„„, = probability that genotypes cor form to expectations of Hardy- Weinbei-R oqui librium. Legen ds to .sampl e localities are in Appen dix Table 2 Microsatellite (allele) KEY2 KEY^ MCI BCG SAR' SAR2 SAR^ SAR^ SAR5 PCY Sca-14 n 29 50 55 35 46 55 56 70 70 50 ^DC 0.517 0.500 0.473 0.371 0.543 0.345 0.429 0.414 0.500 0.520 ^im 0.760 0.143 0.246 0.042 0.562 0.403 0.690 0.540 0.824 0.336 Sca-23 n 29 50 55 35 47 60 55 69 68 50 ^nc 0.897 0.760 0.745 0.886 0.723 0.800 0.855 0.783 0.838 0.880 ^HW 0.416 0.106 0.032 0.947 0.006 0.039 0.846 0.000 0.647 0.365 Sca-37 n 29 50 55 35 47 60 55 69 68 50 ^DC 0.552 0.469 0.491 0.486 0.447 0.617 0.446 0.571 0.543 0.419 ^HW 0.092 0.647 0.384 0.851 0.079 0.431 0.817 0.499 0.785 0.235 Sca-44 n 29 50 54 35 47 60 56 70 70 50 ^DC 0.828 0.560 0.611 0.686 0.723 0.633 0.696 0.714 0.739 0.700 ^HW 0.686 0.219 0.953 0.722 0.353 0.495 0.335 0.813 0.132 0.050 Sca-49 n 29 50 55 35 46 48 56 70 70 50 ^DC 0.586 0.580 0.727 0.771 0.630 0.625 0.714 0.500 0.686 0.640 ^HW 0.667 0.055 0.679 0.153 0.098 0.806 0.303 0.246 0.847 0.856 Sca-61 n 29 50 55 35 46 58 55 70 70 50 ^DC 0.138 0.480 0.218 0.343 0.326 0.379 0.327 0.314 0.371 0.280 ^HW 1.000 0.486 0.390 0.073 1.000 0.348 0.016 0.105 0.364 1.000 Sea -65 n 29 50 55 35 47 56 56 70 70 50 ^DC 0.690 0.800 0.727 0.743 0.681 0.696 0.839 0.843 0.886 0.880 ^HW 0.581 0.656 0.317 0.615 0.250 0.425 0.526 0.879 0.183 0.942 510 Abstract-Catch rates in the South Af- rican rock lobster iJasus lalandii) fish- ery declined after 1989 in response to reduced adult somatic growth rates and a consequent reduction in recruit- ment to the fishable population. Al- though spatial and temporal trends in adult growth are well described, little is known about how juvenile growth rates have been affected. In our study, growth rates of juvenile rock lobster on Cape Town harbor wall were com- pared with those recorded at the same site more than 25 years prior to our study, and with those on a nearby natural nursery reef We found that indices of somatic growth measured during 1996-97 at the harbor wall had declined significantly since 1971-72. Furthermore, growth was slower among juvenile J. lalandii at the harbor wall than those at the natural nursery reef These results suggest that growth rates of juvenile and adult J. lalandii exhibit similar types of spatiotemporal pat- terns. Thus, the recent coastwide decline in adult somatic growth rates might also encompass smaller size classes. Do fluctuations in the somatic growth rate of rock lobster Uasus lalandii) encompass all size classes? A re-assessment of juvenile growth R. W. Anthony Hazell Department ol Zoology University ol the Western Cape Private Bag X17 Bellville, 7535, South Africa David S. Schoeman Department ol Zoology University of Port Elizabeth PO Box 1600 Port Elizabeth, 6000, South Africa E-mail address (for D S Schoeman, coniacl aultior) zladssiSzooupeacza Mark N. Noffke Marine and Coastal Management Private Bag X2 Rogge Bay 8012 South Africa Manuscript accepted 2 May 2001. Fish. Bull. 100:510-518 (2002). The South African fishery for rock lobster (Jasus lalandii) began in the late nineteenth century and expanded rapidly so that catches peaked at approximately 10,000 metric tons (t) per annum between 1950 and 1965 (Pollock, 1986). However, subsequent deterioration in catches in the 1960s and 1970s was arrested only in the mid-1980s by the implementation of a management approach centered on an annual total allowable catch (TAC). Over the following five years, catches were maintained at apparently sustainable levels of 3500-4000 t per annum (Cockcroft and Payne, 1999), still representing the largest yield of any Jasus species at that time (Pollock, 1986). However, the period of stability ended after 1989, when catch rates declined in response to reduced adult somatic growth rates and a concomi- tant reduction in recruitment to that part of the population larger than the minimum legal size (Melville-Smith et al., 1995; Goosen and Cockcroft, 1995; Cockcroft, 1997). Adult somatic growth rates for this resource are assessed by means of a tag- recapture program. This is relatively simple because mature individuals molt only once a year, just prior to the start of the commercial fishing season, and samples of these animals can be tagged shortly before they molt. Annual gi'owth increments can therefore be calculated from postmolt- tagged animals recaptured during the subsequent commercial fishing season. This type of tag-recapture data is routinely collected from most of the fishing grounds that are important to the South African commercial fish- ery (Fig. 1, inset). This tagging pro- gram has yielded one of the most com- prehensive rock lobster tagging-for- growth databases in the world; it con- tains continuous time-series for most areas since 1986 and broken time- series for some sites dating back to the 1969-70 season (Goosen and Cockcroft, 1995). Consequently, temporal and spatial trends in adult growth are well described (Melville-Smith et al, 1995; Goosen and Cockcroft, 1995; Cockcroft, 1997; Cruywagen, 1997; Pollock et al., 1997). By contrast, little is known about what may affect juvenile growth rates. Of particular concern is the lack of information regarding the way in which juvenile growth rates might have been affected by factors that Hazell et aL: Somatic growth rate of lasus lalandii 511 Figure 1 Map of Table Bay showing the position of the two study sites. Inset: map of southern Africa showing the distribution of rock lobster, Jasus lalandii. caused the decline in adult growth rates during the early 1990s. A decrease in the size of females at 50% maturity provided indirect evidence that juvenile growth rates may also have been adversely affected (Pollock, 1987; Cockcroft and Goosen, 1995; Pollock, 1995; Pollock, 1997), but this decline in juvenile growth rates has not been confirmed by direct measurement. Despite both this substantial gap in our knowledge of the ecology of J. lalandii and the growing international recognition of the importance of understanding the early life history stages of species with complex life cycles I Herrnkind et al. , 1994 ), juvenile growth of this species was last assessed more than 25 years ago ( Pollock, 1973 ). It was therefore essential that juvenile growth be re-examined by repeating Pollock's (1973) study at Cape Town harbor wall. This site was originally selected because of the large numbers of juvenile rock lobster present on the vertical face of the harbor wall and because these lobsters were easy to collect. In our study, a second, natural nursery reef site at nearby Mouille Point (Fig. 1) was selected to compare with the original artificial harbor wall site. This paper addresses two hypotheses. The first is that somatic growth rates of juveniles on the harbor wall have not declined since Pollock's (1973) study. The second is that juvenile growth rates of rock lobster on the harbor wall are not different from those of rock lobster on the nearby natural nursery reef at Mouille Point. Methods Juvenile rock lobsters were collected between March 1996 and February 1997 at two sites: Cape Town harbor wall (seaward side) and Mouille Point (Fig. 1). Only the harbor wall was sampled in March and intervals between sam- ples were approximately one month, although it was not possible to sample either site every month. Two autumn months (one for Mouille Point), two winter months, two spring months, and three summer months were sampled, thereby allowing for seasonal variation in growth. Juvenile rock lobsters were sampled by two SCUBA div- ers who collected specimens for roughly 20-30 minutes, or by one diver for about twice as long (Table 1). At the harbor wall, divers started at the base of the wall (>10 m depth) and worked their way up to the surface, collecting every lobster encountered, where possible. At Mouille 512 Fishery Bulletin 100(3) Table 1 Summary of data relating to the field sampling of juvenile rock lobster at Mouille Point ar d Cape harbo r wall, 1996-97 Postmolt Premolt Water Intermolt Number of Effort temperature' Soft Hard Hard Soft Site Sample date divers (diver minutes) (°C) new new Hard old old Total Mouille Point 4 Apr 1996 1 40 11.2 2 4 202 65 0 273 21 Jun 1996 1 50 13.5 2 3 167 67 0 239 24 Jul 1996 2 30 13.3 8 15 174 56 0 253 17 Oct 1996 2 75 12.5 2 10 239 80 4 335 27 Nov 1996 2 50 9.4 4 12 291 58 0 365 20 Dec 1996 1 40 10.0 4 14 212 60 1 291 30 Jan 1997 2 50 8.8 2 19 224 52 0 297 25 Feb 1997 2 65 9.2 5 17 176 77 0 275 Harbor wall 15 Mar 1996 1 60 10.0 12 6 325 121 0 464 24 Apr 1996 2 55 12.5 3 7 228 52 0 290 21 Jun 1996 1 20 13.5 1 1 210 42 2 256 24 Jul 1996 1 45 13.3 8 13 243 29 6 299 17 Oct 1996 2 60 12.5 3 15 205 80 0 303 29 Nov 1996 2 45 12.9 4 9 262 30 0 305 20 Dec 1996 1 40 10.0 6 7 275 38 0 326 30 Jan 1997 2 50 8.8 2 13 280 38 2 335 25 Feb 1997 2 64 9.2 10 34 210 76 0 330 ' Temperature readings were taken from a tempera ure logger located ir 10 m of water approximately 2 km east of Mouille Point Point, divers swam around the reef area, covering different habitat types (ledges, vertical walls, cracks, etc.), and attempted to catch every lobster encountered. The catch was maintained in a bin of seawater until the sex, shell state (sensu Pollock, 1973) and carapace length (CD of each animal could be recorded. Hard old and soft old shell states were considered to represent premolt animals, soft new and hard new to represent postmolts, and those animals in the hard shell state were considered to be in the intermolt phase of the molt cycle. Postmolt and intermolt specimens were released, whereas premolt animals were measured to the nearest 0.1 mm and transported to the laboratory. Whenever relatively few premolt specimens of a particular size-category were captured during routine searches, additional dive time was spent collecting these. In the laboratory, captive specimens were transferred to perforated plastic jars (either 250 or 500 cm-'' in volume, with square perforations of approximately 25 mm-), which were placed in a 2-m'' holding tank. To mimic natural conditions (Table 1), aquarium water temperature was controlled between 11° and 14°C during holding, with a mean of 12.6°C (SD=0.48°C), and salinity was maintained at 35-36%f by periodic addition of freshwater or partial replacement of seawater. Stocking density did not exceed three specimens per jar, and larger specimens were held singly or in pairs. By using a combination of jar number, sex, premolt carapace length, and the pattern of missing limbs, it was possible to identify each individual. Because J. lalandii are unable to feed shortly before molting (Zou-tendyk, 1988), it was assumed that no feeding was required. Jars were checked for molting individuals at least every second day. Cast exoskeletons were examined to identify those molting, and each new carapace length, as well as any limb regeneration, was recorded once the shell had hardened. Increment data recorded for December 1996 were discarded because of aquarium failure: raised temperatures and lower oxygen levels resulted in high mortality during molting. Because increments from spe- cimens maintained in the laboratory for more than 18 days were considered unreliable, only animals that molted within this period were included in analyses. Tests of the hypothesis that somatic growth rates of juveniles on the harbor wall had not declined since Pollock's (1973) study were complicated by the absence of his original data. All that remain are mean molt in- crements for male and for female specimens categorized into 5-mm-CL size classes, their standard deviations, and their sample sizes. Calculation of combined mean increments for males and females of a given size is simple, but their standard deviations had to be approximated by adding corresponding sums of squares for males and females and dividing this value by the overall degrees of freedom within the size class (Table 2). It had been our intention to compare these reconstructed data with those from our study by using a two-way AN OVA (site x size class), but this was not possible, given the limitations of Pollocks ( 1973) data. Instead, the data were subjected to a one-way ANOVA and subsequent Student-Newman-Keuls Hazell el al : Somatic growth rate of Jasus lalandii 513 Table 2 Mean molt increments and their standard deviations (mm and 1996-97, and at Mouille Point in 1996-97. CD of juveni le Jasus lalandii measured on the harbor wall n 1971-72 Size class (mm CD Harbor wall 19' ■1-72' Harbor wall 1996- -97 Mouille Point 1996-97 n Mean SD2 n Mean SD n Mean SD 15-20 14 2.65 0.76 35 1.94 0.56 21 2.10 0.46 20-25 37 3.12 0.82 98 2.45 0.56 74 2.72 0.64 25-30 28 3.71 1.08 95 2.78 0.64 80 2.93 0.68 30-35 14 4.36 0.73 64 3.07 0.75 93 2.86 0.96 35-40 17 4.71 0.54 33 3.39 0.93 44 3.56 0.90 40-45 23 4.72 0.93 13 3.66 0.93 29 3.74 0.77 45-50 23 5.02 0.62 6 3.50 0.79 3 4.00 0.69 ' After Pollock (1973). - An approximation of the standard deviation of freedom. calculated by add ng the sums of squares for mal 's and females ar dd ividing by the overall degrees multiple range post hoc tests (SigmaStat, version 2.03, [SPSS Inc., 1997]). Of greatest interest in this analysis were the results of post hoc tests in which means of the corresponding size classes of the two sample periods were compared. For comparison of length-related growth increments at the two sites in our study, individual molt increments were plotted against premolt carapace lengths, for each combination of sex and site, and their linear regressions were compared by using a two-way ANCOVA (Zar, 1984; Norman and Streiner, 1994). Investigating relative intermolt periods from field samples requires identifying those specimens that are considered to be molting. Rock lobster juveniles in the hard old shell state isensii Pollock 11973]) were easily identified as such by visual inspection, and we confirmed this by breaking the tip off an antenna and inspecting the condition of the underlying integument. Furthermore, on returning these specimens to the laboratory, they generally molted within 20 days. Because field samples were taken at intervals at least this long (Table 1), the proportion of each sample found to be in the hard old shell state was a reasonable indication of the relative number of juveniles molting during that sample month. This proportion, in turn, was used as a proxy for intermolt period under the assumption that shorter intermolt periods would result in relatively higher numbers of juveniles that were molting. The numbers of specimens with soft old or soft new shells did not accurately reflect the relative rate of molting because, as a result of behavioral modifications, rock lobster in these conditions may have had a lower catchability than those with hard shells. Using this index of relative intermolt period, we per- formed a log-linear analysis (Zar, 1984; Norman and Strei- ner, 1994) to assess the interdependence of the factors site, size-category, sample month, and molt state with regard to the frequency of observations in each subcategory. Only size classes smaller than 40 mm carapace length were used in our analysis because the decreasing molt frequency with increasing carapace length means that larger sample sizes are required to accurately estimate the proportion of animals of the size class in which rock lobsters are molting in the population. However, the length distribution of ju- venile lobsters meant that sample sizes decreased above 40 mm CL. Results An average of 323 juvenile J. lalandii was collected per sample at the harbor wall (256-464), and 291 at Mouille Point (239-365). After directed collections for additional premolt specimens of various sizes, a total of 375 juveniles from the harbor wall and 383 from Mouille Point were maintained in the laboratory for analysis of growth rates. Mortality rates were low (<1%) among these specimens, except during December 1996, when aquarium malfunc- tion resulted in extensive fatalities during molting. Can- nibalism was not observed during laboratory trials. Individual molt increments amongst the captive spec- imens were highly variable, ranging from 0.4 to 5.5 mm. However, mean molt increments for each of seven con- secutive 5-mm-CL size-intervals ( Table 2 ) were consistently and significantly smaller in our study (1996-97 season) than those measured in 1971-72 (Table 3, Fig. 2). Thus, the hypothesis that somatic growth rates of juveniles on the harbor wall had not declined since Pollock's (1973) study could be rejected at one level. Unfortunately, there were no comparative data to provide accurate information on corresponding intermolt periods. In terms of our samples from Mouille Point and the harbor wall, significant, positive linear relationships were found between molt increment and premolt carapace length for both males and females at both sites (Fig. 2, 514 Fishery Bulletin 100(3) Table 3 Results of one-way ANOVA comparing growth increments among size-intervals at the harbor wall from the two sample periods. Subsequent post hoc Student-Newman-Keuls multiple-range test results are summarized in the matrix below; an asterisk indi- cates that molt increments differ significantly (P<0.05) between the two samples being compared. Source of variation df 88 MS Between groups Within groups Total 13 486 499 332.57 258.48 591.06 25.58 0.53 48.1 <0.001 Size class (mm CLi 1971-72 1996-97 15-20 20-25 25-30 30-35 35-40 40-45 45-50 15-20 20-25 25-30 30-35 35-40 40-45 4.5-50 1971-72 1996-97 15-20 20-25 25-30 30-35 35-40 40-45 45^50 15-20 20-25 25-30 30-35 35-40 40-45 45-50 P<0.0001 in all cases). Three females at the harbor wall >50 mm CL were excluded from all analyses (unshaded circles, Fig. 2A) because they were the only specimens of this size. Further, females >50 mm CL might have had smaller than expected molt increments due to approach- ing maturity (Pollock, 1973). Two-way, fixed effects ANCOVA comparing these linear regressions indicated no interaction between the factors site and sex; nor was there a significant sex-effect (Table 4). However, there was a significant site-effect: the rela- tionship for samples from Mouille Point had a significantly higher regression constant than that for samples from the harbor wall (Table 4). These results imply first that there were no differences in length-specific growth increments between the sexes within sites. The second implication is that molt increment increased with carapace length at the same rate, irrespective of sex or site. Finally, juvenile lobsters of any given size or sex at Mouille Point grew sig- nificantly more during each molt stage than coiTesponding specimens at the harbor wall (see also Table 2 and Fig. 2). Log-linear analysis indicated significant first-, second-, and third-order interactions among the factors site, size- category, sample month, and molt state (Table 5). How- ever, the only significant third-order interaction was that among the independent factors site, size-category, and sample month. This interaction simply confirmed that size-frequency distributions differed between sites over Table 4 Summary of ANCOVA results from length-specific molt increments cl assified according to sex and site. Factor ''/://.■. df.rr.., F P Slope Site 1 597 0.70 0.40 Sex 1 597 0.66 0.42 Site X sex 3 595 0.61 0.61 Elevation Site 1 598 4.63 0.03 Sex 1 598 1 11 0.29 Site X sex 1 598 0.01 0.93 the sample period, as would be expected if growth rates differed. Of the three remaining nonsignificant third- order interactions, the most important for our purposes were those including the site factor. These imply that the temporal pattern of molting is similar between sites, as is the size-specific pattern of molting. These results are sup- ported by graphical representations of the data (Fig. 3), which suggest that for both sites 1) the proportion of molt- ers per size-category decreases as size increases, and 2) Hazell et al.: Somatic growth rale of Jasus lalandn 515 5- A • 5- B 4- 3- J^ 4- 3- 0 2- E 1- E^ 4^. • ' • y»=0 05t119 • r-= 0 21 2 1- * yj<= 0.07 + 0 83 r = 0 31 c E 20 30 40 50 20 30 40 50 Molt inc en c 5- " ^'- 4- 3- 2- 1 ■ •*• • y»= 0 05 + 1 35 r=0 19 4- 3- 2- 1- y»= 0 06 t 1 36 f=0 17 20 30 40 50 20 30 40 50 Premolt carapace length (mm) Figure 2 Relationship between molt increment and premolt carapace length of juvenile lobsters at the harbor wall and Mouille Point: (A) harbor wall females, (B) harbor wall males, (C) Mouille Point females, (D) Mouille Point males. Shaded circles represent data included in analyses; unshaded circles represent data excluded from analyses (see text for details). AMJJASONDJF Time (months) ^ — I B 17.5 22.5 27.5 32,5 37 5 Carapace length (mm) Figure 3 Site-specific patterns of the proportions of sampled juvenile rock lobster that were considered to be molt- ing with respect to (A) sample month and (B) 5-mm-CL size classes. MP = Mouille Point; HW = harbor wall. the proportion of lobsters in the molt state fluctuates over time. In both cases, the proportion of lobsters in the molt state was consistently lower at the harbor wall than at Mouille Point (with the exception of the sample taken in October 1996). This apparent dependence of molt state on each of the factors site, size-category, and sample date is confirmed by the corresponding significant second-order interactions (Table 5). Together with the molt-increment data, these results give reason to reject the hypothesis that current juvenile growth rates on the harbor wall are no different from those on the nearby natural nursery reef in Mouille Point; growth is significantly faster at the latter. Discussion The significance of puerulus and postpuerulus ecology to the management of rock lobster fisheries is now widely accepted (Herrnkind et al., 1994). In the South African J. lalandii resource, the current controversy surrounding the question of whether or not temporal trends in juvenile growth rates reflect those of adults provides a good exam- ple of why so much emphasis should be placed on early life history stages. For this species, adult somatic growth rates have declined substantially since the mid-1980s (Melville- Smith et al., 1995; Cockcroft, 1997, Cockcroft and Payne, 1999). Although the causes of this phenomenon are not yet clearly understood, its widespread nature is indica- tive of a large-scale environmental perturbation. This has prompted hypotheses that the anomalous El Nino years of 1990-93 may have resulted in dramatic changes in the productivity of the southern Benguela Current (Pol- lock et al., 1997; Pollock et al., 2000). Alternatively, it is plausible that heavy, size-selective fishing of this resource may have caused a decline in growth rates by removing individuals genetically predisposed to more rapid growth (see for example Stokes and Law 12000]). Notwithstand- ing the causes, if it is incorrectly assumed that juvenile growth rates mirror the trends of the regularly monitored adults, overly conservative estimates of recruitment into the fishery might result, whereas assuming the converse could lead to a higher risk of overfishing. Results from our re-examination of the growth rates of juvenile J. lalandii provided some evidence that temporal trends in their growth rates are consistent with those of the adults of the species. Specifically, juvenile molt increments at the Cape Town harbor wall were smaller in 516 Fishery Bulletin 100(3) Table 5 Results of log-linear analysis investigating the mterdependcnce of the factors s te (S), size -category (SO, sample month (M) and molt state (MS) wi h regard to the frequency of observations in each subcategory, x' statistics are reported for conventional tests | of observed versus expected freq uencies and have been calculated for model fit s with different numbers of levels of interactions | as well as for partial and marginal associations among combinations of factors. In the latter sense. partial associations take into account the effect of all factors at each given evel, whereas marginal associations take into account only those factors listed | (Norman and Streiner, 1994). Tests of hypotheses that all A'-level interactions are simultaneously zero. Level (A') df x' P 1 13 1833.48 <0.01 2 51 626.90 <0.01 3 67 168.45 <0.01 4 28 39.51 0.07 Tests of marginal and partial association Factor df Partial association Margina 1 association X' P X- P Site (S) 1 6.60 0.01 Size-category (SO 4 500.95 <0.01 Sample month (M) 7 42.78 <0.01 Molt state (MS) 1 1198.48 <0.01 SxSC 4 142.77 <0.01 125.72 <0.01 SxM 7 23.89 <0.01 21.05 <0.01 SxMS 1 52.47 <0.01 35.73 <0.01 SCxM 28 320.55 <0.01 307.70 <0.01 SCxMS 4 68.53 <0.01 41.78 <0.01 MxMS 7 68.24 <0.01 55.70 <0.01 SxSCxM 28 121.80 <0.01 129.12 <0.01 SxSCxMS 4 7.99 0.09 4.91 0.30 SxMxMS 7 16.04 0.03 11.51 0.12 SCxMx MS 28 28.66 0.43 32.55 0.25 1996-97 than those recorded in 1971-72 (Pollock, 1973), irrespective of size. Although corresponding measures of intermolt period are not available, established trends in juvenile rock lobster biology indicate that such data would probably not have exhibited conflicting trends. In general, their growth rates tend to respond first by changes in molt frequency and then, under more extreme conditions, by changes in molt increments (Chittleborough, 1975; Serfling and Ford, 1975; Philhps et al., 1977; Dennis et al., 1997). Therefore, the trend to smaller growth increments among juvenile J. lalandii at the Cape Town harbor wall between 1971-72 and 1996-97 was most plausibly accompanied by an increase in intermolt period. Adding credence to this deduction are Cockcroft and Goosen's (1995) results, which demonstrated that the size at which female J. lalandii reach sexual maturity on a range of fishing grounds had decreased significantly over the two decades leading up to the mid-1990s. Using simulation models, Pollock (1987) demonstrated that reduced juvenile growth rates would lead to a smaller size at sexual maturity. In combination, these conclusions provided additional rationale for rejecting the hypothesis that somatic growth rates of juvenile J. lalandii on the harbor wall have not declined since the last study. Pollock (1973) found that a modal size of 38 mm CL was attained approximately 1.6 years after settling. This alone implies a slower growth rate than that for other temperate lobster species (Jernakoff et al., 1994). Because our data suggest that current growth rates are even slower than this, and that they may decline further with falling adult growth rates, recruitment to the fishable part of the J. lalandii resource must be seen as a real management concern (Bergh and Johnston, 1992). Therefore, the precautionary assumption made in the operational management pro- cedure designed for this resource (Cockcroft and Payne, 1999; Pollock et al., 2000) that juvenile growth mirrors adult growth is justified. Another feature of trends in growth rates shared be- tween adult and juvenile J. lalandii is small-scale spatial variation. During the monitoring period, juveniles of this species grew relatively faster at Mouille Point than at the harbor wall, both in terms of molt increment Hazell et al.: Somatic growth rate of Jasus lalandii 517 and intermolt period. Amongst adults, this variability has been attributed to patterns of food availability and gradients in environmental characteristics, particularly oxygen content (Newman and Pollocii, 1974; Pollock and Beyers, 1981; Pollock, 1982; Pollock and Shannon, 1987; Pollock et al., 1997; Mayfield, 1998). Although there is some spatial separation of juvenile and adult J. lalandii, it is likely that the biotic structure of their environment influences juvenile growth rates in a manner similar to that of adults. In a study of the diets of juvenile lobsters at the Cape Town harbor wall and Mouille Point, Mayfield (1998) found that diets were similar at the two sites, but that the benthic communities differed. He concluded that lobsters on the harbor wall would have to spend more time and energy feeding to maintain a diet similar to those at Mouille Point, where more favored food species (e.g. mussels, barnacles) were better represented. Similar patterns were evident when comparing diets of adults at Olifantsbos and Dassen Island (Mayfield, 1998), a slow- and fast-growth site, respectively (Cockcroft and Goosen, 1995). Subsequently, a more detailed study of the diets of two size classes (<35 mm CL and 40-60 mm CD of juvenile J. lalandii at the Cape Town harbor wall and at Mouille Point was able to detect a significant difference between the diets of inhabitants of the two sites (Mayfield et al., 2000). Juveniles on the harbor wall consumed significantly less black mussel, their preferred prey species. This result, however, does not contradict the suggested link between food distribution and juvenile growth rates; specimens forced to subsist off suboptimal food resources could be expected to grow more slowly than those at liberty to pursue their preferred prey. The differences in growth rates between juvenile J. lalandii at the artificial harbor wall and those at the natural nursery reef at Mouille Point do not necessarily preclude the use of the harbor wall as a legitimate site to continue monitoring juvenile growth. The existence of historical data (albeit for only two seasons) provides a good reference point, and the easy and consistent sampling conditions permit reliable data collection. These sampling conditions are of particular importance, given the established tendencies of early juveniles of other species of rock lobster on natural reefs to be less vulnerable to sampling than later stages (Annala and Bycroft, 1985; Marx and Herrnkind, 1985; Breen and Booth, 1989; Jernakoff, 1990; Childress and Herrnkind, 1994; Forcucci et al., 1994; Jernakoff et al., 1994). Furthermore, knowledge of how growth rates change over time is in itself important to the management of the fishery (Cockcroft, 1997). Should spatial and temporal variability in juvenile growth continue to be a contentious issue, a juvenile tag- ging program would provide simpler and more rigorous comparisons. It might also provide reliable estimates of the intermolt period. Mitigating against this, however, are the low rate of recapture of tagged specimens in the wild and potential tag-related growth retardation and mortal- ity. Nevertheless, continued monitoring of juvenile growth may elucidate how close this link is between adult and juvenile growth, thereby improving our ability to manage this valuable resource. Acknowledgments Numerous people assisted on sampling trips and in mea- suring molt increments in the aquarium, but especially Steven McCue, Marico Vercuiel, Eric Simpson, and Danie van Zyl. UCT Zoology Department provided us with the aquarium space that allowed us to collect molt increment data. Many thanks also go to Dave Pollock, Andy Cock- croft, and Retha Hofmeyer, who provided useful comments in finalizing this manuscript. Literature cited Annala, J. H., and B. L. Bycroft. 1985. Growth rate of juvenile rock lobsters {Jasus edward- sii) at Stewart Island, New Zealand. NZ J. Mar. Freshwa- ter Res. 19:445^55. Bergh, M. O., and S. J. Johnston. 1992. A size-structured model for renewable resource man- agement, with application to resources of rock lobster in the south-east Atlantic. S. Afr J. Mar Sci. 12:1005-1016. Breen, P. A., and J. D. Booth. 1989. Puerulus and juvenile abundance in the rock lobster Jasus edwardsii at Stewart Island, New Zealand (Note). NZ J. Mar. Freshwater Res. 23:519-523. Childress, M. J,, and W. F. Herrnkind. 1994. The behavior of juvenile Caribbean spiny lobster in Florida Bay: seasonality, ontogeny and sociality. Bull. Mar Sci. 54:819-827. Chittleborough, R. G. 1975. Environmental factors affecting growth and sur- vival of juvenile western rock lobsters Panulirus longipes (Milne-Edwards). Aust. J. Mar Freshwater Res. 26:177- 196. Cockcroft, A. C. 1997. Biochemical composition as a growth predictor in male west-coast rock lobster {Jasus lalandii I. Mar Fresh- water Res. 48:845-856. Cockcroft. A. C, and P. C. Goosen. 1995. Shrinkage at moulting in the rock lobster Jasus lal- andii and associated changes in reproductive parameters. S. Afr. J Mar Sci. 16:195-203. Cockcroft, A. C. and A. I. L. Payne. 1999. A cautious fisheries management in South Africa: the fisheries for rock lobster Marine Policy 23(6):587-600. Cruywagen, G. C. 1997. The use of generalized linear modelling to determine inter-annual and inter-area variation of growth rates: the Cape rock lobster as example. Fish. Res. 29:119-131. Dennis, D. M.,T D. Skewes, and C. R. Pitcher 1997. Habitat use and growth of juvenile ornate rock lob- sters, Panulirus ornatus (Fabricius, 1798), in Torres Strait, Australia. Mar Freshwater Res. 48:663-670. Forcucci, D., M. J. Butler, and J. H. Hunt. 1994. Population dynamics of juvenile Caribbean spiny lob- ster, Panulirus argus, in Florida Bay, Florida. Bull. Mar Sci. 54:805-818. Goosen, P. C, and A. C. Cockcroft. 1995. Mean annual growth increments for male west coast 518 Fishery Bulletin 100(3) rock lobster Jasus lalandii. 1969-1993. S. Afr. J. Mar Sci. 16:377-386. Herrnkind, W. F., P. Jernakoff, and M. J. Butler 1994. Puerulus and post-puerulus ecology. In Spiny lob- ster management (B. F. Phillips, J. S. Cobb, and J. Kittaka eds. I, p. 213-229. Blackwell Scientific, London. Jernakoff, P. 1990. Distribution of newly settled western rock lobsters Panulirus cygnus. Mar Ecol. Prog. Ser. 66:63-74. Jernakoff, P. J. J. Fitzpatnck, B. F Phillips, and E. De Boer 1994. Density and growth in populations of juvenile west- ern rock lobsters, Panulirus cygnus (George). Aust. J. Mar Freshwater Res. 45:69-81. Marx, J. M., and W. F. Herrnkind. 1985. Macroalgae (Rhodophyta: Laurencia spp. ) as habitat for young juvenile spiny lobsters, Panulirus argus. Bull. Mar Sci. 36:423-431. Mayfield, S. 1998. Assessment of predation by the West Coast rock lob- ster (Jasus lalandii ): relationships among growth rate, diet and benthic community composition, with implications for the survival of juvenile abalone (Haliotis midae). Ph.D. diss., 213 p. Univ. Cape Town, Cape Town, South Africa. Mayfield, S., L. A. Atkinson, G. M. Branch, and A. C. Cockcroft. 2000. Variation in the diet of the West Coast Rock Lobster, Jasus lalandii: influence of rock-lobster size, sex, capture depth, latitude and moult stage. S. Afr. J. Mar Sci. 22: 57-69. Melville-Smith, R., P C. Goosen, and T. J. Stewart. 1995. The spiny lobster Jasus lalandii iH. Milne Edwards, 1837) off the South African coast: inter-annual variations in male growth and female fecundity. Crustaceana 68: 174-183. Newman, G. G., and D. E. Pollock. 1974. Growth of the rock lobster Josi/s lalandii and its rela- tionship to benthos. Mar Biol. 24:339-346. Norman, G. R., and D. L. Streiner 1994. Biostatistics the bare essentials. 260 p. Mosby, Chi- cago, IL. Phillips, B. F, N. A. Campbell, and W. A. Rea. 1977. Laboratory growth of early juveniles of the western rock lobster Panulirus longipes cygnus. Mar Biol. 39: 31-39. Pollock, D. E. 1973. Growth of juvenile rock lobster Jasus lalandii. In- vestl. Rep. Div Sea Fish. S. Afr 106:1-16. 1982. The fishery for and population dynamics of West Coast rock lobster related to the environment in the Lam- bert's Bay and Port Nolloth areas. Investl. Rep. Sea Fish. Inst. S. Afr 124:1-57. 1986. Review of the fishery for and biology of the Cape Rock Lobster Jasus lalandii with notes on larval recruitment. Can. J. Fish. Aquat. Sci. 43:2107-2117. 1987. Simulation models of rock lobster populations from areas of widely divergent yields on the Cape West Coast. S. Afr. J. Mar Sci. 5:531-545. 1995. Changes in maturation ages and sizes in crustacean and fish populations. S. Afr. J. Mar Sci. 15:99-103. 1997. Egg production and life-history strategies in some clawed and spiny lobster populations. Bull. Mar. Sci. 61: 97-109. Pollock, D. E., and C. J. De B. Beyers. 1981. Environment, distribution and growth rates of West Coast rock-lobster Jasus lalandii (H. Milne Edwards). Trans. R. Soc. S. Afr 44:379-400. Pollock, D. E., A. C. Cockcroft, and R C. Goosen. 1997. A note on reduced rock lobster growth rates and related environmental anomalies in the southern Ben- guela, 1988-1995. S. Afr J. Mar Sci. 18:287-293. Pollock. D. E., A. C. Cockcroft, J. C. Groeneveld. and D. S. Schoeman. 2000. The Commercial fisheries for Jasus and Palinurus species in the south-east Atlantic and south-west Indian Oceans. In Spiny lobsters: fisheries and culture (B. F. Phillips and J. Kittaka, eds.), p. 105-120. Blackwell Sci- entific Publications, Oxford. Pollock, D. E., and L. V. Shannon. 1987. Response of rock-lobster populations in the Benguela ecosystem to environmental change — a hypothesis. S. Afr. J. Mar Sci. 5:887-899. Serfling, S. A., and R. R Rord. 1975. Laboratory culture of juvenile stages of the California spiny lobster Panulirus mterruptus (Randall) at elevated temperatures. Aquaculture 6:377-387. SPSS Inc. 1997. SigmaStat for Windows, version 2.03. SPSS Inc., Chicago, IL. Stokes, K. and R. Law. 2000. Fishing as an evolutionary force. Mar Ecol. Prog. Ser 208:307-309. Zar, J. H. 1984. Biostatistical analysis (2"'* ed.), 718 p. Prentice-Hall, Inc. Englewood Cliffs, NJ. Zoutendyk, P. 1988. Consumption rates of captive Cape Rock lobster Jasus lalandii. S. Afr J. Mar Sci. 6:267-271. 519 Abstract— Stock-rebuildinR time iso- |)li'ths relate constant levels of fishing; mortality iF), stock biomass, and man- agement goals to rebuilding times for overfished stocks. We used simulation models with uncertainty about Fj^f^y and variability in annual intrinsic growth rates (r^ ) to calculate rebuilding time isopleths for Georges Bank yellow- tail flounder, Limanda ferruginea, and cowcod roekfish. Sebastes levis, in the Southern California Bight. Stock-re- building time distributions from sto- chastic models were variable and right- skewed, indicating that rebuilding may take less or substantially more time than expected. The probability of long rebuilding times increased with lower biomass, higher F, uncertainty about F^igy. and autocorrelation in r^ values. Uncertainty about Fysi- had the great- est effect on rebuilding times. Median recovery times from simulations were insensitive to model assumptions about uncertainty and variability, suggesting that median recovery times should be considered in rebuilding plans. Iso- pleths calculated in previous studies by deterministic models approximate median, rather than mean, rebuild- ing times. Stochastic models allow managers to specify and evaluate the risk (measured as a probability) of not achieving a rebuilding goal according to schedule. Rebuilding time isopleths can be used for stocks with a range of life histories and can be based on any type of population dynamics model. They are directly applicable with con- stant F rebuilding plans but are also useful in other cases. We used new algorithms for simulating autocor- related process errors from a gamma distribution and evaluated sensitivity to statistical distributions assumed for r^,. Uncertainty about current biomass and fishing mortality rates can be con- sidered with rebuilding time isopleths in evaluating and designing constant-F rebuilding plans. Stock-rebuilding time isopleths and constant-F stock-rebuilding plans for overfished stocks Larry D. Jacobson Steven X. Cadrin Northeast Fisheries Science Center National Marine Fishenes Service 166 Water Street Woods Hole, MA 02543 Email address (lor L D Jacobson) Larry JacobsomgiNOAA gov Manuscript accepted 12 Febraury 2002. Fish. Bull. 100:519-536 (2002). Stock-rebuilding plans proposed for overfished stocks are best evaluated by stock-specific simulation analysis (e.g. PFMC^). However, general approaches are also valuable because many stocks are overfished (NMFS, 1999) and default or generic rebuilding plans can be used without extensive analyses for each species (e.g. PFMC'^; Applegate et al.'^). In this article we show how stock- rebuilding time isopleths can be used to design, evaluate, and monitor prog- ress of "constant F" and other types of rebuilding plans. Constant-F stock- rebuilding plans maintain fishing mor- tality at a fixed level until the stock is rebuilt, and are relatively simple and easy to analyze. The isopleth approach is easy to use as both a general and stock-specific tool. The U.S. Sustainable Fisheries Act (SFA) mandates rebuilding plans for overfished stocks (DOC, 1996, 1998). Federally managed stocks are consid- ered overfished when stock biomass is less than the biomass threshold (fiT-zir,.,/,. gij) defined in the Fishery Manage- ment Plan (FMP). National Standard 1 (DOC, 1998) for the SFA indicates that fi7-/,res/,o/rf should be the greater of one-half of fi,vf.sy *'-he theoretical biomass level for maximum sustained yield, MSY) or the minimum biomass from which rebuilding to B^j^y could be expected to occur within ten years if the stock is exploited at Frhreshoid- Typically, Fj,^,„;,„w = ^w.^j- (the theo- retical fishing mortality rate for MSY) when current biomass is at or above ^Threshold' and F7.^„,,;,„w < P'msy at lower biomass levels. A common approach (Fig. 1, and Thompson, 1999) reduces ^Threshold from the F^i^Y level linearly to zero as biomass declines from B7'/,„.,/,oW- In cases where Bg^gy, and F^^gy can not be estimated, reasonable proxy values (e.g. one-half unfished biomass or F^ ,) are typically used instead. The goal for most rebuilding plans under the SFA is to achieve the target biomass level iB^jgy or an acceptable proxy level) in ten years or less. Even with zero fishing mortality, ten years may not be sufficient to rebuild some overfished stocks. In such cases, the Guidelines for National Standard 1 al- low a rebuilding time period no longer than one mean generation time (Re- strepo et al., 1998) plus the expected time to recovery in the absence of fish- ing mortality (DOC, 1998). ' PFMC (Pacific Fishery Management Coun- cil). 1999. The coastal pelagic species fishery management plan, Amendment 8, 405 p. Pacific Fishery Management Coun- cil, 7700 NE Ambassador Place, Portland, OR, 97220-1384. - PFMC (Pacific Fishery Management Coun- cil). 1999. Status "of the Pacific Coast groundfish fishery through 1999 and rec- ommended acceptable biological catch for 2000 stock assessment and fishery evalua- tion, 230 p. Pacific Fishery Management Council, 7700 NE Ambassador Place, Port- land, OR, 97220-1384. 3 Applegate, A., S. Cadrin, J. Hoenig, C. Moore, S. Murawski, and E. Pikitch. 1998. Evaluation of existing overfishing definitions and recommendations for new overfishing definitions to comply with the Sustainable Fisheries Act, 179 p. New England Fishery Management Council, 50 Water Street, Mill 2, Newburyport, MA 01950. 520 Fishery Bulletin 100(3) 10-year Isopleth ■• • 5-year Isopleth ^ 4-year Isopleth ^— 3-year Isopleth ^^ 2-year Isopleth i^» Control Rule ^ VPA 0 4 0 6 Relative biomass (S/S^^jy.) Figure 1 Isopleths for median rebuilding times based on a deterministic logistic population growth model (model type 11 for Georges Bank yellowtail flounder with Ff^^^^^O.'i. Also shown are a common harvest control rule, and the biomass-F trajectory during 1996-99 for Georges Bank yellow- tail flounder. The harvest control rule specifies a maximum (threshold) F as a function of stock biomass level. The biomass-F trajectory shows a time series of F and biomass estimates from virtual population analysis (VPA, Cadrin'"'!. Stock-rebuilding time isopleths Catirin ( 1999 ) calculated theoretical recovery times for Georges Bank yellowtail flounder (Limanda ferruginea ) and used rebuilding time isopleths to depict trends in stock biomass in relation to fishing mortality. Calculations were based on a deterministic logistic population growth model with a range of constant annual fishing mortality rates (F=zero to Fy^y) and a range of initial biomass levels less than the target level (fi,|=zero to Bj^jgy). Recovery time was the number of years required for stock biomass to increase from an initial overfished biomass level 'fioFg (above or to the left of the isopleth) would rebuild the stock later. Rebuilding time isopleths were used to develop overfish- ing definition options for nine overfished New England groundfish stocks (Applegate et al.-*). In this article, we calculate stock-rebuilding time iso- pleths based on stochastic population dynamics models and characterize statistical distributions (mean, median, and percentiles) of stock-rebuilding times under different assumptions about uncertainty and process error (pro- cess errors are uncertainty in population dynamics due to natural variability in growth, recruitment, and other biological factors, Hilborn and Walters, 1992). Like Cadrin (1999), we use logistic population growth models, but our analysis includes uncertainty about F\,,,.y and autocorre- lated process errors in production. We analyze rebuilding times for two stocks (cowcod rockfish, Sebastes levis, and Georges Bank yellowtail flounder) with different life histo- ries, levels of Fycjy, and autocorrelation in production pro- cess errors (the calculations are examples only and not for use by managers). We also describe how stock-rebuilding time isopleths from deterministic and stochastic models can be used to develop and evaluate rebuilding plans and to monitor their progi-ess. Materials and methods Following Prager ( 1994), we used the continuous time ver- sion of the logistic population dynamic model in simula- tion calculations.^ In particular, for the logistic population growth parameter r^, (subscripted to represent the value in yeary) carrying capacity K, b = r IK, and a = r -F^O: B„ So. a+bRJe"' -1) (1) SAS simulation program code available from the senior author Jacobson and Cadrin Stock-rebuilding time isopleths and constant f stock-rebuilding plans for overfished stocks 521 Table 1 Six types of simulation models used to estimate rebuilding time isopleths. of uncertainty about F^sy- a"d variance and autocorrelation in production Model types include all of the meaningful combinations process errors. Model type CV for uncertainty Variance in process errors (%) Autocorrelation in process errors (p) Description 1 Zero 0 zero No uncertainty about F^sy; deterministic production 2 20 0 zero Uncertainty about F^i,.j-; deterministic production 3 Zero from stock assessment zero No uncertainty about /^vr.s}- stochastic production; no autocorrelation 4 20 from stock assessment zero Uncertainty about Fj^f^y'^ stochastic production; no autocorrelation 5 Zero from stock assessment from stock assessment No uncertainty about F^^^y, stochastic production; with autocorrelation 6 20 From stock assessment from stock assessment Uncertainty about F^.,,,; stochastic production; with autocorrelation When fishing and the intrinsic rate of increase exactly bal- ance fa^= r^-F^.=0): ^v+l = '"' l + b.B, (2) Note that biomass decHnes lB^,^j estim ited with fixed y = -M by maximum likelihood According to Cadrin,'' the carrying capacity is | A:=99,000 metric tons (t) (95% confidence interval 84,500-103,000 t). Parameter ^■=84,500 t A:=99,400 t K-- = 103,000 t Number B^ values sga-Zr K 25 25 25 Sample mean 0.58 0.65 0..58 Sample variance 0.036 0.037 0.036 Autocorrelation p 0.34 0.33 0.34 a 15.7 17.5 15.7 n 0.0498 0.0488 0.0499 y=-M -0.2 -0.2 -0.2 Mean 0.58 0.65 0.,58 Variance 0.0.39 0.039 0.039 Mode 0.53 0.61 0.53 r = G„ + R-M (7) where G^ = the instantaneous rate of somatic gi'owth; and R^ = an instantaneous rate for recruitment (in units of biomass). Somatic growth and recruitment may be density depen- dent but are usually positive (G,>0 and i?,>0). Thus, r^=-M is possible in the extreme case of zero growth and zero recruitment. Production process errors were simulated by drawing random numbers from a three-parameter gamma prob- ability distribution (Johnson et al., 1994, Appendix 1). Runs with autocorrelated process errors used one of two algorithms based on gamma distributions with adjusted parameter estimates (Appendix 2). assumed M=0.2 /yr), relatively productive (r=0.58-0.65) stock with some autocorrelation (p =0.33-0.34) in pro- duction process errors (Table 2). Empirical, and gamma distributions fit by maximum likelihood and the method of moments had similar means and variances (Table 2). Surplus production and biomass are related for Georges Bank yellowtail flounder, with P^ reduced at the low B^, levels (Fig. 2A). Variability in estimated r^, values indicate autocorrelation in process errors (Fig. 2B). The distribu- tion of r^, values (Fig. 2C) was skewed to the left and there were no negative values. Gamma distributions fitted by maximum likelihood and the method of moments (Appen- dix 1) were similar in shape (Fig. 2C). In simulations for yellowtail flounder, we used F^^^.j^0.30 (from ASPIC) with CT;^ = 0.037andp = 0.33. Cowcod rockfish Georges Bank yellowtail flounder Cadrin'' used virtual population analysis (VPA, cali- brated by using survey data) to estimate stock biomass for Georges Bank yellowtail flounder during 1973-98. In the same assessment, a surplus production model (ASPIC [stock-production model incorporating covariates], Prager, 1994) was used to estimate 7^=93,700 metric tons (t) (80'7r bootstrap confidence interval 87,700-97,000 t) and F,^^^.5— 0.30/yr (80% bootstrap confidence interval 0.27- 0.32/yr'). Cadrin's stock assessment^ and our production calcula- tions indicate that Georges Bank yellowtail flounder is a moderately long-lived (maximum observed age 14 yr. Cadrin, S. X. 2000. Georges Bank yellowtail flounder In Northern demersal working group: assessment of 11 northeast groundfish stocks through 1999. In Northeast Fisheries Science Center Reference Document 00-05, p. 45-64. Northeast Fisher- ies Science Center 166 Water Street, Woods Hole, MA, 02543. Butler et &\.^ (see also Butler et al.') estimated A:=3400 t (95% CI 2800-4000 t) with a delay-difference biomass dynamic model for cowcod rockfish in the Southern Cali- fornia Bight. Annual biomass estimates from the same source were used to calculate surplus production during 1951-97 when the stock was fished down from about 3200 t to 240 t (about 1% of virgin biomass). Butler et al.^ and our production calculations indicate that cowcod rockfish are a long-lived (maximum observed ^ Butler, J. L., L. D. Jacobson, J. T. Barnes, and H. G. Moser 2002. Manuscript in revew. Biology and population dynam- ics of cowcod rockfish iSebastes leuis) in the southern California Bight. ' Butler, J. L., L. D. Jacobson, J. T. Barnes, H. G. Moser, and R. Collins. 1999. Stock assessment of cowcod. In Appendix to the state of the Pacific Coast groundfish fishery through 1999 and recommended acceptable biological catch for 2000 stock assessment and fishery evaluation, p. Vi-113 (section 5). Pacific Fishery Management Council, 7700 NE Ambassa- dor Place, Portland, OR, 97220-1384. 524 Fishery Bulletin 100(3) 5. 8000 - W 4000 - 0 10000 20000 30000 40000 Stock biomass (t) 0 20 ■ I I ' I ' r I 1970 1975 1980 1985 1990 1995 2000 Year — •— Observed O Normal -▼- LognormalJ Gamma 015 - 0 00 -0 25 0 00 0 25 0 50 0 75 1 GO 1 25 Intrinsic population growth rate {rj Figure 2 (A) Surplus production estimates P^, and biomass estimates B^, for Georges Bank yellowtail flounder during 1973-97, The biomass estimates and parabolic curve fitted to production estimates are from Cadrin.'' (B) Time series of annual intrin- sic rate of growth (r^, I parameter value estimates from Equation 6. (C) Probability distributions for r^, values used in simula- tions for Georges Bank yellowtail flounder. age 55 yr, assumed M=0.055/yr), relatively unproductive stock (r^,=0. 027-0. 037). Cowcod rockfish are much less pro- ductive than Georges Bank yellowtail flounder because of their long lives and slower growth and because adult habitat is limited to steep rocky areas in relatively deep water (90-500 m, Butler et al.^). Production process errors show a high level of autocorrelation (p=0.83-0.94. Table 3). Means, variances and autocorrelations for r^, were not very sensitive to assumptions about 7<^ (Table 3). There was 1000 2000 3000 4000 Stock biomass (t) 1950 1960 1970 1980 1990 2000 Year -•— Observed O Normal ▼- Lognormal — Gamma -0 02 0 00 0 02 0 04 0 06 0 08 0 10 Intrinsic population growth rate {r^) Figure 3 (A) Surplus production estimates P^ and biomass estimates B^, from Butler et al.'' for cowcod rockfish during 1951-97 in the Southern California Bight. The parabolic curve fitted to production estimates (P^,) shows trends only and is not recommended for management purposes. (B) Time series of annual intrinsic rate of growth (r^,) parameter value estimates from Equation 6. (C) Probability distributions for r^, values used in simulations for cowcod rockfish. no clear relationship between surplus production and bio- mass, but Pj, was lowest at the highest and lowest B^, lev- els and autocorrelation in production process errors was obvious (Fig. 3A). The distribution of r^, values (Fig. 3B) was skewed to the right and there were no negative val- ues. Gamma distributions fitted by maximum likelihood and the method of moments (Appendix 1) were similar in shape (Fig. 3). Jacobson and Cadrin: Stock-rebuilding time isopleths and constant Fstocl0.05 A', as long as fishing morality rates are less than 0.2/yr. Seventy-five year stock-rebuilding time isopleths for cow- cod rockfish were sensitive to assumptions about the distri- bution of production process errors but 10-year isopleths for yellowtail flounder were not (Figs. 10-11). We hypothesize that differences among statistical distributions r^, assumed in simulations were magnified for cowcod by long (e.g. 75 yr) rebuilding times (see "Discussion" section). Rebuilding isopleths for cowcod based on i\, values from a gamma distribution had higher F, at a given biomass, than rebuilding isopleths based on the distribution of ob- served !\, values (Fig. 11). In other words, results based on the gamma distribution suggest a more productive cowcod stock, presumably because the distribution of r^, values for cowcod had more mass than the gamma distribution over low r^ values (<0.03/yr, Fig. 3). Discussion Managers should consider using median rebuilding time goals, in addition to mean or other quantiles, in develop- Jacobson and Cadrin: Stock-rebuilding time isopleths and constant-f stock-rebuilding plans for overfished stocks 527 0 60 120 180 240 300 1.0 0.5 0.0 1 " ^M Model Type 2 i"~^ ^m Model Type 4 ^^ Model Type 5 ^ Model Type 6 1.0 0.5 0.0 1.0 0.5 0.0 0 60 120 180 240 300 Rebuilding time (yr) Figure 5 Distribution of simulated recovery times for coweod rockfish in the Southern CaUfornia Bight from six types of logistic population growth models ( type- 1 deterministic, others stochastic, see Table 1 ) with F^js,.)~0. 0.018 /yr, CV for uncertainty in F^^g^— zero or 20%, variance for production process errors zero or 0.037, and autocorrelation in process errors p=zero or 0.9. Results are for 2000 model runs starting from an initial biomass ot B/ B^f^y =0.19 and constant F=0.011/yr. ing and evaluating rebuilding plans. This recommenda- tion is based on a narrow technical consideration, i.e. that median recovery time calculations are less sensitive to model assumptions. A median rebuilding time plan is risk neutral in the sense that the probability of rebuilding times less than intended is the same as the probability of rebuilding times longer than intended (i.e. both 50%). Of course, rebuilding plans based on median rebuilding time goals will be more liberal in terms of short-term catch (i.e. have higher F) and have longer rebuilding times on average than plans based on mean or, for example, Qggcr, rebuilding time goals. In summary, a manager who is willing to accept a 50% chance of rebuilding times greater that desired and who is concerned about model uncertainties, might choose a median rebuilding time approach. A potential advantage in using percentiles other than the median is that managers can specify risk levels in try- ing to rebuild stocks. For example, managers could choose and evaluate rebuilding plans based on a Qg^cy, isopleth to insure at least a 90% chance of rebuilding in specified time period. However, Qg^,- isopleths may be sensitive to model assumptions. Approaches to using rebuilding time isopleths without relying on uncertain estimates of rebuilding time distribu- tions (e.g. mean or Qgg,- ), are an important area for future research. Cadrin (1999) used deterministic rebuilding time isopleths with the tenth percentile of the estimate for r. Isopleths for median rebuilding time might be used with lower bounds on confidence intervals for Bg and Bg/K or upper bounds on current F, F^gy, or F/F^jgy Estimates of uncertainty in these parameters are often available (Prager, 1994) and can be incorporated in an ad hoc fash- ion. For example Butler et al.^ suggest that Bj^^glK for coweod rockfish is 7% (CV about 30%) and that the F in 1998 was 0.085/yr (CV 34% ). A risk-averse manager might implement a rebuilding plan that reduces that Fjf^^^^f^^,^ to a point on the 75-year rebuilding time isopleth that lies above the lower boundary of a 95% confidence interval for biomass. Similarly, a risk-averse manager might select a rebuilding plan that reduces ^T-^r.s/io/rf '-° account for the upper bound on uncertainty in estimating F. Uncertainty in Fj^jgy is an important factor to consider in rebuilding plans as F increases from low levels towards Fj^^gy Expected rebuilding times increase at high F be- cause F may exceed true F^fgy (assumed known but with 528 Fishery Bulletin 100(3) o CO d ^ o CM d /^''''■^''^'^'^' o 1/ T— If o f o o Mean ^ o o ^ ,- f - ^ " "■'' o / ^ ' " ^"^ CM 1 , - .^ O / ' '. ' O /' >-'.^^'-'' /' >r.-- O //' / ^ t, //' o o // Q99% Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Q10% 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Relative biomass (Bo/K) Figure 6 Isopleths for mean, median, mode, Qm,,, Qgo-j. ^nci Qggr,, ten-year stock- rebuilding time.s simulated with six model types (Table 1) for Georges Bank yellowtail flounder Isopleths for all statistics and one type of model are shown in each panel. Model types 2, 4, and 6 include uncer- tainty in F,,,.,.. error) in a high proportion of cases (Figs. 8-9). Fortunately, process errors and autocorrelation may reduce this prob- lem because stock growth rates increase in some years so that the true Fj^^gy exceeds the manager's estimate. Stock-rebuilding times calculated with deterministic models approximate median rebuilding times from sto- chastic models. For example, rebuilding times in Cadrin (1999) calculated with a deterministic model for Georges Bank yellowtail flounder are very close to median rebuild- ing times from our stochastic models. From our results, we hypothesize that rebuilding time isopleths for other species (Applegate et al.'^) based on Cadrin's ( 1999) deter- ministic model should also be viewed as approximations to isopleths for median rebuilding times. Our simulation analyses indicate that rebuilding times for overfished stocks (with a range of life history charac- teristics, initial biomass levels and fishing mortality rates) tend to be skewed and can be highly variable (Figs. 4-5). Hence, rebuilding in any specific case may be quicker or take much longer than expected, particularly if expecta- tions are based on deterministic models that approximate median rebuilding times. For example, probabilities of re- building times twice as long as the goal for Georges Bank yellowtail flounder (10 yr) and cowcod (75 yr) were 4% and 8'7( and probabilities of rebuilding times half as long were 40% and 1%. Modeling choices Stochastic models are necessary when estimates of mean rebuilding times or quantiles other than the deterministic approximation to the median are needed. Rebuilding time Jacobson and Cadrin: Stock-rebuilding lime isopleths and constants stock-rebuilding plans for overfished stocks 529 o o o o ^ y ^r / / y / / / y / ^ 1'' / Q99% IVIodel 1 Model 2 Model 3 Model 4 Model 5 Model 6 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Relative biomass (BiyK) Figure 7 Isopleths for mean, median, mode, Qior,, Q90.-. and Q99,. 75-year stock- rebuilding times simulated with six model types (Table 1) for cowcod rocktish in the Southern California Bight. Isopleths for all statistics and one type of model are shown in each panel. Model types 2, 4, and 6 include uncertainty in F^^^- isopleths from deterministic model type 1 were quite dif- ferent from isopleths (other than for median rebuilding times) from stochastic models. Rebuilding time isopleths can be used with any type of population dynamics model. We used the logistic popula- tion growth model in this paper because it is clearly linked to Sj^jsy and F^^y, incorporates density dependence, is easy to apply to a wide range of stocks (with varying amounts of information), and computationally efficient. However, re- building time isopleths could have been calculated by us- ing Cadrin's'' age-structured model for Georges Bank yel- lowtail flounder or Butler et al.'s'' biomass dynamic model for cowcod rockfish. Age-structured models might be best for calculating rebuilding time isopleths if the rebuilding time frame is relatively short and estimates of abundance are available for several incoming year classes because age-structured models account for transient conditions (e.g. recruitment and growth patterns) that are important in the short term. Age-based projections were used to test the expected performance of rebuilding targets for six New England groundfish stocks (NDWG**). The original rebuilding targets were derived from five-year rebuilding time iso- pleths calculated with a deterministic logistic growth models, but the targets incorporated estimation uncer- tainty by assuming the tenth percentile of the estimate 8 NDWG (Northern Demersal Working Group). 2000. Assess- ment of 11 northeast groundfish stocks through 1999. North- east Fisheries Science Center Reference Document 00-05, 175 p. Northeast Fisheries Science Center, 166 Water Street, Woods Hole, MA, 02543. 530 Fishery Bulletin 100(3) o CO o o CNJ o o Mean Q10% Median Q90% Q99% Mode Model Type 1 CO o Z^-''--' o CNJ f/^'' ' o 1 1 ' 1 / / o V' T— o I o o Model Type 3 CO • ->-^ ^ o ^'J^ — ^■^'^^ (-1 /Jr ^— " CSJ J y^ o I ^ '"' o \ 1 ' r- \l f o t o o Model Type 5 - " " — • •• ■ //""" Model Type 2 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Relative biomass (Bo/K) Figure 8 Isopleths for mean, median, mode, Qxir.^ Qgo';- ^""^ Q99", ten-year stock- rebuilding times simulated with six model types (Table 1) for Georges Bank yellowtail flounder. Isopleths for one statistic and all six model types are shown in each panel. Isopleths of mean and modal rebuilding times in results for models with uncertainty in Fv,j,.j. (model types 2, 4, and 6) may be distorted (flat) at relatively high fishing mortality levels because fishing mortality exceeds the simulated true F(,,,.j- in some simulations, so that the simulated stock may never rebuild. of r (Cadrin, 1999; Applegate et al.'). For the six stocks, starting biomass in 1999 ranged from 25% to 93% of B^^-y- and estimates of F^gy ranged from 0.5 to 0.8. Estimated rebuilding times averaged 3.5 years (ranging from 1 to 7 yr) for 50% probability of attaining B^j,.;. (NOWG^). There- fore, the age-based simulations indicated that isopleths based on deterministic biomass dynamic models generally performed well for overfished New England groundfish stocks. Brodziak et al. (2001) analyzed stock-recruit data and concluded that "(1) time horizons for rebuilding will be uncertain, owing to recruitment variability, (2) some productive stocks (haddock, yellowtail flounder) have seri- al correlation in recruitment and this may either enhance or diminish chances for stock recovery." Thus, results with surplus production models, age-structured models, and stock-recruit analyses highlight the fundamental similarities between a wide range of modeling approaches (Sissenwine and Shepherd, 1987). The best choice of stochastic simulation model for devel- oping and evaluating rebuilding plans will depend on the situation. Sainsbury (1993) concluded that models incor- porating simple assumptions about population dynamics were more appropriate for evaluating performance of con- trol rules than models with more complex assumptions. PFMC used a simple model incorporating environmental effects on recruitment with useful results. However, Bell Jacobson and Cadnn: Stock rebuilding time isoplelhs and constant f stock rebuilding plans for overfished stocks 531 O O o o o CN o ra O S O O O Mean Q10% Median Q90% Q99% Mode Model Type 1 // / / / / Model Type 2 Model Type 3 / // y / '/ / ' // ^ ' ' ^ ^ Model Type 4 o o ,. _^^ ^ ' tdS^^ ^ ^ ' ' ■■■'yf^^^y 1— jr^''..'" o *r/ '^ o //'' ' t / / I/I o i'> Model Type 5 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Relative biomass [B^K) Figure 9 Isopleths for mean, median, mode, Q,orj, Qgo^j, and Q^^,,, 75-year stock- rebuilding times simulated with six model types (Table 1 ) for cowcod rock- fish in the Southern California Bight. Isopleths for one statistic and all six model types are shown in each panel. Model types 2, 4, and 6 include uncertainty in f^gj-. Isopleths of mean and modal rebuilding times in results for models with uncertainty in F_„>,.j. (model types 2, 4, and 6) may be distorted (flat) at relatively high fishing mortality levels because fish- ing mortality exceeds the simulated true F^^gy '" some simulations so that the simulated stock may never rebuild. and Stefansson^ and Patterson (1999J used more complex simulation models with success. Distributional assumptions Simulation analyses (Figs. 10-11) indicate that the choice of statistical distribution for simulating process errors in model parameters (e.g. r^) may be important, ^ Bell, E. D., and G. Stefansson. 1998. Performance of some harvest control rules. NAFO (North Atlantic Fisheries Orga- nization! SCR Doc. 98/7, 1-19. Northwest Atlantic Fisheries Organization, 2 Morris Drive, P. O. Box 638, Dartmouth, Nova Scotia, B2Y 3Y9, Canada. particularly when rebuilding times are long (e.g. those for cowcod rockfish) due to low stock productivity, low stock biomass, unproductive stock dynamics, or autocorrelation in process errors. The choice of statistical distributions for simulating r^, involves choosing between theoretical distri- butions supported by theory (e.g. autocorrelated gamma distribution with negative values bounded below at -M) or bootstrap distributions of observed values. The program- ming and work required to experiment with alternative distributions is not overwhelming and we recommend sensitivity analyses in cases where distributional assump- tions may be important. Theoretical distributions for stochastic parameters are flexible because many types of distributions are available, most can be modified to include autocorrelation, most can 532 Fishery Bulletin 100(3) o CO o O O d Mean [- Gamma Bootstrap " Normal — Lognormal Q10% CO d --z^ o ^ ^^'^'^ (M ^ ^^^^ O f y^ If O if d jf o o Q99% 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 Relative biomass (BtJK) Figure 10 Mean, median, mode, Qi,,,;. Qy,,.;. and Qj,^,, ten year stock-rebuilding time isopleths for Georges Bank yellowtail flounder, simulated with model type 3 (no uncertainty about F^j^y and uncorrelated r^ values). Stochastic r^, values were from a gamma distribution (same as Figs. 6 and 8), normal distribution, lognormal distribution, or bootstrap of observed r values. All distributions had the same mean and variance. be modified to include negative or extreme values not evident in short observed time series, and most can give the same mean, variance and autocorrelation levels as estimated from available data. However, as in the case of Georges Bank yellowtail flounder and cowcod, the shape of theoretical and observed distributions may not match closely (Figs. 2-3). In comparing theoretical and observed distributions for model parameters (e.g. Figs. 2-3), it is important to remember that most observed distributions are based on relatively few observations (Table 2). Furthermore, ob- served values may be autocorrelated (p=0.33 for Georges Bank yellowtail flounder and p=0.94 for cowcod rockfish. Tables 2-3). High levels of autocorrelation reduce the "effective" number of observations dramatically so that observed values may provide a poor estimate of the shape of their distribution (Bartlett, 1946; Bayley and Ham- mersley, 1946). For example. Equation 16 in Bayley and Hammersley, with estimated autocorrelations (lags of 1-13 years) for Georges Bank yellowtail flounder, gives an effective sample size n*=Yl (compared to ;!=25 r^, values). For cowcod rockfish (with autocorrelations for lags 1-17), n*=l\ (compared to n=47 r^, values). Thus, autocorrela- tion in r^, values may reduce the effective sample size and information used to estimated the shape, mean and vari- ance of statistical distributions for r^, values by about 32% for Georges Bank yellowtail flounder and 77% for cowcod rockfish. Jacobson and Cadrin: Stock rebuilding time isopleths and constant f stock rebuilding plans for overfished stocks 533 CM O O O o o o ^ ' ./^'^^ / ^' /<' //,' //' /'' //' Mean m C\J (D C ) (T) o ^^ Q. ^^^ y^'''' ■r— ro o Af' o F o Ol //' 1. o o 1/' Median CNJ O o o o o Gamma Bootstrap Normal LognormaL Q99% X / / X / ' //'' //' //' /// . //' . " ll' 1, Ih Q10% Q90% Mode 0.00 0.25 0.50 0.00 0.25 0.50 Relative biomass (Sq/K) Figure 11 Mean, median, mode, Qiq,-,, Qgo^-,. and Qggr-, 75-year stock-rebuilding time isopleths for cowcod rockfish, simulated with model type 3 (no uncer- tainty about Fysy and uncorrelated r^ values). Stochastic r^, values were from a gamma distribution (same as Figs. 7 and 9>. normal distribution, lognormal distribution or bootstrap of observed r values. All distribu- tions had the same mean and variance. Developing, monitoring, and evaluating stock-rebuilding programs Once the management goal, desired probability of achiev- ing the stock- rebuilding goal, and the time frame for rebuilding are identified (e.g.lO-yr median rebuilding time to a B^j^y target), the simplest way to use stock-rebuilding time isopleths in designing a rebuilding plan is to choose a constant-Fg level from the appropriate rebuilding time isopleth, based on a current estimate of Bq. Cadrin (1999) has provided an example of this approach. Stock-rebuilding time isopleths can be used to monitor the progress of any rebuilding plan although interpreta- tion is clearest with constant-F values (Cadrin, 1999). For example, the point defined by current biomass and F for Georges Bank yellowtail flounder in the second year (1997) of a hypothetical five-year rebuilding plan begin- ning in 1996 should lie near or within the 3-year rebuild- ing time isopleth (Fig. 1). If the point lies far outside the 3-year isopleth, then managers could be sure that the rebuilding plan was behind schedule. Evaluating harvest control rules as stock-rebuilding programs It may be necessary to evaluate harvest control rules that allow F to vary with biomass (e.g. the common harvest control rule in Fig. 1) as a rebuilding plan. Rebuilding isopleths provide guidance in this situation because they can be used to reject some harvest control rules based on a single necessary criterion. However, the test is weak because a harvest control rule that passes the test may or 534 Fishery Bulletin 100(3) may not be sufficient as a rebuilding plan. In this context, it is important to remember that rebuilding time isopleths are constructed based on the assumption of constant fish- ing mortality rates during the rebuilding program. More complicated rebuilding plans, that allow F to vary accord- ing to changes in biomass or other factors, are best evalu- ated by stock-specific simulations. The test is based on the notion that control rules that al- low F levels above the stock-rebuilding time isopleth for bio- mass levels above Bq are unlikely to rebuild the stock with desired probability in the desired time frame. Therefore, as a minimum requirement for meeting rebuilding time goals, harvest control rules used as rebuilding plans should lie on or under the corresponding rebuilding time isopleth for all biomass levels above Bq. Consider a hypothetical overfished stock for which there is a ten-year median rebuilding time goal. Assume that a harvest control rule proposed as a rebuilding plan has the typical shape (i.e. F increases or stays the same as biomass increases, as in Fig. 1). If the control rule lies above the isopleth for some critical biomass level between B^^ and B^/s-y then the rule will allow fishing mortality rates that are generally too high to meet manage- ment goals once biomass reaches the critical level. The example of Georges Bank yellowtail flounder Georges Bank yellowtail flounder (Fig. 1) can be used to illustrate how rebuilding time isopleths might have been used to evaluate stock-rebuilding plans for hypothetical implementation during 1996 and how stock-rebuilding time isopleths can be used to monitor progress in rebuild- ing overfished stocks. The discussion is hypothetical, however, because the examples evaluate management approaches that have not been used in rebuilding the stock. In reality, managers kept F for Georges Bank yel- lowtail flounder during 1996-99 nearly constant at a level well the below the 10-year isopleth, and the Georges Bank yellowtail flounder stock was almost rebuilt to the Bj^jgy target level in 1999 after only four years (Fig. 1). Based on this example, the hai-vest control rule in Figure 1 would have been marginal for use in a hypothetical medi- an ten-year stock-rebuilding plan for Georges Bank yellow- tail flounder starting in 1996, because the rule lies slightly above the 10-year isopleth for biomass levels of 47*7^ B^jgy. The harvest rule might have been rejected outright as a five-year rebuilding plan because the rule lies well above the 5-year rebuilding time isopleth. The rebuilding trajec- tory for Georges Bank yellowtail flounder (Fig. 1) shows that the five-year rebuilding plan, which began in 1996, was on schedule during 1996-98 because fishing mortality and estimated biomass were within the 5-year, 4-year, and 3-year isopleths during successive years. The Georges Bank yellowtail flounder example provides an important final lesson about uncertainty in actual re- building times, even if statistical distributions of potential rebuilding times are characterized accurately. During 1996-99, the stock was managed at a relatively constant F level that was well above the entire 4-year median rebuilding time isopleth (Fig. 1). Despite the relatively high F level, Georges Bank yellowtail flounder reached a biomass level near fi^gy in 1999, after four years. We attri- bute this fortunate chain of events to stochastic variation in process errors stemming primarily from recruitment and growth of the strong 1997 year class (Cadrin'^). Acknowledgments We thank P. Rago, D. Hart, M. Sissenwine (National Ma- rine Fisheries Service, Woods Hole, MA), M. Prager (National Marine Fisheries Service, Beaufort, NC), and three anonymous reviewers for information and advice. S. Murawski suggested sensitivity analyses. J. Boreman, S. Murawski, and F. Serchuk (National Marine Fisheries Sei-vice, Woods Hole, MA) made editorial suggestions. Literature cited Bartlett, M. S. 1946. On the theoretical specifications and sampling prop- erties of autocorrelated time series. J. R. Stat. Soc. Ser B (methodoI.)8;27-41. Bayley, G. V., and J. M. Hammersley. 1946. The "effective" number of independent observations in an autocorrelated time series. J. R. Stat. Soc. Ser B (methodol.) 8:184-197. Beauchamp, J., and J. Olson. 1973. Corrections for bias in regression estimates after log- arithmic transformation. Ecology (Washington DC) 54: 1403-1407. Beddington, J. R., and R. M. May. 1977. Harvesting natural populations in a randomly fluctu- ating environment. Science 197:463-465. Brodziak, J. K. T., W. J. Overholtz, and P. J. Rago. 2001. Does spawning stock affect recruitment of New Eng- land groundfish. Can. J. Fish. Aquat. Sci. 58: 306-318. Cadrin, S. X. 1999. A precautionary approach to fishery control rules based on surplus production modeling. In Proceedings of the fifth national NMFS stock assessment workshop (V. Restrepo, ed. ), p. 17-22. U.S. Dep. Commer . NOAA Tech. Memo. NMFS-F/SPO-40. Clark, \V. G. 1991. Groundfish exploitation rate based on life history parameters. Can. J. Fish. Aquat. Sci. 48:734-750. Cleveland, W. S., and S. J. Devlin. 1988. Locally weighted regression: an approach to regression analysis by local fitting. J. Am. Stat. Assoc. 83:596-610. Cleveland, W. S., S. J. Devlin, and E. Grosse. 1988. Regression by local fitting: methods, properties and computational algorithms. J. Econometrics 37:87-114. Dennis, B., and G. P. Patil. 1984. The gamma distribution and weighted multimodal gamma distributions as models of population abundance. Math. Biosci. 68:187-212. Deriso, R. B. 1982. Relationship of fishing mortality and growth and the level of ma.ximum sustainable yield. Can. J. Fish. Aquat. Sci. 39:1054-1058. DOC (Department of Commerce). 1996. Magnuson Stevens Fishery Conservation and Manage- ment Act as amended through October 11, 1996. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/SPO-23, 121 p. Jacobson and Cadrin: Stock rebuilding time isopleths and constant-f stock-rebuilding plans for overfished stocks 535 1998. Mapnuson Stovpns Act Provisions, National Standard Guidelines. Federal Register 63(841:24212-24237. Gov- ernment Printing Office. Washington D.C. Doubleday. W. G. 1976. Environmental fluctuations and fisheries manage- ment. Int. Comm. Northwest Atl. Fish. 1: 141-150. i:fron, B, 1982. The jackknil'e, the bootstrap and other resampling plans, 92 p. Society for Industrial and Applied Mathemat- ics. Philadelphia. PA. Getz. W.M. 1984. Production models for nonlinear stocha.stic age-struc- tured fisheries. Math. Biosci. 69:11-30. ( ileit, A. 1978. Optimal harvesting in continuous time with stochas- tic growth. Math. Biosci. 41:111-123. Hilborn, R., and C. Walters 1992. Quantitative fisheries stock assessment. Choice, dy- namics and uncertainty. 570 p. Chapman and Hall. New York. m'. Jacobson. L. D., J. A. A. De Oliveira. M. Barange. M. A. Cisneros-Mata, R. Felix-Uraga, J. R. Hunter. J.Y. Kim. Y. Matsuura, M. Niquen, C. Porteiro, B. Rothschild. R. P. Sanchez, R. Serra, A. Uriarte, and T. Wada. 2001. Surplus production, variability and climate change in the great sardine and anchovy fisheries. Can. J. Fish. Aquat.Sei. 58:1891-1903. Jacobson. L. D.. N. C. H. Lo. and J. T. Barnes. 1994. A biomass based assessment model for northern anchovy. Engraulis mordax. Fish. Bull. 92: 711-724. Johnson, N .L., S. Kotz, and N. Balakrishnan. 1994. Continuous univariate distributions, vol. 1. 719 p. John Wiley and Sons. Ltd., New York, NY. Ludwig D. 1981. Har\'esting strategies for a randomly fluctuating population. J. Cons. Int. Explor. Mer 39:168-174. Ludwig, D., C. J. Walters, J. Cooke. 1988. Comparison of two models and two estimation methods for catch and effort data. Nat. Res. Modeling: 2:457-498. May, R. M.. J. R. Beddington, J. W Horwood, and J. G. Shepherd. 1978. Exploiting natural populations in an uncertain world. Math. Biosci. 42:219-252. 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: 2404-2419. Nelson, C. R. 1973. Applied time series analysis for managerial fore- casting, 231 p. Holden-Day, Inc., San Francisco, CA. NMFS (National Marine Fisheries Service). 1999. Report on the status of U.S. living marine resources, 1999. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-F/ SPO-41,301p. Patterson, K. R. 1999. Evaluating uncertainty in harvest control law catches using Bayesian Markov chain Monte Carlo virtual populations analysis with adaptive rejection sampling and including structural uncertainty. Can. J. Fish. Aquat. Sci. 56:208-221. Prager, M. H. 1994. A suite of extensions to a nonequilibrium surplus-pro- duction model. Fish. Bull. 92:374-389. Restrepo, V. R., G. G. Thompson, R M. Mace, W. L. Gabnel, L. L. Low. A. D. MacCall. R. D. Methot, J.E. Powers. B. L. Taylor. R R. Wade, and J F Witzig. 1998. Technical guidance on the use of precautionary ap- proaches to implementing National Standard 1 of the Magnuson-Stevens Fishery Conservation and Manage- ment Act. U.S. Dep. Commer., NOAA Tech. Memo. NMFS- F/SPO-31,.54p. Sainsbury, K. 1993. The use of simulation to evaluate the performance of stock rebuilding strategies, including the use of reference points. In Defining overfishing-defining stock rebuilding (A. Rosenberg, ed.), p. 30-39. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/SPO-8. Schaefer, M. B. 1954. Some aspects of the dynamics of populations impor- tant to the management of the commercial marine fish- eries. Inter-Am. Trop. Tuna Comm. l(2):27-56. Shepherd, J. G., and J. W. Horwood. 1980. The sensitivity of exploited populations to environ- mental "noise", and the implications for management. J. Cons. Explor Mer 38:318-323. Sissenwine, M. P. 1977. The effect of random fluctuations on a hypothetical fishery. ICNAF (International Commission for the North- west Atlantic Fisheries) Sel. Pap. 2:137-144. Sissenwine, M. P., M. J. Fogarty, and W. J. Overholtz. 1988. Some fisheries management implications of recruit- ment variability. In Fish population dynamics (J. A. Gulland. ed. ), p. 129-152. John Wiley and Sons. Ltd, New York. NY. Sissenwine. M. P., and J. G. Shepherd. 1987. An alternate perspective on recruitment overfishing and biological reference points. Can. J. Fish. Aquat. Sci. 44:913-918. Thompson, G. G. 1999. Optimizing harvest control rules in the presence of natural variability and parameter uncertainty. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-F/SPO 40:124-143. Appendix 1 —Independent process errors We simulated independent production process errors using a three-parameter gamma probability distribution (Johnson et al., 1994). The gamma probability density function for independent (no autocorrelation) r^, values in one simulation (denoted as s) was P(/;,Ja,,/3„7) = (r. ,a, -1 -(r^.-yllp. p:'naj (1) where ViaJ is the gamma function and the parameters of the gamma distribution are a^ >0, jS^ >0 and y. The expected ( mean ) value for r^, values from the three- parameter gamma distribution is P^ a^ + yand the variance is P^oig. The parameter /defines a minimum value for the distribution of r^, values; therefore we set y= -M (see text). There were too little data to directly estimate the lower- bound 7 for either of the species in our analysis because years with negative production occurred infrequently for stocks in our analysis. The three-parameter gamma dis- tribution has a single mode at y-n pJa-\) if a, >1. When a, <1, the probability distribution function declines mono- tonically as r^ increses from the minimum value at r^=y. 536 Fishery Bulletin 100(3) The mode for empirical r^, values in our study was always larger than -M so that a^ was larger than one. Maximum likelihood estimates of the parameters a and ji for each stock (conditional on the assumption y=-M) were obtained iteratively in a spreadsheet by maximum likelihood with observed r^ values as data. In simulations, we used the simpler method of moments to calculate gam- ma distribution parameters. Given y= -M and estimates of the mean (jj) and variance (a~) for r^ values from real data, the method of moments solves the two equations ^=(Ba+y} and &-= a/3- for the two unknowns a and fi. In particular {i=a'^l iii-y) and a=&^/ P'^. The maximum likeli- hood approach and method of moments gave similar pa- rameter estimates for real data sets (Figs. 2-3) suggesting that the method of moments was acceptable for simula- tions. In what comes later, maximum likelihood parameter estimates from data (a, j3, y= -M) are distinguished from simulated values calculated by the method of moments (dj, P^,y=-M). In each simulation run, the assumed "true" value of ^MSY^ was used to calculate T\ = 2F^^,jy . In simulation runs with no process error the simple calculation r^^= r^ was used. In runs with process error, r^, was the mean of the distribution of stochastic r,^, values (see below). For runs with independent process errors (no autocor- relation I, annual logistic parameter values r^ ^, were drawn from a gamma distribution with parameters (a, fi, y) cal- culated by the method of moments from the mean r^ and variance cr^ estimated from empirical data. average parameter ( G) was 2p (2) with 6 in the range (0,1). The random numbers d^^ in algorithm 1 were drawn from a gamma distribution with parameters: - (1-i-r) a, =a, 5" A. = A-*'"'' (i+r) (3) (4) and y=-M. The adjusted parameter values (d,, P^, y) make the mean and variance of autocorrelated r^ ^, values from algorithm 1 the same as for an independent series drawn from a gamma distribution with parameters a^,p^,y (see Appendix 1). Algorithm 2 was for autocorrelations (p) in the range (0.5,1): Z^»o-i-. J=l (5) Appendix 2— Autocorrelated process errors This appendix describes two algorithms for generating autocorrelated production process errors from gamma distributions. The algorithms are based on first-order autoregressive and moving average error structures used in time-series analysis (Nelson, 1973). The shapes of origi- nal uncorrelated gamma distributions and new, correlated probability distributions for both algorithms appeared identical in plots. Algorithm 1 was for autocorrelations (p) in the range zero to 0.5: 's.v ~ "s,V +"'^s,y-l' (1) where r^^= was the logistic population growth parameter for yeary in simulation run s, and the moving where the random numbers d^ ,, were drawn from gamma distributions with parameters d,^ = a.^JL,L an integer >3, ?>sr- A^r^' ^■^'^ y- '^he adjusted parameter values make the mean and variance of the autocorrelated and independent r^ ^, values the same. The autocorrelation in algorithm 2 is p = (L-l)IL. For simulations, we chose the smallest value of L that gave an autocorrelation that was at least as large as the value desired. Extremely high autocorrelations (e.g. p>0.9) can give dj,, = O-^Jl^ ^ 1 when a^^ is near L. This was a minor problem in some cases because gamma distributions for rfj.^., and autocorrelated values of r^^,, have no mode when dj^ < 1. We avoided this, where necessary, by setting the maximum value of L to 10 (p=0. 90) and constraining dj,^,< 1 in simulations. 537 Abstract— I ndepondont molecular mar- kers based on mitochondrial and nucle- ar DNA were developed to provide pos- itive identification of istiophorid and xiphiid billfishes (marlins, spearfishes. sailfish, and swordfish). Both classes of markers were based on amplifica- tion of short segments (<1.7 kb) of DNA by the polymerase chain reaction and subsequent digestion with informative restriction endonucleases. Candidate markers were evaluated for their abil- ity to discriminate among the different species and the level of intraspecific variation they exhibited. The selected markers require no more than two restriction digestions to allow unam- biguous identification, although it was not possible to distinguish between white marlin and striped marlin with any of the genetic characters screened in our study. Individuals collected from throughout each species' range were surveyed with the selected markers demonstrating low levels of intraspecific character variation within species. The resulting keys provide two independent means for the forensic identification of fillets and for specific identification of early life history stages. Nuclear and mitochondrial DNA markers for specific identification of istiophorid and xiphiid billfishes^ Jan R. McDowell John E. Graves School of Marine Science Virginia Institute of Marine Science College of William and Mary Rt 1208 Create Rd Gloucester Point, Virginia 23062 E mail address (for J McDowell), McDowell(q'VIM5 edu Manuscript accepted 18 September 2001. Fish. Bull. 100:537-544 (2002). Species-level identification of most ma- rine fishes is typically based on adult characters. However, distinguishing char- acters may be removed from adults when they are processed for market or personal consumption, making such identification problematic. Furthermore, species identification from early life history stages of many marine fishes is not possible because diagnostic mor- phological characters at these stages are not currently known. Consequently, alternative means of identification are needed. A variety of molecular genetic char- acters have been used to provide iden- tifications of marine fishes. Marine fish eggs and larvae have been identified by using allozymes (Mork, et al., 1983; Graves et al., 1988), restriction analy- sis of whole mitochondrial (mt) DNA (Daniel and Graves, 1994), restriction analysis of specific mtDNA gene regions (Luczkovich et al.'), and specific ampli- fication of mtDNA gene regions (Rocha- Olivares, 1998 ). A similar suite of molec- ular markers has been used to provide positive identification of adult marine fish tissues, with recent emphasis on restriction analysis of amplified regions of the mitochondrial genome (Chow et al., 1993; Chow 1994, Chow and Kishino, 1995; Heist and Gold, 1997; Innes et al., 1998; Cordesetal., 2001). To be effective, a diagnostic molecular marker must demonstrate consistent differences among closely related spe- cies and exhibit very limited intraspe- cific variation. Restriction analyses of regions of the mitochondrial genome have met these criteria for several marine fishes. However, reliance on a single, maternally inherited character (mtDNA) can provide misleading results in situations where there is a possibility of hybridization or introgression, and analyses of both nuclear and mitochon- drial markers are therefore desirable. The istiophorid and xiphiid billfishes (marlins, spearfishes, sailfish and swordfish) represent an important commercial and recreational fisheries resource. Because of depleted stock levels, current regulations within the United States prohibit the sale of is- tiophorid billfish taken in the Atlantic Ocean. Although adult billfishes are easily identified on the basis of morpho- logical characters, these characters are typically removed during processing, preventing morphological identifica- tion. In addition, the early life history stages of istiophorid billfishes are not well known, and specific identification is problematic (Nakamura, 1985). Chow (1994) used 13 restriction en- zymes in a restriction fragment length polymorphism (RFLP) analysis of a 350-bp region of the mtDNA cytochrome b gene to discriminate among ten nomi- nal species of billfishes; however sam- ples sizes were small for several species * Contribution 2470 of the Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA 23062. ' Luczkovich, J. J. ,H. J. Daniel, M.W. Sprague, S. E. Johnson, R. C. PuUinger, T Jenkins, and M.Hutchinson. 1999. Characteriza- tion of critical spawning habitats of weak- fish, spotted seatrout and red drum in Pam- lico Sound using hydrophone surveys. Fi- nal report to the North Carolina Division of Marine Fisheries under grant numbers F-62- 1 and F-62-2, 128 p. North Carolina Department of Environment and Natural Resources, Division of Marine Fisheries. Morehead City, NC 28557. 538 Fishery Bulletin 100(3) and banding patterns differed by as little as 15 base pairs, making alternate patterns difficult to distinguish. Innes et al. (1998) were able to discriminate among seven species of billfish found in Australian waters with RFLP analysis of a 1400-bp region of the mtDNA control region (D-loop). Their analysis, which employed four restriction enzymes, revealed relatively high levels of intraspecific variation of the diagnostic characters within some species, and there was some overlap of banding patterns between species. In neither study was an independent nuclear marker developed to corroborate specific identifications based on analyses of mtDNA. In this article we present a molecular key to the iden- tification of istiophorid and xiphiid billfishes using RFLP analyses of independent mitochondrial and nuclear DNA regions. We demonstrate low intraspecific variation of the characters within large collections of individuals sampled from throughout each species' range and show the utility of the markers for the identification of fillets and early life history stages. Materials and methods Collections of striped marlin (Tetraptiiriis audax), white marlin (Tetraptu/'us albidus). blue marlin iMakaira nigri- cans), and sailfish ilstiophortis platypterus) were available from previous analyses of stock structure (Graves and McDowell 1994, 1995; Graves, 1998), and individuals from locations throughout each species' range were selected for the present study (Table 1). These DNA samples con- sisted of the nuclear and mitochondrial bands resulting from mtDNA purifications with the equilibrium den- sity gradient centrifugation protocols of Lansman et al. (1981). Samples of black marlin iMakaira indica), longbill spearfish (Tetrapturus pfluergeri), shortbill spearfish (Tet- rapturus angustirostris). and swordfish iXiphias gladius), were obtained from recreational and commercial fisher- men (Table 1) and consisted of either frozen heart tissue or white muscle tissue preserved in DMSO storage buffer (Seutin et al., 1991). DNA was extracted from these tissues following the protocols of Winnepenninckx et al. (1993). Evaluation of candidate mitochondrial and nuclear loci involved a two-step process. The first was to ensure consis- tent amplification by the polymerase chain reaction (PCR) of a similar-size product across all taxa. The second step was to screen those loci that successfully amplified across all billfish species with a panel of restriction endonucleas- es to identify enzymes that discriminated among species and revealed limited intraspecific variation. Several candidate mitochondrial and nuclear gene re- gions were amplified by PCR (Table 2). The 25 ijL PCR reactions consisted of 0.25 luL template DNA, 2.5 ^L lOX PCR buffer plus magnesium, 0.5 ;iL dNTP mix, 0.25 pL forward primer, 0.25 i.tL reverse primer, 0.125 pL Tag DNA polymerase, and 21.125 pL PCR grade water Primers were ordered from either Life Technologies (Gaithersburg, MD) or Genosys Biotechnologies Inc. (The Woodlands, TX), and PCR reactions were carried out in an MJ Research Corporation PTC-200 Peltier thermal cycler (Watertown, Table 1 Collection in formation for billfish samples surveyed. Species Location Number Total Sailfish Brazil 34 99 Mexico 24 Ecuador 17 Australia 24 White marlin Brazil .38 99 Morocco 36 Venezuela 25 Striped marlin Mexico 28 96 Ecuador 38 Australia 30 Black marlin Ecuador 12 60 Australia 48 Spearfish Venezuela 12 16 Hawaii 4 Blue marlin Mexico 24 150 Australia 4 Ecuador 20 Hawaii 63 Jamaica 39 Swordfish Hawaii 20 20 MA) by using the Life Technologies PCR reagent system (Gaithersburg, MD). Initial screening demonstrated that the mitochondrial ND4 gene region and the nuclear MN32-2 locus produced the most reliable amplifications across taxa and PCR conditions were optimized for these loci. The cycling parameters for the ND4 gene region were an initial denaturation at 95°C for 5 min., followed by 35 cycles of 94°C for 1 min., 47°C for 1 min., 65°C for 3 min., and a final extension at 65°C for 7 min. Amplification of MN32-2 proceeded with an initial denaturation at 95°C for 5 min., 40 cycles of 94°C for 1 min., 57°C for 1 min., 65°C for 3 min., and a final extension at 72°C for 7 min. Amplified products were held at 4°C until use. The size of each amplification product was determined on a 1% aga- rose gel run in TBE (45 mM Tris, 45 mM boric acid, I mM EDTA) at 100 volts for 1 hour ND4 amplification resulted in a product of approximately 1.7 kb and MN32-2 amplifi- cation resulted in a product of approximately 1.2 kb. Amplified products were screened with a panel of re- striction endonucleases to identify those that discrimi- nated among species, and revealed a minimum level of variation within species. All enzymes were purchased from Gibco/BRL Life Technologies Inc. (Bethesda, MD) with the exception of Banl, which was purchased from Promega (Madison, WI). All were used according to the manufacturers' instructions. Restriction fragments were separated on 2.5'/r hori- zontal agarose gels made from 1.25% UltraPure agarose (Life Technologies Inc.. Bethesda, MD) and 1.25% NuSeive McDowell and Graves; Nuclear and mitochondrial DNA markers for identification of istiophorid and xiphiid billfisfies 539 Table 2 Primer pairs used to amplify regions evaluated in this study. Locus Primer sequence (5-3') Source Cytochrome b: CYTB-F TGGGSNCARATGTCNTWYTG Joseph Quattro, personal commun.' CYTB-R GCRAANAGRAARTACCAYTC ATPase 6: ATPase L8331 TAAGCRNYAGCCTTTTAAG Joseph Quattro, per.sonal commun.' ATPase H8969 GGGGNCGRATRAANAGRCT D-Loop: CB3R-L CATATTAAACCCGAATGATATTT Palumbi et al., 1991 12SAR-H ATAGTGGGGTATCTAATCCCAGTT ND4: ND4 ARG-BL CAAGAOCCTTGATl'TCGGCTCA Bielawski and Gold, 1996 ND4 LEU CCAGAGTTTCAGGCTCCTAAGACCA ITS: ITS-3 TATGCTTAAATTCAGCGGGT Goggin, C.L, 1994 ITS-5 CGTAGGTGAACCTGCGGAAGG SACTIN: SACSMSF-F CGGACGCCCCCGTCACCAGGTAC This study SACIN-R CCAGAGGCATACAGGGACAGCACAGC MN32-2: MN32-2F GTAGCAAGGGGCTGTTGCATAG Buonaccorsi et aL, 1999 MN32-2R GAGTCAGTGGTTCGGGATTTTATC MN47: MN47-F GCTGTTGACCCAAACAATCCGG Buonaccorsi et al., 1999 MN47-R GGGCATAAATGCTCAGGACACTT MN81: MN81-F CACTCAAACAGGTGAATCCTGGC Buonaccorsi et al., 1999 MN81-R CAAAACAACAGATGCCGCTAAGG WM08: WM08F AGCAGCTAGGGACACACGATTCC Buonaccorsi et al., 1999 WM08R GGCAAACCTTACACTGAGGGGATG ' Quattrro.J. 1995. Personal commun. Department of Biological Sciences, University of South Carolina College of Science and Mathematics. Columbia, SC, 29208. GTG agarose (FMC BioProducts, Rockland, ME), and vi- sualized under UV light after having been stained with ethidium bromide. Fragment sizes were estimated by comparison with a 1-kb size standard (Life Technologies Inc., Bethesda, MD) using RFLPScan Plus 3.0 (Scanalyt- ics, Billeriea. MA). Results Mitochondrial marker Four mtDNA regions (cytochrome b. D-loop, ND4, and ATPase) were included in the initial screening (Table 2). The ATPase region was tested with eight potentially use- ful enzjTnes based on published sequences and was found to have an extremely low level of interspecific variation (many species exhibiting identical banding patterns). The cytochrome b region was screened with four enzymes based on published sequences, but because of the small size of the amplification product (350bp), differences in banding patterns were small and difficult to distinguish. The D-loop region was screened with a total of 40 enzymes. Of these, Bel I, Ahi I, Rsa I, and Hinf I were tested with up to 50 individuals from each species. Banding patterns that were initially thought to be diagnostic for blue marlin based on Rsa I were found to occur at low frequency in sailfish. This overlap combined with the large amount of intraspecific variation in some species made this region unsuitable for use as a forensic marker. Finally, the ND4 region was screened with a total of 47 restriction enzymes. Of these, 17 were tested more extensively, and the combi- 540 Fishery Bulletin 100(3) IkbDNA Ladder Black Marlin Blue Marlin White/Stnped Marlin Sailfish Spearfish Ikb DNA Ladder Black Marlin Blue Marlin Sailfish White/Striped Marlin Speartish 2,036- 1 ,636 — 1.018- ^ mt^ ^Z — — ^_ 1.018 — ^^^ —_ 506,517- 396- 344 — 298- III 1 I II II 1 II 1 III 506,517- 298- ^ 220.201- 154,134- ^ 154,134 — — ~ ND4 cut with Hae III ND4 cut with Ban I Figure 1 ND4 gel. Most common restriction fragment patterns of the ND4 mitochondrial gene region of istio- phorid billfishes. (A) Digestion with Hae III. From left to right, 1-kb DNA ladder, black marlin (pat- tern D), blue marlin (pattern A), white and striped marlin (pattern B), sailfish (pattern B), shortbill spearfish (pattern E), and longbill spearfish (pattern E). (B) Digestion with Ban I. From left to right. 1-kb DNA ladder, black marlin (pattern Ai. blue marlin (pattern A), white and striped marlin (pat- tern B), sailfish (pattern A), shortbill spearfish (pattern B). and longbill spearfish (pattern B). Table 3 Restriction fragment patterns of the mitochondrial ND4 region of istiophorid and xiphiid billfishes. Only diagnostic bands are shown. (A) Digestions with Hae III. A = blue marlin; B = striped marlin, white marlin, sailfish; C = white marlin, sail- fish; D = black marlin; E = spearfish; F = white marlin; G = spearfish; H = swordfish. A B C D E F G H 800 .570 800 570 405 570 405 950 320 405 405 530 380 405 320 300 270 320 320 320 320 380 270 290 (B) Digestions with Ba/i I. A = blue marlin. sailfish, black marlin, spearfish; B = striped marlin, white marlin, spearfish; C = swordfish. B C 850 1500 700 650 400 650 400 400 nation of Ban I and Hae III was found to be diagnostic and to reveal a low level of intraspecific variation. After finding diagnostic enzymes for use with tlie ND4 region, a total of 540 billfish samples was screened (Table 1) to evaluate the accuracy of the marker. Samples from a broad geographic range, including both the Atlantic and Indo-Pacific, were used for each species whenever possible. Of these, the white marlin, spearfishes and sailfish each exhibited one alternative restriction pattern for Hae III at low frequency (3.6%, 20.0% and 7.4% respectively) but in no case was the alternate pattern the same as a pattern seen in another species (Table 3). In addition, spearfishes exhibited an alternate pattern for the enzyme Ban I at a frequency of 40%; however, because Ban I was used only to discriminate white and striped marlin from sailfish in the ND4 identification key, this pattern did not affect the results (Figs. 1 and 2). Nuclear marker Six nuclear markers were screened in the preliminary analysis (Table 2). These included the short actin intron, the internal transcribed spacer (ITS) region (Goggin, 1994), and four anonymous single copy nuclear (scnDNA) markers. The scnDNA markers MN32-2, BM47, BM81, and WM08 were originally developed for analyses of popu- McDowell and Graves: Nuclear and mitochondrial DNA markers for identification of istiopfiond and xiptniid blilfisfies 541 lation structure in blue marlin (Buonaccorsi et al.. 1999). The short actin intron primers were modified from "universal" actin gene primers "480" and "483" (Siddall et al„ 2001). Both the short actin and the ITS marker were rejected because neither marker amplified reli- ably across species. For the scnDNA markers, the program GeneJockey (Taylor, 1996) was used to search for the presence of restriction sites in sequences previously generated for blue marlin. The WM08 marker was screened with a total of ten enzymes, each of which produced identical patterns across species. BM47 and BM81 were also screened with ten enzymes each. For BM47, the combination of enzymes Bel I and Dele I appeared to be diagnostic in a preliminary screening. However, upon further analy- sis, it was discovered that this combination of locus and enzymes produced confounding patterns for blue marlin and sailfish; the most common pattern for blue marlin was seen as a rare pattern for sailfish. Likewise, the BM81 lo- cus did not distinguish between blue marlin and sailfish or between white and striped marlin and spearfish with any of the enzymes used. The MN32-2 locus was screened with a total of nine restriction endonucleases. The combination of Dra I and Dde I was found to allow for unambiguous identification of billfish species (Figs. 3 and 4). As with the ND4 locus, after determining a diagnostic enzyme-locus combination, we screened a total of 540 billfish samples from a broad geographic range to evalu- ate the accuracy of the marker (Table 1 ). The enzyme Dra I was found to have two alternate alleles: "D" and "E" for blue marlin at a frequency of 19% and .5.5%, respectively. All other species appeared to be fixed for different (ho- mozygous) alleles with respect to this enzyme. For Dde I, spearfishes had an alternate allele, "E," at a frequency of 36%. In addition, blue marlin had two alternate alleles "H" and "I" at frequencies of 16% and 40%, respectively (Table 4). Although the "H" allele in blue marlin was the only allele seen in black marlin, use of Dde I was not nec- essary to distinguish the two species since they are easily differentiated by Dra I (Figs. 3 and 4). All other species appeared to be homozygous for different alleles. Discussion The purpose of this study was to develop a key to the iden- tification of billfish species based on independent mito- chondrial and nuclear markers. Our goal was to make the process streamlined and capable of being performed in a modestly equipped genetics laboratory. Specific identifica- tion can be accomplished with a single PCR amplification of either the mitochondrial ND4 locus or nuclear MN32-2 locus and two restriction digestions. Previous methods with other mitochondrial gene regions have required the use of either four or 13 restriction digestions (Innes et al., 1998 and Chow 1994, respectively). To facilitate specific identification, an objective of this study, was to develop diagnostic markers that exhibited Swordlish //idlll Blue Marlin M>4 Black Marlin ^P^--^'^''^'' «.,.,, r ^^ Striped Marlin White Marlin Striped Ivlarlin 1 White Marlin 1 Sa.lfisli L ^— Sailfish Figure 2 Key to di-stinguish species of billfishes based on the mitochondrial locus ND4. Table 4 Restriction fragment patterns of the nuclear gene region BM32-2 of istiophorid billfishes. This locus did not amplify in the swordfish. (Al Digestions with Dra I. A = blue marlin; B = striped marlin, white marlin, sailfish, black marlin; C = spearfish; D = blue marlin; E = blue marlin. A/D and A/E heterozygotes were seen in blue marlin cut with Dra I. Individuals of all other species were fixed homozygous. B C D 650 640 640 650 450 280 280 400 260 220 50 1200 (B) Digestions with Dde I. A = blue marlin; B = striped marlin, white marlin; C = spearfish; D = sailfish; E = spearfish; H = blue marlin; I = blue marlin. A/I A/H and H/I heterozygotes were seen in blue marlin cut with Dde I. C/E heterozygotes were seen in spearfish cut with Dde I. Individuals of all other species were fixed homozygous. A B C D E H I 700 850 700 475 475 775 700 300 425 475 370 370 425 425 125 125 280 130 205 125 125 limited intraspecific variation. Analysis of large sample sizes (60-100 or more) of sailfish, white marlin, striped marlin, blue marlin, and black marlin from throughout each species' range revealed minimal variation of the spe- cies-specific characters. Most species displayed a single genotype for digestions with the two enzymes used to cleave either the mitochondrial or nuclear amplification products, and no species exhibited more than three geno- types for any locus-restriction enzyme combination. In 542 Fishery Bulletin 100(3) < D ^. ^ 5 1 ?^ g. s. c !:i CO ;/) 1/5 :s : 1 = B 1/) CQ t j: ^ ■a C/5 O 2 -J < z D S3 ■o K. ■£ i75 s i = s ^ K. K C/5 t/5 r. r 0/1 O c i5 B ^ ^ S, ^ 2 1,018-^ 506,517- 396- 344 — 298* — ~ 220.201- 154,134 — Bm32-2cut with Dra I Bm32-2 cut with Dde I Figure 3 BM32-2 gel. Most common restriction fragment patterns of the nuclear gene region BM32-2 of istiophorid biUfishes. This locus did not amplify in the swordfish. (A) Digestions with Dra I. From left to right, 1-kb DNA ladder, blue marlin (pattern A), white and striped marlin (pattern B). sailfish (pattern B). black marlin (pat- tern B). shortbill spearfish (pattern C), and longbill spearfish (pattern C). (B) Digestions with Dde I. From left to right, 1-kb DNA ladder, blue marlin (pattern A), white and striped marlin (pattern B), sailfish (pattern D), black marlin (pattern H), shortbill spearfish (pattern C), and longbill spearfish (pattern Ci. Dra\ BM32-2 Spearfish Blue Marlin Slriped Marlin White Marhn Black Marlin Sailfish Di/c\ £ Striped Marlin White Marlin Black Marlin Sailfish Figure 4 Key to distinguish species of billfishes based on the single copy nuclear locus MN32-2. contrast, Innes et al. (1998) reported ten composite hap- lotypes among 47 black marlin, six composite haplotypes among 26 blue marlin, six composite haplotypes among 46 striped marlin, and six haplotypes among 21 swordfish, all from the southwest Pacific. From the level of intraspecific variation in relation to the sample sizes and the regional nature of their collections, it is reasonable to assume that Innes et al. ( 1998) may have missed several composite gen- otypes characteristic of the different species. In fact, from the level of variation exhibited by black marlin, striped marlin, and swordfish. Innes et al. (1998) suggested that their diagnostic species markers could be of potential use in population structure analyses. It occurs to us that if a genetic character exhibits sufficient intraspecific variation to be useful for analyses of stock structure, it is probably not a good candidate for species identification. A high degree of genetic similarity was noted among white marlin and striped marlin in our study None of the molecular markers evaluated in this study was able to unambiguously distinguish between the two species. McDowell and Graves: Nuclear and mitochondrial DNA markers for identification of istiopfiorid and xiphiid billfisfies 543 I 2 3 Figure 5 Larval billfish and its corresponding specific identification as a sailfish based on tlie BM32-2 locus. Lane 1 Dra I, lane 2. 1-kb plus DNA ladder (Gibco/BRL Life Technologies Inc., Bethesda MDl, lane3, Ddc/. Chow (1994) was also unable to distinguish between the two species based on RFLP analysis of the cytochrome b gene, and Innes et al. ( 1998) did not consider white marlin in their investigation because it does not occur in Austra- lian waters. RFLP analysis of the whole mtDNA molecule indicated that white and striped marlin share composite haplotypes, although there are highly significant frequen- cy differences between the species (Graves and McDowell, 1995; Graves, 1998). Sequence analysis of the mtDNA cytochrome b gene also demonstrated a lack of genetic divergence among white and striped marlin (Finnerty and Block, 1995), and a further genetic analysis of the species' relationships is warranted. To evaluate the utility of the methods outlined in our study with those of other investigators, detailed protocols and six unknown billfish samples were sent to the South- east Fisheries Science Center's (now National Ocean Survey's) Charleston, SC, laboratory. Scientists at the Charleston Laboratory analyzed both mitochondrial and nuclear DNA markers for each sample and arrived at con- sistent, correct identifications for each of the samples of unknown billfish. In addition, samples of juvenile billfish collected by investigators at the University of Miami were analyzed in our laboratory with these molecular markers. Samples consisting of one eye taken from a 3-mm juvenile billfish provided sufficient DNA to amplify the mtDNA and nuclear markers, allowing specific identification (Fig. 5). The technique is currently being used to deter- mine the temporal occurrence of istiophorid larvae in the Florida Straits iLuthy and McDowell'-). Although the methods presented in our study allow the specific identification of billfish species, more sensitive mo- lecular markers are required to distinguish among ocean populations of some istiophorid species. Amendment I to the Fishery Management Plan for Atlantic Billfishes prohibits the sale of blue marlin, white marlin, and sailfish taken in the Atlantic Ocean, although it is legal to market blue mar- lin, striped marlin, and sailfish from the Indian or Pacific oceans. Enforcement of this regulation requires the ability to discriminate between Atlantic and Indo-Pacific individu- als of blue marlin, sailfish, and white and striped marlin. Examination of our results suggests that there are several other molecular markers, which while not used in this study, occur at relatively high frequencies in Atlantic blue marlin, sailfish, and white marlin but do not occur in their Pacific conspecifics. These molecular markers could potentially be used to identify some Atlantic individuals without misclas- sifying a Pacific fish, thereby allowing the enforcement of the management plan. Additional work will be required to develop a database that would support such analyses. Acknowledgments We thank all those who so generously helped with collec- tions including Julian Pepperell, Guy Harvey, Ed Everett, Eric Prince, and Mike Judge, as well as the anglers at Hotel Spa Buenavista, Baja California Sur, Mexico, and The Mid-Atlantic 500,000 Tournament, Cape May, New Jersey. We also thank Cheryl Woodley and the National Ocean Services' Charleston Laboratory for helping to vali- date our results. This project was funded by the National Oceanic and Atmospheric Association's Saltonstall-Ken- nedy Program grant number NA67FD00038. 2 Luthy, S., and J. McDowell. 2001. A molecular approach to the identification of larval billfishes. Submitted to the Austra- lian Journal of Marine and Freshwater Research as part of the y^ international billfish symposium. CSIRO Publishing, 150 Oxford St. Collingwood, Vic. 3066, Australia. Literature cited Bielawski, J. P., and J R. Gold. 1996. Unequal synonymous substitution rates within and 544 Fishery Bulletin 100(3) between two protein-coding mitochondrial genes. Mol. Biol Evol. 13:889-892 Buonaccorsi, V. P.. K. S., Reece, L.W. Morgan, and J. E. Graves. 1999. Geographic distribution of molecular variance within the blue marlin iMakaira nigricans): a hierarchical analy- sis of allozyme, single-copy nuclear DNA, and mitochon- drial DNA markers. Evolution 53:568-579. Chow. S. 1994. Identification of biUfish species using mitochondrial cytochrome b gene fragment amplified by polymerase chain reaction. Int. Comm. Conserv. Atl. Tunas. Report of the 2nd ICCAT billfish workshop. ICCAT 41:549-56. Chow, S., M. E. Clarke, and P. J. Walsh. 1993. PCR-RFLP analysis on thirteen western Atlantic snappers (subfamily Lutjaninae): a simple method for spe- cies and stock identification. Fish. Bull. 91:619-627. Chow, S., and K. Hazama. 1998. Universal PCR primers for S7 ribosomal protein gene introns in fish. Mol. Ecol. 7:1247-1263. Chow, S., and H. Kishino. 1995. Phylogenetic relationships between tuna species of the genus Thiinnus (Scombridae: Teleostei): Inconsistent implications from morphology, nuclear and mitochondrial genomes. J. Mol. Evol. 41:741-748. Cordes, J. F. A., 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. Estu- aries. 24:49-58. Daniel, L. B., Ill, and J. E. Graves. 1994. Morphometric and genetic identification of eggs of spring-spawning sciaenids in lower Chesapeake Bay. Fish. Bull. 92:254-261. Finnerty, J. R.. and B. A. Block. 1995. Evolution of cytochrome b in the Scombroidei (Tele- ostei): molecular insights into billfish (Istiophoridae and Xiphiidae) relationships. Fish. Bull. 9:78-96. Goggin, C. L. 1994. Variation in the two internal transcribed spacers and 5.8S ribosomal RNA from five isolates of the marine parasite Perkinsiis (Protista, Apicomplexa). Mol. Biochem. Parasi- tology 65 (19941179-182. Graves, J. E. 1998. Molecular insights into the population structures of cosmopolitan marine fishes. J. Heredity 89:427-437. Graves, J. E., and J. R. McDowell. 1994. Genetic analysis of striped marlin Tetrapturus audax population structure in the Pacific Ocean Can. J. Fish. Aquat.Sci., 51:1762-1768. 1995. Inter-ocean genetic divergence of istiophorid billfishes. Mar Biol. 122:193-204. Graves, J. E., M. A. Simovich, and K. M. Schaefer 1988. Electrophoretic identification of early juvenile yellow- fin tuna. Fish. Bull. 86:835-838 Heist, E. J., and J. R. Gold. 1997. Genetic identification of sharks in the U.S. Atlantic large-coastal 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 Freshwater Res. 49:383-388. Lansman, R. A., R. O. Shade, J. F. Shapira, and J. C. Avise. 1981. The use of restriction endonucleases to measure mitoc- hondrial DNA sequence relatedness in natural popu- lations. J. Mol. Evol. 17:214-226. Mork, J., P. Solemdal, and G. Sundnes. 1983. Identification of marine fish eggs: a biochemical gen- etics approach. Can. J. Fish. Aquat. Sci. 40: 361-369. Nakamura, I. 1985. FAO species catalogue. Vol. 5: Billfishes of the worid: an annotated and illustrated catalogue of marlins, sail- fishes, spearfishes and swordfishes known to date. FAO Fish Synop 125. FAO, Rome. Palumbi. S., A. Martin, S. Romano, W.O. McMillan, L. Stice, et al., 1991. The simple fool's guide to PCR, 47 p. Department of Zoology and Kewalo Marine Laboratory, Univ of Hawaii, Honolulu, HI. Rocha-Olivares, A. 1998. Multiplex haplotype-specific PCR: a new approach for species identification of the early life stages of rockfishes of the species-rich genus Sebastes Cuvier J. Exp. Mar Biol. Ecol. 231:279-290. Seutin, G., B. N. White, and P. T. Boag. 1991. Preservation of avian blood and tissue samples for DNA analysis. Can. J. Zool. 69:82-90. Siddall, M. E., K. S. Reece, T A. Nerad, and E. M. Burreson. 2001. Molecular determiniationofthe phylogenetic position of a species in the genus Colpodella (Alveolata). American Museum Novitiates 3314:1-10. Taylor, P L. 1996. GeneJockey-II sequence processor Software distrib- uted by BIOSOFT Cambridge, UK. Winnepenninckx, B., T. Backeljau, and R. De Wachter 1993. Extraction of high molecular weight DNA from mol- lusks. Trends Genet. 9:407. 545 Abstract-Samples of 11.000 King George whiting iSillagi nodes punctata ) from the South Austrahan commercial and recreational catch, supplemented hy research samples, were aged from otoliths. Samples were analyzed from three coastal regions and by sex. Most sampling was undertaken at fish pro- cessing plants, from which only fish longer than the legal minimum length were obtained. A left-truncated normal distribution of lengths at monthly age was therefore employed as model likelihood. Mean length-at-monthly-age was described by a generalized von Bertalanffy formula with sinusoidal seasonality. Likelihood standard devia- tion was modeled to vary allometrically with mean length. A range of related formulas (with 6 to 8 parameters) for seasonal mean length at age were com- pared. In addition to likelihood ratio tests of relative fit, model selection cri- teria were a minimum occurrence of high uncertainties (>20'7t SE), of high correlations O0.9, >0.95, and >0.99) and of parameter estimates at their biologi- cal limits, and we sought a model with a minimum number of parameters. A gen- eralized von Bertalanffy formula with ^^ fixed at 0 was chosen. The truncated likelihood alleviated the overestimation bias of mean length at age that would otherwise accrue from catch samples being restricted to legal sizes. Seasonal growth of King George whiting (Sillaginodes punctata) estimated from length-at-age samples of the legal-size harvest Richard McGarvey Anthony J. Fowler SARDI Aquatic Sciences 2 Hamra Avenue West Beach South Australia 5024, Australia E-mail address (for R, McGarvey) mcgarvey.richard@saugov.sa.gov.au Manuscript accepted 15 February 2002. Fish. Bull. 100:545-558 (2002). In this stu(iy, we estimate(i growth curves from six data sets of fish sampled from the commercial and recreational catch of an important fish species in South Australia (Fig. 1), King George whit- ing (Sillaginodes punctata). Like many temperate fish populations, recruitment and growth can be assumed to follow a yearly cycle. A variety of methods have been used to measure seasonal variation in fish growth including mark-recapture (Francis, 1988; Coggan, 1997) and oto- lith annulus-diameter increments in combination with tetracycline marking (Panfili et al., 1994; Fabre and St. Paul, 1998; Francis et al, 1999). The method used in our study is the most widely employed, of fitting to lengths-at-age, where the age of each sampled fish is read as a count of yearly otolith annuli from samples of the harvest. The model-fitting algorithm had three key objectives, each to represent a specific feature of the growth of South Australian King George whiting, or of the data set. These were expressed mathematically as deviations from a standard 3-parameter von Bertalanffy model fitted with a normal likelihood. The first two objectives were elabo- rations on the standard von Berta- lanffy growth curve (understood as a deterministic function of mean length versus age). The first objective was to make seasonality explicit in the growth curve (Pitcher and MacDonald, 1973; Somers, 1988; Hoenig and Hanumara, 1990; Pauly and GaschiitzM and the second objective was to allow a wider range of curvatures by using the r-ex- ponent (Schnute, 1981). For application of this growth descrip- tion in length- and age-based stock as- sessment modeling, we also estimated the shape of the length-at-age probabil- ity density function (pdH, which quanti- fies the distribution of fish of different lengths at each age. The principal regulation of both rec- reational and commercial King George whiting harvest is by legal minimum length (LML); fish smaller than the LML cannot be landed and must be returned to the sea. Therefore fish obtained from the catch are a biased sample. The third objective was to ex- plicitly account for the absence of fish below the LML, for which a truncated pdf was used. This truncated normal density, the pdf of observed lengths-at- age, was also used as the likelihood for each individual sample. A range of models were examined that met these three objectives. These included a seasonal version of the Richards (1959) model proposed by Akamine (1993), which makes differ- ent use of the /-exponent. Therefore a fourth objective was to apply standard statistical model selection techniques. Methods to choose the most appropri- ate model for a given growth data set were reviewed by Quinn and Deriso (1999). Hierarchical model fits are sta- tistically comparable using their likeli- ■ Pauly, D., and G. Gaschutz. 1979. A simple method for fitting oscillating length growth data, with a program for pocket calculators. Demersal Fish Committee, ICES council meeting 1979/G:24. ICES, PaliEgade 2-4 DK-1261 Copenhagen K, Denmark. 546 Fishery Bulletin 100(3) Figure 1 South Australian coastal waters. King George whiting iSillaginodes punctata) are found predominantly in the two Gulfs, and along the West Coast. hood ratio (Ivimura, 1980; Rice, 1995). We sought to reduce the original 8-parameter model and employed likelihood ratio tests to evaluate the relative fit of the simpler mod- els tested. We specify and apply a set of selection criteria below, seeking one common model form for six data sets to describe the length-at-age distributions. A 7-parameter generalized seasonal von Bertalanffy model, with t^ fixed at zero, was chosen. In addition, we applied an allometric relationship (a pow- er curve) to fit weight versus length for this population. That the samples are not uniform by length in commer- cial catch is common to most fisheries. The absence of fish below LML or below the lengths selected by the fishing gear, inevitable when sampling is from the landed catch, creates an overestimation bias of mean length-at-age, because fast- er growing fish reach the legal stock sooner and are thus over-represented in those samples. This sampling bias was taken into consideration by the use of a truncated model likelihood assuming zero probability of capture below legal size. In practice, the clear advantage of samples from the landed (commercial or recreational) harvest is a substan- tially lower cost than the alternative of fishery-independent samples gathered, usually by researchers, at sea. The specific length-dependent bias in sampling changed during the time of the study. The LML was raised from 280 mm to 300 mm. Also, some of the samples were gathered by researchers who landed all, notably smaller fish. By applying one of three different likelihoods to any given sample, having different truncation lengths of 280 mm, 300 mm, or no truncation, all fish samples were combined in a single length-at-age estimation for each of three re- gional subpopulations, and by sex. Materials and methods A total of 11,164 King George whiting were sampled between 1995 and 1998 from across the main fishery area of South Australia. Most samples were obtained by subsampling the commercial catch at local fish processing plants or by purchasing fish from commercial processors. Some samples were provided as frozen fish frames (fillets removed) by recreational fishermen, and the remainder were caught on scientific cruises. Each fish was measured for total (TL) and standard lengths (SL) to the nearest mm, and weighed to 0.1 g. Gonads were removed, sexed, and weighed. The sagittae, the largest pair of otoliths, were removed from each fish for age determination. Samples were subdivided into six data sets: three spa- tial regions and by sex (Table 1). Movement studies based McGarvey and Fowler: Seasonal growth of Sillagtnodes punctata 547 Table 1 Summary of data sets, each fi sh measured I )r age, length, and weighl in the King G eorge whiting an alysis. Samples from com- mercial and recreational fisheries were regu ated by a 28()-mm legal minimum length (I.ML) before 1 September 199.5, a 300-mm minimum length thereafter. Area Beginning Sample size Fisheries Fisheries Area abbreviation Sex date End date Researchers 280 LML 300 LML GulfSt. Vincent and GSV male 4 Aug 94 17 Feb 99 293 503 1139 nortern Kangaroo Island female 4 Aug 94 17 Feb 99 373 615 1343 Spencer Gulf SG male 23 Apr 94 30 Sep 97 267 298 924 female 23 Apr 94 30 Sep 97 363 441 1300 South Austrahan West Coast WC male 27 Apr 94 18 May 97 167 805 608 female 27 Apr 94 19 May 97 226 857 642 on tag recoveries over three decades (Fowler and March^) indicated three largely self-sustaining subpopulations: GulfSt. Vincent, Spencer Gulf and West Coast (Fig. 1). Young fish (two- and three-year-olds) were aged by interpretation of the macrostructure of the whole sagit- tae. For fish with more complex otoliths, the otolith was snapped in two across the posterior-anterior axis through the center, exposing the transverse face of both halves. One of these halves was burnt in a bunsen flame and then examined with a binocular dissecting microscope at 6-20x magnification. The surface being examined was smeared with immersion oil. The alternating opaque and trans- lucent zones were counted. The periodicity of formation of this macrostructure in sagittae of King George whit- ing has been validated and the following algorithm was developed for conversion of ring count to age (Fowler and Short, 1998): a = 12N + w^ -I- m^„ where a = age in months; N = number of opaque zones; nif^ = 8 = number of months from universal birth date (i.e. 1 May) midway through spawning season to the end of the year; and m^ = number of months from the start of year to the month of capture. Some samples offish were scaled in the commercial pro- cessing plant prior to weighing and measuring for length. Although this process did not affect their lengths, it did result in an appreciable loss of weight. Consequently, for estimation of weight-length relationships we corrected the ■ Fowler, A. J., and W. A. March. 2000. Adult movement pat- terns. In Development of an integrated fisheries manage- ment model for King George whiting (Sillaginodes punctata I in South Australia fA. J. Fowler and R. McGarvey, eds.), p. 83-104. FRDC Final Project Report 95/008. Fisheries Research and Development Corporation, PO Box 222, Deakin West ACT 2600, Australia. weights of scaled fish using a linear relationship that was derived by weighing 155 fish before and after scaling. This linear relationship was {corrected iveight) = 1.0176 x (scaled weight) + 3,5835 (r2=0.99, P<0.001, df=154). Growth: length-at-age In order to make explicit the absence of fish samples smaller than LML, a truncated normal probability density function of length was used to describe the probability of capture of individual fish in each monthly age. This pdf was employed as the likelihood of observation of each indi- vidual, given its age. Truncation implies a zero predicted probability of observing a commercially or recreationally sampled fish less than LML. A few sublegal fish were mea- sured and, being unrepresentative, were removed from the six data sets. During the period of sampling, in September 1995, the regulated size of LML for commercial and recreational fishery samples was increased from 280 to 300 mm. The likelihood truncation length was thus a function of date of capture. This necessitated two forms of likelihood, for LMLs of 280 and 300 mm. Smaller numbers of samples obtained on scientific cruises were not subject to LML size controls. Thus a third, regular untruncated, likelihood was used to model research samples. A normal likelihood was fitted to model the distribution of lengths at each age: 1 2jr oia, ) -exp J. u, 2 -/(g,) aia,) (1) where /, = length offish sample ;'; and a, = age of fish sample i obtained from count of its otolith annuli. 548 Fishery Bulletin 100(3) The mean length-at-age lia,)-- L_ • 1 - exp -Ki a, -tn sin(2;r(a ,-w)/12) 12 2;r sin(2;r(/o-«)/12) • (2) was modeled by a seasonally periodic von Bertalanffy growth formula, generalized by the inclusion of an expo- nent, r. Age (a,) was in integral units of months, with May, the assumed date of birth at mean time of spawning of South Australian King George whiting, being month 1. Division by 12 in the mean length formula preserves the usual units of ii" where age is in years. The seasonality function is sinusoidal, although more complex seasonality functions can be postulated that al- low asymmetry in growth through the year. Values of the seasonality amplitude parameter ;/ > 1 imply decreasing length in the yearly time of minimum growth (Pauly and GaschiitzM. We therefore constrained u < 1, assuming that no shrinking in length occurs. With sine, the phase pa- rameter (ft)) gives the month of maximum growth, where months of age 1, 13, 25, etc. denote the birth month. May. The likelihood standard deviation (a) was modeled as an allometric function of mean length: Parameters were estimated by minimizing the negative sum of log-likelihoods with the AD Model Builder estima- tion software (Otter Research Ltd., Sidney, B.C., Canada): O -£ln(L, (5) cr(a, ): Ha, (3) Initial parameter values for all models tested were LML, [LML, 0, if/, < LML, postulates a probability cut-off to zero for landed samples less than LML and a normal probability, integrating to 1, for the range of legal lengths. LML is subscripted by the fish sample data point (/) to indicate that LML is either 280 or 300 mm depending on the date of capture of the fish. For research samples, for which all lengths could be observed, the full untruncated normal likelihood (Eq. 1) was used. 1 Search for parameters that frequently show high cor- relations with other parameters, have high variances, or in estimation hit preset upper and lower bounds, beyond which biologically unrealistic values are being inferred. 2 Set those parameters to fixed likely values. This assumes a biologically likely value can be postulated. If not, use a mean value among the range of estimates obtained when the parameter is allowed to vary freely. A parameter may also be selected if it varies little among the range of data sets for which estimates are obtained. 3 Test for the change in negative log-likelihood to deter- mine whether the reduced model is significantly less well fitting by using chi-square likelihood ratios. 4 Check that confidence intervals and correlations are reduced. In addition, we seek model that converges reliably for all data sets. Satisfactory convergence should not depend on the initial parameter values chosen, but in some cases a new set of initial values can be tried for any given non- converging data set. Convergence can also depend on the minimization algorithm or subroutine employed. However, for purposes of model selection, failure to converge for some data sets analyzed is an indication that the likeli- hood surface for that model is less smooth and the true global maximum may not always be obtained. Reliable convergence is particularly desirable when 1) bootstrap- ping, where nonconvergence leaves resampled data sets McGarvey and Fowler: Seasonal growth of Sillaginodes punctata 549 unestimated, and 2) for integration of growth into overall fishery stock assessment estimators, where finding more appropriate initial values for the growth parameters may be nontrivial, and the cause of nonconvergence of the overall estimation may not be easily traced to the growth submodel. In addition to reduced versions of the model presented above, we fitted a related model proposed by Akamine (1993), which incorporates into the Richards (1959) growth curve a sinusoidal seasonality function like that of Equation 2. Weight-at-length Mean (corrected) weight versus total length was modeled by an allometric relationship: m,) = ail' (6) A normal likelihood was again used. The standard devia- tion of the likelihood (i.e. of the fitted spread of observed weights about the mean given in Eq. 6) was assumed to vary linearly with length: ^Jl,'l=(^uO+^,J>' (7) Parameter confidence bounds were estimated by a boot- strap of 1000 runs. Results Growth model choice The generalized von Bertalanffy model (Eq. 2, abbreviated as gVB) gave the best fit with five of six data sets (Table 2). However for both gVB and the other 8-parameter model, Akamine-Richards (AR), correlations among parameters were unacceptably frequent and high (Table 2). Other evidence of overparameterization included frequent occurrence of wide confidence bounds (SE >209c. Table 2). High correlations and parameter uncertainties were most widespread for Sg and s,. The exponent, r. and seasonality amplitude, u. also occurred frequently with high uncer- tainty, and tf^ and r often were found in high correlations; Iq hit both upper (7.99) and lower (-30) confidence bounds for some models with west coast data sets. As with stan- dard von Bertalanffy models, ?„ quantifies the age (here, in months) at which length extrapolates to 0. The occurrence of ;/ estimates hitting their upper bound of 1 is not due to model overparameterization but reflects the biological modeling decision to exclude shrinking. The Akamine-Richards ( AR) model varied widely among data sets in its relative closeness of fit, failed to converge for West Coast females, and did not yield a positive defi- nite hessian for Gulf St. Vincent females. It was less well fitting than the gVB for all but one data set. The gVB model was therefore chosen as the better 8-parameter model, and subsequent reduced models were based on it in preference to AR. From indicators summarized above, four parameters for defining overall mean length K, L^, t„ and r, appear to be too many. Two obvious candidates for fixing to constant values were t^ and r. These occurred frequently in indica- tors of model overparameterization. Both have intuitive biological default values at which they can be fixed, 0 and 1 respectively. Thus two 7-parameter models were run, with /„ and r fixed, the latter (with r=l) reducing to a seasonal von Bertalanffy curve (Somers, 1988; Hoenig and Hanumara, 1990). The fit with 1^ = 0 fixed was not significantly different from the full 8-parameter gVB for three of six data sets. This fit was tested by chi-square likelihood ratios at 95% confidence, indicated (Table 2) by successive likelihood-ra- tios of 1 .92 or less, for comparing fits of hierarchical models differing by one in number of freely estimated parameters (Rice, 1995). The regular seasonal von Bertalanffy model (r=l, abbreviated as "reg VB") fitted less closely than gVB with t^1 = 0 for all six data sets; significantly worse for all Spencer Gulf and West Coast data sets. In addition, not shown in Table 2, the estimates of tg were frequently far from the realistic biological range near 0, often estimated at 10 to 20 months above or below. The frequency and magnitude of high correlations were substantially reduced with both 7-parameter models. Thus we choose to fix t^ = 0 and let r freely vary. The occurrence of very high correlation between Sf, and Sj for all data sets and models (Table 2) suggested fixing one of them, in particular, the exponent, Sj. We set Sj = 0.3 which fell near the average among the range of estimated values. The outcome was that t^ (now free) hit its bound in three of six cases, thus exacerbating that pathology. When both ^q = 0 and s^ = 0.3 were fixed, the fits were uni- formly poor (Table 2). Thus, the high correlation between the two allometric standard deviation parameters met with no obvious solution. However, these posed no wider problem because these parameters did not interact with those describing the mean length at age and the correla- tion between pairs of allometric parameters (such as a and p in the weight-length model, Eq. 6) is common and often unavoidable. Because a significantly better fit for the length-at-age standard deviation is sought, we let both Sq and Sj vary freely. The gVB with ^q = 0 fixed provided an optimal trade-off in reduced overparameterization and good fit: it achieved the objective of substantially reduced interparameter correlations without significantly worse fits in three of six cases. The remaining three data sets (Spencer Gulf females and males, West Coast females) had yielded un- realistic estimates of tg in full model gVB (of-17.9, -21.7, and the upper bound of 7.99 respectively). This model was therefore chosen and results from it presented below. Growth: length-at-age Estimated length at age showed seasonal periodic trends for the three regions and two sexes (Figs. 2-4). Estimates of;/ (seasonality amplitude) were constrained at the maxi- mum allowed value of u = 1 for three of six data sets ( Table 3 ). The peak month of maximum growth occurred in mid- 55C Fishery Bulletin 100(3) 600 -1 500 400 - 500 400 300 200 100 females LML=300 mm LML=280 mm researcher caught fisher caught males LML=300 mm LML=280 mm 13 25 37 49 61 73 85 97 109 121 133 145 157 169 181 Age (months) Figure 2 Lengths at age. beginning at 10 months, for Gulf St. Vincent King George whiting of both sexes. The sohd line is the generalized seasonal von Ber- talanffy mean length cui-ve lEq. 2) with t^, = 0 fixed. Dashed lines indicate confidence bounds as normal likelihood standard deviation (Eq. 3). Dots represent individual obser\'ations of length-at-age. summer (December-February) for all data sets except Gulf St. Vincent males. Estimated (likelihood maximum) and observed distribu- tions of lengths-at-age were plotted for Gulf St. Vincent females (Fig. 5) and males (Fig. 6). Because the estima- tion likelihood was fitted to formulas for mean (Eq. 2) and standard deviation (Eq. 3) of all ages at once, close fit to the majority of individual length-at-age distributions in- dicated a growth description mutually consistent among ages and a satisfactory approximation to monthly lengths at age overall. For younger ages (28-35 months), the sampled frequen- cies were modeled by the truncated right-hand component of the normal pdf The LML truncation lengths for each normal curve (280 or 300 mm in Figs. 5 and 6) indicated the lower limit of samples from the fishery. These samples are gener- ally well fitted (Figs. 5 and 6), suggesting these truncated younger samples do contribute to the overall growth fit and that the truncated likelihood is effective in describing the monthly growth of the length-at-age cohort across LML into legal sizes. One exception yielding a less close fit were female fish aged 28 months taken under a LML of 280 mm (Fig. 5). Weight-at-length Parameter estimates for a and [i (Eq. 6) covaried strongly. When all four parameters were allowed to freely vary, /3 differed little from 3.2 for all data sets (Table 4). Likeli- hood-ratio tests were therefore carried out to determine whether the model allowing p to vary freely yielded sig- nificantly better fits than one with three parameters and P fixed at 3.2. For all but Spencer Gulf males, likelihood- McGarvey and Fowler Seasonal growth of Sillaginodes punctata 551 c -a _i c -a 3 -T3 o 5- >> XI CO a> 3 o T. a CJ c„ - .. "^ -C o 'S " i, u] C ■ CD O O C '^ II § -^ *. »» t^ 05 5 ^ ° -a e ^ o — w a; ra < ■:: s -a c 3 - ^ I ai cu - t> ^ 01 ":• iS - 2 .S 0, s- H S -0 S 1 °- =" " J - "^1% ^ 1 tC r; c: Q, n rt Q £ ;- c in *-> > ^n fe c 1 05 ;i o '-S o o s o fll L- ;:i CO L- c GJ O o c C^I O tuc ^ JS CQ o o X > o r/l u ti, T) c o c fl) 0. c < e s iTi X ■a X u £: -a ^ C O « S <£ >> c: CTl o S « n _c u T3 tl 03 e c u o c o wJ CJ o c f/1 3 o ^ bD J3 0) c r e +-» o to r; n" 6 tli o ^ U A 0; t/J i= X 6 *-» C m a o A u .'n CI) t> s C9 r; CU S 3 J O a> o> II o (3> O) o 05 (35 CO ^ ^ ^'" t:~ C- II II II II II II o 3 3 3 o c ^ ^ ec -H ^ V- en tn O '3 3 c? ^ c„=^ t«; ^ I :^ ^ < < •k < \ \ i«j ^ i- ^ in (TJ en 00 CTJ CO o o to o in t— c Tf 00 O to CJl en Tf in in CJ5 in o CO (35 to CM 05 CD ^ to " C^ o •* o o o ^ to 00 CSl 00 o to o o in to CM to -* 05 o in IT- CO CD o 00 CO CM to in to in CJ> to CJ5 in 00 c35 o CO in 00 to o to CO 00 CM CM 00 in in CJJ O to 00 CO CM in CM to in C^ in (35 O c-q r- c^ ^ to CO CO o CO in to CD to 00 to in o in in in in CM CO to CM CO to CM CO to CM CO to CO to CO O) CO CM CJl CO o C35 CM o CM to to to 00 CM O CO CM 00 (35 in C35 3 00 00 o CO o CO 00 o 00 00 OOOOO OOOOOOOOOO 0505(350505 CDtOtOtOt^ t^OC- XC^tDt^oO xxc^oto xc^toxc^ XC^XC--tO XXt-t^tD xr^t^tox o II =3 o II o II (0 =8 =a o II o o II II ^ -= If o 11^ CO °s .=3 o II o 11 - o „ II o II = 3 O B O S c2i (^ "3 o a (S e c*-i "3 o o (J 552 Fishery Bulletin 100(3) 600 500 400 - 300 200 females LML=300 mm LML=280 mm researcher caught fishermen caught males Figure 3 Lengths versus age of King George whiting sampled from Spencer Gulf, plotted as in Figure 2. ratio tests showed that /3 did not differ significantly from 3.2 (Table 4). With /} = 3.2 fi.xed. values of a among the six data sets differed little (Table 5, Fig. 7). Discussion In this study we estimated parameters of length at age for six data sets, comparing the fits of different models by using four standard diagnostic tools, notably likelihood ratios, correlations, standard errors, and parameter esti- mates among data sets either varying widely or varying little, to evaluate the estimators being compared. Simi- larly, for weight versus length in the same population, we assessed six data sets, and used likelihood ratio to com- pare model fits. For both length at age and weight versus length, the diagnostic analysis suggested the initial full models were overparameterized and indicated which parameters should be fixed and what values. Likelihood ratio tests allow comparisons only between hierarchical models, that is in comparing full and reduced models, the latter a subcase of the full model obtained by fixing one or more parameters in the full model. In the case of generalized von Bertalanffy and Akamine- Richards, setting r = 1 yielded the same model, namely seasonal von Bertalanffy. The other two reduced models analyzed (gVB with /q=0 and with both fg=0 and Sj=0.3) were reduced models only of generalized von Bertalanffy. Although other comparison tests are possible for hier- archical models (Quinn and Deriso, 1999), the Neyman Pearson lemma assures that in the situation where it ap- plies, the likelihood-ratio test is optimal in that it yields the most powerful test for any given choice of significance level, a (Rice, 1995). Fits of nonhierarchical models can be compared by us- ing the Akaike or Bayes information criteria (Quinn and Deriso, 1999). Fournieret al. (1998) applied the Aitkin pos- terior Bayes factors for hypothesis model comparisons in McGaA/ey and Fowler: Seasonal growth of Sillaglnodes punctata 553 500 400 - 200 females LML=280 mm LML=300 mm researcher caught fishermen caught 0 500 ^ 400 ■ 300 200 100 males LML=280mm Llv1L=300 mm 13 25 37 49 Age (months) 73 Figure 4 Lengths at age for the South Australian West Coast, plotted as in Figure 2. a Bayesian context. Recently, Buckland et al. (1997) have proposed averaging the estimates from the range of mod- els examined, weighting each estimate by functions of the Akaike or Hayes information criteria. This mitigates the need to choose one specific model as we have done, and like the Akaike or Bayes criteria, applies to nonhierarchically related models. However, in situations where the growth model will be incorporated into a larger stock assessment estimation, a range of growth submodels would be chal- lenging to implement. Moreover in cases where evidence of overparameterization is given, a reduced-parameter model can reduce both bias and the influence of sample variation. Sinusoidal seasonality (Pauly and Gaschutz^) was represented in a now standard way that preserves the interpretation of ?q as the age at which length equals zero (Somers, 1988; Hoenig and Hanumara, 1990). Pawlak and Hanumara (1991) showed that this form of seasonality model yielded statistical advantage. Hyndes et al. (1998) aged samples of King George whiting in southwestern Australia that grew faster and reached larger maximum lengths than those from South Australia. The analysis of Hyndes et al. (1998) did not describe the distribution of lengths-at-age or consider seasonality, which was less evi- dent in their plotted data. The exponent, r, allowed nonlinear variation away from the strict von Bertalanffy form. With the South Australian King George whiting data sets, it improved the model de- scription; fitted r's ranged from 1.25 to 4.59, outside the range describable by the unmodified seasonal von Berta- lanffy model where implicitly r = 1 fixed. We are not aware of previous attempts to use a truncat- ed likelihood in an age-based description although Smith and Botsford ( 1998) applied size truncation in fitting a von Foerster equation to length samples. Truncation proved effective in alleviating this potential large source of bias. The extent of the bias is evident in the growth curve scat- terplots (notably Figs. 2 and 4) where large numbers of 554 Fishery Bulletin 100(3) Table 3 Length-age parameters and derived estimates (with t„=0 fixed). The 95% the asjonptotic standard error; for strictly positive parameters, these are confidence intervals shown in parentheses are shown as percentages of the estimate value. 1.96 times Parameter Gulf St. Vincent Spencer Gulf West Coast females males females males females males L„ 467.0 418.4 492.6 416.1 454.4 387.0 (±1.5%) (±1.1%) (±2.2%) (±1.5%) (±3.2%) (±2.8%) K 0.48 0.61 0.49 0.77 0.70 1.17 (+6.4%) (±5.8%) (±6.9%) (±6.6%) (±9.4%) (±15%) u 0.59 0.46 1 1 1 0.79 (±41%) (±68%) (±0.05%) (±0.08%) (±0.14%) (±57%) (0 -1.55 (Feb) 24.0 (Apr) 8.50 (Dec) 8.67 (Dec) 8.96 (Dec) 9.02 (Jan) (±0.77) (+0.004) (±0.32) (±0.39) (±0.42) (±0.59) So 8.97 4.17 7.34 27.6 3.53 3.93 (+58%) (±47%) (±52%) (±53%) (±96%) (±100%) Sl 0.19 0.32 0.23 -0.005 0,35 0.31 (±0.10) (±0.082) (±0.091) (±0.095) (±0.17) (±0.17) r 1.25 1.41 1.66 2.35 2.33 4.59 (±6.9%) (+7.8%) (±6.9%) (±9.8%) (±11%) (+40%) Table 4 Likelihood-ratio tests companng differences in the fits between weight /3 = 3.2 fixed. Significant difference is indicated by P < 0.05. -length models which allow P to vary and those which kept Gulf St. Vincent Spencer Gulf West Coast females males females males females males j3-estimate (free to vary) L-ratio statistic P-significance (models differ) 3.200255 0.677 0.41 3.200008 0.000 0.99 3.200031 3.198721 0.005 6.094 0.94 0.01* 3.200050 0.033 0.86 3.200021 0.006 0.94 Weight (g) are 1.96 ti versus length (mm TLl parameters from Equat lies the bootstrap standard deviation over 1000 Table 5 on 6 and derived estimates with p = 3.2 fixed. 95% confidence intervals runs given in parentheses as percentages of the estimate value. Parameter Gulf St Vincent Spencer Gulf West Coast females males females males females males a 1.858-10-'= (±0.51%) 1.874-10-s (±0.637c) 1.880-10-« (±0.77%) 1.888-10-« (+0.87%) 1.861-10-6 (+0.40%) 1.887-10-'= (+0.32%) P 3.2 3.2 3.2 3.2 3.2 3.2 <7„.o -7.76 (±15%) -6.74 (±12'/,) -6.94 (±13%) -5.21 (±24%) -6.38 (±13%) -6.32 (+6.2%) <^..l 0.094 (+8.0%) 0.080 (±7.2%) 0.087 (+12%) 0.078 (±19%) 0.067 (±8.0%) 0.063 (±4.6%) McGarvey and Fowler: Seasonal growth of Stilagtnodcs punctata 555 o 28 months n =24 29 months n =15 47 months n =152 48 months 'i,„=39 34 months n.„=50 49 months H- 2aO 300 320 MO 340 IW 400 410 440 400 200 300 320 340 300 380 40O 420 440 460 200 300 320 340 160 300 400 420 440 460 Total length (m) Figure 5 Female Gulf St. Vincent length-frequency histograms expressed as proportion of fish caught per 5-mm length class. Ages (by monthi illustrated were those of largest sample size. Black bars correspond to fish caught when the fishery was subject to a 280-mm size limit, and gray bars for fish caught under a 300-mm size limit. The fitted normal likelihood probability density cur\'es (Eq. 4i calculated for each size limit have been overlaid for comparison. samples were obtained in the ages (-30 months) at which the faster fish are entering the legal-size stock. These clumps of points lie well above the estimated mean length at age. In the absence of the truncation method, the mean curves would pass through these points. Only the Akamine-Richards model failed to converge with the South Australian King George whiting data sets. Because the Richards form of the model includes both the exponent and its inverse in different places in the formula, it is likely to result in a more numerically challenging 556 Fishery Bulletin 100(3) 29 months n. =10 0 20 - 0 15 - 37 months 38 months - "300=32 39 months n =41 300 0 10 - 005 - 000 - r J i hL fl r \ bK 1 llil iiW^ 47 months 49 months n =50 ~i 1 1 \ 1 1 1 1 1 1 1 1 1 1 1 r — 280 300 320 340 360 380 400 420 440 260 300 320 340 360 360 400 420 440 280 300 320 340 360 380 400 420 440 Total length (mm) Figure 6 Male Gulf St. Vincent length-frequency histograms and fitted truncated normal density curves, as in Figure 5. minimization algorithm. The two occurrences of the pa- rameter, r, in the Richards formula L = L./ll + re''^" 'Ol"^, as a factor in front of the negative exponential and as its reciprocal exponent around the entire function apart from L^ can act in contrary fashion. For /• > 1, increases in r im- ply an increase in length-at-age because of the parameter as a reciprocal exponent, and the opposite for the second role the parameter plays in describing mean length. We speculate that the Richards growth formula was less nu- merically robust because r can rise and fall at different parameter ranges along the path to convergence, resulting in a more complex likelihood surface. Modeling the full length distribution at each monthly age is valuable for assessing populations of heavily ex- McGarvey and Fowler: Seasonal growth of Siltaginodes punctata 557 700 females ; ' 600 - ■ •■ y • / / 500 - ■ yi':-y^J^-'Je'' ■ " ,^^^^^f^^- 200 - "■^SJ*^^^- O) - 100 - I^-^^^"*"'^ "^ £: Ol a> 1 0- 0 B 700 - males QJ O " 600 - y. . 'y ^^ ' /."^^y 500 - ■ yy^y yCyy.y ■ -yjcy 400 - ■- .*^>^. ■ • ^^^Jc^'y^ ' •■, -^aJr^^..- '^&^^'- ■ 300 ■ _•■ ■ .-^ji^^^ ^' ' ' 200 - 100 • =.$0) : area + depth stratum + l-Pr(CPUE>0), gear + (area x depth stratum ) + fix year. where the number of positive catches (CPUE>0) was assumed to follow a binomial distribution. The most par- simonious models for each species were used to test the null hypothesis that there was no significant (linear) trend in the catch rate (CPUE-where-present) or frequency of occurrence (Pr{CPUE>0|) over time. The null hypothesis was rejected if the slope P was significantly different from zero at the 5% level. Results Arrowtooth flounder had the highest average CPUE (kg/km^) and the highest frequency of occurrence during all surveys (Table 2). Walleye pollock was second in most years, followed by Pacific cod and Pacific halibut. However, in 1996, the estimated CPUE of Pacific Ocean perch exceeded that of pollock, cod, and halibut. Sablefish, five other flatfish species, two Sebastes species, and Atka mackerel (Pleurogrammus monopteiygius) were other important species by CPUE (> 200 kg/km'-, Table 2). The most abundant species, including gadids and most of the flatfishes, were generally also the most widespread spe- cies in the survey (>50'7f frequency of occurrence). In spite of relatively high CPUEs, rockfishes, Atka mackerel, and yellowfin sole (Limanda aspera) had low frequencies of occurrence (4-40%), indicating high local abundances and a more restricted spatial distribution. Independent variables used in the analysis of species richness, diversity, CPUE, and species composition were moderately correlated (Table 3). The largest correlations were between area swept and year (-0.452), due to a reduction in haul duration after 1987, between depth and temperature (-0.407), and between Julian day and alongshore distance (-0.383). Depth and bottom tempera- ture on the shelf and slope are invariably confounded and their effects may generally be difficult to separate. The confounding between Julian day and AD was extreme in some years (e.g. r=-0.79 in 1993) because sampling vessels traveled from west to east during most surveys. However Mueter and Norcross Spatial and temporal patterns in the demersal fish community off Alaska 563 Table 2 All K''oiiiidfisli tax;i iiuUided in analysis and their estimated iveraRe ('PUK (kg/km'' ) by year and for all years combined. CPUEs 1 wtTi' computi'd tor oach haii , averaged by stratum, weighted by stratum area, and combine d to obtain area-wide averages FO donotos overall frcciucncy of iccurrence for all years combined in percent. Scientific name Common name 1984 1987 1990 1993 1996 Mean FO Alhcrcsllw.'i slomiu.\ ' arrowtoolh flounder 3790 3339 6499 5298 5574 4900 91 lltcwfiiv chuk<)f>niinmii ' walleye pollock 2431 3139 2873 2586 2320 2670 74 (iaJiiy nuicroci'i)luilu.\ ' Pacific cod 1876 1516 1424 1397 1821 1607 73 llil>ini\;l(issus stenolepis ' Pacific halibut 141S 1410 1118 2007 1936 1577 11 ScIhisU's (ihiuis ' Pacific Ocean perch 7.-^3 823 535 1649 2629 1278 40 Aiioploponui fitnbriii ' sablefish 753 1379 731 844 492 840 52 HipixiitUiiSoiJes eliiy.sotlim ' flathead sole 912 783 828 643 689 771 57 Lfpidopsetta spp. ' rock sole .515 668 531 590 695 600 44 Sehiistes polyspinis ' northern rockfish 135 475 365 356 337 334 20 PU'urograinmus mimopicni^ius Atka mackerel 153 98 101 73 1179 .321 5 Gtyptocephahts zuchinis ' rex sole 187 269 335 297 245 266 67 Microslomus pacificus ' Dover sole 164 263 329 291 266 263 55 Linuintla tispera ' yellowfin sole 313 192 196 277 162 228 7 Scha.sh's cilialus ' dusky rockfish 87 505 91 194 255 227 20 Sehastes aleutianus ' roughcye rockfish 153 227 165 212 156 183 34 Rajidac unident. skate (unidentified) 139 129 164 205 274 182 33 Clupca pcillasi Pacific herring 186 511 60 53 3 163 8 Sehusies zacenmis sharpchin rockfish 23 274 131 81 220 146 10 Alhalnissia peclomlis giant grenadier 115 96 81 155 175 124 3 Sehasuilohiis alascuiuis shortspinc thornyhead 126 127 68 114 177 122 24 Scbasles variegaliis harlequin rockfish 9 247 60 32 68 83 II Tluileichthys pacificus eulaehon 24 56 95 119 110 81 32 Isopsetia isotepis ' butter sole 80 71 59 102 71 77 7 Phiticlnhys stellatus ' stany flounder 50 63 35 137 93 76 5 Sehasles borealis shortraker rockfish 59 137 43 69 69 75 8 Sebasles proriger redstripe rockfish 18 90 92 101 51 71 4 Sqiuihis acanthias spiny dogfish 34 34 65 114 96 69 16 Ht'inilepidotits jonUiiu yellow Irish lord 49 46 40 40 61 47 27 Sehastes hrevispinis silvergray rockfish 16 18 48 65 82 46 7 Ophiodon elongauis lingcod 15 32 35 53 80 43 1(1 Heniilnpienis holini bigmouth sculpin 54 35 29 19 14 30 18 Laitiiui dilriipis salmon shark 27 43 42 26 11 30 1 Mxiixocephahis spp. sculpin (unidentified) 37 23 18 IT 30 26 9 Pawphi-ys vetulus ' English sole 11 28 26 28 15 -)~i 7 Siininiiisiis pacificus Pacific sleeper shark 1 1 6 29 72 22 1 Ziipriira sileniis prowfish 13 24 15 28 23 21 11 Pleuniiiectes c/uadriluljerciilatus ' Alaska plaice 7 16 20 9 17 14 4 Microgadus proximtis Pacific tomcod 5.03 34.67 6.32 9.01 5.78 12.16 -) Sehastes babcoclii redbanded rockfish 4.88 6.21 11.20 12.53 15-54 10.07 11 Hxdrolagus colliei spotted ratfish 13.06 8.16 6.32 7,08 12.65 9.45 6 Hexagrammns decagrainmus kelp greenling 3.89 8,17 13.87 8.21 6.69 8.16 8 Sebasles reedi yellowmouth rockfish 1.69 0.89 6.39 12.14 3.15 4.85 1 Sehastes ruhenimus yelloweye rockfish 2.10 10.17 3.29 3.88 3.70 4.63 3 Sebastes helvomaculatus rosethom rockfish 1.99 5.12 2.50 2.54 7.71 3.97 4 Sehastes melanops black rockfish 1.14 3.75 4.65 0.86 7.89 3.66 1 Batliymasler signatus searcher 2.08 2.05 5.39 3.04 4.38 3.39 15 Trichodon trichodon Pacific sandfish 7.56 3.71 2.45 1.76 0.52 3.20 3 Malacocottus spp. sculpin (unidentified) 4.33 2.97 2.40 3.55 1.63 2.97 14 Cnpiacanthcdes giganteus giant wrymouth 0.76 3.26 3.98 1.17 2.77 2.39 1 Lwiides paleans watlled eelpout 0.86 1.75 3.10 3.16 2.30 2.23 II continued 564 Fishery Bulletin 100(3) Table 2 (continued) Scientific name Common name 1984 1987 1990 1993 1996 Mean FO Eopsetia jonlani petrale sole 0.97 0.83 0.61 2.77 3.68 1.77 T Malloltts villosus capelin 1.47 0.17 0.51 0.42 4.98 1.51 7 LxcoJes brevipes shortfin eelpout 0.13 0.84 0.79 2.41 2.17 1.27 8 DasYcotlus seliger spinyhead sculpin 1.33 0.49 0.89 1.88 0.83 1.08 11 Aptocvcius veniricflsiis smooth lumpsucker 0.74 1.29 1.77 0.37 0.56 0 45 t Podotheciis acipensenmts sturgeon poacher 0.19 0.42 1.23 2.. 34 0.33 0.90 5 Sebasies ehmgunis greenstriped rockfish 0.05 0.22 0.59 0.91 1.20 0.59 1 Sebasles wilsoni pygmy rockfish 0.00 1.38 0.30 0.01 0.96 0.53 1 Myctophidae lantemfish (unidentified) 0.19 0.19 0.11 1.52 0.53 0.51 4 Trii^htps spp. sculpin (unidentified) 0.37 0.32 0.68 0,40 0.61 0,48 6 Stichaeidae prickleback (unidentified) 0.58 0.05 0.51 0.70 0.41 0.45 4 Sebasles crameri darkblotched rockfish o.o: 0.13 0.59 0.99 0.41 0.43 1 Hexagrammos slelleri whitespotted greenling 0.17 0.56 0.64 0.33 0.27 0.39 1 L\opsena exilis slender sole 0.10 0.07 0.29 0.77 0.59 0.37 5 Cyclopteridae (Liparidinael snailfish 0.30 0.23 0..30 0.28 0.04 0.23 3 Hemilepidolus hemilepiJotiis red Irish lord 0.07 0.29 0.05 0.32 0.10 0.16 1 Cottidae sculpin (unidentified) 0.25 0.00 0.28 0.08 0.01 0.12 1 Gymnoctiiilluis spp. sculpin (unidentified) 0.16 0.03 0.00 0.06 0.06 0.06 1 Eumicrntremus orbis Pacific spiny lumpsucker 0.02 0.08 0.03 0.07 0.04 0.05 -> Lycodes duiplerus black eelpout 0.08 0.05 0.00 0.01 0.02 0.03 2 Sarnlor freitaUis sawback poacher 0.03 0.03 0.01 0.04 0.04 0.03 2 Balhxagojuts lugripiiuns blackfin poacher 0.02 0.01 0.01 0.01 0.00 0.01 1 Total 14,7X6 7,286 17.4.^6 IX,4()() 20,662 17,714 ' Catches for 1984 and 1487 were adjusted bv using hstiing power coefficients in Tables 2S and ."^ 1 in Munro and Hoff ( 1995) Other fiati shes were adjusted by using rock sole coefficients Table 3 Paii^isf Pearson s correlation coefficients among independent variables used in the statistical analysis. Depth Alongshore distance Gear temperature Julian day Year Alongshore distance -0.326' Gear temperature -0.407* -0.203* Julian day 0.129* -.383* 0.330* Year -0.026 -0.035 -0.266* -0.306* Area swept 0.012 0.184* 0.019 0.145* -0.452* ♦ Highly significant |P= TOll the direction of sampling was reversed in some years, pro- viding some contrast. Species richness, as measured by the number of spe- cies per haul, was initially modeled as a Poisson variable (counts). However, an analysis of the distribution of the observed numbers and of residuals from the fitted models suggested that it could more appropriately be modeled as a normally distributed variable (see below). We fitted models assuming either a Poisson or a normal distribu- tion and found that the estimated patterns were virtually identical and in both cases the same model was selected as best model. Only results based on the normal distribution are presented. Species richness was highly variable and the best model explained a small portion of the overall variability (Pseudo-r2=0.25). The average number of spe- cies increased with area swept (Fig. 2), which may be ex- pected if species area relationships established for terres- trial ecosystems also hold in the marine environment. The number of species tended to peak at intermediate depths (200-300 m) and was highest in the eastern GOA (Table 4), decreasing steadily west of Prince William Sound (km 1100, approximately 147''W). Species richness appeared Mueter and Norcross Spatial and temporal patterns in the demersal fish community off Alaska 565 2 f ^ \\ 0 i / \S ^ \\ -/ V\n -2 / \\ " \\ -4 \ 0 100 200 300 400 500 Depth (m) 2' \ 0 2 \s^^ \ 4 140 180 220 260 Julian day 2^ 0 4*10' SMO' 12*10' Area swept (m2) SE Yak Kod O Shu 2 0 2 s v^ - \ 1000 2000 Alongshore distance (km) ■84 '87 '90 '93 '96 Year 160 172 Gear type Figure 2 Estimated trends in number of species per haul by depth, alongshore distance, Julian day, year, area swept, and gear type. Alongshore dis- tance was measured along the 200-m depth contour from east to west. Major geographic regions from east to west are Southeast (SE), Yaku- tat (Yak), Kodiak (Kod), Chirikof (Chi), and Shumagin (Shu). Dashed lines indicate approximate 95% confidence limits of the regression lines. Horizontal lines in lower right plots indicate mean response with 95% pointwise confidence intervals. Width of bars is proportional to number of observations. Fitted lines in each panel are adjusted for the effects of all other variables. Standardized effects in each plot {y-£ixes, no units) are on the same scale for comparison. Effects are standardized because the estimated CPUE at a given value of a vari- able is dependent upon the levels of all other variables. to decrease during the last month of the survey season and showed some variation across years (Fig. 2). There was no consistent trend in species richness with tempera- ture (not shown). One of the Japanese trawls (gear type 717, Fig. 2) tended to have a much smaller number of species per haul on average, suggesting a low catchability for at least some species. Residuals from the best model were close to normally distributed (Fig. 3, A and B) and showed no apparent trends with any of the covariates or over time. Only a very small portion of the variability in species diversity (Shannon-Wiener index) was accounted for by the best regression (Pseudo-7-2=0.17). Depth had the stron- gest effect on species diversity with highest diversities ob- served at intermediate depths (Fig. 4). Diversity generally increased with the number of species, which showed a very similar trend with depth (Fig. 2). Diversity was higher in the eastern Gulf and decreased west of Prince William Sound (Figs. 4 and 5). It showed little variation over time, both within the survey period and across years. However, 566 Fishery Bulletin 100(3) B - 6 D 8 10 12 14 (N 1.4 1,6 1,8 2.0 2,2 24 ^ F CNJ -.^n:;^*^^. CSI , '$fT^WW t . •'" ' ; . ■„ <9 -2 0 2 Quantiles of standard normal 80 85 90 95 Fined values 100 Figure 3 Quantile-quantile plots (left), showing sorted residuals (points) against corresponding quantiles of the standard normal distribution, and plots of residuals against fitted values with smoothing spline (right) for best-fit models of species richness (A, B), species diversity (C, D), and total CPUE (E, F). Points in Q-Q plots fall on the indicated straight line if data are normally distributed. Table 4 Average number of groundfish species caught per haul by depth stratum and geographic area. not adj usted for effects of other variables. Depth stratum (m) No. of species/haul (SE) Area No. of species /haul (SE) 0-100 9.5(0.12) Shumagin 9.7(0.11) 100-200 11.5(0.08) Chirikof 10.5 (0.12) 200-300 13.2(0.13) Kodiak 11.6(0.11) 300-500 10.9(0.15) Yakutat 12.5(0.14) Southeast 12.8 (0.20) Mueter and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 567 ,^ 0.4- 0 SE Yak Kod O Shu ^^^'^ \\ A e -0.4- > / \\ -0.4- \ \ S 0 100 200 300 400 500 0 1000 2000 g. Depth (m) Alongshore distance (km) to o 1 0.4 LU 0.4' \ 0 \^ . - - ' 0 — ; — -*- -r -»- -0.4- -0.4 140 180 220 260 '84 '87 '90 '93 '96 Julian day Year Figure 4 Trends in species diversity (Shannon-Wiener index) by depth, along- shore distance, Julian day, and year (area swept and gear type were not significant). For details see Figure 2. 95% confidence intervals suggest a significant drop in species diversity between 1993 and 1996. Area swept and temperature did not enter the best model, suggesting they had no strong effect on species diversity. Diagnostic plots indicated approximate normality and no apparent trends in the residuals (Fig. 3, C and D). Total biomass (CPUE) showed strong trends with depth and alongshore distance (Fig. 6) but had high variability around these trends (Pseudo-r-=0.14). Average CPUE in- creased sharply with depth to about 150 m — a trend that reflected, on average, approximately a doubling of CPUE between the shallowest sampling stations and stations at 150-200 m depth. A similarly strong gradient exists in the alongshore direction between Yakutat (km 700) and the Kodiak Island area (km 1500), where total CPUE has a pronounced maximum. Estimates of the spatial trend suggested that highest CPUEs are found around Kodiak Island, particularly in Shelikof Strait, along the shelf break and upper slope, and on the banks and in the gullies northeast of Kodiak Island (Fig. 5). Total CPUE appeared to decrease significantly after approximately Julian day 240 (Aug. 28); however, Julian day was confounded with alongshore distance (r =-0.38, Table 3) and the effect could in part be due to lower abundance in the eastern GOA, which was often sampled later than other areas. Gear effects were significant and indicated a lower catch rate for one of the Japanese trawls (gear 717, Fig. 6). The esti- mated time trend (Fig. 6) indicated a significant increase in total CPUE between 1984 and 1996 (P<0.001), based on a chi-square test of the difference in residual deviances (Hastie and Tibshirani, 1990). We estimated that the com- bined CPUE of all groundfish species included in the anal- ysis increased from a gulf-wide average of 14,786 kg/km^ in 1984 to 20,600 kg/km- in 1996 (Table 2), representing an increase of 40%. No significant effect of temperature on total CPUE was found. Diagnostic plots indicated approxi- mate normality and no apparent trends in the residuals (Fig. 3, E and F). The biomass of 72 groundfish species in 240 stratum/ year combinations could effectively be summarized by an NMDS ordination in three dimensions (stress=0.085). Our findings suggested that the main gradients along which species composition varied were depth and alongshore distance, whereas temperature and temporal gradients (within and between years) had relatively minor effects on these indices of species composition. The first two dimen- sions (axes 1 and 2) of the ordination accounted for 67% and 19% of the overall variation respectively. An ordina- tion plot of the first two axes indicated that the sample strata were most clearly separated along the depth gradi- ent (Fig. 7). The alongshore gradient was roughly perpen- dicular to the depth gradient and clearly distinguished ar- eas Southeast from the other areas (Fig. 7). Yakutat strata occupy an intermediate position between the Southeast strata and strata in other areas. Kodiak, Chiniak, and Shumagin were similar in species composition, as sug- 568 Fishery Bulletin 100(3) o CD 00 in CD in CM in o CD 00 in CD in in c\j . in Estimated spatial trend in species diversity -170 -160 -150 -140 Estimated spatial trend in total CPUE (log scale) -130 Bering Sea -170 -160 -150 Longitude ("W) -140 -130 Figure 5 Spatial trends in species diversity (Shannon-Wiener index) and total CPUE. The trends are based on pre- dictions from the best regression models for diversity and total CPUE (see Fig. 4 and Fig. 6). gested by a substantial overlap in the location of strata in the ordination diagram. A three-way analysis of variance for each of the three axes by depth stratum, area, and year indicated highly significant differences for all three axes between depth strata and geographic areas, but weak and nonsignificant differences among years (Table 5), suggest- ing a relatively stable species composition from 1984 to 1996. Residuals from the ANOVA for axis 1 and 3 were approximately normally distributed and showed no appar- ent violations of the ANOVA assumptions. Residuals for axis 2 had a very long-tailed distribution, but an analysis of variance based on ranks yielded very similar results; therefore the conclusions appear to be robust. Ordinations of individual hauls within each year con- firmed that depth and alongshore distance explained much of the variation in species composition within years. The species composition in each year was effectively sum- marized by ordinations in five dimensions with Kruskal's stress values that were very close to 0.1 in all years (axes 1-5 in Table 6). The first axis of all five ordinations ac- counted for A2-Al'7c of the variation in species composi- tion. This axis, as well as most of the other axes, was most strongly related to the depth gradient in all years (Table 6). Alongshore distance appeared to explain a relatively small proportion of the variance in species composition. However the pseudo-coefficients of partial determination for alongshore distance typically increased substantially if Julian day was excluded from the model because of high correlations between Julian day and alongshore distances. Regressions of the axes on depth and alongshore distance alone resulted in pseudo-r^ values that were very close to those from models that included Julian day. Furthermore, residuals from these models were not significantly related to Julian day for any year (P>0.5) based on linear regres- Mueler and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 569 U.b / /-?<:.- ''^ 0.0 ^^--. 1.0 / 0.5- / SE Yak Kod O Shu 0.4 1 r^!^ 1 7 ^ bv_/ 0.0 / ~^v 1 V / \ 0.4 ^J /'^ 0.8 1 100 200 300 400 500 Depth (m) 0 1000 2000 Alongstiore distance (km) a o 0.2 0.2- -:^ -zr .r^ -5- ~ 0.6 \\ \ 1.0 0.6- 0.2- -0.2- -0.6- 140 180 220 260 Julian day '84 '87 '90 '93 '96 Year 0.4 0.0 -0.4 -0.8 160 172 Gear type Figure 6 Trends in total CPUE by depth, alongshore distance, Julian day, year, and gear type. For details see Figure 2 o 0-100 m • 100-200 m X 200-300 m A 300-500 m -06 -0.4 -0.2 0 0.2 04 06 Axis 1 Tt D Shumagin o Cliirikof • Kodiak X Yakutat A Southeast A A A • .• 2'*>^''*^A Jo O A O •0 6 -0.4 -0 2 0 0.2 04 0 6 Axis 1 Figure 7 Plots of the first two axes from an NMDS (mul- tidimensional scaling) ordination of 48 strata sampled during each of fivfe years. Distances between two points (strata) in the ordination dia- gram approximately reflect their dissimilarity in terms of species composition. Symbols indicate depth strata (A) and geographic area (B). sions). These findings suggest that species composition within the summer remains relatively stable over the 14—16 week survey period. To examine effects of temperature on species composi- tion we used only those hauls from the 1993 and 1996 data for which temperature measurements were avail- able (n=726 and «=716 respectively, Table 1). Results for other years are not reported because of the relatively small number of temperature measurements. The avail- able temperature data for earlier years were often local- ized in certain strata, whereas large areas had few if any measurements. The lack of contrast did not allow definite conclusions with regards to the temperature effect. For 1993 and 1996, we repeated the GAM analysis for the five axes of species composition with temperature included. Temperature effects were apparently large. However, wide confidence intervals and inconsistent patterns between years suggest that the apparent effects were at least in part due to confounding of temperature with alongshore distance and depth. Therefore we first adjusted for the effects of depth and alongshore distance using the re- gression models summarized in Table 6. Temperature explained a small but significant portion of the remaining residual variation in three of the ordination axes in both 1993 and 1996 (Pseudo-r^ values <0.1). To identify which species were most strongly related to the major gradients we computed rank correlations between each species and the three axes of species compo- sition based on the NMDS ordination of strata averages. We considered all species that had a positive or negative rank correlation of at least 0.4 with one of the axes for further examination, based on a visual inspection of scat- terplots. Species associated with the first two axes were clearly separated along gradients of depth and alongshore distance (Figs. 8 and 9). The first axis was positively cor- related with a number of deep-water species that were 570 Fishery Bulletin 100(3) Table S Analysis of variance of three axes of species composition by statistical area, depth stratum, and year Effect df Sum of squares Mean squares F-value P-value Axis 1 Depth stratum 3 129.793 43.264 613.856 0.000 Area 4 10.350 2.588 36.713 0.000 Year 4 0.665 0.166 2.358 0.056 Depth stratum:area 11 1.756 0.160 2.265 0.014 Depth stratum:year 12 0.412 0.034 0.487 0.920 Area:year 16 0.291 0.018 0.258 0.998 Depth stratum:area:year 44 0.839 0.019 0.271 1.000 Residuals 145 10.220 0.070 Axis 2 Depth stratum 3 6.601 2.200 37. .581 0.000 Area 4 27.771 6.943 118.587 0.000 Year 4 0.346 0.087 1.478 0.212 Depth stratum:area 11 1.091 0.099 1.694 0.080 Depth stratum:year 12 0.366 0.030 0.521 0.899 Area:year 16 1.017 0.064 1.086 0.374 Depth stratum:area:year 44 0.815 0.019 0.317 1.000 Residuals 145 8.489 0.059 Axis 3 Depth stratum 3 0.478 0.159 1.231 0.301 Area 4 9.541 2.385 18.423 0.000 Year 4 0.615 0.154 1.188 0.318 Depth stratum:area 11 3.080 0.280 2.163 0.019 Depth stratumiyear 12 0.212 0.018 0.137 1.000 Area:year 16 0.698 0.044 0.337 0.992 Depth stratum;area:year 44 1.365 0.031 0.240 1.000 Residuals 145 18.773 0.129 more abundant in the eastern GOA, but generally had a broad geographic distribution. The deep-water species included several rockfish species, rex sole iGlyptocephalus zachiriis), Dover sole {Microstomiis pacifictis), sablefish, and myctophids. The alongshore distribution of CPUE suggests that many species displayed local maxima in biomass on the scale of 100-200 km (Fig. 8). The first axis was negatively correlated with a group of shallow-water species that were typically more abundant in the west- ern GOA. The shallow-water group included a number of flatfish species, gadids, and sculpins. Many of the shallow water species had pronounced peaks in biomass at the lon- gitude of Kodiak Island (km 1400-1700) and between the Shumagin Islands and Sanak Island (km 2000-2200). The second axis was positively correlated with a "shelf break" group. Most of the species in this group had a pronounced peak in biomass near 200 m and were found primarily in the eastern GOA (Fig. 9). The group included a number of rockfish species, as well as lingcod iOphiodon elongatus), petrale sole (Eopsetta jordani), and slender sole {Lyopsetta exilis). Species that were negatively or positively associated with the third axis (not shown) were not differentiated along gradients of depth and alongshore distance, suggesting that the third axis, which accounted for 14% of the overall variation, was related to other, un- identified gradients. Finally, we examined trends in species composition over the 12-year period from the first survey in 1984 to the most recent survey in 1996 in more detail based on the NMDS ordination of species CPUEs averaged by stratum (Fig. 7). Although the individual indices of species composition were not significantly different among years (Table 5), a canonical correlation between the ordination axes and time (survey years) indicated a highly significant trend (P<0.001) in species composition. The linear combi- nation of ordination axes that maximized the correlation with survey year was used as an index for the trend in species composition ("time index"). The index increased in most areas and depth strata between 1984 and 1996. Eight species that were positively correlated with the time index showed significant and increasing trends in either CPUE-where-present or frequency of occurrence for at least two area-depth stratum combinations (Fig. 10). Skates were most strongly associated with the index and showed a widespread increase in CPUE-where-present and frequency of occurrence, particularly in the Chirikof, Mueter and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 571 Table 6 Ki'sults from NMDS l multidimensional scaling) ordinations of species CPUEs in five dimensions (axes 1-5) by year, and GAM (generalized additive models! results for regressions of axis scores on depth, alongshore distance, and Julian day. Variance indicates proportion of overall variation accounted for by each axis. Stress indicates Kruskal's stress values for each ordination. Numbers for the full model indicate pseudo-;- values for the best model fit ( 1 - deviance of the best fitting model/deviance of the null model). Numbers following individual variables indicate pseudo-coefficient of partial determination, computed from a reduced model that excluded the variable (= 1 - deviance of best model / deviance of reduced model). Only coefficients of determination exceeding 0.1 are shown. Coefficient of determination for Julian day did not exceed 0.1 in any year or for any axis. AD = alongshore distances. Year A.xis I Axis ; Axi.s 3 Axi.s 4 Axis 5 Stress 1>)S4 1487 I QMO 1993 1946 Variance I'ull model Depth AD Variance Full model Depth AD Variance Full model Depth AD Variance Full model Depth AD Variance Full model Depth 44% 19% 14%' 12% 11% 0.84 0.28 0.26 0.58 0.13 0.74 U.I 9 0.42 0.12 0,13 0.12 42% 22% 14%. 12% 10% 0.77 0.22 0.32 0.36 0.21 0.67 0.11 0.28 0.19 0.17 0.15 0.24 42% 21% 1.5% 12% 10% 0.83 0.37 0.29 0.34 0.16 0.69 0.23 0.17 0.27 0.10 O.Il 0.12 0.17 47% 18% 13% 11% 11% 0.81 G.24 0.37 0.49 0.31 0.71 0.26 0.42 0.21 0.27 44% 20% 14% 12% 10% 0.78 0.22 0.31 0.55 0.34 0.71 0.11 0.18 0.40 0.17 0,102 O.IOO 0.096 0.098 0.099 Kodiak, and Yakutat areas ( Figs. 10 and 11). Another elas- mobranch (Pacific sleeper shark, Somniosus pacificus), two osmerids (capeHn, Mallotus villosus, and eulachon, Thaleichthys pacificus), three flatfish species (Dover sole, rex sole, and arrowtooth flounder), and one rockfish spe- cies (Pacific Ocean perch, Sebastes alutus) increased sig- nificantly in one or more areas and depth strata (Fig. 10). The frequency of occurrence of capelin in the trawl survey increased significantly, whereas CPUE-where-present did not change or. in some cases, decreased significantly. Clos- er examination revealed that the increase in frequency of occurrence occurred primarily between 1993 and 1996 (Fig. 12). Capelin were caught in 14% of all hauls in 1996, compared to only T^, in 1993. Changes in the CPUE- where-present of three flatfish species differed among species. Although Dover sole increased primarily in the eastern GOA (Yakutat and Southeast), rex sole increased most strongly in the Chirikof area (Fig. 10). In contrast, arrowtooth flounder increased in all areas, but only in the 0-100 m depth stratum. Although all three flatfish species increased significantly over time, estimates of their gulf- wide average CPUEs were highest in 1990 and declined from 1990 to 1996 (Table 2). Both the frequency of occur- rence and CPUE-where-present of Pacific Ocean perch increased strongly in the Chirikof area, and frequency of occurrence decreased in the Yakutat area. CPUE-where-present of the four species that were nega- tively correlated with the time index did not decrease sig- nificantly, and, in the case of rock sole (Lepidopsetta spp.), even increased in some areas (Fig. 13). However, frequency of occurrence decreased significantly for three of the spe- cies, particularly for bigmouth sculpin (Hemitripterus holini. Fig. 13). Discussion Research survey data provide standardized indices of rela- tive abundance that often track abundance trends more accurately than methods that use commercial catch data (Pennington and Stromme, 1998). However, results need to be interpreted with caution because the catchability of species may change over time owing to changes in survey gear, vessel type, spatial distribution, or the length composition of a population. In our study we were par- ticularly concerned that differences in gear type (Table 1) might bias our results. We opted to include gear type directly into the analysis as a "nuisance" variable, instead 572 Fishery Bulletin 100(3) Sebasto/obus afascanus Mafacocotfus sp Micrustomus pacfficus Sebastes ateutianus Sebastes babcocki Sebastes boreafis Sebastes a/ufus Anopfopoma fimbria Myctophidae A/batrossfa pectorafrs Lyopsetfa exfTrs Sebastes zacentrus Gfyptocepfia/us zachiws Sebastes hetvomacufatus BGymnocantftus sp Bafhymaster sfgnatus Trigtops sp. Parophrys vetufus Tncliodon trichodon Mrcrogadus projcrmus MafTofus wTTosus Rajidae unident Theragra chafcogramma Podothecus aclpenserinus Pfeuronectes quadrftubercufatus Pfatfchthys stefTatus Hexagrammos decagrammus Lrmanda aspera Isopsetta fsofepis Gadus macrocephafus Hippogfossus stenofeprs Myoxocephafus sp Hippogtossoides efassodon Hemdepfdotus Jordan! Lepfdopsetta sp Kt>">iW 4 ■■ ♦■ t£ *- ■♦«► ¥^ if»#Hn.>- I I t^ii 200 400 Depth (m) 0 1000 2000 Alongshore distance (km) Figure 8 Distribution of CPUE by depth and alongshore distance for all species that had a strong positive (A) or negative (Bl association with the first index of species composition derived from an NMDS ordination of abun- dances averaged by strata. Widths of dark bands are proportional to average CPUE of a given species. Average CPUE as a function of depth and alongshore distance was estimated by using a scatterplot smoother (cubic smoothing spline). Degree of smoothing was determmed by cross- validation for each species separately. of standardizing the CPUE of each species to a common gear standard prior to analysis because of large uncertain- ties in the estimation of fishing power coefficients (Munro and Hoff, 1995). The estimated differences among gear types agreed qualitatively with results in Munro and Hoff ( 1995 ) for those species for which these authors estimated fishing power. Nevertheless, differences among gear types may have biased some of our results. In spite of these problems, the survey data used in our study provided the best available indicator of relative changes in species com- position and distribution for many species of ecological or commercial interest. The observed trends in species richness, diversity, and biomass, as well as in the CPUE of individual species, showed that depth is an important gradient structuring the groundfish community in the GOA, and that the shelf break and upper slope (at 150-300 m) is a particularly important depth range. Strong depth-dependent gradi- ents are found in many other demersal fish communities inhabiting shelf and upper slope regions (Colvocoresses and Musick, 1984; Gomes et al., 1992; Blaber et al., 1994; Fujita et al.. 1995; Jay, 1996; Farina et al, 1997; Mahon et al., 1998). Distinct depth preferences of many individual species (Fig. 8) result in a turnover of species along the depth gradient and lead to the observed patterns in rich- ness, diversity, and biomass. Similar to the altitude gra- dient in terrestrial environments (Brown and Lomolino, 1998), depth appears to be the major ecological gradient structuring benthic communities in the ocean from shal- low, nearshore areas (Mueter and Norcross, 1999) to the abyssal plain (Merrett, 1992). Our results suggest a pronounced peak in species rich- ness, diversity, and total biomass at intermediate depths. Species richness as well as biomass of demersal fish com- Mueter and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 573 Ophiodon e/ongafus Squafus acanfhras Sebastes brevJspinls Sebasfes vahegatus Sebasfes zacenfrus Sebastes hefvomacufafus Parophrys vefufus Sebastes proriger Sebasfes rubemmus Lyopseffa exilis Sebastes efongatus Hydrofagus coffiei Sebastes cflFafus Eopsetta jordani B Sebastes ateutfanus Hemitripterus bofmi A/batrossia pectora/is 200 400 Depth (m) 0 1000 2000 Alongshore distance (km) Figure 9 Distribution of CPUE by depth and alongshore distance for all spe- cies that had a strong positive (A) or negative (B) association with the second index of species composition derived from an NMDS ordination of abundances averaged by strata. For details see Figure 8. munities are generally higher on the continental shelf than on adjacent slope regions, reflecting a strong gradient with depth (Day and Pearcy, 1968; Colvocoresses and Mu- sick, 1984; Farina et al., 1997). In contrast to our results for the GOA, these studies did not find peaks in richness or biomass at intermediate depths. This may have been partly the result of aggregating data into relatively large depth ranges in those studies (and the associated loss in resolution), whereas our study examined patterns in rich- ness and biomass along a continuous depth gradient. We suggest that the high biomass of demersal fishes near the shelf break (in particular the biomass of flat- fishes, rockfishes, and gadids) reflects favorable feeding conditions at this depth range. Favorable feeding condi- tions may result from enhanced benthic productivity or availability of suitable prey at this depth. The region along the shelf break and over the upper slope has high primary productivity that may be driven by shelf-break fronts, seasonal upwelling, strong alongshore currents, and tidal mixing (Parsons, 1986). Shelf-break fronts have frequently been observed in the GOA (Weingartner^) and can enhance phytoplankton and zooplankton biomass, particularly when coupled with upwelling (Mann and Lazier, 1991), which is common during the summer in the western GOA (Reed and Schumacher, 1986). We speculate that increased production in the water column becomes available to the benthos when sinking particles are consumed by larger 2 Weingartner, T. J. 2000. Personal commun. Institute of Ma- rine Sciences, University of Alaska Fairbanks, Fairbanks, Alaska 99775. organisms such as shrimp or euphausiids, which are often concentrated near the shelf break and are consumed by demersal fish (Mackas et al., 1997; Robinson and Gomez- Gutierrez, 1998). The most abundant demersal fishes in the GOA, arrowtooth flounder and walleye pollock, as well as several rockfish species, have been shown to feed near or off the bottom on euphausiids, mysids, copepods, shrimps, and other fishes (Yang, 1993). Thus, a region of enhanced benthic productivity near the shelf break is consistent with the total biomass of groundfish being highest just inshore of the shelf break between 150 and 200 m (Fig. 6). In addition to depth (or elevation), latitude is a major ecological gradient in both marine and terrestrial environ- ments. Demersal fishes on the continental shelf off both the east coast of North America (Mahon et al.. 1998) and the U.S. west coast (Jay, 1996) form loose assemblages that are clearly separated by latitude. Although the latitudinal extent of our study region is relatively small, we found strong alongshore (roughly corresponding to longitudi- nal) differences in community structure from the eastern GOA to the Aleutian Islands, with the steepest gradient between the Yakutat and Kodiak areas. This gradient was apparent in species richness and diversity (Figs. 2 and 4), total CPUE (Fig. 6), and indices of species composition (Fig. 7). The alongshore gradient is perpendicular to the depth gradient, with depth generally increasing in the offshore direction. However, these two variables were moderately correlated (-0.32, ranging from -0.20 to -0.44 in individual years), primarily due to the larger number of shallow stations in the western part of the study area and a relative lack of shallow stations in Southeast Alaska. 574 Fishery Bulletin 100(3) Rajidae (skates) Somniosus pacifiais Depth (m) 5 )huniagin ( 'hirikot Kodiak Yakut; tsouUiea\l 0- 100 CPUE FO ++ ++ 100-200 CPUE FO + ++ + ++ ++ ++ 200 - 300 CPUE FO ++ ++ + + ++ 300 - 500 CPUE 1 FO -H- + 1 ''■'•^■' tz hkillotus villosiis Tludcichilns /)iiLifuii\ Sliumagin Cliinkof Kodiak \akutalSoiitheast Shumagin Chirikot Kodiak ^^akulal SniulKasi )nuniagin LIliriKOI KOUiaK 0- 100 CPUE FO -- -I-+ + ++ + + 100-200 CPUE FO + -l~l- -t-h ++ 200 - 300 CPUE FO 1 -- -- ++ 1 ++ ++ 300 - 500 CPU WKKKKM FO tmmymmm^^^ Micros toinus pacijkiis (jlyphKepludus zachinis Depth (111) Shumagin Chirikof Kodiak 'I'akutatSoutheast Shumagin Chinkof Kodiak Yakutat Southeast '- ' 1 * . iiiii!i»i.»ii»m I ' , , 1 0- 100 CPUE FO + - 100-200 CPUE FO ++ -I-+ + + ++ - 200 - 300 CPUE -H- FO 300 - 500 CPUE FO -f -H- + - + -l-l- -l-l- ++ Atheresthes s!omiii.s Sehustcs alutiis Shumagin Chirikol Kodiak YakutatSoutheast Shumamn Chinkof Kodiak ^'akutat Southeast 0 - 1 00 CPUE FO ++ ++ 4-f -I-+ -f+ -1- 100-200 CPUE FO 200 - 300 CPUE FO 300 - 500 CPUE FO -1- _ -1- -H- -H- -l~(- _ + -t-l- -H- - T Figure 10 Changes in CPUE-where-present and frequency of occurrence (FO) by depth stratum and area for eight species that increased significantly over time. Significant changes at the b"i level are indicated by '+' or '-'. Changes that were significant at the Y7t level are indicated by '++' or '-'. Areas-depth strata where a species was not caught in any of the five survey years are shaded. Significance was tested by using generalized linear models including gear effects (see text). Therefore, the alongshore trends may be partly due to differences in the distribution of sampling depth between the eastern and western GOA resulting from differences in topography. Nevertheless, we do not believe that the magnitude of correlations (Table 3) presents a serious problem for the general patterns that were observed for CPUE, diversity, and species composition. Alongshore patterns in CPUE (Figs. 5 and 6) indicate a higher biomass of demersal fishes in the western GOA, particularly in the Kodiak Island area. If the higher bio- mass is related to the availability of food, this pattern may reflect higher benthic productivity in the Kodiak Island region. There are several pieces of evidence that suggest a higher productivity in this region. First, there is a higher benthic biomass and productivity of infaunal organisms on the Chirikof and Kodiak shelf compared with the Yakutat and Southeast areas (Semenov, 1965, as cited in Feder and Jewett, 1986). Second, enhanced primary pro- ductivity in the western GOA is evident in estimates from coastsl zone color scanner (CZCS) data and from more Mueter and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 575 60°N ^HV^ W^ ■31 H 58N • ^ 56N ^-*"' 1984 *. 54N \ 52 N • 60 N 58 N 56N 54°N 52-N 1 70 W 160W 150W 140'W 60 N 58N 56"N ^ ^ s • 1990 ■ * •• b4N 52N • \ 170W 160'W ISO'W 140°W ''tS, %^ :t?«" ». •. 1996 170"W 160'W 150"W MO'W Figure 11 All sampling locations l • ) where CPUE of skates (Rajidae) exceeded 300 kg/km2 in 1984, 1990, and 1996. recent SeaWifs data (Falkowski et al., 1998, see web site: http://seawifs.gsfc.nasa.gov/SEAWIFS.html). Third, the east-west gradient in productivity and biomass is consis- tent with differences in upwelling, which is generally more pronounced and more frequent in the western GOA ( Reed and Schumacher, 1986). Fourth, the presence of strong tidal currents in the vicinity of Kodiak Island (Kowalik*) (see also Isaji and Spaulding, 1987) may further enhance demersal productivity (Mann and Lazier, 1991). Fifth, strong alongshore currents, advection, and nutrient input from major rivers may contribute to a higher productivity in the western GOA and around Kodiak Island. Finally, lower productivity in the eastern GOA may be related to the limited shelf area in this part of the Gulf Broader shelf areas such as that in the central Gulf of Alaska around Kodiak Island are generally believed to be more productive because of increased nutrient enrichment. •^ Kowalik, Z. 2000. Personal commun. Institute of Marine Sciences, University of Alaska Fairbanks, Fairbanks, Alaska 99775. There is evidence from many areas around the world that high productivity is associated with a high biomass of demersal fishes and vice versa. For example, Farina et al. (1997) showed that on the continental shelf and upper slope in the Mediterranean Sea upwelling areas with high primary productivity correlate with a high biomass in the upper trophic levels, including demersal species, and with low species diversity. In contrast, low productivity areas have the greatest species richness and a smaller biomass of demersal fish. Our study suggests that the same gen- eral patterns observed by Farina et al. (1997) may hold for the continental shelf and upper slope of the GOA. The productive western GOA is characterized by a high demer- sal biomass and relatively low species diversity, which is attributable to the presence of a few highly abundant spe- cies of flatfish and gadids (including arrowtooth flounder. Pacific cod, walleye pollock). In contrast, the eastern GOA appears to be less productive and is characterized by a low biomass of demersal fishes and high species diversity, due to the presence of a large number of rockfish species with relatively low abundances. 576 Fishery Bulletin 100(3) 60°N - 58 N - 56 N - 54'N - 52'N • * • 1984 170"W 160°W 150'W 60 N - 58°N -j 56"N - 54°N - 52 N ■ 140°W >.. 1990 1 70 W leo'w 150W 140°W 60°N - 58 N ■ 56'N - 54'N - 52 N ■ ..^■'' 'iff *• 1996 1 70"W 160°W 150W 140'W Figure 12 All sampling locations ( • ) where capelin [Mallotus villosus) were caught in 1984. 1990. and 1996. In contrast to strong spatial and depth gradients, there was Httle evidence for strong temporal gradients in the GOA gToundfish community between 1984 and 1996, except for the pronounced increase in total CPUE and a potential decrease in species diversity after 1993. The species composition of the groundfish community in the GOA, as reflected in our indices, has remained relatively stable from 1984 through 1996. Furthermore, a visual comparison of the indices mapped by year (not shown) suggests that relative species composition is character- ized by spatial patterns that were stable from 1984 to 1996. Similarly, no apparent shifts in individual species distributions were found. The apparent stability of spatial patterns in the GOA may simply reflect stability in species composition over the time period of the study, as well as a period of relatively moderate variations in the physical environment. It also suggests that recent levels of fishing may not have had a strong impact on the relative species composition, or possibly had a stabilizing effect. The ob- served composition of the groundfish community, which is currently dominated by relatively long-lived species, may also contribute to its own stability. Some of the most pronounced changes over time oc- curred in noncommercial species such as skates and capelin (Fig. 10). Whether these changes are directly or in- directly linked to fishing or are the result of environmen- tal changes is currently unknown. Interestingly, capelin decreased in abundance in some areas and depth ranges but have been caught at an increasing proportion of the sampling stations in more recent surveys. These changes appear to reflect a "spreading out" of the population at relatively fixed numbers, possibly in response to the in- creasing abundance of large predatory fishes. Other studies of gi'oundfish communities have docu- mented relatively stable spatial patterns in species com- position over time periods of 10-20 years (Colvocoresses and Musick, 1984; Gabriel, 1992; Gomes et al., 1992; Jay, 1996). In contrast. Gomes et al. (1995) observed a sharp decline in the biomass and abundance of a number of com- mercial groundfish species on the Newfoundland-Labra- Mueter and Norcross: Spatial and temporal patterns in ttie demersal fish community off Alaska 577 Hemithptcrus holini Depth (m) Slnim.igii Chirikof Kodi.ik Yakutat SouthtMst (1-100 CPUE FO ■■■ . 1 00-200 CPUl FO -- -- ~ 200-300 CPUE FO .. 300-500 CPUE FO -- - - 0-100 CPUE FO Myoxocephalus spp. SInimagin Chirikof Kodiak Yakiil:il Southeast St-^ 100-200 CPUE FO 200-300 CPUl FO 300-500 CPUl FO Lcpidupselta spp. Sh umagin Chirikof K idiak Y akulat Southeasl 0-100 CPUE FO +-I- -1- . + 100-200 CPUE FO . .. 200-300 CPUE FO _ . 1 r ^"^ 300-500 CPUE FO Figure 13 Changes in CPUE-where-present and frequency of occur- rence (FO) by depth stratum and area for three species that decreased significantly over time. Significant changes at the 59c level are indicated by '+' or '-'. Changes that were signifi- cant at the IVc level are indicated by '-i-i-' or '-'. Areas-depth strata where a species was not caught in any of the five survey years are shaded. Significance was tested by using generalized linear models including gear effects (see text). dor shelf from 1978 to 1991 and dramatic changes in the spatial distribution of groundfish assemblages after 1987. They attributed the changes primarily to intense exploita- tion, although environmental effects were likely at work as well. Similarly, on the continental shelf off Northwest Spain, significant decreases in commercial groundfish spe- cies, and a concurrent increase of several noncommercial species, can be attributed to a combination of fishing and environmental changes (Farina et al., 1997). The recent period of stability in the GOA can be con- trasted with a period of dramatic changes in the late 1970s and early 1980s that affected many parts of the North Pacific ecosystem including groundfish communi- ties (e.g. Beamish, 1995). There is strong evidence of a shift in species composition of at least the nearshore demersal communities during this time period (Piatt and Anderson, 1996; Anderson and Piatt, 1999; Mueter and Norcross, 2000). Whether similarly dramatic changes oc- curred in offshore demersal communities is not known due to the lack of long time series for most species. Estimated biomass trends for important commercial species based on recent stock assessment data (NPFMC^) and historical surveys (Ronholt et al.'') suggest that the Ronholt, L. L., H. H. Shippen, and E. S. Brown. 1978. Demer- sal fish and shellfish resources of the Gulf of Alaska from Cape Spencer to Unimak Pass 1948-1976: a historical review. Proc. rep., vols. 1-3, 872 p. Northwest and Alaska Fisheries Center, National Marine Fisheries Service, NOAA, 7600 Sand Point Way N.E., Seattle, WA. 578 Fishery Bulletin 100(3) 1961 (n =555) 1973-76 (n=310) 1 984/87 n=1625) 1 993/96 (n=1581) Figure 14 Proportional biomass composition of Gulf of Alaska groundfish communi- ties from the 1960s through the 1990s as estimated from bottom trawl surveys. All surveys used similar gear, but sampling in 1961 and 1973-76 was limited to 0-400 m depth. Total number of hauls is indicated below each time period. composition of the groundfish community on the shelf and upper slope has changed substantially since the early 1960s. Resource assessment surveys were conducted in 1961 and 1973-76 with similar gear and covering much of the same area that was sampled during the later sur- veys. Survey data from the 1960s and 1970s have to be interpreted carefully because the sampling design, gear, and the spatial coverage of these sui-veys differ from the surveys done in the 1980s and 1990s.'' In particular, no stations below 400 m or east of 136°W were sampled in the early surveys. In spite of these caveats, changes in species composition among surveys are likely to reflect ac- tual changes, at least for taxa that occur primarily above 400 m and west of 136°W. The data suggest a dramatic decrease in the relative abundance of red king crab iPara- lithodes camtschaticus) and Tanner crab iChionoecetes spp. ), which together accounted for 22% of total CPUE in 1961, and for less than 1% after the 1970s (Fig. 14). The data indicated that walleye pollock became the dominant species in the mid-1970s, peaked in the early 1980s, and has declined in relative abundance since then. Arrow- tooth flounder replaced walleye pollock as the dominant groundfish species in the 1980s and 1990s. Pacific Ocean perch and other rockfishes have increased substantially in biomass since the 1970s. Sablefish and Pacific halibut, two of the commercially most valuable species, made up a substantially larger proportion of the total CPUE in the 1980s and 90s compared to earlier decades (Fig. 14). The relative abundance of sculpins declined from about 8% in the early 1960s to less than O.l'/r in the 1980s and 1990s. The observed difference is likely to reflect a real decline in sculpin abundance of several orders of magnitude between the 1960s and 1980s. The decline appears to continue into the 1990s, as suggested by a decrease in the frequency of occurrence and biomass of two of the most abundant scul- pin taxa (bigmouth sculpin and Myoxocephalus spp., see Fig. 13, Table 2). The observed increase in overall groundfish biomass in the GOA from the 1980s to the 1990s (Fig. 6) is consistent with increases in the productivity of the northeast Pacific Mueter and Norcross: Spatial and temporal patterns in the demersal fish community off Alaska 579 Ocean between the period of the 1960s and 70s and the 1980s and 90s. Polovina et al. (1995) suggested that the increase in primary and secondary productivity in the NE Pacific Ocean resulted from a shoahng of the mixed layer in the GOA by 20-309f associated with variations in the strengtli and position of the Aleutian low pressure system from the late 1970s to the late 1980s. Over the same time period large increases occurred in chlorophyll a levels (Venrick, 1995). zooplankton stocks (Brodeur and Pearcy, 1992; Brodeur et al., 1996), and upper trophic level species (Brodeur and Ware, 1995; Francis et al., 1998). Our study suggests that there has been a parallel increase in the bio- mass of demersal fish communities on the GOA shelf and upper slope since 1984, in response to the overall increase in productivity in the NE Pacific Ocean. Given the increasing emphasis on ecosystem-based management, managers will have to rely on a variety of indicators in addition to traditional single-species biomass estimates to assess the status and health of an ecosystem (Yaffee et al., 1996). Indices of species composition like those used in this study, as well as survey-based assess- ments of noncommercial species, provide indicators that can help researchers in assessing changes in the commu- nity and can help managers in responding to such changes in a timely fashion. Species richness and diversity are particularly simple, yet potentially very useful indices. The concept of marine biodiversity has received considerable attention in recent years (NRC, 1995) and should be included in planning and policy-making (Bengtsson et al., 1997). There is some evidence that bottom fishing can reduce the diversity of benthic communities (Collie et al., 1997) and, in turn, that changes in diversity can impact ecosystem function and productivity (Naeem et al., 1994). There was some indication in our study that the number of species and the diversity of the groundfish community in the GOA as mea- sured by the Shannon-Wiener index decreased between 1993 and 1996 (Fig. 4). Although gear effects may have confounded differences among earlier surveys, the same gear and identical survey designs were used in both 1993 and 1996; thus the differences in species richness and di- versity cannot be attributed to sampling effects and merit further study. The simple diversity indices used in our study were primarily chosen for practical considerations. Other indices, e.g. indices of beta-diversity, may be more appropriate for describing diversity of marine ecosystems and should be explored. Multivariate techniques that summarize the major variation in species composition with a small number of indices (dimensions) provide another useful tool for monitoring changes in groundfish communities. Such in- dices provide sensitive indicators of change in community structure and can identify general trends that may not be apparent when using univariate measures (Austen and Warwick, 1989). The indices reduce the noise from random fluctuations of many individual stocks and the resulting "signal" may be used to test statistically for differences in species composition among areas or time periods, as well as helping to identify those species that show significant trends. Another important part of an ecosy.stem approach to management is the monitoring of the abundance of non- commercial species, in addition to commercial species. Changes in the abundance of skates, sculpins, and other noncommercial species are examples of potentially im- portant trends that may go undetected if management agencies focus on the assessment of commercial species only. Although these species comprise a small proportion of total biomass, such trends may be an early indication of changes in the groundfish community due to changes in the environment or fishing. Such monitoring, along with the monitoring of environmental variables, will further help to understand how and why groundfish communities change in response to environmental variation. Current fisheries management often fails to account for the ecological complexities inherent in fish communities (Roberts, 1997). This is hardly surprising because of the focus of most stock assessment on individual stocks and on stock-recruitment relationships. Prediction becomes increasingly difficult as other species and environmental re- lationships are added to models. Nevertheless, the complexi- ties should not discourage researchers from trying to under- stand species relationships and the response of species and communities to environmental variation and harvesting. Knowledge of such relationships can improve manage- ment in several ways. First, it can help to explain why stocks are fluctuating in particular ways and help manag- ers devise management strategies that take into account such fluctuations. Second, it can improve our understand- ing of multispecies or community-level relationships and can lead to better assessment methods that take such relationships into account. Third, it can lead to predictive hypotheses that can be explored and tested through rigor- ous research programs and adaptive management proce- dures. Ultimately, improved understanding of ecosystem processes will improve fisheries management only if it can reduce uncertainties in prediction. Thus the challenge re- mains to develop improved models that account for species relationships and environmental influences. Acknowledgments We thank Eric Brown and Michael Martin at the Alaska Fisheries Science Center, NOAA, Seattle, for granting us access to Gulf of Alaska bottom trawl survey data and for help with retrieving the data. The manuscript was greatly improved thanks to thorough reviews of earlier drafts by Ron Barry, Sue Hills, Mark Johnson, Terry Quinn, Bob Foy, and five anonymous reviewers. This work was funded in part by NOAA Saltonstall-Kennedy grant no. 50ABNF700086. Literature cited Anderson, P. J., and J. F. Piatt. 1999. Community reorganization in the Gulf of Alaska fol- lowing ocean climate regime shift. Mar. Ecol. Prog. Sen 189: 117-123. 580 Fishery Bulletin 100(3) Austen, M. C, and R. M. Warwick. 1989. Comparison of univariate and multivariate aspects of estuarine meiobenthic community structure. Estuarine Coastal Shelf Sci. 29:23-32. Beamish, R. J. 1995. Climate change and northern fish populations. Can. Spec. Publ. Fish. Aquat. Sci. 121, 739 p. Bengtsson, J., H. Jones, and H. Setala. 1997. The value of biodiversity. Trends Ecol. Evol. 12:334- 336. Blaber, S. J. M., D. T Brewer, and A. N. Harris. 1994. Distribution, biomass and community structure of demersal fishes of the Gulf of Carpentaria, Australia. Aust. J. Mar Freshwater Res. 45:375-396. Brodeur, R. D., and W. G. Pearcy. 1992. Effects of environmental variability on trophic inter- actions and food web structure in a pelagic upwelling ecosystem. Mar Ecol. Prog. Ser 84:101-119. Brodeur, R. D., and D. M. Ware. 1995. Interdecadal variability in distribution and catch rates of epipelagic nekton in the Northeast Pacific Ocean. In Climate change and northern fish populations iR. J. Bea- mish, ed.l, p. 329-356. Can. Spec. Publ. Fish. Aquat. Sci. 121. Brodeur, R., B. Frost, S. Hare, R. Francis, and W. J. Ingraham. 1996. Interannual variations in zooplankton biomass in the Gulf of Alaska, and covariation with California Current zooplankton biomass. CALCOFI Rep. 37:80-99. Brown, E. S. 1986. Preliminary results of the 1984 U.S. -Japan Coopera- tive bottom trawl survey of the central and western Gulf of Alaska. In Condition of groundfish resources of the Gulf of Alaska region as assessed in 1985 (R. L. Major, ed), p. 259-296. U.S. Dep. Commer, NOAA Tech. Memo. NMFS F/NWC-106. Brown, J. H., and M. V. Lomolino. 1998. Biogeography, 691 p. Sinauer Associates Inc., Pubis., Sunderland. MA. Collie, J. S.. G. A. Escanero, and P. C, Valentine. 1997. Effects of bottom fishing on the benthic megafauna of Georges Bank. Mar Ecol. Prog. Ser. 155:159-172. Colvocoresses, J. A., and J. A. Musick. 1984. Species associations and community composition of Middle Atlantic Bight continental shelf demersal fishes. Fish. Bull. 82:295-313. Day, D. S., and W. G. Pearcy. 1968. Species associations of benthic fishes on the continen- tal shelf and slope off Oregon. J. Fish. Res. Board Can. 25: 2665-2675. Falkowski, P G., M. J. Behrenfeld, W. E. Esaias, W. Balch, J. W. Campbell, R. L. Iverson, D. A.. Kiefer, A. Morel, and J. A. Yoder 1998. Satellite primary productivity data and algorithm development: a science plan for mission to Planet Earth. NASA Goddard Space Flight Center NASA Tech. Memo. 104566, vol. 42, 36 p. Farina, A., J. Freire, and E. Gonzalez-Gurriaran. 1997. Demersal fish assemblages in the Galician continen- tal shelf and upper slope (NW Spain): spatial structure and long-term changes. Estuarine Coastal Shelf Sci. 44: 435-454. Feder, H. M., and S. C. Jewett. 1986. The subtidal benthos, /n The Gulf of Alaska: physi- cal environment and biological resources (D. W. Hood and S. T. Zimmerman, eds), p. 347-396. Minerals Manage- ment Service, Springfield, VA. Francis, R. C, S. R. Hare, A. B. Hollowed, and W. S. Wooster. 1998. Effect of interdecadal climate variability on the oce- anic ecosystems of the NE Pacific. Fish. Oceanogr. 7:1-21. Fujita, T, T. Inada, T, and Y. Ishito. 1995. Depth-gradient structure of the demersal fish com- munity on the continental shelf and upper slope off Sendai Bay, Japan. Mar Ecol. Prog. Ser 118:13-23. Gabriel. W. 1992. Persistence of demersal fish assemblages between Cape Hatteras and Nova Scotia, Northwest Atlantic. J. Northwest Atl. Fish. Sci. 14:29-46. Gabriel, W. L., and A. V. Tyler 1980. Preliminary analysis of Pacific Coast demersal fish assemblages. Mar. Fish. Rev. 42:83-88. Gomes, M., R. Haedrich, and J. Rice. 1992. Biogeography of groundfish assemblages on the Grand Bank. J. Northwest Atl. Fish. Sci. 14:13-27. Gomes, M. C, R. I. Haedrich, and M. G. Villagarcia. 1995. Spatial and temporal changes in the groundfish as- semblages on the north-east Newfoundland/Labrador Shelf, north-west Atlantic, 1978-1991. Fish. Oceanogr. 4:85-101. Hastie, T. J., and R. J. Tibshirani. 1990. Generalized additive models. Monographs on statistics and applied probability, 335 p. Chapman & Hall, London. Isaji, T, and M. L. Spaulding. 1987. A numerical model of the M2 and Kj tide in the north- western Gulf of Alaska. J. Geophys. Res. 17:698-704. Jay. C. 1996. Distribution of bottom-trawl fish assemblages over the continental shelf and upper slope of the U.S. west coast, 1977-1992. Can. J. Fish. Aquat. Sci. 53:1203-1225. Langton, R. W., and R. L. Haedrich. 1997. Ecosystem-based management. In Northwest Atlan- tic groundfish: perspectives on a fishery collapse, Bethesda, Maryland (J. Boreman, B. S. Nakashima, J. A. Wilson and R.L. Kendall, eds. 1. p. 15.3-158. Am. Fish. Soc, Bethesda, MD. Mackas, D. L,, R. Kieser, M. Saunders, D. R. Yelland, R.M. Brown, and D. F Moore. 1997. Aggregation of euphausiids and Pacific hake iMerluc- cius productus) along the outer continental shelf off Van- couver Island. Can. J. Fish. Aquat. Sci. 54:2080-2096. Mahon, R., S. K. Brown, K. C. T Zwanenburg, D. B. Atkinson, K. R. Buja, L. Claflin, G. D. Howell, M. E. Monaco, R. N. O'Boyle, and M. Sinclair 1998. Assemblages and biogeography of demersal fishes of the east coast of North America. Can. J. Fish. Aquat. Sci. 55:1704-1738. Mann, K H., and J. R. N. Lazier 1991. Dynamics of marine ecosystems: biological — physi- cal interactions in the oceans, 466 p. Blackwell Scientific Publications, Boston, MA. Martin. M. H. 1997. Data Report: 1996 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer NOAA Tech. Memo. NMFS AFSC 82, 235 p. Martin, M. H., and D. M. Clausen. 1995. Data Report: 1993 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer NOAA Tech. Memo. NMFS AFSC 59,217 p. McCullagh, P, and J. A. Nelder 1989. Generalized linear models. Monographs on Statistics and Applied Probability. Chapman and Hall, London, 511 p. Merrett, N. R. 1992. Demersal ichthyofaunal distribution in the abyssal Eastern North Atlantic, with special reference to Cory- phaeniodes iNematonurus) Armatus (Macrouridae). J. Mar Biol. Assoc. U.K. 72:5-24. Mueter and Norcross; Spatial and temporal patterns in the demersal fish community off Alaska 581 Mueter, F. J., and B. L. Norcross. 1999. Linking community .'^trucluro of small demersal fishes around Kodiak Island, Alaska, to environmental variables. Mar Ecol. Prog. Scr 190:37-51. 2000. Changes in species composition of the demersal fish community in nearshore waters of Kodiak Island, Alaska. Can. J. Fish. Aquat. Sci. 57:1 169-1180. Munro, P. T., and R. Z. Hoff. 1995. Two demersal trawl surveys in the Gulf of Alaska: Implications of survey design and methods. U.S. Dep. Comnier NOAA Tech. Memo. NMFS AFSC 50, 139 p. Naeem, S., L. J. Thompson, S. P. Lawler, J. H. Lawton, and R. M. Woodfin. 1994. Declining biodiversity can alter the performance of ecosystems. Nature 368:734-738. NRC (National Research Council). 1995. Understanding marine biodiversity, 114 p. National Academy Press, Washington D. C. Parsons, T. R. 1986. Ecological relations, /n The Gulf of Alaska: physical environment and biological resources (D. W, Hood and S. T. Zimmerman, eds), p. 561-570. Minerals Management Service, Springfield, VA. Pennington, M., and T Stromme. 1998. Surveys as a research tool for managing dynamic stocks. Fish. Res. 37:97-106. Piatt, J. F., and P. Anderson. 1996. Response of common murres to Exxon Valdez oil spill and long-term changes in the Gulf of Alaska marine eco- system. Am. Fish. Soc. Symp. 18:720-737. Polovina, J., G. Mitchum, and G. Evans. 1995. Decadal and basin-scale variation in mixed layer depth and the impact on biological production in the Central and North Pacific, 1960-88. Deep-Sea Res. 42:1701-1716, Reed, R. K., and J. D. Schumacher. 1986. Physical oceanography. In The Gulf of Alaska: phys- ical environment and biological resources (D. W. Hood and S. T Zimmerman, eds). p. 57-75. Minerals Management Service Springfield, VA. Roberts, C. M. 1997. Ecologicaladvicefortheglobalfisheriescrisis. Trends Ecol. Evol. 12:35-38. Robinson, C. J., and J. Gomez-Gutierrez. 1998. Daily vertical migration of dense deep scattering layers related to the shelf-break area along the northwest coast of Baja California, Mexico. J. Plankton Res. 20: 1679-1697. Rogers, J. B., and E. K. Pikitch. 1992. Numerical definition ofgroundfishassemblagescaught off the coasts of Oregon and Washington using commercial fishing strategies. Can. J. Fish. Aquat. Sci. 49:2648- 2656. Stark, J. W., and D. M. Clausen. 1995. Data Report: 1990 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer. NOAA Tech. Memo. NMFS AFSC 49, 221 p. Swartzman, G., C. Huang, and S. Kaluzny. 1992. Spatial analysis of Bering Sea groundfish survey data using generalized additive models. Can. J. Fish. Aquat. Sci. 49:1366-1378. Venrick, E. L. 1995. Scales of variability in a stable environment: phy- toplankton in the central North Pacific. In Ecological time series (T. M. Powell and J. H. Steele, eds), p. 150-180. Chapman & Hall, New York, NY. Yaffee, S. L., A. F Phillips, I. C. Frentz, P W. Hardy, S. M. Maleki, and B. E.Thorpe. 1996. Ecosystem management in the United States: an as- sessment of current experience, 352 p. Island Press., Washington, D.C. Yang, M. 1993. Food habits of the commercially important ground- fishes in the Gulf of Alaska in 1990. U.S. Dep. Commer. NOAA Tech. Memo. NMFS-AFSC-22, 150 p. 582 Abstract— Life-history dynamics of pin- fish (Lagodon rhomboides) were exam- ined from data derived from random- station surveys conducted in Tampa Bay and adjacent Gulf of Mexico waters during 1993-97. In addition, patterns in spatial distribution and abundance in Gulf of Mexico waters were inves- tigated. Ages determined from whole otoliths ranged from 0 to 7 years, and von Bertalanffy growth models for males and females were not signifi- cantly different. Von Bertalanffy growth model parameters were L„=219.9 mm SL, k =0.33/yr, and ^,| =-1.10 years for all fish combined. High gonadosomatic indices during October-December indi- cated that some spawning may occur in Tampa Bay. Estimated lengths at 50'7f maturity were 132 mm SL for males and 131 mm SL for females. Total instantaneous mortality rates derived from the Chapman-Robson estimator ranged from 0.88 to 1.08/yr, and natural mortality was estimated to be 0.78/vr. In Gulf of Mexico waters, pinfish catch rates declined with increasing depth, and most pinfish were caught in <17 m of water Length distributions showed that pinfish segregate by size with increasing depth. Age, growth, mortality, and distribution of pinfish (Lagodon rhomboides) in Tampa Bay and adjacent Gulf of Mexico waters Gary A. Nelson Ronda Fish and Wildlife Conservation Commission Flonda Manne Research Institute 100 Eighth Avenue SE St. Petersburg, Flonda 33701 Present address: Massachusetts Division of Manne Fisheries Annisquam River Marine Fishenes Station 30 Emerson Avenue Gloucester, Massachusetts 01930 E mail address Gary.NelsoniSistatema us Manuscript accepted 15 February 2002. Fish. Bull. 100:,582-592 (2002). The pinfish (Lagodon rhomboides) is an ecologically important sparid that inhabits estuarine and offshore waters of the United States from Massachu- setts to Texas (Darcy, 1985). Young- of-the-year pinfish are predators on a range of invertebrates in estuaries, often to the degree that entire assem- blages of macrobenthic fauna are affected (Young et al., 1976; Young and Young, 1977; Nelson. 1978), They also consume seagrasses during their estuarine phase, which makes them an important link between primary and secondary production (Stoner, 1982; Weinstein et al., 1982; Montgomery and Targett, 1992). Both young-of-the- year and adult pinfish are prey for other fishes (Gunter, 1945; Darnell, 1958; Carr and Adams, 1973; Seaman and Collins, 1983) and are used as bait for many recreationally and commer- cially important species. Despite their ecological and economi- cal importance, the population biol- ogy and dynamics of pinfish in Florida have not been adequately examined. Most knowledge about growth (Reid, 1954; Caldwell, 1957; Hellier, 1962; Nelson, 1998), distribution and migra- tion (Caldwell, 1957; Hansen, 1970; Nelson, 1998), and mortality (Nelson, 1998) comes froin studies of young-of- the year pinfish in estuaries. The only information available on older indi- viduals comes from estuarine studies that describe size and age at spawning (Caldwell, 1957; Hansen, 1970), gross fecundity (Caldwell, 1957), and lon- gevity and size at age (Hansen, 1970). However, that information may contain biases because sampling did not extend into offshore waters where larger, and presumably older, pinfish are thought to occur, particularly during cooler months (Darcy, 1985). In this study, age, growth, mortality, and maturation of pinfish were evalu- ated by studying data from multigear fisheries-independent surveys conduct- ed in Tampa Bay, Florida, and in off- shore waters off west-central Flonda. In addition, patterns in abundance and spatial distribution of pinfish in the Gulf of Mexico were examined. Methods Biology Pinfish for aging, maturity, and mor- tality analyses {n=711) were collected opportunistically from fall 1993 to spring 1997 during multiple fisheries surveys conducted in Tampa Bay and offshore Gulf waters by the Florida Marine Research Institute (Fig. 1). Pinfish from Tampa Bay were collected primarily during multigear seasonal (spring and fall) and monthly fisher- ies-independent surveys designed to estimate relative abundance of small estuarine fishes and evaluate life-his- tories parameters of large estuarine fishes (Table 1). During the surveys, sites were chosen by randomly selecting 1" latitude x 1" longitude microgrids, representing the site to be sampled, within randomly selected 1' latitude x Nelson: Age, growth; mortality, and distribution ol Lagodon ihomboides 583 Table 1 lAMit,nli summary statistics (by area, year. and gear) for pinfi sh collected during biological sampling. Exp. gill net = experimental gill net. Pinfish SL (mm) Stretched Year Gear mesh (mm) Minimum Maximum Mean n Tampa Bay 1993 trammel net 76, 305 179 255 211.8 31 1995 exp. gill net 50-152 170 185 180.0 3 183-m seine 38 73 202 121.1 50 trammel net 76, 305 173 216 191.4 8 21-m seine 3.2 73 150 112.1 24 61-m seine 25.4 72 165 132.6 19 1996 183-m seine 38 66 207 126.9 313 purse seine 50 100 181 125.9 34 trammel net 76, 305 203 203 203.0 1 21-m seine 3.2 34 174 89.0 28 hook and line — 198 205 202.0 3 1997 183-m seine 38 83 155 111.3 62 purse seine 50 91 156 108.9 16 21-m seine 3.2 25 104 71.7 18 Gulf of Mexico 1996 hook and line — 116 149 135.2 17 1997 bottom trawl 254 95 175 128.8 84 27 1 ' longitude grids. At each site, gears were deployed with standardized procedures (see Nelson [1998] and Tremain and Adams [1995] for detailed description of the survey design and deployment techniques). Selected individuals were immediately placed on ice for transport and frozen whole at the laboratory. Pinfish from the Gulf of Mexico were collected primarily during a bottom-trawl survey conducted in spring in the Gulf of Mexico to document baitfish abundance and distribution. The Gulf trawl sui-vey was conducted during April from 1994 to 1997 at stations randomly selected along line transects ran- domly placed perpendicular to the west coast of the Florida peninsula adjacent to Tampa Bay. At each station, a 19.8-m, 254-mm stretched-mesh modified ballon trawl with 25.4-mm codend liner and 2.2-m^ China V-doors was towed at approximately 3 knots for 30 min during daylight hours in depths ranging be- tween 6 and 31 m. At each station, trawl catches were sorted by species, counted, weighed collectively, and 50 individuals per haul were measured to the nearest 1 mm fork length (see Pierce and Mahmoudi (20011) for a detailed description of the survey design). Depth (m) at the start and end of the trawl tow and surface salinity and temperature were recorded at all stations. To match data collected from the Tampa Bay samples, all length data were converted to standard length by using conversion equations derived from this study. Individual fish were immediately frozen whole and returned to the 30 N Figure 1 Maps of Florida showing study area. laboratory. Occasionally, supplemental samples were col- lected by hook-and-line fishing in Tampa Bay and the offshore waters (Table 1). In the laboratory, standard length (SL ±1 mm), fork length (FL ±1 mm), total length (TL ±1 mm) and total 584 Fishery Bulletin 100(3) weight (TW ±0.01 g; measured after each individual was blotted dry) were recorded for all thawed, undamaged individuals. Gonads were excised from a subsample of in- dividuals (-80%) of each collection, identified macroscopi- cally (and microscopically if needed) as ovaries or testes, blotted dry on a paper towel, and weighed (±0.01 g). Sea- sonal conversion equations for length measurements were generated by least-squares regression (SAS, 1990a). Age and growth Sagittal otoliths for aging studies were excised, cleaned of extraneous tissue, and stored dry in vials. Otolith length was measured under a binocular dissecting microscope at lOx with a calibrated ocular micrometer, and an otolith length (mm) versus fish standard-length relationship was derived by using least squares regression. For age determination, one whole otolith (either left or right) was placed on black velvet to enhance the visibility of opaque zones, submerged in a 60% glycerin solution, and viewed at 32x with reflected light. Otolith edges were classified as either opaque or hyaline to determine time of opaque ring formation. The number of opaque rings was considered to equal the age in years of the pinfish. Age determinations were repeated three times on all whole otoliths. If two of the three age readings were in agreement, the value was accepted. If age varied between all readings, the structure was not included in the analysis. The accuracy with which whole otoliths can be used to age pinfish was confirmed by comparing whole otolith ages from a subsample of the largest pinfish to age readings made on thin sections of the same otoliths. Methods of Hood and Johnson (1999) were used to section otoliths. To verify the consistency of the opaque ring counts, a random subsample ( 13% ) of all whole otoliths was aged once by an experienced second reader and the results were compared to the final age determination. Pinfish larvae are present off the west coast of Florida between November and February (Darcy, 198.5); therefore a 1 January birthdate was assumed for all pinfish. Age in years was designated with a decimal exten- sion that represented the date of capture in days from the 1 January birthdate. Growth was modeled by fitting length-at-age for all years combined to the von Bertalanffy equation: L„,=L„.,*(l-exp<-*''"^'-'"") + f,^, where L, , = the standard length of the / th individual at aget: L^ = the asymptotic maximum length; k = a growth constant; Iq = the hypothetical age at which length is zero; and £ s = independent, identically distributed, normal random errors. Parameters L, k, and /„ were estimated by using SAS non- linear regression (Proc NLIN) with the Marquardt method (SAS Institute, 1990b). Lack of fit was assessed by using residuals plots (Bates and Watts, 1988). Growth-rate differences between sexes were investi- gated by using an approximate randomization test to compare growth curves (Helser, 1996). Essentially, differ- ences were investigated by comparing an observed test statistic to an empirical probability density function of a test statistic under the null hypothesis of no difference. The observed test statistic for pinfish was developed by fit- ting the von Bertalanffy growth functions to length-at-age data for sexes combined and to length-at-age for each sex separately. The sums of the residual sum of squares from the two sex-specific models were then used to calculate the observed test statistic t(x, )=E"''-''''-X'''' -/ r 't.p' i.i.p where tiXg) = the test statistic; Z, = the predicted length-at-age for all sexes combined; and Itp = the predicted length-at-age for each sex (p=l,2) (Helser, 1996). To generate the empirical probability density function (pdf), length-at-age data for both sexes were pooled and then assigned randomly without replacement to two groups with sample sizes equal to the original number of observations per sex. Growth curves were then fitted separately to the length-at-age data of the randomized groups, and the test statistic under the null hypothesis of no difference was calculated as tix}- (/,,-/,)- 't.'.P ■l.'.P^ >.t,P where tix) = the difference in residual sum of squares between the von Bertalanffy fits to the entire pooled data set and von Bertalanffy fits to the randomized groups (denoted by *). The randomization procedure was repeated 1000 times to obtain the pdf of tix). The null hypothesis of no difference was rejected if /(.V|,)>Z(.v) at a=0.05. Reproduction, sex ratios, and maturity Gonads excised from selected individuals were classified macroscopically as either immature or mature by using the maturation criteria of Nikolsky (1963) and Cody and Bortone (1992). Seasonality of reproduction was deter- mined by noting when changes in the gonad condition took place. A gonadosomatic index (GSI) was calculated to show changes in gonad weight in relation to gonad-free total body weight. The index was computed as GSI = gonad weight/itotal body weight - gonad weight) x 100. Length at 50% maturity was estimated for pinfish cap- tured during the season of gonadal maturation. The prob- Nelson: Age, growth; mortality; and distribution of Lagodon rhombotdes 585 ability of pinfish being mature was modeled as a logistic function of the following form: V(Z) = sa-sf p.= exp 1+exp iu+fi*x, - + £,< where P = the response probability; -v = the standard length (mm) of the / th fish; a = the intercept; P = the slope coefficient of standard length; and f, = the error term. Model regression coefficients were estimated using maxi- mum likelihood (SAS, 1997). Goodness-of-fit was assessed with the chi-square test. Chi-square analyses were used to test for deviations from a theoretical 50:50 sex ratio. Length-weight relationships Because of the seasonality of reproduction, data on body length and weight were separated into two groups rep- resenting two time periods to construct length-weight relationships that reflected seasonal changes in gonad weight. Data collected during March-August composed the data group representing the spring-summer period, and data collected during September-February composed the data group representing the fall-winter period. Sea- sonal regression equations for logj^-transformed standard length and body weight (total and gonad-free weight) were generated by the least-squares regression. One-way analysis of covariance (Sokal and Rohlf, 1981), conducted by using SAS PROC MIXED (Littell et al, 1996), was used to test for differences between regression slopes and adjusted means of the length-weight relationships by sex and season. Mortality Total instantaneous mortality (Z) was estimated in Tampa Bay and the Gulf of Mexico from pinfish numbers-at-age data by using the Chapman-Robson equation for survival (S) estimation: -Z = ln(S) = ln I<^ \ i=i J the natural log; , - the number of years the / th fish is older than where In x^ the age-at full-recruitment S = annual survival; and n = the total number of fully recruited fish (Chap- man and Robson, 1960). The variance of Z was estimated from the variance of S by (Jensen, 1985). The Chapman-Robson estimator was used because it is more robust to sample siz.e variation in num- bers-at-age (Jensen, 1996; Murphy, 1997). In addition to Z, natural mortality (A/) of pinfish was estimated by using a multiple regression equation relat- ing M to L. (TL in cm) and k (per yr) of the von Berta- lanffy equation, and to mean annual water temperature ("C)(Pauly, 1980). To estimate the growth parameters used in the equation, the von Bertalanffy function was refitted to total length and age data. Mean annual water tempera- ture was estimated from temperature data of the Tampa Bay survey. Distribution To determine if pinfish catch rates in the Gulf of Mexico were associated with depth, a generalized linear model (McCullagh and Nelder, 1989) was used to model the random trawl catches during the April survey from 1994 to 1997. Catch data were transformed by using ln(.r+l) to stabilize the variance and reduce the influence of sam- pling variability between tows. Year, average depth of the trawl tow, and their first-order interaction were included in the model. Only depth was analyzed because surface readings of temperature and salinity were thought to not reflect bottom conditions where the trawl tows were made. Size structure of pinfish caught during the Gulf of Mexico survey was also examined by calculating summary statistics (mean and percentiles) of length data to identify patterns related to depth. Results Biology Length-conversion equations are listed in Table 2. All slopes and intercepts were significantly different from zero. Age and growth Age determination was based on sagittae from 658 pinfish (66-255 mm SL). Alternating opaque and hyaline zones, which composed an annual growth increment, were evi- dent on whole pinfish otoliths (Fig. 2A). Only 13 otoliths were considered unreadable. Agreement between the number of opaque bands counted on whole otoliths and the number counted on sections of the same otoliths (Fig. 2B) was 96.3% (26/27), indicating that whole otoliths can be used to reliably age pinfish. The high percentage of agreement (89.6%) between the author's final annulus counts and those made by a second reader, and no appar- ent bias towards a particular aging error, indicated that ages were consistently interpreted. Plots of the monthly proportions of otolith with opaque edges for pinfish ages 1 to 3 indicated that an opaque ring 586 Fishery Bulletin 100(3) Figure 2 A whole (A) and sectioned (B) sagitta from a 7-year-old l230-mm-SL) pinfish caught in Tampa Bay in November 1993. is formed only once a year, during late winter to late spring. Deposition of opaque material began as early as January; it peaked during March, April, or May depending on age; and it was completed, or nearly so. by June (Fig. 3). A single regression equation characterized the relation- ship between otolith length (OT in mm) and standard length for a random sample of pinfish 71-230 mm SL (standard errors are in parentheses) used in age analysis: or =8.03(1.07) -I- 0.39(0.008) xSL; \r- = 0.95,n = 132]. The intercept and slope coefficients were significantly dif- ferent from zero (P<0.001). Ages of pinfish ranged from 0 to 7 years (66-255 mm SL) in Tampa Bay and from 1 to 6 years (95-175 mm SL) in the Gulf of Mexico (Fig. 4A). The majority (88.1%) of aged pinfish were <2 years. Residual plots from the growth models showed near-random patterns, indicating that the von Bertalanffy model adequately described the growth of male and female pinfish. Results of the randomization test did not indicate a significant difference between male and female growth models (^(.to)=2,691;p(«:»:)>«j:o))=0.27); therefore, a von Bertalanffy model was fitted to all pinfish data combined I Fig. 4B, Table 3). Growth rate of pinfish was rapid for the first 1-3 years of life but gradually de- creased thereafter (Fig. 4B). Reproduction, maturity, and sex ratios Seasonality of reproduction was indicated by monthly changes in the GSIs of all pinfish over time. In Tampa Bay and the Gulf of Mexico, GSIs for males and females were <0.5'7f' during the spring and summer months (Fig. 5). Nelson: Age, growth; mortality, and distribution of Lagodon rhomboides 587 Table 2 Linear regression statistics showing the length conversion and length-weight relationships for pinfish. For the length conversions, the model is.v = n+ /3x.v. For the weight-length relationships, the model is log,D(,y) = a-h /3x log,,^. a is the regression intercept, ^ is the slope coefficient, r'^ is the coefficient of determination, n is the sample size, MSE is the mean square error, FL is fork length (mm), SL is standard length (mm), TL is total length (mm>. GFW is gonad-free body weight, and TW is total body weight (g). SP-SU is spring-summer and FA-Wl is fall-winter. All slopes and intercepts were significantly different (P<0.05) from zero. Slopes between TW-SL relationships were significantly different (P<0.01). .V X Season a SE(a) /3 SE(/3) r2 n MSE FL SL All 2.65 0.368 1.14 0.003 0.996 603 4.736 TL SL All -2.20 0.542 1.33 0.004 0.994 625 13.745 TL FL All -3.73 0.412 1.15 0.003 0.997 591 5.695 GFW SL SP-SU -4.56 0.061 3.04 0.029 0.977 257 0.0020 TW SL SP-SU -4.53 0.045 3.03 0.021 0.983 343 0.0020 GFW SL FA-WI -4.57 0.064 3.07 0.030 0.977 244 0.0024 TW SL FA-WI -4.68 0.043 3.11 0.021 0.987 302 0.0023 14 80 Age 1 5 22 23 11 19 9 Age 2 2 5 3 12 1 Age 3 12 3 3 9 1 DJ FMAMJ JASONDJ Montti Figure 3 Monthly proportions of whole otoliths with opaque edges for pinfish ages 1-3. ±1 standard error bars are shown. GSIs began to increase in October (females) and Novem- ber (males) in Tampa Bay. The highest GSIs for males and females in Tampa Bay were 5% and 8%, respectively, observed in December. High and moderate GSIs for females and males during January in the Gulf of Mex- ico indicated continued spawning, but low values in April suggested that spawning ended prior to that month (Fig. 5). Maturity of male and female pinfish was related sig- nificantly to length (male: Wald x^=18.8, P<0.001; female: 0.5 0,4 c o 5 0.3 Q. O " 02 0-1 \ 0.0 E - 200 150 100 50 300 1 250 Tampa Bay, /i=549 Gulf of Mexico, n=96 E i^ -t^^-pi B 012345678 Age (years) Figure 4 Plots of (A) age composition from Tampa Bay and Gulf of Mexico samples and (B) von Bertalanffy growth model fitted to standard length and age data for all sexes combined. Wald x^=32.1, P<0.001) by logistic equations, and there was no evidence of lack of model fit (male: x^=81.1, df=94, P=0.82, max-rescaled r2=0.38; female: x^=109.2, df=121. 588 Fishery Bulletin 100(3) Table 3 Estimates of the von BertalanfTy growth parameters for pinfish by sex and for all sexed and unsexed fish combined. MSE is the mean square error, r'- is the coefficient of deter- mination, and n is the sample size. Asymptotic standard errors are shown in parentheses. Sex MSE Male 228.5 0.31 -1.21 517.926 0.57 285 (19.733) (0.079) (0.331) Female 212.0 0.33 -1.16 551.020 0.51 330 (16.504) (0.082) (0.309) All fish 219.9 0.33 -1.10 537.490 0.55 645 (12.204) (0.055) (0.203) en Tampa Bay Male o o 0 o fl n - - i^ f n t t-+-^ Female 8 0 „ =•= : 8 B ° J FMAMJ JASOND Gulf of Mexico J FMAMJ JASOND Month Figure 5 Plots of monthly gonadosomatic indices (GSI) for male and female pinfish from Tampa Bay and the Gulf of Mexico. The solid line connects median values. P=0.77, max-rescaled r^^CGO). Parameter estimates (stan- dard error) were a = -7.78 (1.745) and /3 = 0.06 (0.014) for males, and a = -9.83 (1.721) and (i = 0.08 (0.013) for fe- males. The estimates of length at 50% maturity ±95% C.I. were similar for both sexes: 132 ±8 mm SL for males and 131 ±8 mm SL for females. The smallest mature male was 112 mm SL and the smallest mature female was 92 mm SL. The largest immature male pinfish was 245 mm SL and female pinfish was 173 mm SL. Chi-square tests indicated that sex ratios did not devi- ate significantly (P>0.05) from unity in any season. Length-weight relationships No significant differences (P>0.10) in slopes or inter- cepts of standard-length on body-weight relationships were observed between sexes in any period; therefore all data were pooled for the subsequent analyses. Seasonal length-weight regressions for both sexes combined are listed in Table 2 for total and gonad-free body weight. Regression slopes were significantly different between periods only for the standard-length on total-body-weight relationship (P<0.01). The regression slope was highest in fall-winter, which indicates that pinfish gain more weight per unit increase in length during this period than in spring-summer. Adjusted means for the standard-length on gonad-free body-weight relationship were not signifi- cantly different between periods (P>0.05). Mortality Based on frequency plots by age, age-at-full-recruitment was age 1 for pinfish in Tampa Bay and age 2 for those in the Gulf of Mexico. Estimates of Z were 1.08/yr for Tampa Bay pinfish and 0.88/yr for the Gulf of Mexico pinfish (Table 4). Estimates of Z for all areas combined, assuming full recruitment at age 2, were intermediate to the Z esti- mates made for each area separately (Table 4). The estimate of A/ was 0.78/yr from Pauly's ( 1980) mul- tiple regression equation by using L^ = 30.1 cm, ^=0.31/yr, and mean annual temperature of 24°C. Distribution In the Gulf of Mexico, pinfish were captured at 90% ( 55/6 1 ) of sites where trawls were pulled in depths of 6 to 30 m during the baitfish survey (Fig. 6). Catch rates of pinfish in the Gulf of Mexico varied significantly between years and were associated with depth, but the significant interaction term indicated that the relationship between depth and catches varied between years (Table 5). Plots of cumula- tive proportions of total catches by year showed that most trawl catches (95%) occurred in waters <15 m during the 1994, 1995, and 1997 surveys, but in 1996, 95%^ occurred in waters <17 m (Fig. 7A). Length summary plots for pinfish caught during the baitfish survey showed that pinfish became segregated by size as depth increased. Median length increased from 109 mm SL in the 6-10-m depth range to 152 mm SL in the 26-30 m depth range (Fig. 7B). The smallest (80-mm-SL ) Nelson: Age, growth, mortality, and distribution of Lagodon rhomboides 589 28,20 28 10 2800 - 27.90 - a. 27.80 "D 3 5 27,70 ■ 27,60 ■ 27,50 ■ 27,40 27,30 27,20 No per sq-nm N A 0 • 1 -500 • W1-5 00U • e>ooi M 01.1(1 • ■■Si) (ion -8340 -83,20 -83 00 -82 80 Longitude -8260 -82,40 Figure 6 Map of distribution of pinfish catches from the baitfish trawl survey in Gulf of Mexico waters adjacent to Tampa Bay, 1994-97. Latitude and longitude values are given in decimal degrees. Table 4 Estimates of annual survival (S) and total instantaneous mortality iZ) of estimator for Tampa Bay, the Gulf of Mexico, and both areas combined, n is the number of age classes included. pinfish determined by using the Chapman-Robson ( 1960) is the sample size at each age of full recruitment and n^ Age (yr) at full recruitment "r "a Estimate Variance S Z S z Tampa Bay (n=549) 1 Gulf of Mexico (n =96) 2 Combined (n=645) 2 379 7 70 5 189 6 0.34 0.41 0.40 1.08 0.88 0.90 0.0004 0.0020 0.0008 0.0033 0.0120 0.0043 and the largest (203-mm-SL) pinfishes were captured in the 6-10 m and 11-15 m depth ranges, respectively. Discussion Age and growth This study is the first to report otolith-derived age esti- mates for pinfish. Hansen (1970) reported scale-derived age estimates for pinfish from a Florida Panhandle estuary. In the current study, annuli were identified and counted on whole otoliths and were then identified and counted on thin sections. The high initial agreement (89.6'7f) between the age determinations of the author and a second reader suggests a high degree of precision can be attained without sectioning the otolith. Given the high degree of error generally associated with estimating ages of fish from their scales ( Jearld, 1983; Beamish and McFarlane, 1987), I believe that age estimations made from pinfish otoliths are far more likely to be accurate. Deposition of opaque material in the sagittal otoliths of pinfish occurs over a protracted late-winter to early-spring period and indirectly validates the formation of annuli in 590 Fishery Bulletin 100(3) Table 5 Results of the generalized linear model analysis of In (.v+1) | pinfish catches (in number per nmi- from the baitfish cruises in the Gulf of Mexico 1994-97. ■**=P<0.001. Source df MS F r^ Model 7 79.9 le.r ■ 0.680 Year 3 25.5 5.1*- Depth 1 389.3 78.4"- Year < depth 3 31.3 6.3'" Error 53 5.0 Corrected total 60 13.7 pinfish. Similarly, Hansen (1970) found that annuli on scales are generally formed once a year, in April. The maximum age of pinfish determined in this study was 7 years, which exceeded the maximum age of 2 years reported by Hansen (1970). The difference in maximum age is probably due to Hansen's inadequate sampling of the entire size range of pinfish in Pensacola Bay — he col- lected pinfish at only two stations using a single, selective trawl. In addition, Hansen's use of scales as the primary aging structure may have contributed to the underesti- mate of the maximum age of pinfish (Beamish and Mc- Farlane, 1987). In contrast, pinfish collected in this study were sampled from numerous locations in Tampa Bay and adjacent Gulf waters with multiple types of gears that had various mesh sizes, and otoliths, which are more reliable aging structures than scales (Beamish and McFarlane, 1987), were used for age determination. Coincidentally, Caldwell (1957) projected from the growth rates of age-1 pinfish that a 328-mm-SL pinfish may have been as old as 7 years; my results provide direct evidence that supports his estimate. Reproduction, sex ratios, and maturity The monthly changes in GSIs show that pinfish are late- autumn to early-spring spawners in Tampa Bay and adjacent Gulf of Mexico waters; this conclusion was also reported by Darcy (1985) and Cody and Bortone (1992). Pinfish spawning is suspected to occur in offshore oceanic waters (Hansen, 1970; Darcy, 1985); however, the high GSIs of pinfish collected in Tampa Bay and the occurrences of several of these pinfish with late-developing gonads in waters about 3-6 km into the bay during October-Decem- ber suggest that some spawning activity may occur in this estuary. This is not surprising because most studies that have examined gonadal activity in pinfish have used sam- ples only from offshore waters (Franks et al., 1972; Stott et al., 1980, 1981), did not use gear with various mesh sizes to adequately sample the entire size range of pinfish found in estuaries (Caldwell, 1957; Hansen, 1970), or did not compare data on an estuary-offshore basis (Cody and Bortone, 1992). *- 1994 o 1995 ▼- 1996 ^ - 1997 10 15 20 Depth (m) 210 ^ 180 £ 150 • 120 ■ 90 • 60 B 282 115 T 42 6-10 11-15 16-20 21-25 26-30 Depth range (m) Figure 7 Plots of (A) cumulative catch versus depth by year and (B) size frequency of pinfish by depth range from the baitfish trawl survey, 1994-97. In B, the lower and upper dots rep- resent the 5"^ and 95"^ percentiles of the length distribu- tion, the whiskers represent the 10"^ and 90'*' percentiles, the box represents the 25"' and 75"' percentiles, and the horizontal line represents the median length. Length-weight relationships The length-weight relationships were affected by gonadal maturation. During the September-February period, go- nad weight of both sexes increased, which resulted in a standard-length to total-body-weight relationship that was higher than it was in March-August period. Differences in predicted total weight at length were 2-8% between the September-February and March-August periods and illustrate the effect of maturation on these relationships. The slopes of the length-weight relationships from both periods (3.03-3.11) were higher than the slopes of length on weight relationships (2.91 and 2.90, respectively) de- veloped by Caldwell (1957) and Cameron (1969). Nelson Age; growth, mortality; and distribution of Lagodon ihomboides 591 Mortality This study is the first to report total instantaneous mortal- ity (Z) rates (0.88-1.08) for pinfish older than young-of-the- year. It is difficult to conclude whether the mortality rate in Tampa Bay is really higher than that in the Gulf The small mesh characteristics of the sampling gear, the restricted spa- tial coverage of sampling from Tampa Bay to the shallows of the estuary where young pinfish dominate (most gears were employed in waters <3 m), or the migration of pinfish from estuaries to offshore waters to spawn may have contributed to the older (>age 1) fish being underrepresented in the sampling of bay waters (Hansen, 1970; Darcy, 1985; Nelson, 1998). By assuming that pinfish are fully recruited at age 2 (Table 4), I estimated that Z from data for both areas combined is probably more realistic because the numerical proportions of pinfish ages 2 and older in Tampa Bay were nearly identical to the numerical proportions of pinfish age 2 and older from the Gulf of Mexico. The estimate of Z from data for both areas combined (0.90/vr) was close to the natu- ral mortality estimate of 0.78/yr, suggesting that the portion of annual mortality attributed to fisher activity, calculated by (Z-M)/Z, is low [13%]. Distribution Darovec (1995). Franks et al. (1972), and Darcy and Gutherz (1984) reported that pinfish from the Gulf of Mexico occurred in trawls to depths between 73 m and 93 m. However, most pinfish captured in Gulf waters off west-central Florida were captured from depths <19 m (Darovec, 1995). The results from the baitfish survey were very similar in that most pinfish (95%) were caught in waters <17 m, but the abundance-depth relationship varied between years, and larger pinfish were generally associated with greater depths. The propensity of pinfish to limit their depth distribution and alter their size dis- tribution with depth is probably due to the distribution of their epibenthic prey (Stoner, 1980; Nelson and Bortone, 1996), the distribution of their predators, and (because pinfish are visual feeders) reduced light intensity with depth (Kjelson and Johnson, 1976; Luczkovich, 1988). In summary, ages determined from whole otoliths ranged from 0 to 7 years, and von Bertalanffy growth model parameters for all aged fish were L^=219.9 mm SL, k =0.33/yr, and .rf / ? ;■ e,. — ■'- Sort JVew England // .<4*' ' *^- Southern" ^ *» ^*^";/' M^7 ,i^ - he- y Figure 1 Regional map of the Mid-Atlantic. New England, and Canada area showing states, the providence of Nova Scotia, offshore place names, and the 100-m depth contour Shoals region during 1979-84 (Lough et al., 1985; Smith and Morse, 1993). Presently pelagic fishes such as Atlantic herring and mackerel (Scomber scombrus) have recovered to historic high abundances (Clark, 1998). Atlantic herring have re- covered to pre- 1960s biomass and abundance and are now becoming the subject of renewed interest by fleets from the United States and Canada (Clark, 1998; NEFSC-). During a recent survey (spring 1999), herring were caught in 47% of the standardized tows and represented one of the top five species in terms of abundance and biomass (NEFSC3). 2 NEFSC (Northeast Fisheries Science Center). 1998. Atlantic herring: report of the 27"' Northeast regional stock assessment workshop (27"' SAW). Stock Assessment Review Committee (SARC) consensus summary of assessments, 27 p. Northeast Fisheries Science Center, Woods Hole, MA 02.543. The presence and increase of spawning fish and lar- vae in historically important areas of Nantucket Shoals (Great South Channel area), Georges Bank, and along the coast of Maine have been documented ((Zinkevich, 1967; Stephenson and Kornfield, 1990; Smith and Morse, 1993; Stevenson'*). The purpose of this paper is to describe the magnitude of the decline and recovery of the herring complex in the Gulf of Maine-Georges Bank region and the chronology of '■> NEFSC (Northeast Fisheries Science Center). 1999. Fisher- man's report: bottom trawl survey. Woods Hole Laboratory, National Marine Fisheries Service, March 1-April 22, 1999, 20 p. Northeast Fisheries Science Center, Woods Hole, MA 02543. ■* Stevensen, D. K. 1989. Spawning locations and times for Atlan- tic herring on the Maine Coast. Maine Department of Marine Resources Research Reference Document 89/5, 24 p. Maine Department of Marine Resources, Boothbay Harbor, ME 04575. Overholfz and Friedland: Recovery of the Gulf of Maine-Georges Bank complex 595 400 ^ 350 - t — 300 - -a c 1 250 ■ o - 200 ■ i 150 ■ c 5 100 - 1 V * /■; Georges , Bank '^ ^ Gulf of Ivlaine 50 ■ \Ai^^^A.^-^ \ ^^^^^-y — ~-^ ^ f 1960 1965 1970 1975 1980 1985 1990 1995 Year Figure 2 Landings of Atlantic herring (thousands oft) from the Gulf of Maine and Georges Bank, 1960-98. the differential recovery by area by using NEFSC bottom trawl data from 1963 to 1998. We also wanted to examine two theories regarding the recovery of Georges Bank her- ring: 1) that it was due to a resurgence of extant spawners (Stephenson and Kornfield, 1990) or 2) that is was due to recolonization from adjacent components of the complex (Smith and Morse, 1993). Methods Details of the statistical design, areal coverage, standard- ization methods, gear type, and species catchability in the NEFSC bottom trawl surveys are provided in Grosslein ( 1969); Azarovitz ( 1981); Byrne et al. ( 1981), and a NEFSC report (NEFSC, 1988). Briefly, the surveys are based on a stratified random research design that covers the region from Cape Hatteras to the Scotian Shelf at depths between 27 and 200 m (15 to >100 fathoms), each spring and autumn. We used data from spring ( 1968-98) and autumn ( 1963- 98) bottom trawl surveys to produce indices of abundance, and plot distribution patterns for the Gulf of Maine- Georges Bank Atlantic herring complex. For spring, we used the same strata set (strata 1-30, 33-40, and 61-76) (Fig. 3) as used in the assessment of the herring complex (NEFSC^). We produced indices of total abundance (swept area numbers Imillions]) to follow changes in the stock complex over time, assuming that a standardized survey tow covers 0.1 nmi and that all herring are available to the gear For autumn we used a smaller strata set (1-30, 33-40) (Fig. 3) to generate indices of abundance (area-swept num- bers) because herring are seldom found south and west of Long Island during this season (NEFSC^). Changes in abundance by subregion (reflecting abundance trends by spawning components) were assessed by using indices for the Gulf of Maine (strata 26-28, 37-40), Nantucket Shoals (Great South Channel area) (strata 9-11, 23-25), and Georges Bank (strata 13-14, 16-17, 19-22, 29-30) (Fig. 3). We produced box plots that represent the median sur- vey catch location aggregated over a latitudinal gradient (35-45°N latitude) collapsed over the entire east-west longitude of the region. We repeated the same procedure for the longitudinal gradient (64-76°W longitude). These plots show the median, inter quartiles (25% and 75%), whiskers ( 1.5 times the absolute value of the interquartile range) and outliers of the data. We also constructed maps of spring survey catch locations for 1970, 1975, 1980, 1985, 1990, and 1995 to show trends in herring distribution and abundance over the time period. Changes in distribution and abundance during autumn 1963-98 are depicted in plots of Atlantic herring catch locations for 1965, 1970, 1975, 1980, 1985, 1990, and 1995. We also produced box plots of bottom temperature for the sites where herring were captured during both seasons. For spring and autumn, we also plotted the pro- portion of survey tows with catches of Atlantic herring as a measure of how often surveys encountered Atlantic herring over the time series. We used LOWESS (locally weighted regression smoothing) (Cleveland, 1979) to make the temporal trends in both of these time series easier to follow. Results Atlantic herring abundance indices from spring surveys changed markedly during 1968-98 (Fig. 4A). A period of relatively high abundance (area swept, millions offish), in 596 Fishery Bulletin 100(3) 78° 76° 74° 72° 70° 68° 66° 64° 36°- Figure 3 Survey strata used in the Northeast Fisheries Science Center research bottom trawl surveys for the region from Cape Hatteras, NC, to Nova Scotia, Canada. the late 1960s was followed by a low abundance from 1971 to 1985. After 1985, abundance indices increased steadily to values that were 2-3 times larger ( 1993, 1996-981 than those at the beginning of the time series (Fig. 4A). Atlantic herring abundance was six times higher during 1992-98 than during 1968-75. Herring spawn between mid-September and mid-Octo- ber, and the timing is dependent on location and year The NEFSC autumn research survey is usually conducted just after the spawning season for Atlantic herring, but this survey is a useful indicator of the abundance of spawners in the various component areas. Herring abundance in the autumn surveys was low (range 2.5-17.7 million fish) dur- ing 1963-67, very low from 1968-1986 (range 0.0-4.4 mil- lion fish), and much higher since 1987 (range 45.6-604.2 million fish) (Fig. 4B). The herring complex occupied a consistently wider latitu- dinal range in the spring during the late 1980s and 1990s than in prior years. The complex was centered between 39°-41°N latitude during 1968-70 and then gradually shifted northwards (Fig. 5A). During 1982-84, the median latitude of the complex had moved north to 42° (north of Cape Cod); by 1984, the distribution of the stock complex was confined between 40°30'-44°10'N latitude, about a 50% decrease in latitudinal range as compared to 1968 (Fig. 5A). As abundance increased in the late 1980s, the center of the spring distribution of Atlantic herring moved southwards and the range extended; since 1989 the medial latitudinal position of the complex has remained at 41°N (Fig. 5A) Major shifts in the longitudinal position of the com- plex have also occurred. During 1968-70, herring were generally west of 70°W longitude (Fig. 5B). Subsequently (through 1986) the complex shifted farther to the east. As abundance increased, in the late 1980s, the longitudi- nal center of the complex moved westward remaining at about 70°W longitude during 1989-98. During this time period, the east-west extent of the complex expanded greatly (Fig. 5B). The median, interquartile range, and overall range in. latitude of the herring complex in autumn is character- Overholtz and Friedland: Recovery of the Gulf of Maine Georges Bank complex 597 1968 1974 1980 1986 1992 1998 Y 700 I 600 - 500 - 400 - 300 200 100 h B 0^*S*M»^ 1963 1970 1977 1984 Year 1991 1998 Figure 4 Abundance indices (swept area, millions offish) of Atlantic herring from (A) spring bottom trawl surveys during 1968- 98 and IB) autumn bottom trawl surveys during 1963-98. ized by rapid changes and a general movement to the north during 1963-75 (Fig. 6A). During 1963-64, the stock complex was centered around 42°, moved a full degree to the south by 1967, and then shifted north to almost 43° by 1975. Survey abundance was so low during 1976-81 that any trends in median latitude were obscured (Fig. 6A). As the stock recovered, median latitude became stable, rang- ing between about 42°00' and 42°30'. The median longitude of the complex hovered around 68°15'-68°45' during 1963-72, then shifted to the west of 69° during 1973-86 (Fig. 6B). When recovery began, the median longitude of the complex moved slightly to the east of 69°, stabilized between 69°30' and 69°50' during 45 r A * + * 43 - , M MiM 1 J, ill 1 11 1 [1 39 JJ 1 ^ + ■¥ * , + 37 - %* 1 ♦ * * * * :: .-^.s 1 1 1 1 2000 2000 Figure 5 Median (notch), interquartile range, whiskers (±1.5 times the absolute value of the interquartile range), values out- side the whiskers ( * ), and far outside the whiskers (o) of the (A) latitude and (B) longitude of positive catches of Atlan- tic herring from the spring NEFSC bottom trawl survey during 1968-98. 1991-95, and moved farther east during 1996-98 to about 68°35'(Fig. 6B), Since 1968, median bottom water temperatures for the entire continental shelf during spring have fluctuated between 5° and 7.5°C, with individual bottom locations ranging between 1° and 14°C and a few values falling out- side this range (NMFS^). Herring maintained their pref- •'' NMFS (National Marine Fisheries Service). 2001. Unpubl. data. Northeast Fisheries Science Center, Woods Hole, MA 02543. 598 Fishery Bulletin 100(3) 45 44 43 I 42 _l 41 40 r A 39 1960 73 72 71 70 69 68 67 66 1970 1980 1990 2000 65 1960 _L 1970 1980 Year 1990 2000 Figure 6 Median (notch), interquartile range, whiskers (±1.5 times the absolute value of the interquartile range), values out- side the whiskers (*), and far outside the whiskers (o) of the (A) latitude and (B) longitude of positive catches of Atlantic herring from the autumn NEFSC bottom trawl survey during 1963-98. erence for water temperatures that were cooler than the annual shelf-wide ranges; most herring were caught at bottom temperatures around 5°C during 1968-94 and at about 6°C thereafter (Fig. 7A). During 1972-91, the inter- quartile range in capture temperature was very narrow, about 1-2°C, expanding afterwards to 2-3°C (Fig. 7A). The overall range in capture temperatures increased greatly in the 1990s. The median bottom water temperature for the conti- nental shelf during autumn 1963-98 fluctuated between 1968 1976 1984 1992 2000 20 r B 15 ™ 10 _L 1960 1970 1980 Year 1990 2000 Figure 7 Median (notch), interquartile range, whiskers (±1.5 times the absolute value of the interquartile range), values out- side the whiskers (*), and far outside the whiskers (o), of the bottom temperature of positive catches of Atlantic herring from the (A) spring NEFSC bottom trawl survey during 1968-98 and (B) autumn NEFSC bottom trawl survey during 1963-98. 8° and 12°C, and ranged between 3° and 25°C with a few higher observations (NMFS^). Herring were generally captured in the autumn surveys at bottom water tempera- tures around 8°C during 1963-76 and at about 7-8°C from 1983 to 1998 (Fig. 7B). The interquartile range in capture temperature was about 2-4°C (Fig. 7B), again much cooler than the shelf-wide ranges. The proportional number of locations where Atlantic herring were captured during spring surveys declined Overholtz and Fnedland: Recovery of the Gulf of Maine-Georges Bank complex 599 from 0.30 in 1968 to 0.09 in 1982 (Fig. 8A). During 1983- 93, herring were captured in an increasing proportion of survey hauls; by the 1990s, the herring proportion in all survey tows was over 0.60 (Fig. 8A). Autumn surveys dur- ing the 1960s and 1970s also showed a declining trend in the proportion of stations containing herring. This decline was even more pronounced than in spring, with the pro- portion of herring at less than 0.05 of the sample sites during 1979-81 (Fig. SB). This trend began to change in 1982 and by the 1990s the proportion of tows with herring was well over 0.40 (Fig. 8B). Spring distribution maps from selected years ( 1970-951 show that significant changes in herring distribution oc- curred during 1968-98. In 1970 herring were encountered at 74 sites in the central Gulf of Maine (GOM), on south- ern Georges Bank (GB), and across the continental shelf from south of Nantucket to Chesapeake Bay (Fig. 9). Dur- ing 1975, fewer sampling locations (39 sites) had herring and the distribution was constricted. The complex was confined to a few areas in the central GOM, on southern GB, south of Cape Cod, and off of Long Island (Fig. 9). In 1980. fish were encountered (89 sites) along eastern GB and found across western-southern GB to south of Long Island. During 1985. the spring distribution (48 sites) was more restricted than in 1980. By 1990 Atlantic herring were found at 101 sites throughout the GOM, western and southern GB. and across the shelf from Cape Cod to Chesapeake Bay. In 1995, herring were even more widely dispersed (108 sites) across the continental shelf Abundance of herring from these selected spring re- search survey years reflect the general decline in the stock in the late 1970s and recovery to much higher levels in the late 1990s (Fig. 9). In 1970, herring were abundant over the entire continental shelf from Nantucket to Ches- apeake Bay with the highest abundance off Long Island and New Jersey (Fig. 9). During the late 1970s, through the mid 1980s, herring were less abundant and the largest catches (101-500 fish per station) were generally taken south of Cape Cod (Fig. 9). By the 1990s, herring were abundant over a wider area and catches in the 101-500 fish-per-tow range were common. Maps of autumn survey catch locations depict significant spatial changes in herring distribution during the past 30 years (Fig. 10). In 1965 postspawning fish were located (47 sites) in the central GOM, on Jeffreys Ledge, Nantucket Shoals, and the shelf-slope break along western to north- ern GB (Fig. 10). In 1970 herring were caught at 25 sites (mostly in the GOM and on eastern GB), in 1975 at only 18 sites, and in 1980 at only one location (along the coast of Maine). The number of capture sites increased in 1985 (17) when herring were found primarily in the Stellwagen Bank-Jeffreys Ledge region, and at a few sites on Georges Bank (Fig. 10). In 1990, herring were widely dispersed (31 sites) along the coast of Maine, Jeffreys Ledge, and Nantucket shoals, and in 1995 herring were dispersed throughout the entire region (75 sites) (Fig. 10). In the early part of the autumn survey time series, her- ring were widely dispersed, but not very abundant. In 1965, only a few tows had catches greater than 50 fish (Fig. 10). During 1970-85 very few herring were taken 0.7 A 0.6 - 0.5 - c o 0.4 _ r o Q. O a. 0.3- • •^^v^ 0.2 • 0.1 - _L 0.0 1968 1974 1980 1986 1992 1998 0-7 rB 1963 1970 1977 1984 1991 1998 Year Figure 8 Proportion of survey tows with positive occurrences of Atlantic herring with a LOWESS curve for the data from (A) spring bottom trawl surveys during 1968-98 and (B) autumn bottom trawl surveys during 1963-98. and most tows represented only single fish (Fig. 10). By 1990 a general recovery was apparent and catches of 50-t- fish per station were common. Atlantic herring were again abundant along the coast of Maine, on Jeffreys Ledge, and in the Great South Channel area ( Fig. 10 ). In 1995, herring were abundant over the entire Gulf of Maine-CJeorges Bank region. Survey abundance indices of herring in the Gulf of Maine, on Nantucket Shoals, and on Georges Bank, in- creased greatly since the mid 1980s, but at different rates (Fig. 11). Increases in the GOM (range 2.8-85.9 million fish, 1987-98) and GB (range 2.1-120 million fish, 1991-98) regions appeared to be about equal in magnitude although the recovery in the GOM began earlier than on GB (Fig. 600 Fishery Bulletin 100(3) 38' Abundance 1970 . 1-25 . 26-50 . 51-100 • 10-500 • 5004- 38° 76° 74° 72° 70° 68° 66° Abundance 1975 1-25 . 26-50 . 51-100 • 101-500 • 500+ 70° 68° 66° Figure 9 Catch locations and abundance (number of fish per station) for Atlantic herring for selected years (1970, 1975, 1980, 1985, 1990, 1995) from spring bottom trawl surveys conducted during 1968-98. Overholtz and Friedland: Recovery of the Gulf of Maine-Georges Bank complex 601 72° 70° 68° Figure 9 (continued) 602 Fishery Bulletin 100(3) 72° 70° Figure 9 (continued) Overholtz and Fnedland: Recovery of the Gulf of Maine-Georges Bank complex 603 7^° 74° 7?° 7p° 6?° _ 6$: Ml. 44° 42» 40° 38° 36° Figure 10 Catch locations and abundance (number of fish per station) for Atlantic herring for selected years (1965, 1970, 1975, 1980, 1985, 1990, 1995) from autumn bottom trawl surveys conducted during 1963-98. 604 Fishery Bulletin 100(3) 7^° 79° 6g° 6f _64^ 76° 74° 72° 70' 76° 74° 72° 70° 68° 66° Figure 10 (continued) 64° Overholtz and Friedland: Recovery of the Gulf of Maine-Georges Bank complex 605 76° 74° 72° 70° 68° Figure 10 (continued) Abundance 1990 • 1-25 . 26-50 • 51-100 • 101-500 • 5004- 44° 42° 40° 38° ^136° 66° 64° 606 Fishery Bulletin 100(3) Abundance 1995 . 1-25 . 26-50 • 51-100 • 101-500 5004- 44' 42° 40° 38° 36° 76° 74° 72° 70° 68° Figure 10 (continued) 66° 64° 11). The recovery in the Gulf of Maine started in 1984 and continued through to 1998 (Fig. 1 lA). The largest overall in- crease in abundance has occurred in the Nantucket Shoals region (range 28.8-398.3 million fish) (Fig. IIB). Similarly, after a long hiatus from 1963-1986, this stock component increased considerably during 1987-98. The Georges Bank component was the last to recover, with significant increas- es in abundance beginning in 1992 (Fig. IIC). Discussion Atlantic herring are currently abundant in the Gulf of Maine-Georges Bank region as evidenced by research survey results and recent stock assessments (NEFSC^). This resource was heavily exploited during 1961-76 by distant-water fishing vessels with catches that were not sustainable, resulting in the collapse of the Georges Bank component in 1977. The causes of this collapse have been ascribed to excessive fishing mortality and concentration of effort on spawning areas (Anthony and Waring, 1980). ICNAF routinely set total allowable catches (TACs) in excess of scientific recommendations and these TACs were often exceeded (Anthony and Waring, 1980). Spawning con- centrations were heavily fished, and egg and larval produc- tion declined steadily, followed by the complete absence of larvae on Georges Bank for an entire decade (Anthony and Waring, 1980; Lough et al., 1985; Smith and Morse, 1993). Full recovery of the Atlantic herring resource has re- quired almost two decades (NEFSC^). Spawning area closures and restrictions in the Gulf of Maine beginning in the 1980s may have been influential, but the relative impact of these indirect measures cannot be assessed (Ste- venson^). During 1978-94 there were almost no offshore landings and abundance and biomass for the GB compo- nent improved steadily (NEFSC-). The distribution range of the herring complex was greatly reduced after each of the three stock components were heav- ily fished; the extent of the spring distribution was much reduced by the early 1980s and autumn spawning activity was confined to the western Gulf of Maine. The relative en- counter rate in both research surveys, as measured by the proportion of tows with herring, also declined during this period. As the complex began to recover, its range extended and the proportion of survey tows with herring increased dramatically. The complex was more widely dispersed by the mid 1990s with the recovery of all three components to, or above, historic abundance and biomass (NEFSC-). Murawski (1993) found that temperature is a very im- portant factor in determining the distribution of pelagic fishes, particularly Atlantic herring. Herring appear to have maintained a preferred thermal regime despite el- evated and variable overall shelf temperatures in spring and autumn, and large changes in abundance. The ±5°C median spring temperature for 1968-94 from our study is in agreement with the average of 4.8°C from Murawski (1993). After the complex recovered, the median spring bottom temperature increased slightly to about 6°, prob- ably indicative of the much broader range of the complex. Murawski ( 1993) calculated an autumn average of 8.5° for Atlantic herring and the median from the current study was 7-8°. This autumn preference reflects the distribu- Overholtz and Friedland Recovery of the Gulf of Maine-Georges Bank complex 607 90 80 70 60 50 40 30 20 10 0 1963 1970 1977 1984 1991 1998 W^ «*»t«..,i 400 I 300 E 0) o I 200 2 100 B Oi««^*i>«»i«««>»»i»»»«««i« 1963 1970 1977 1984 1991 1998 150 100- 50- otw^.i i i 1963 1970 1977 1984 Year 1991 1998 Figure 11 Abundance (swept area, millions of fish) of Atlantic herring from (A) the Gulf of Maine, (B) Nantucket Shoals, and (C) Georges Bank from autumn bottom trawl surveys during 1963-98. Several independent sources of information suggest that the recovery of this stock occurred in a stepwise manner with the Gulf of Maine component recovering first, followed by the Nantucket Shoals component, and finally the Georges Bank component. The autumn bottom trawl survey data showed an improved trend in the Gulf of Maine in 1984, the Nantucket Shoals component in 1987, and Georges Bank in 1992. Larval surveys conducted dur- ing 1971-90 showed a progression of larval abundance from the Cape Cod Bay-Stellwagen Bank area in 1976-84, to Nantucket Shoals in 1985-87, and on Georges Bank in 1988-90 (Smith and Morse, 1993). Larval distribution data suggested that larvae from Jeffreys Ledge spawners were probably transported to Nantucket Shoals and that in subsequent years larvae from these areas were trans- ported or adults moved to western Georges Bank and finally eastern Georges Bank (Smith and Morse, 1993). The general pattern of water circulation in the region (McGillicuddy et. al, 1988) also favors this hypothesis for the recovery. Smith and Morse (1993) suggested that the larval sur- vey data support the hypothesis that a recolonization of Atlantic herring occurred on Nantucket Shoals and Georg- es Bank. The data from their study showed a chronological pattern in larval abundance and distribution, beginning in the Gulf of Maine and progressing toward the Nantuck- et shoals region and finally to Georges Bank. In contrast, Stephenson and Kornfield (1990) asserted that the reap- pearance of herring on Georges Bank was more related to a resurgence of low numbers of stock-specific endemic fish. The autumn NMFS bottom trawl survey data tend to support the conclusions of Smith and Morse ( 1993) in both direction and timing. Survey indices show a progressive recovery in herring abundance from the Gulf of Maine to Nantucket Shoals, and finally to Georges Bank. Significant changes in stock distribution, shifts in cen- ters of primary abundance, and regional declines or disap- pearance of historically important spawning contingents are evident from the data. Autumn indices show that the stock component in the Gulf of Maine reached very low abundances during 1973-83 and that the Nantucket Shoals and Georges Bank components were nearly, if not actually, extirpated. Recovery times for these two compo- nents of the stock were relatively long — on the order of 10-15 years. Without a local supply of larvae, recovery was delayed considerably. Recovery was probably contingent on a supply of larval herring from the Gulf of Maine that eventually recolonized the offshore areas. Spawning re- strictions along the coast of Maine and on Jeffreys Ledge in the 1980s may also have contributed to the recovery of Gulf of Maine component, thereby setting in motion the recovery of the entire stock complex. tion of postspawning fish in the cooler waters of the Gulf of Maine and Northern Georges Bank. Although thermal conditions play a major role in the zoogeography of fishes, the influence of abundance was probably overwhelming in producing the recovery of the historic range in distribution observed in the Atlantic herring complex. Acknowledgments We thank the dedicated personnel from the Northeast Fisheries Science Center and the many volunteers who worked tirelessly to collect the bottom trawl survey data during 1963-98. 608 Fishery Bulletin 100(3) Literature cited Anthony, V. C, and G. Waring. 1980. The assessment and management of the Georges Bank herring fishery. Rapp. P.-V. Reunion Cons. Int. Explor. Mer 177:72-111. Azarovitz, T. R. 1981. A brief historical review of the Woods Hole laboratory trawl time series. In Bottom trawl surveys (W. G. Double- day and D. Rivard, ed.), p. 62-69. Can. Spec. Pub. Fish. Aquat. Sci. 58. Byrne, C. J., T. R. Azarovitz, and M. P. Sissenwine. 1981. Factors affecting variability of research vessel trawl surveys. In Bottom trawl surveys (W G. Doubleday and D. Rivard, ed.), p. 258-273. Can. Spec. Pub. Fish. Aquat. Sci. 58. Clark, S.H.(ed). 1998. Status of the fishery resources off the Northeastern United States for 1998. U.S. Dep. Commer, NOAA Tech. memo. NMFS-NE-115, 149 p. Cleveland, W. S. 1979. Robust locally weighted regression and smoothing scatterplots. J. Am. Stat. Assoc. 74:829:836. Grosslein, M. D. 1969. Groundfish survey program of BCF Woods Hole. Comm. Fish. Rev 31:22-35. Hennemuth, R. C, and S. Rockwell. 1987. History of fisheries conservation and management: Georges Bank (R. H. Backus and D. W. Borne, eds.), 593 p. MIT Press, Cambridge, Massachusetts, and London, England. Lough, R. G., G. R. Bolz, M.Pennington, and M. D. Grosslein. 1985. Larval abundance and mortality of Atlantic herring {Clupea harengus L.I spawned in the Georges Bank and Nantucket shoals areas, 1971-78 seasons, in relation to spawning stock size. J. Northwest Atl. Fish. Sci. 6:21-35. McGillicuddy, D. J.. D. R. Lynch, A. M. Moore, W C. Gentleman, and C. S. Davis. 1988. An adjoint data assimilation approach to the estima- tion o{ Pseudocalanus spp. population dynamics in the Gulf of Maine-Georges Bank region Fish. Oceanog. 7:205-218. Murawski, S. A. 1993. Climate change and marine fish distributions: fore- casting from historical analogy. Trans. Am. Fish. Soc. 122: 647-658. NEFSC (Northeast Fisheries Science Center). 1988. An evaluation of the bottom trawl survey program of the Northeast Fisheries Center U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/NEC-52, 83 p. Smith, W. G., and W. W. Morse. 1993. Larval distribution patterns: early signals for the collapse/recovery of Atlantic herring Clupea harengus in the Georges Bank area. Fish. Bull. 91:338-347. Stephenson, R. L., and I. Kornfield. 1990. Reappearance of spawning Atlantic herring iClupea harengus harengus) on Georges Bank: population resurgence not recolonization. Can. J. Fish. Aquat. Sci. 47:1060-1064. Zinkevich, V. N. 1967. Observations on the distribution of herring, Clupea harengus L., on Georges Bank and in adjacent waters in 1962-1965. ICNAF (International Commission for the Northwest Fisheries) Res. Bull. 4:101-115. 609 Abstract— Ago, size, abundance, and birthdaU' distributions were compared for larval Atlantic menhaden (Brevoor- tia tyrannus) collected weekly during their estuarine recruitment seasons in 1989-90, 1990-91, and 1992-93 in lower estuaries near Beaufort, North Carolina, and Tuckcrton, New Jersey, to determine the source of these larvae. Larval recruitment in New Jersey ex- tended for 9 months beginning in Octo- ber but was discontinuous and was punctuated by periods of no catch that were associated with low water temper- atures. In North Carolina, recruitment was continuous for 5-6 months begin- ning in November Total yearly larval density in North Carolina was higher (15-39x) than in New Jersey for each of the 3 years. Larvae collected in North Carolina generally grew faster than larvae collected in New Jersey and were, on average, older and larger Birthdate distributions (back-calculated from sag- ittal otolith ages) overlapped between sites and included many larvae that were spaw ned in winter Early spawned (through October) larvae caught in the New Jersey estuary were prob- ably spawned off New Jersey. Larvae spawned later (November-April) and collected in the same estuary were probably from south of Cape Hatteras because only there are winter water temperatures warm enough (>16°Cl to allow spawning and larval develop- ment. The percentage contribution of these late-spawned larvae from south of Cape Hatteras were an important, but variable fraction (10% in 1992-93 to 9,T7( in 1989-90) of the total number of larvae recruited to this New Jersey estuary. Thus, this study provides evi- dence that some B. tyrannus spawned south of Cape Hatteras may reach New Jersey estuarine nurseries. Recruitment of larval Atlantic menhaden (Brevoortia tyrannus) to North Carolina and New Jersey estuaries: evidence for larval transport northward along the east coast of the United States* Stanley M. Warlen Center lor Coastal Fisheries and Habitat Research National Ocean Service, NCAA Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516-9722 Kenneth W. Able Marine Field Station Institute of Marine and Coastal Sciences Rutgers University 800 Great Bay Boulevard Tuckerton, New Jersey 08087 Elisabeth H. Laban Center For Coastal Fishenes and Habitat Research National Ocean Service, NOAA Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516-9722 E-mail address (for E H Laban, contact author): Elisabeth Labamainoaa gov Manuscript accepted 23 January 2002. Fish. Bull. 100:609-623 (2002). " The larvae of some species of fishes that spawn in the South Atlantic Bight (SAB; Cape Hatteras, North Caro- hna to Cape Canaveral, Florida) are transported northward to the Middle Atlantic Bight (MAB; Cape Cod, Mas- sachusetts to Cape Hatteras, North Carolina) where they eventually use estuaries as juvenile nurseries. Kendall and Walford ( 1979) first suggested that spring-spawned bluefish, Pomatomus saltatrix, from the SAB are trans- ported several hundreds of kilometers northward where they eventually enter estuaries in the MAB (Hare and Cowen, 1996). Nyman and Conover (1988) and McBride and Conover (1991) used oto- lith analyses to confirm the presence of a spring-spawned component from the SAB in the juvenile bluefish population in MAB estuaries. The larvae of several other species from more southerly areas are also expatriated to the MAB (Hare and Cowen, 1991; Cowen et al., 1993; McBride and Able, 1998). Tropical and subtropical planktonic invertebrates are also known to occur in the MAB (Cox and Wiebe, 1979). Some seemingly unusual occurrences of Atlantic men- haden larvae in New Jersey during the winter suggested that the same pattern might occur for this species. Atlantic menhaden, Brevoortia tyran- nus, is a clupeid that migrates along the coast from the Gulf of Maine to Florida and adults spawn throughout this range. Studies (summarized in Warlen, 1994; Epifanio and Garvine, 2001) have shown that this species spawns off New England from late spring into summer and again in early fall, off the mid-Atlantic states in spring and fall, and in the SAB from October to March. Maximum numbers of Atlantic menhaden probably spawn during winter in offshore waters south of Cape Hatteras (Reintjes, 1969; Judy and Lewis, 1983) and North Carolina waters are likely one of the * Contribution 2002-04 of the Institute of Marine and Coastal Sciences, Rutgers Uni- versity, Tuckerton, New Jersey 08087. 610 Fishery Bulletin 100(3) major spawning grounds (Higham and Nicholson, 1964). The larvae are recruited to estuaries in North Carolina (Lewis and Mann, 1971; Warlen and Burke, 1990; Hettler and Barker, 1993; Warlen, 1994). For the purposes of this article, we define estuarine recruitment as the ingress or immigration of larvae to the estuary from the ocean. A variety of field and laboratory observations suggest that spawning is not likely to occur at the temperatures found from late fall through spring in the MAB. Field observations in Naragansett Bay, Rhode Island (GovoniM, have indicated that peak densities of Atlantic menhaden eggs are found at 18-20°C and very few eggs are found at temperatures below 16°C. Kendall and Reintjes (1975) found Atlantic menhaden eggs at only one of 92 stations during one (9 November-14 December) of three late fall- winter cruises in the MAB. A summary of the Marine Monitoring Assessment and Prediction (MARMAP) col- lections, for November-April of the 1979-87 surveys in the MAB west of longitude 72°W from Montauk Point, New York, to Cape Hatteras, North Carolina, showed late- stage eggs in only 25 of 2247 plankton hauls (Berrien and Sibunka, 1999). Except in one haul in January near Cape Hatteras, all other Atlantic menhaden eggs found in this MAB survey were collected in November. Atlantic menha- den eggs were collected in Onslow Bay, North Carolina, in December 1992 only at temperatures of 17-23. 5°C, and most were found at about 22°C (Peters'-). During South Atlantic recruitment experiment studies (SABRE), eggs were found between the Gulf Stream and mid-shelf fronts (17-23°C) in Onslow Bay (Checkley et al., 1999). Reduced temperatures (14.8-15.7°C) appeared to diminish the abil- ity to induce spawning of Atlantic menhaden in the labora- tory (Fitzhugh and Hettler, 1995) as compared to tempera- tures >17°C. Even if spawning can occur at temperatures <15°C, there must be successful hatching and larval development to ensure larval survival. At 16°C, early larval growth in dry weight was about one-half that at 20°C (Powell, 1993). Therefore, we concluded that opti- mum temperature for hatching and larval survival and growth is probably >16°C. From the above findings, we concluded that late-fall to early-spring water temperatures in the MAB are unsuit- ably low for Atlantic menhaden spawning. Water tempera- tures in the MAB to at least 300 km offshore are usually 7-14°C from mid-November through April (Benway et al., 1993a, 1993b). However, Atlantic menhaden larvae, prob- ably spawned during that period, recruit to New Jersey es- tuaries from winter to spring (Witting et al., 1999). These two facts suggest a warmer water (i.e. southern) origin of the wintertime larval Atlantic menhaden recruits in New Jersey and prompted us to ask whether larval transport from the SAB could explain this occurrence. To test this hypothesis, we examined synoptic collections of larval At- ' Govoni, J. J. 1996. Personal commun. Center for Coastal Fisheries and Habitat Research, National Ocean Service, NOAA, 101 Fivers Island Rd., Beaufort, NC 28516. ^ Peters, D. S. 1996. Personal commun. Center for Coastal Fisheries and Habitat Research. National Ocean Service, NOAA, 101 Pivers Island Rd., Beaufort, NC 28516. lantic menhaden recruiting to estuaries in North Carolina and in New Jersey from fall to spring. We determined the duration of recruitment, age and size, relative abundance, spawning season, and relative contribution of cohorts of larvae spawned in a given calendar week to the estuarine recruitment of Atlantic menhaden larvae to an estuary in each area. These data were used to estimate the percent- age of larval Atlantic menhaden collected in the New Jer- sey estuary that may have originated in the SAB. Methods Larval occurrence and abundance Sampling for larval Atlantic menhaden was conducted in North Carolina and New Jersey as these larvae recruited to the estuaries from the Atlantic Ocean. In North Caro- lina, larvae were collected at a station adjacent to Pivers Island in the lower Newport River estuary about 2 km inside Beaufort Inlet (Fig. 1). Nighttime sampling was conducted weekly at mid-flood tide during the expected recruitment period in three seasons (15 November 1989-2 May 1990, 14 November 1990-24 April 1991, and 19 November 1992-11 May 1993). Initial seasonal sampling was based on previous years' collections (Lewis and Mann, 1971; Warlen, 1994) that showed that Atlantic menhaden larval recruitment usually began no earlier than mid- November. End-of-year sampling terminated when larval density dropped to zero. Larvae were collected in four consecutive sets of a 1x2 m neuston net with 947-pm mesh fished (most sets 5-7 minutes long) just under the surface from a bridge platform. A flow meter was attached to the net to estimate the amount of water filtered. Details of the sampling protocol are given in Warlen (1994). Ich- thyoplankton samples were preserved in 95% ethanol and diluted so that the final concentration was at least 70% ethanol. Catches were standardized as the number of larvae/100 m-^ of water filtered. The mean of the density data for the four net sets on a given night was used as the density estimate of Atlantic menhaden larvae recruiting during the flood tide. In New Jersey, gear type and sampling effort were similar in concept but differed in some details (Witting et al., 1999). Ichthyoplankton were collected with a 1-m diameter ( 1-mm mesh) plankton net fitted with a flow me- ter The net was fished during nighttime flood tides from a bridge spanning Little Sheepshead Creek about 3 km upstream from the mouth that is approximately 2.5 km inside Little Egg Inlet (Fig. 1). Water depth at this bridge and the Pivers Island bridge was approximately 4 m. Sampling was conducted over the same years as in North Carolina, and sampling was conducted year round in New Jersey. A mean larval density (number larvae/100 m-* wa- ter fished) was calculated from all samples on each night ( five surface and five bottom sets in 1989-90, three surface and three bottom sets in 1990-91, three mid-water sets in 1992-93) and was used to estimate Atlantic menhaden larval recruitment during the flood tide. Each net set last- ed 0.5 hour Current speeds were generally 25-50 cm/s. Warlen et al ; Recruitment of larval Bievoortia tyrannus to North Carolina and New Jersey estuaries 611 Figure 1 Location of sampling sites for larval Atlantic menhaden (.Brevoortia tyrannus) collected at Little Sheepshead Creek, New Jersey, and Fivers Island, North Carolina, during 1989-90, 1990-91, and 1992-93. and approximated those measured in North Carolina. As in North Carolina, samples were preserved in 95% ethanol. We assumed that larval fish densities within and among years at a site were comparable (see Witting et al., 1999): therefore year-to-year relative abundances could be esti- mated at that site. However, differences between sites are not easily compared because the catch efficiency of the two types of passive nets is not known and there is no indepen- dent estimate of absolute abundance at either site. Age, growth, and birthdate determination Larvae were randomly subsampled from individual weekly net sets at both locations in proportion to their contribu- tion to the total nightly catch. All larvae in catches up to 20 fish were used. In catches of >20 fish, subsample sizes were proportional to catch but generally no more than 50 fish were aged per week from any one site. Experienced otolith readers at the Beaufort Laboratory determined the ages of 1435 larvae from North Carolina and 444 from New Jersey for our study. Larvae were measured to the nearest 0. 1 mm standard length (SL) with an ocular micrometer Estimated age was the number of sagittal otolith growth increments (Maillet and Checkley, 1990) plus an empirically derived value for the number of days (five) from spawning to first increment formation (Warlen, 1992). Additionally, late- larval to early-juvenile Atlantic menhaden reared in the laboratory through the winter at ambient water tempera- tures, formed daily otolith growth increments even when temperatures declined to 3°C (Ahrenholz et al., 2000). We assumed that the age at initial increment deposition in larval otoliths did not vary and that otolith increment deposition rate was constant within and between sampling seasons and the two sampling sites. To reduce the chance of underestimating ages of lar- vae, we conducted a detailed examination of the otolith microstructure of larvae collected in the 1990-91 season. Otoliths from fish with our assigned January-February hatching dates were re-examined and their growth incre- ments measured on an image analysis system. One reader made all counts and measurements in close consultation with two other experienced otolith readers. In contrast to the North Carolina larvae. New Jersey larval otoliths had areas in the middle portion of the counting path with narrower increments. However, New Jersey larvae did not have checks on their otoliths, which indicated that otolith growth did not stop. The otolith radius, standard length, and age relations suggested that New Jersey fish grew more slowly than North Carolina fish but did not indicate a systematic underestimation of the age of New Jersey fish due to narrow increments. Average growth of larvae was described by the Laird version (Laird et al., 1965) of the Gompertz growth equa- tion (Zweifel and Lasker, 1976). The model was fitted to 612 Fishery Bulletin 100(3) data for SL and estimated age at time of capture. To stabi- lize the variance of length over the observed age interval, length data were log-transformed and model parameters were estimated from the log-transformed version of the growth equation (Warlen, 1992). Because the conventional 3-parameter fit produced two cases where the estimates of the model intercept (L^^^) were biologically unreason- able, all data sets were rerun by using a 2-parameter (i.e. A|Q, and a) fitting procedure and by setting the intercept (third parameter) at 3.5 mm. This size at hatching was an intermediate value between the estimates of Powell and Phonlor { 1986) for Atlantic menhaden hatching at 16° and 24°C. The overall average growth rate from hatching to a given age was the quotient of the predicted size at that age minus the size at hatching (3.5 mm) divided by that age. Differences in population growth curves between sampling locations in each year were tested by using a 2-parameter Hotelling's T'^ test of the model parameters (Bernard |1981| as modified by Hoenig and Hanumara-^). The birthdate (=spawning date) of each larva was back- calculated by subtracting its estimated age from the date of capture. Birthweek cohorts were defined as larvae spawned in a given calendar week. Density values and percentage age composition of each week's catch were used to estimate the estuarine recruitment of birthweek cohorts for each year and each location by using the meth- ods of Warlen (1994). Birthdate distributions were used to estimate the percentage of larval recruits to the New Jersey estuary that could have originated in the SAB. The menhaden spawning season in North Carolina and New Jersey was estimated from the birthdate distributions of larvae that survived to enter the estuaries. The percent- age distribution of spawning by week was based on the relative abundance of larvae collected throughout each recruitment year We assumed that larval Atlantic menhaden caught each week were newly recruited to the estuary and that they were in transit past the sampling sites to upper portions of the estuaries. This assumption is supported by the arguments in Warlen (1994), Churchill et al. (1999), and Forward et al. (1999) for the North Carolina sampling site and the patterns observed for other shelf-spawned estuarine dependent species such as summer flounder iParalichthys dentatus) (Keefe and Able, 1993; Able and Kaiser, 1994) and other species for New Jersey (Witting etal., 1999). ResuKs Timing and abundance of larval recruitment The magnitude of larval Atlantic menhaden recruitment appeared to differ substantially between the two loca- tions (Fig. 2). The sum of the weekly mean larval densities ' Hoenig, N. A, and R. G. Hanumara. 1983. Statistical consid- erations in fitting seasonal growth models for fishes, 25 p. ICES council meeting 1983/D. Sep 'Nov 'Jan' 'Mar' 'Hay' Aug Oct Dec Feb Ape Jun Date Figure 2 Weekly mean density (larvae/100 m^) (line graph) and birthdate frequency distributions (bar graph) as calculated from relative abundances of larval Atlantic menhaden iBrevoortia tyrannus) in collec- tions at Pivers Island, North Carolina, and Little Sheepshead Creek, New Jersey, during 1989-90, 1990-91, and 1992-93. Warlen et al.: Recruitment of larval Brevooitia tyiannus to North Carolina and New Jersey estuaries 613 24 20 — 16 2 12 8 4 0 -2 24 20 16 12 8 4 0 -2 1989-90 NOV ' DEC ' JAN ' FEB ' MAR ' APR ' MAY NOV ' DEC JAN FEB MAR ' APR ' MAY NOV DEC ' JAN ' FEB ' MAR ' APR ' MAY ' Collection date Figure 3 Surface water temperature at Pivers Island, North Carolina, and Little Sheepshead Creek, New Jersey, during collections in 1989-90, 1990-91, and 1992-93. over all collections was a measure of the total abundance during a year. The total yearly densities (larvae/100 m') for North Carolina were 173 (1989-90), 128 (1990-91), and 286 ( 1992-93) and for New Jersey were 5.5 ( 1989-90), 8.4 ( 1990-91 ). and 7.3 ( 1992-93). Although there appeared to be generally large differences in overall abundance between the two areas, we could not determine how much of the 15-39x difference was due to the geographic loca- tion, the catch efficiency of the two types of gear, or to dif- ferences in sampling effort. Nightly mean densities were always low in New Jersey and, with one exception (20 October 1992), never exceeded 2.0 larvae/100 m^, whereas in North Carolina nightly catches could exceed 40 larvae/ 100 m' every season (Fig. 2). There were apparent differences in the timing of larval Atlantic menhaden entering estuaries in North Carolina and New Jersey (Fig. 2). Larval recruitment in New Jer- sey, which extended over a longer period (8-9 months) than that in North Carolina (5-6 months), began in Octo- ber and ended in June (Fig. 2). Recruitment in New Jersey was discontinuous and was punctuated by periods of no catch, usually in late fall to early spring. The absences of 614 Fishery Bulletin 100(3) larval catches in some periods in New Jersey are associ- ated with low water temperatures. There were no larvae caught in December^anuary of 1989-90 and 1990-91 and in February-March 1993 when water temperatures remained between 1.0 and 5.0°C (Fig. 3). In contrast, re- cruitment in North Carolina began in November and was continuous until ending in April or May (Fig. 2). Water temperature in North Carolina was warmer than in New Jersey and only twice dropped below 6°C (Fig. 3). In all three years in North Carolina, recruitment was character- ized by a late season peak that occurred in March or April (Fig. 2). Predominant recruitment in New Jersey occurred either early (1992-93), late (1989-90) or with peaks early and late in the season (1990-91). Larval age and size Menhaden lai-vae entered both estuaries after development in the ocean. Most larvae were >1 month old and > 17 mm long and there were no recently hatched specimens. How- ever, the age and size of larvae entering the two estuaries differed within and between sites. The weekly mean age of Atlantic menhaden larvae collected in North Carolina increased linearly from early in the season to about the end of March in each year (Fig. 4). The weekly mean age over this period increased by a factor of 2-3x. In two years (1990, 1993) the mean age declined after March to levels approximating those early in the recruitment season. Larvae collected in the North Carolina estuary during peak recruitment were also the older (and larger) larvae. In New Jersey, although there was a general increase in larval age between October and December, there was no late season decline in mean age as we observed for North Carolina larvae. Larvae collected in New Jersey were gen- erally younger than those collected in North Carolina for each of the three years (Fig. 5). The overall mean age of larvae collected in North Carolina for 1989-90, 1990-91, and 1992-93 were 60.7, 58.6, and 69.3 d as compared to 54.3, 52.7, and 50.9 d for New Jersey, respectively. The size of larvae showed seasonal trends similar to those for larval ages in each of the years, i.e. larval length increased and then decreased through time in North Carolina, whereas in New Jersey it increased but did not decline (Fig. 6). Larvae recruited in North Carolina were generally similar in size but were on average larger (24.9, 25.4, and 25.8 mm) than larvae recruited in New Jersey (22.2, 21.3, and 22.1 mm) in all three years (Figs. 6 and 7). Growth of larvae Larvae collected in North Carolina were generally older and larger and grew faster than larvae in New Jersey (Fig. 8). Population growth curves were significantly different between sampling locations in all years (Table 1). The pre- dicted overall average growth rate from hatching to 65 d for larvae recruited in North Carolina was similar among years (0.35 mnVd for 1989-90, 0.36 mm/d for 1990-91, and 0.35 mm/d for 1992-93). Corresponding rates for larvae recruited to New Jersey were 0.30 mm/d for 1989-90, 0.31mm/d for 1990-91, and 0.32 mm/d for 1992-93. Larval birthdates Back-calculated spawning dates varied between years in both locations but were most variable in New Jersey (Fig. 2). In New Jersey in the 1989-90 season there were two distinct groups of larvae. About 13'7f of the larvae were spawned early (September-October), then none to the middle of December, and the remaining larvae (87%) were spawned thereafter to the end of March. Larvae captured in North Carolina were spawned from the week ending 21 October 1989 continuously through the week ending 3 March 1990 and overlapped the later spawning dates for New Jersey larvae. Larvae spawned in March recruited to New Jersey but virtually none recruited to North Carolina. In the 1990-91 season, spawning dates of larvae collected in New Jersey (7 months) totally overlapped those collected in North Carolina (4 months) (Fig. 2). About 45% of the New Jersey larvae were spawned early (through week ending 27 October 1990) and the balance was distributed over the last five months with a mode in the week ending 23 February 1991. As in 1989-90, March-spawned larvae recruited to New Jersey but not to North Carolina. The highest frequen- cy of spawning dates of larvae caught in North Carolina occurred from mid-January to the end of the season and peaked during the week ending 26 January 1991. The spawning date distributions in 1992-93 were dif- ferent from those in the previous two sampling seasons. About 77% of the New Jersey larvae were spawned by the end of October An additional 18'7( of the total larvae entering the estuary were spawned in November to mid- December and only about 5% of the total were spawned in March-April. Larvae recruited to North Carolina had been spawned from October to April but there was no dis- tinct mode as observed in the other two years. Late season spawning (March-April) contributed larvae to both New Jersey (=5% of total) and North Carolina (=13% of total). In both earlier years there were no April-spawned larvae and March-spawned larvae were only found in New Jersey. Discussion Atlantic menhaden larvae recruited to the New Jersey estuary probably originate from two sources: locally (MAB) and from the SAB. Evidence of local spawning is seen in the larvae recruited to the New Jersey estuary from October through early December (Fig. 2). These larvae probably originated from local (i.e. New Jersey) spawning as adults begin their southward migration out of the MAB area in about November of each year (Higham and Nicholson, 1964). Schools of spawning-size adults (200 mm-i-) are captured in the commercial fishery in the fall and early winter as they riTove south along the coast from Maryland to North Carolina (Smith-*). Spawning in the SAB produces larvae, some of which are transported northward and contribute recruits to New Jersey later in ^ Smith, J. W. 1996. Personal commun. Center for Coastal Fisheries and Habitat Research, National Ocean Service, NOAA, 101 Pivers Island Rd., Beaufort, NC. Warlen et a\ Recruitment of larval Bievoortia tyrannus to North Carolina and New Jersey estuaries 615 NORTH CAROLINA 1989-90 NEW JERSEY 1989-90 OCT , NOV I DEC I J»N I .EB I M.« I .PR I „»» I JUN I "' OCT I NOV I DEC I MN I FEB I MAR I ^^ I MAT I J^ I JO 100 NORTH CAROLINA 1990-91 NEW JERSEY 1990-91 OCT I NOV I DEC I JAN I fEB I MAR I APH I MAY I JUN I OCT I NOV I DEC I JAN I FEB I MAR I APR I UAT I JUN I lOOr NORTH CAROLINA 1992-93 NEW JERSEY 1992-93 OCT I NOV I DEC I JAN I FEB I MAR I APR I MAT I JUN I OCT I NOV I DEC I JAN I FEB I MAR I APR I MAT I JUN I Collection date Figure 4 Mean of estimated age (days) with 95^^ confidence limits of larval Atlantic menhaden (Brevoortia tyran- nus) collected weekly at Pivers Island. North Carolina, and Little Sheepshead Creek, New Jersey, during their estuarine recruitment in 1989-90, 1990-91, and 1992-93. the recruitment season. Some larvae originating in the southernmost portion of the MAB in fall may even contrib- ute some recruits to the SAB as three-dimensional circula- tion models have predicted (Quinlan et al., 1999; Rice et al., 1999; Stegmann et al.. 1999). However, any adults that might remain in the MAB after November would experi- ence ocean temperatures on the continental shelf that are not only too cold to allow menhaden spawning but that alSo might prevent eggs from hatching or larvae from develop- ing. Even if we accept the minimum temperature of 13°C 616 Fishery Bulletin 100(3) New Jersey 1989-1990 North Carolina 1989-1990 15 25 35 45 55 65 75 85 95 105 115 125 Estimated age (days) Figure 5 Percentage distribution of estimated age (5-d intervals) for larval Atlantic menhaden (Brevoortia tyrannus) recruited to Fivers Island, North Carolina, and Little Sheepshead Creek, New Jersey, in 1989-90, 1990-91, and 1992-93. for Atlantic menhaden spawning suggested by Stegmann et al. (1999), there were no water temperatures above 12°C in the MAB in December through March (Fig. 9) during the three years of our study. Two differing hypotheses could explain the occurrence of menhaden larvae recruiting to the New Jersey estuary in winter and spring. One is that larvae spawned in the MAB in October or November remain in the MAB until they move into estuaries in the spring. This hypothesis would require a retention mechanism to keep larvae on the continental shelf over the winter in water <10°C. No such mechanism is likely as field and modeling studies indicate persistent southwestward transport (Pietrafesa et al., 1994; Churchill and Berger, 1998; Werner et al., 1999). If larvae overwintered on the continental shelf, they would be about 150-210 d old when they recruited to the estuary 5-7 months after spawning. No Atlantic menhaden this old were found in any of the samples, or from any other studies (Warlen, 1994; Rice et al., 1999), and underaging of larval otoliths in our study did not appear to be a problem. An alternative but more feasible hypothesis for the ob- served winter-spawned larvae collected in New Jersey from December to the end of the recruitment season (May-June) is that they were spawned in the SAB and were transported northward to the MAB. We used the density-weighted, back- calculated spawning date distributions of larvae collected in New Jersey that were spawned from November to April as the best estimator of the percentage of the total recnaits that were spawned in the SAB. If we use the portion of lar- vae spawned December or later as the most conservative es- timator, then the contribution of larvae from the SAB, of all New Jersey recruits, was 87% in 1989-90, 47% in 1990-91, and 10% in 1992-93. If larvae spawned November or later are considered to be from the SAB where there was spawn- ing, then the estimate (87%) is the same for 1989-90 but increases to 55% in 1990-91 and to 23% in 1992-93. Warlen et al Recruitment of larval Bicvoortia tyrannus to North Carolina and New Jersey estuaries 617 OCT I NOV I DEC 1 JAN I FEB I MAR I APR I MAY I JUN NORTH CAROLINA 1990-9 1 OCT 1 NOV I OEC I JAN t FEB I MAR : APR I MAY I JUN I NORTH CAROLINA 1992-93 Otf I NOV I OEC I JAN I FEB I MAR I APR I MAY I JUN I NEW JERSEY ias«-oo bif I NOV I DEC I jaN I m I UK* I KK I MJIV I JIM I NEW JERSEY 1990-9 1 k^ ' oet I NOV I ott I jAn I fa I mAD I AM I UIV I JUN I atT I n4v I bli I iM I rtt I UAH I aM I UAV I JUW I Collection date Figure 6 Mean standard length (mm) with 95% confidence Umits of larval Atlantic menhaden iBrevoortia tyrannus) collected weekly at Pivers Island, North Carolina, and Little Sheepshead Creek, New Jersey, during their estuarine recruitment in 1989-90, 1990-91, and 1992-93. Although the estimated percentage of larvae contributed to the New Jersey estuary from the SAB varied between \Q'7( and &1^7(^ the actual numbers of contributed larvae were relatively low, over an order of magnitude less than that observed for North Carolina, because of the overall low number of larvae recruited there. However, the total contribution of SAB larvae to New Jersey recruitment over the entire season may actually be greater than esti- mated from the birthdate distributions. The contribution of larvae spawned in the MAE to recruitment in New Jer- sey is based on the presence of early season larvae, but for these fish to contribute to the population the next spring, they must survive the winter in the estuary. If the winter water temperature in New Jersey estuaries drops below the lethal limit for Atlantic menhaden larvae (<5.0°C; Lewis, 1965), there will be decreased survival for the fall 618 Fishery Bulletin 100(3) 30 r 20 10 0 10 20 30 30 20 10 0 10 20 30 30 20 10 0 10 20 30 New J«rs«y 1989-1990 North Carolina 1989-1990 New Jersey 1990-1991 I I I North Carolina 1990-1991 New Jersey 1992-1993 North Carolina 1992-1993 12 14 16 18 20 22 24 26 28 30 32 34 Standard length(mm) Figure 7 Percentage distribution of standard length (1-nim length intervals! for larval Atlantic menhaden iBrevoortia tyrannus) collected during their recruitment at Pivers Island, North Carolina, and Little Sheepshead Creek, New Jersey, in 1989-90, 1990-91, and 1992-93. recruited larvae. Temperatures this low are often found during winter in New Jersey estuaries (Able et al., 1992) and result in mortality of several other estuarine species (Szedlmayer et al., 1992; Keefe and Able, 1993; Hales and Able, 2001). We did not catch Atlantic menhaden larvae in the New Jersey estuary at any time when the surface wa- ter temperature was <5°C. Temperatures <3°C deterred larval Atlantic menhaden entry into estuaries, inhibited movements into tributaries, and caused mass mortalities in Indian River, Delaware (Reintjes and Pacheco, 1966). Alternatively, larvae recruiting in March or later have less chance of encountering lethal temperatures and experi- encing cold-related mortality and hence their proportion of the total year's recruitment may increase. The precise mechanisms responsible for larval Atlantic menhaden transport from the SAB to the MAB are un- known. However, it seems likely that larvae may be trans- ported northward by the Gulf Stream. Atlantic menhaden larvae that are spawned on the outer continental shelf of North Carolina (Checkley et al., 1988; Warlen, 1992; Govoni and Pietrafesa, 1994) may be entrained in mean- ders that impinge onto the shelf (Pietrafesa et al., 1985) and carried northward out of the SAB. Govoni and Spach (1999) estimated offshore fluxes of Atlantic menhaden larvae from the North Carolina shelf to the Gulf Stream during the winter Modeling results supported these em- pirical findings and indicated that export from the North Carolina shelf is influenced by both wind and bathymetry (Hare et al., 1999). Once in the Gulf Stream, larvae can be transported rapidly (=100 cm/s) to the northeast (Hare^). Hare, J. A. 1999. Personal commun. Center for Coastal Fisheries and Habitat Research, National Ocean Service, NOAA, 101 Pivers Island Rd., Beaufort, NC. Warlen el a\: Recruitment of larval Bievooitia tyi annus to North Carolina and New Jersey estuanes 619 1989-1990 A ^ < < ^A^^^^tt^ It* t * a » ■ ■MB 1 ■= a a * NC n=418 ^- NJ n=I87 A 0 20 30 40 50 60 70 80 90 100 110 120 6 E £ 25- c 0) "O ra 20 ■o c nj 03 1990-1991 NC n=449 NJ n=171 10 20 30 40 50 60 70 80 90 100 110 120 3St 1992-1993 30- -/ 1 NC n=568 NJ n= 86 10 20 30 40 SO 60 70 80 90 100 110 120 Estimated age (days) Figure 8 Summary plots of standard length (mm) on estimated age (daysl for larval Atlantic menhaden [Brevoortia tyran- nusi collected at Pivers Island. North Carolina, and Little Sheepshead Creek, New Jersey, in 1989-90, 1990-9, and 1992-93. o 789 10 1112 123456 IVIonth STATION o 44025 - Long Island A 44029 - Delaware Bay O CHLV2 - Chesapeake Bay Figure 9 Maximum monthly sea surface temperatures recorded at three NOAA stations in the Middle Atlantic Bight that encompassed the time peri- ods of our three sampling years. Minimum tem- perature that Brevoortia tyrannus eggs had been caught during MARMAP surveys (l.S'C; Steg- mann et al., 1999) are indicated by a horizontal dashed line. In months with maximum tempera- tures below 13°C no spawning of B. tyrannus north of Cape Hatteras was expected. NOAA Data Buoy 4402.5 located -45 km south of Long Island, New York at 40°15'1"N, 73°10'0"W. NOAA Data Buoy 44029 is located -45 km southeast of Delaware Bay at 38''27'49"N, 74°42'7"W. NOAA C-MAN station CHLV2 is located --20km east of Chesapeake Bay at 36°54'18"N, 75°42'48"W. Warm core ring streamers and surface intrusions of warmer water could assist in transporting larvae from the Gulf Stream to the MAB continental slope (Hare and Cowen, 1991, 1996). Larvae must then cross the shelf- slope front and once larvae are in shelf waters they may be transported south and westward along the coast and to- ward shore by the predominant winds from the northeast which induce an Eckman drift westward toward shore. Biological mechanisms such as vertical migrations and directional swimming behavior, can modulate transport by placing larvae in water masses favorable to cross slope or cross shelf transport (Hare and Cowan, 1996). Cowen et al. (1993) and Hare and Cowen (1993) have described this general scenario for several species that may be trans- ported to the MAB from the SAB. Regardless of the type of transport mechanisms, it is clear that larvae of a number of other species are trans- ported from the SAB to the MAB. The larvae ofAnguilla rostrata, which originate in the Sargasso Sea (Schmidt, 620 Fishery Bulletin 100(3) Table 1 Estimates of Gompertz growth model parameters iA^g,, a) with standard errors in parentheses for larval Atlantic menhaden col- lected in 1989-90, 1990 -91 and 1992-93 in North Carolina and New Jersey and results of Hotelling's T^ tests comparing growth models between sites for each year. In each model fit, L^g, was set at 3.5 mm Standard length. * Significant at 0.01. Statistics Year 1989-90 1990-91 1992-93 Growth model' North Carolina Aoi 0.113(0.0021 0.101(0.001) 0.113(0.002) a 0.054(0.001) 0.048 (0.001) 0.055(0.001) New Jersey ^lOI 0.129(0.0081 0.092(0.003) 0.104(0.005) a 0,068 (0.005) 0.046(0.002) 0.051(0.003) Hotelling's T- Sample size ;! 1— NC 418 449 568 ;!,— NJ 187 171 85 F(0.01,/',,/:,)- 5.30 5.30 5.30 '^ (calcutated> 153.87* 127.81* 7.12* ' L,Q, = length at hatching. '^(O)" = specific growth rate at hatching, o = exponential decline in A^^-j^- - 0.01 = level of testing, /■( = np =no. of parameters )./^ = n^+n.-np- 1. 1922; McCleave and Miller, 1994), occur in and near the Gulf Stream in the SAB and appear as glass eels in the New Jersey collections at approximately the same time as menhaden larvae (Witting et al., 1999). The same general track is presumed for Conger oceanicus which also spawn in the Sargasso Sea (Miller, 1995) and are also collected at the New Jersey study site, although they occur later (May^une) (Able and Fahay, 1998; Witting et al., 1999). Similarly, other species which presumably spawn in the SAB such as Chaetodon spp. (McBride and Able, 1998), Mugil curema (Collins and Stender, 1989), and Lutjanus griseus (Able and Fahay, 1998) also enter the New Jersey study site (Able et al., 1997; Witting et al., 1999). Larval spot, Leiostomus xanthurus, spawned south of Cape Hat- teras may also be transported to estuaries in the MAB (Norcross and Bodolus, 1991). In summary, although the SAB may be a very important spawning area for Atlantic menhaden, all larvae produced in the SAB are not retained there; some are transported to the MAB. We believe the birthdate distribution data sug- gest that Atlantic menhaden larvae spawned in the SAB are an important but variable source of recruits (10-87%) for the MAB based on the collections from New Jersey. The number of winter-spawned larvae ultimately recruited to New Jersey probably depends on the initial number of larvae transported northward from the SAB, the efficacy of the transport mechanisms, and the mortality of lar- vae during the process. Obviously, the estimates for this single New Jersey estuary may differ from other areas in the MAB. As a result of the above variables and recent insights from the SABRE program, it appears that larval Atlantic menhaden supply is very complex with frequent exchanges between the MAB and the SAB, as well as the contribution from local spawning. Ultimately, we need to know which spawning seasons and sites contribute most individuals to nursery habitats (Beck et al., 2001) and to the adult population. The relative contribution of SAB menhaden lai^ae, and those of other species, to other estuaries in the MAB might clarify the importance of MAB estuaries and contribute to an improved understanding of recruitment mechanisms for shelf-spawned estuarine-dependent fishes in the MAB. Additional evidence for the contribution of SAB larvae to the MAB could come from analyses of the elemental composition of the primordium (nucleus) of larval otoliths from both New Jersey and North Carolina. The laser ablation inductively coupled plasma mass spectroscopy (ICPMS) technique (Campana et al., 1994; Thorrold et al., 2001) might be used to discriminate among larvae to determine if they have similar or different geographic ori- gins. Similarities in elemental composition of New Jersey and North Carolina larvae with the same winter-spawned birthdate would be an important step toward validating the common geographic origin of the larvae recruited to New Jersey and perhaps elsewhere in the MAB. Acknowledgments Many individuals from Rutgers University Marine Field Station and the Beaufort Laboratory assisted in the col- lection and analysis of material. We thank William Rugen Warlen et al.: Recruitment of larval Brevoonia tyi annus to North Carolina and New Jersey estuaries 621 who assisted in aping larvae and Dave Witting and Mike Fahay who provided valuable assistance in several other phases of this study. Douglas Vaughan ran the Hotelling's T- test and offered comments. Dean Ahronholz, David Peters, Jonathan Hare, and three anonymous referees provided valuable critical reviews of earlier versions of the manuscript. Support for portions of this study was received from the South Atlantic Bight Recruitment Experiment (SABRE) program of the National Oceanic and Atmospheric Administration's (NOAA) Coastal Fish- eries Ecosystem/Coastal Ocean Program and from NOAA's Office of Sea Grant NA89AA-D-SSG057 (project no. R/F-42) and NA36-RGO505 (project no. R/F-65) and the Institute of Marine and Coastal Sciences (IMCS), Rutgers University. 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 Univ. Press, New Brunswick. NJ. Able, K. W.. R. Hoden, D. A. Witting, and J. B. Durand. 1992. Physical parameters of the Great Bay-MuUica River Estuary (with a list of research publications). Tech. Rep. 92-06, .38 p. Rutgers Univ., Institute of Marine and Coastal Sciences, Tuckerton, NJ. Able, K. W., and S. C. Kaiser 1994. Synthesis of summer flounder habitat parameters. NOAA Coastal Ocean Program Decision Analysis Series 1, 68 p. NOAA Coastal Ocean Office, Silver Spring, MD. Able, K. W., A. Kustka, D. A. Witting, R. A. Roundtree, K. J. Smith, and R. S. McBride. 1997. Fishes of Great Bay, New Jersey: larvae and juveniles collected by nightlighting. Tech. Rep. 97-05, 30 p. Insti- tute of Marine and Coastal Sciences, Rutgers Univ., Tucker- ton, NJ. Ahrenholz, D. W.. D. D. Squires. J. A. Rice, S. W. Ni.xon. and G. R. Fitzhugh. 2000. Periodicity of increment formation in otoliths of over- wintering postlarval and prejuvenile Atlantic menhaden, Breuoortia tyrannus. Fish. Bull. 98:421-426. 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. Shendan, M. P. Weinstein. 2001. The identification, conservation and management of estuarine and marine nurseries for fish and invertebrates. Bioscience 51(8):633-641. Benway R. L., J. W. Jossi, K. P. Thomas, and J. R. Goulet. 1993a. Variability of temperature and salinity in the Middle Atlantic Bight and Gulf of Maine. U.S. Dep. Commer, NOAA Tech. Rep. NMFS 112, 108 p. Benway, R. L., K. P. Thomas, and J. W. Jossi. 1993b. Water-column thermal structure in the Middle Atlan- tic Bight and Gulf of Maine during 1978-92. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/NEC-97, 1.54 p. Bernard, D. R. 1981. Multivariate analysis as a means of comparing growth in fish. Can. J. Fish. Aquat. Sci. 38:23.3-236. Berrien, P., and J. Sibunka. 1999. Distribution patterns of fish eggs in the U.S. north- east continental shelf ecosystem. 1977-1987. U.S. Dep. Commer, NOAA Tech. Rep. NMFS 145, 310 p. Campana, S. E., A. J. Fowler, and C. M. Jones. 1994. Otolith elemental fingerprinting for stock identifica- tion of Atlantic cod iGadus inorhua) using laser ablation ICPMS. Can. J Fish. Aquat. Sci. 51:1942-1950. Checkley D. M., Jr, S. Raman, G. L. Maillet, and K. M. Ma.son. 1988. Winter storm effects on the spawning and larval drift of a pelagic fish. Nature ( Lond. ) 335:346-348. Checkley D. M., Jr, F B. Ortnor, F. E. Werner, L. R. Settle, and S. R. Cummings. 1999. Spawning habitat of the Atlantic menhaden in On- slow Bay, North Carolina. Fish. Oceanogr. 8 (suppl. 2):22- ,36. Churchill, J. H., and T J. Berger 1998. Transport of Middle Atlantic Bight shelf water to the Gulf Stream near Cape Hatteras. J. Geophy Res. 103 (C13):30605-30621. Churchill, J. H., R. B. Forward, R. A. Luettich, J. L. Hench, W. F Hettler, L. B. Crowder and J. O. Blanton. 1999. Circulation and larval fish transport within a tidally dominated estuary. Fish. Oceanogr. 8 (suppl. 2): 173-1 89. Collins, M. R., and B. W. Stender 1989. Larval striped mullet iMugil cephalus) and white mullet iMugil curema) off the southeastern United States. Bull. Mar Sci. 45:580-589. Cowen, R. K., J. A. Hare, and M. P. Fahay 1993. Beyond hydrography: can physical processes explain larval fish assemblages within the Middle Atlantic Bight? Bull. Mar Sci. 53:567-587. Cox,J.,andPH.Wiebe. 1979. Origins of oceanic plankton in the Middle Atlantic Bight. Estuar Coast. Mar Sci. 9:509-527. Epifanio. C. E., and R. W. Garvine. 2001. Larval transport on the Atlantic continental shelf of North America: a review. Estuar Coast. Mar Sci. 52; 51-77. Fitzhugh, G. R., and W. F Hettler 1995. Temperature infiuence on postovulatory follicle degen- eration in Atlantic menhaden, Brevoortia tyrannus. Fish. Bull. 93:568-572. Forward, R. B., K. A. Reinsel, D. S. Peters, R. A. Tankersley J. H. Churchill, L. B. Crowder W. F Hettler S. M. Warien. and M. D. Green. 1999. Transport of fish larvae through a tidal inlet. Fish. Oceanogr. 8 (suppl. 2): 153-172. Govoni, J. J., and L. J. Pietrafesa. 1994. Eulerian views of layered water currents, vertical dis- tribution of some larval fishes, and inferred advective trans- port over the continental shelf off North Carolina, USA, in winter Fish. Oceanogr. 3:120-132. Govoni, J. J., and H. L. Spach. 1999. Exchange and flux of larval fishes across the western Gulf Stream front south of Cape Hatteras, USA, in winter Fish. Oceanogr. 8 (suppl. 2):77-92. Hales, L. S., Jr, and K. W. Able. 2001. Winter mortality, growth and behavior of young-of- the-year of four coastal marine fishes in New Jersey (USA) waters. Mar Biol. 139:45-54. Hare, J. A., and R. K. Cowen. 1991. Expatriation of Xynchtys nouacula (Pisces: Labridae) larvae: Evidence of rapid cross-slope exchange. J. Mar Res. 49:801-823. 1993. Ecological and evolutionary implications of the larval transport and reproductive strategy of bluefish Pomatomus saltatrix. Mar Ecol. Prog. Ser 98:1-16. 1996. Transport mechanisms of larval and pelagic juve- nile bluefish {Pomatomus saltatnx) from South Atlantic 622 Fishery Bulletin 100(3) Bight spawning grounds to Middle Atlantic Bight nursery habitats. Limnol. Oceanogr. 41:1264-1280. Hare, J. A., J. A. Quinlan, F. E. Werner. B. O. Blanton, J. J. Govoni, R. B. Forward, L. R. Settle, and D. E. Hoss. 1999. Larval transport during winter in the SABRE study area: results of a coupled vertical larval behaviour-three- dimensional circulation model. Fish. Oceanogr. 8 (suppl. 2):57-76. Hettler, W. F., Jr., and D. L. Barker. 1993. Distribution and abundance of larval fishes at two North Carolina inlets. Estuar Coast. Shelf Sci. 37:161- 179. Higham, J. R., and W. R. Nicholson. 1964. Sexual maturation and spawning of Atlantic men- haden. Fish. Bull. 63:255-271. Judy, M. H., and R. M. Lewis. 1983. Distribution of eggs and larvae of Atlantic menhaden, Brevoortia tyrannus, along the Atlantic coast of the United States. U.S. Dep. Commer, NOAA Tech. Rep. NMFS SSRF-774, 23 p. Keefe, M., and K. W. Able. 1993. Patterns of metamorphosis in summer flounder. Para- lichthys dentatus. J. Fish. Biol. 42:71.3-728. Kendall, A. W.. Jr., and J. W. Reintjes. 1975. Geographic and hydrographic distribution of Atlan- tic menhaden eggs and larvae along the Middle Atlantic coast from RV Dolphin cruises, 1965-66. Fish. Bull. 73: 317-335. Kendall, A. W., Jr., and L. A. Walford. 1979. Sources and distribution of bluefish, Pomatomus sal- tatrix. larvae and juveniles off the east coast of the United States. Fish. Bull. 77:213-227. Laird, A. K., S. A. Tyler, and A. D. Barton. 1965. Dynamics of normal growth. Growth 29:23.3-248. Lewis, R. M. 1965. The effect of minimum temperature on the sui-vival of larval Atlantic menhaden, Brevoortia tyrannus. Trans. Am. Fish. Soc. 94:409-412. Lewis, R. M., and W. C. Mann. 1971. Occurrence and abundance of larval Atlantic men- haden, Brevoortia tyrannus, at two North Carolina inlets with notes on associated species. Trans. Am. Fish. Soc. 100:296-.301. Maillet, G. L., and D. M. Checkley Jr 1990. Effects of starvation on the frequency of formation and width of growth increments in sagittae of laboratory- reared Atlantic menhaden Brevoortia tyrannus. Fish. Bull. 88:1.5.5-165. McBride, R. S., and K. W. Able. 1998. Ecology and fate of butterflyfishes, Chaetodon spp., in the temperate, western North Atlantic. Bull. Mar. Sci. 63: 401-416. McBride, R. S., and D. O. Conoven 1991. Recruitment of young-of-the-year bluefish Pomato- mus saltatrix to the New York Bight: variation in abun- dance and growth of spring- and summer-spawned cohorts. Mar. Ecol. Prog. Ser 78:20.5-216. McCleave, J. D., and M. J. Miller. 1994. Spawning of Conger oceanicus and Conger triporiceps (Congridae) in the Sargasso Sea and subsequent distribu- tion of leptocephali. Environ. Biol. Fish. 39:339-3.55. Miller, M. J. 1995. Species assemblages of leptocephali in the Sargasso Sea and Florida Current. Mar Ecol. Prog. Ser. 121:11-26. Norcross. B. L., and D A. Bodolus. 1991. Hypothetical northern spawning limit and larval transport of spot. In Larval fish recruitment and research in the Americas (R. D. Hoyt, ed.), p. 77-88. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 95. Nyman, R. M., and D. O. Conover. 1988. The relation between spawning season and the re- cruitment of young-of-the-year bluefish, Poma(omus salta- trix, to New York. Fish. Bull. 66:237-250. Pietrafesa, L. J., G. S. Janowitz, and P. A. Wittman. 1985. Physical oceanographic processes in the Carolina Capes. In Oceanography of the southeastern U. S. conti- nental shelf (L. P. Atkinson, D. W. Menzel, and K. A. Bush, eds.), p. 23-32. Am. Geophys. Union, Washington, DC. Pietrafesa, L. J., J. M. Morrison, M. P. McCann, J. Churchill, E. Bohm, and R. W. Houghton. 1994. Water mass linkages between the Middle and South Atlantic bights. Deep-Sea Res. 41:365-389. Powell, A. B. 1993. A comparison of early-life-history traits in Atlantic menhaden Brevoortia tyrannus and gulf menhaden B. patronus. Fish. Bull. 91:119-128. Powell, A. B., and G. Phonlor. 1986. Early life history of Atlantic menhaden, Brevoortia tyrannus, and gulf menhaden, B. patronus. Fish. Bull. 84:991-995. Quinlan, J. A., B. O. Blanton, T J. Miller, and F E. Werner. 1999. From spawning grounds to the estuary: using linked individual-based and hydrodynamic models to interpret patterns and processes in the oceanic phase of Atlantic menhaden Brevoortia tyrannus life history. Fish. Ocean- ogr. 8 (suppl. 2):224-246. Reintjes, J. W. 1969. Synopsis of biological data on the Atlantic menhaden, Brevoortia tyrannus. U.S. Fish Wildl. Serv. Cir. 320, p. Dep. Interior, Bur. Comm. Fish. Reintjes, J. W., and A. L. Pacheco. 1966. The relation of menhaden to estuaries. /« A sympos- ium on estuarine fisheries (R. F. Smith, A. H. Swartz, and W. H. Massmann, eds. ), p. 50-58. Am. Fish. Soc. Spec. Publ. 3. Rice, J. A., J. A. Quinlan, S. W. Nixon, W. F Hettler, S. M. Warien, and P. M. Stegmann. 1999. Spawning and transport dynamics of Atlantic menha- den: Inferences from characteristics of immigi-ating larvae and Predictions of a hydrodynamic model. Fish. Ocean- ogr. 8 (suppl. 2):93-110. Schmidt, J. 1922. The breeding places of the eel. Phil. Trans. Royal Soc. London. B211:179-211. Stegmann, P. M., J. A. Quinlan, F E. Werner, B. O. Blanton, and P. Berrien. 1999. Atlantic menhaden recruitment to a southern estu- ary: Defining potential spawning regions. Fish. Oceanogr. 8 (suppl. 21:111-123. Szedlmayer, S. T. K. W. Able and R. A. Roundtree. 1992. Growth and temperature-induced mortality of young- of-the-year summer flounder IParalichthys dentatus) in southern New Jersey. Copeia 1:120-128. Thorrold, S. R., C. Latkoczy, P. K. Swart, and C. M. Jones. 2001. Natal homing in a marine fish metapopulation. Sci- ence 291:297-299. Warien, S. M. 1992. Age, growth, and size distribution of larval Atlantic menhaden off North Carolina. Trans. Am. Fish. Soc. 121: 588-598. 1994. Spawning time and recruitment dynamics of larval Atlantic menhaden, Brevoortia tyrannus, into a North Caro- lina estuary Fish. Bull. 92:420^33. Warlen et al.: Recruitment of larval Brevooitia tyrannus to North Carolina and New Jersey estuanes 623 Warlen. S M., and J. S. Burke. 1990. Immigration of larvae of fall/winter spawninj; marine fishes into a North Carolina estuary. Kstuaries 13:453- 461. Werner, F. E., B. O. Blanton, J. A. Quinlan, and R. A. Luettich Jr. 1999. Physical oceanography of the North Carolina conti- nental shelf during the fall and winter seasons: implica- tions to the transport of lar\al menhaden. Fish. Oceanogr. 8(suppl.2):7-21. Witting, D. A., K. W. Able, and M. P. Fahay 1999. Larval fishes of a Middle Atlantic Bight estuary: assemblage structure and temporal stability. Can. J. Fish. Aquat. Sci. .56:222-230. Zweifel, J. R., and R. Lasker. 1976. Prehatch and posthatch growth of fishes — a general model. Fish. Bull. 74:609-621. 624 Abstract— Loggerhead sea turtles (Ca- retta caretta ) are migratory, long-lived, and slow maturing. They are difficult to study because they are seen rarely and their habitats range over vast stretches of the ocean. Movements of immature turtles between pelagic and coastal de- velopmental habitats are particularly difficult to investigate because of in- adequate tagging technologies and the difficulty in capturing significant num- bers of turtles at sea. However, genetic markers found in mitochondrial DNA (mtDNA) provide a basis for predict- ing the origin of juvenile turtles in developmental habitats. Mixed stock analysis was used to determine which nesting populations were contributing individuals to a foraging aggregation of immature loggerhead turtles (mean 63.3 cm straight carapace length 1SCL| ) captured in coastal waters off Hutchin- son Island, Florida. The results indi- cated that at least three different western Atlantic loggerhead sea turtle subpopulations contribute to this group: south Flonda leg's i, Mexico (20'7f), and northeast Florida-North Carolina (10%). The conservation and manage- ment of these immature sea turtles is complicated by their multinational genetic demographics. Origin of immature loggerhead sea turtles {Caretta caretta) at Hutchinson Island, Florida: evidence from mtDNA markers Wayne N. Witzell National Marine Fisheries Service, NOAA 75 Virginia Beach Drive Miami, Florida 33149 E-mail address maynewitzelhg'noaagov Anna L. Bass Department of Fisheries and Aquatic Sciences 7922 NW 71 St Street Gainesville, Flonda 32653 Michael J. Bresette David A. Singewald Jonathan C. Gorham Quantum Resources, Inc. 6451 South Ocean Dnve Jensen Beach, Flonda 34957 Manuscript accepted 23 January 2002. Fish. Bull. 100:624-631 (2002). " North Atlantic loggerhead sea turtles have extended and complex devel- opmental life histories (Musick and Limpus, 1997). After emerging from their nests, hatchling loggerhead sea turtles enter the surf and eventually move into the pelagic environment for several years before returning to inshore benthic coastal waters. The accepted hypothesis is that these hatchlings are passively transported to the eastern Atlantic by major cur- rent systems and these turtles would eventually return to coastal benthic habitats in the western Atlantic by the North Atlantic gyre when they attain 25-60 cm or an estimated 3-10 years old (Carr, 1986, 1987; Musick and Limpus, 1997). The loggerhead sea turtle (Caretta caretta) is listed as threatened under the United States Endangered Species Act of 1973 and subsequent amend- ments. Although the loggerhead sea turtle nesting population in the south- eastern United States is one of the largest in the world, other distinct nest- ing populations (as defined by genetic divergence) exist in the northwestern Atlantic Ocean. These known sub- populations are found in the Yucatan. northwest Florida, south Florida, and from north Florida to North Carolina. The east central coast of Florida sup- ports the largest nesting subpopulation of loggerhead turtles and is a highly dynamic coastal area with multiple sea turtle species in various postpelagic developmental stages (Witzell. 1987). Sea turtles may hatch in one country, grow through adolescence in a second or more countries, feed and reproduce as adults in a third jurisdiction, and swim through a dozen more territorial waters enroute to and from these des- tinations (Bowen et al., 1995). Tagging studies, unfortunately, are only capable of providing glimpses of these complex changes in developmental habitats because of high tag loss rates and rare opportunities of recapturing tagged turtles thousands of kilometers away in the pelagic environment several years later (Chaloupka and Musick. 1997). The likelihood that turtles from ge- netically distinct stocks share coastal and pelagic developmental habitats may raise doubts regarding the effectiveness of conservation strategies based on geo- graphical or political boundaries (Carr and Stancyk, 1975; Bowen and Witzell, 1996). Consequently, sea turtle biolo- Witzell et al : Origin of Caretta caretta at Hutchinson Island 625 Atlantic Ocean UNITED STATES Figure 1 Locations for which are known loggerhead sea turtle nesting haplotypes and the location of the Hutchinson Island foraging population. gists and marine resource managers are presented with complex challenges that reinforce the need for complete life history information. In particular, the origin of immature loggerhead sea turtles foraging in coastal nearshore waters needs to be determined for the development of effective regional conservation and management strategies. Recent research has demonstrated that most sea turtle nesting colonies are genetically distinct as indicated by mitochon- drial (mt) DNA haplotype frequency shifts. This finding allows the possibility of using mtDNA data to identify rook- ery cohorts on feeding grounds (Bass et al., 1998; Broderick et al., 1994). By using an existing database (Encalada et al., 1998) and molecular techniques, tissue samples from juve- nile marine turtles can be analyzed to estimate the origin of animals inhabiting developmental habitats. These data are collected and analyzed faster than results from tagging studies and may provide information on cryptic migratory behavior (Bowen et al, 1995; Bolten et al.. 1998). Both pe- lagic and coastal benthic zones are believed to be essential developmental habitats for sea turtles, and molecular markers have recently been used to document shifts in the demographic composition between these habitats (Laurent et al., 1998). This article examines the mtDNA composition of juvenile loggerhead sea turtles using the coastal waters off Hutchinson Island, Florida, to determine whether the turtles are primarily from the adjacent nesting subpopu- lation or whether this foraging population is composed of individuals from multiple rookeries. Materials and methods Sea turtles are routinely captured in the canal that sup- plies cooling water at the St. Lucie Power Plant on Hutch- inson Island, Florida (Fig. 1). The power plant intake oper- ates year round — collecting sea turtles from 365 m offshore and providing an excellent opportunity to sample sea turtles in the nearshore developmental habitat (Bass and Witzell, 2000). Power plant biologists collect these turtles daily in an ongoing research and conservation program. Turtles are measured, flipper-tagged, and released in good condition promptly into the adjacent coastal waters. A total of 109 juvenile loggerhead sea turtles were sam- pled at the St. Lucie power plant from January through May 1999 for our study. The minimum straight carapace length (SCL) was measured with calipers. Tissue samples were collected by using a 6-mm biopsy punch and placed in 15 mL of saturated salt preservation buffer developed 626 Fishery Bulletin 100(3) Table 1 Haplotype distribution among loggerhead sea turtle nesting groups compiled by Encalada et al. (1998). Haplotypes K and M were identified previously in the Madeira and Azores foraging assemblages (Bolten et al., 1998). Haplotype N was identified previously in a sample of stranded individuals from the Atlantic U.S. coast (Rankin-Baransky et al., 2001 ). Haplotype NWFL A B C D E F G H 1 J K M N 34 4 2 SFL NEFL-NC 22 24 2 1 1 104 1 Mexico 11 2 Greece Brazil St. Lucie, Florida 19 11 40 45 7 11 by Amos and Hoelzel ( 1991). Samples were transferred to the University of Florida for analysis. Standard phenol and chloroform DNA isolation protocols were conducted on the tissue samples (Hillis et al, 1996). A 380-bp frag- ment of the mitochondrial DNA control region was ampli- fied using primers TCR5 (5'-TTGTACATCTACTTAATTACCAC-3'), and HDCM2 (5-GCAAGTAAAACTACCGTATGCCAGGTTA-3') designed for sea turtles (Encalada et al, 1996; Norman et al., 1994). Cycling parameters were as follows: 94°C 1 min, 25 cycles of (94°C/45s -t- 52°C/30s + 72°C/4.5s) and 3-min extension at 72°C. Individuals were compared to known loggerhead sea turtle haplotypes and assigned a letter designation based on Encalada et al. (1998) and Bolten et al. (1998). To test for statistical differences among haplotype frequencies at rookeries and the foraging location, chi-square analyses were performed with the program CHIRXC (Zaykin and Pu- dovkin, 1993) and probabilities were generated with a Monte Carlo randomization procedure (Roff and Bentzen, 1989). Maximum likelihood (ML) analysis for mixed stock identification (Grant et al., 1980) was used to estimate the contributions of nesting populations to the foraging habitat adjacent to the St. Lucie Power Plant on Hutchin- son Island. This method estimates the most likely con- tributions of source populations based on the haplotype frequencies in the source populations and in the mixed population. The source populations and frequencies of associated haplotypes used in the analysis were those of Encalada et al. (1998). It should be noted that the addition of new nesting population data could change the results presented in our study. The addition of new nesting data will always be a problem when conducting these types of analyses but should not preclude the use of this technique to determine potential contributors to a foraging popula- tion. The maximum likelihood program GIRLSEM was used (Masuda et al., 1991). As a starting point in ML iterations with GIRLSEM, it was assumed that all source populations had an equal probability of contributing (i.e. population size, distance from the foraging location, etc. did not have an impact on the percentage of animals re- cruiting to a particular area). Standard errors and 95% confidence intervals of the point estimates were generated from 100 bootstraps of the stock and mixture data sets with GIRLSEM (Pella et al., 1998). Results Mitochondrial DNA analysis Of the 109 tissue samples collected, 106 produced read- able sequences (Table 1). Eighty five of the samples con- sisted of the most common haplotypes A and B. Haplotype C, another haplotype found in multiple rookeries at low frequency, was found in seven individuals. One indi- vidual was characterized by haplotype G, known only in the northwest Florida (NWFL) and south Florida (SFL) source populations. Two haplotypes "endemic" to Mexico (Yucatan), I and J, were found in six of the individuals sampled. The remaining seven individuals possessed hap- lotypes (K, M, and N) observed previously, but only from surveys of foraging or stranded individuals. The haplotype frequencies of the foraging juveniles were significantly different from all nesting loggerhead sea turtle populations except for the SFL nesting popula- tion (x^=9-51,P=0. 40). Although the haplotype frequencies were similar to that of the SFL nesting population, the presence of haplotypes not associated with the SFL popu- Witzell et al.: Origin of Caretta carctta at Hutchinson Island 627 Table 2 M;ixnmini likelihood estimates of contribution by source populations to immature loggerh( ^ad sea turtles from the St. Lucie Power Plant («=99). Estimates were senerated by usinj; IICON software (M: isudaetal., 1991) St. md ird errors and 957« confidence inter- vals were generated from 100 bootstraps of both the stock and mi xture using GIRLSEM. NWFL = northwest Florida; SFL = south | Florida; and NEFL-NC = northeast Florida to North Carolina. Source population Contribution Standard error 95% Confidence interval NWFL 0.0912 0.1896 0.0479-0.1345 SFL 0.6681 0.3072 0.5979-0.7383 NEFL-NC 0.0366 0.1363 0.0055-0.0677 Yucatan 0.2040 0.1125 0.1783-0.2297 Brazil 0.0000 00000 — Greece 0.0000 0.1555 0.0000-0.0355 Table 3 Maximum likelihood estimates of contribution by source populations to immature loggerhead turtles from the St. Lucie power plant (n=99). Estimates were generated by using UCON software (Masuda et a!., 1991 ). Standard errors and 95% confidence inter- vals were generated from 100 bootstraps of both the stock and mixture by using GIRLSEM. The source populations NWFL, Brazil and Greece, were removed from the analysis. SFL = south Flordia; and NEFL-NC = northeast Florida to North Carolina. Source population Contribution Standard error 95% Confidence interval SFL NEFL-NC Yucatan 0.6965 0.0984 0.2050 0.2211 0.1151 0.1296 0.6527-0.7403 0.07.56-0.1212 0.1793-0.2307 lation indicates that this nesting population is not the sole contributor to this foraging population. Maximum likelihood analysis The initial maximum likelihood analysis yielded esti- mates of contribution that were associated with high standard errors (Table 2). The estimate for NWFL had an extremely high standard error This high standard error is an indication of sampling error associated with the source population (Epifanio et al., 1995). Because of the large standard errors, an attempt was made to reduce the com- plexity of solutions that were involved in the maximum likelihood algorithm's search. Removing a potential source population decreases the potential number of "answers," consequently reducing the standard errors about the mean estimates. The maximum likelihood analysis was repeated, remov- ing the smallest and most distant source populations (>AVFL, Brazil, and Greece) as potential contributors (Table 3). In addition to the statistical assumption that the large variation associated with several estimates is partially due to sampling error, we also made some bio- logical assumptions. By removing the source populations we assumed that they were not contributing to this forag- ing area at levels sufficient to detect in this ML analysis. The nesting effort in NWFL may include 100-200 turtles annually (Meylan et al., 1995), as compared to tens of thousands of turtles that nest in southern Florida. Hence, the Florida panhandle is an important nesting area, but is probably too small to detect with precision in ML analyses. The removal of Greece as a potential source population was based on the results of the first analysis that indi- cated that this source population did not contribute at de- tectable levels (Table 1). Brazil was also removed from the analysis because there were no indications (either from the observed haplotypes or from the initial ML analysis) that this population contributed individuals to the forag- ing population at Hutchinson Island. The results from the final ML analysis provided the most realistic estimate of the genetic composition of this area (Table 3). This analysis indicated that the SFL subpopulations is a major contributor to the Hutchinson Island foraging population. In addition, the Yucatan nest- ing population also appears to contribute a significant percentage of individuals to this area. Although the small northeast Florida to North Carolina (NEFL-NC) rookery appears to contribute some individuals, the standard er- ror is still large; therefore, the estimate is not precise. Individuals from the NEFL-NC nesting population are present, but an accurate estimate (finer than 7-12%) is not possible at this time. The results of the analysis of the St. Lucie foraging pop- ulation are provisional and there are several limitations to the maximum likelihood analysis. A major assumption of this type of analysis is that all potential source populations 628 Fishery Bulletin 100(3) Mean = 63 0 SD = 60 n - 109 iJjjLi u 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 Straight carapace length (cm) Figure 2 Length-frequency distribution for immature loggerhead sea turtles captured at the Hutchinson Island foraging ground (1999). have been identified (Pella and Milner, 1987). Although the source population data set represents most of the major nesting colonies in the Atlantic, other nesting colo- nies exist that may contribute individuals to this foraging ground assemblage. For example, there are small nesting loggerhead populations on the Cape Verde Islands, Cuba, and on the coast of Colombia for which data are not acces- sible and we have been unable to estimate their possible contributions. If these nesting populations were included in the analysis there is a chance that the relative contribu- tions of the source populations included in our study could be altered. A second major assumption is that the source populations were sampled sufficiently to uncover all hap- lotypes (Pella and Milner, 1987). In the case of the south Florida nesting assemblage, this assumption may not hold. As previously mentioned, south Florida supports the largest nesting assemblage in the Atlantic Ocean and it is possible that all haplotypes of the assemblage have not been determined. As with the inclusion of more potential source populations, there is the chance that changes in the frequency and distribution of haplotypes could alter the relative contributions reported in our study. We see these results as a sound starting point for the eventual analysis of foraging loggerhead sea turtles along the coast of the eastern United States. The mean length of the 109 sampled turtles was 63.3 cm (SCL) with a standard deviation of 6.0 cm (Fig. 2). The mean lengths for all haplotype groups were very similar: for haplotypes A, B, C, and G (south Florida), mean length was 62.9 cm (±6.2); for haplotypes I and J (Yucatan), it was 63.8 cm (±8.0); and for haplotypes K, M, and N (other), it was 63.7 cm (±3.8). Discussion Although the estimates of percent contributions of the source populations generated here are provisional (i.e. more sampling of loggerhead nesting subpopulations may eventually uncover previously unrecorded haplotypes), they indicate that sea turtles foraging in the nearshore waters of Hutchinson Island originate from at least three rookeries in the Northwest Atlantic: SFL (69'7f ), Yucatan (20%), and NEFL-NC ( 10%). Although the majority of ani- mals appear to originate from Florida nesting beaches, the maximum likelihood analysis indicates that a substantial proportion of the foraging juveniles are coming from the southern nesting population in the Yucatan. In addition, the northern population, comprising animals nesting on beaches from northeast Florida to North Carolina, also contributes a smaller percentage of individuals. Sears et al. (1995) reported on the demographic com- position of juvenile loggerhead sea turtles off Charleston, South Carolina, and Norrgard and Graves ( 1995) analyzed juvenile loggerhead sea turtles from Chesapeake Bay, Vir- ginia. Restriction fragment length polymorphism (RFLP) analysis of mtDNA was used to identify the genotype of individuals in both studies. The data sets were used in con- junction with a geographic survey limited to nesting loca- tions in the North Atlantic Ocean and Mediterranean Sea by Bowen et al. ( 1993) to estimate the composition of these two areas. The estimates of contribution by nesting popula- tions to these foraging areas (Table 4) appear to differ from the results presented in the present study, particularly the large contributions from the NEFL-NC haplotype. This may either reflect real differences in haplotype composition. Witzell et al.. Origin of Caietta catena at Hutchinson Island 629 Table 4 Doiiiograpliic compositions of coastal juvenili' loggerhead sea turtles from U.S. coastal foraging habitats by using RFLP analysis, n = number of turtles in sample. SFL = south Florida; NEFL-NC = northeast Florida to North Carolina. Location SFL NEFL-NC Literature source Chesapeake Bay South Carolina 0.64 0.50 0.36 0.50 62 33 Norrgard and Graves ( 1995) Scars ctal. (1995) inadequate sample sizes or sampling errors, or possibly changes between year classes. However, the differences are undoubtedly an artifact and are more likely due to the poor resolution and small sample sizes of the RFLP studies. Comparisons of estimates generated by using different methods are problematic for several reasons. Recently, Encalada et al. (1998) published the results of a survey of nesting locations using sequences from the control region of the mtDNA. This region exhibits a sixfold increase in di- vergence rates when compared to estimates of divergence based on RFLP analysis. More variation was detected and consequently more haplotypes were identified than previously in Bowen et al. (1993). Encalada et al. (1998) also included several other nesting subpopulations in the Atlantic Ocean, in particular nesting subpopulations in northwest Florida and the Yucatan of Mexico. The manner and routes of dispersal of posthatchling loggerhead sea turtles from the Yucatan and NWFL are undoubtedly complex and probably follow the same disper- sal scenarios as proposed for the Kemp's ridley sea turtle (Lepidochelys kempii) by Collard and Ogren ( 1990). These loggerhead turtles either stay in the Gulf of Mexico or are transported into the Atlantic by the Florida Current. Once in the Atlantic, they either stay on the Continental shelf area or are entrained in major current systems and are transported to the eastern Atlantic Ocean for an un- determined amount of time before presumably returning to western Atlantic coastal benthic habitats. The SFL and NEFL-NC rookeries are all situated near the western boundary of the Gulf Stream where the hatchlings are quickly transported away from beaches. Juvenile logger- head sea turtles leave their pelagic habitat over a range of sizes, and presumably ages, depending on feeding success. This estimate size range (SCL) was from 25 to 60 cm (Carr, 1986, 1987; Martin et al, 1989; Musick and Limpus, 1997). Hays and Marsh (1997) deduced from drift studies that the pelagic phase ended "at least for some loggerheads in the north Atlantic" when they were around 50 cm. These postpelagic turtles are thought to become coastal benthic crustacean foragers (Musick and Limpus, 1997) and to move up and down the U.S. East Coast seasonally. Movements of juvenile loggerhead sea turtles tagged at the St. Lucie Power Plant indicate that a substantial portion of these turtles appear to reside in the immediate vicinity of Hutchinson Island (M. Bresette, unpubl. data). However, a few turtles have been documented from as far north as North Carolina and as far south as Florida Bay in southwest Florida. Most recaptured juvenile logger- head sea turtles were from the north and may indicate that Hutchinson Island is near the southern boundary of the seasonal migration (Ernest et. al., 1989). Relatively few southern movements were recorded to Florida Bay. Over 47% of 3142 turtles under 85 cm captured at the St. Lucie Power Plant were captured from January through April 1976-2000 (Fig. 3). Captures then tapered off until December, consistent with seasonal north and south move- ments as suggested for immature turtles at nearby Cape Canaveral (Henwood, 1987). There has previously been no tagging evidence to indicate movements of these immature turtles outside the continental United States, yet these mtDNA data indicate a large contribution from Mexico. The Hutchinson Island juvenile loggerhead sea turtles settle into this important coastal habitat at the same size, and presumably age, whatever their genetic origins are. This finding indicates that this particular coastal habitat may be critical for several western Atlantic populations of juvenile sea turtles, both loggerhead and green. Management implications The NEFL-NC population is of particular interest to some U.S. conservation biologists because of an apparent decline in nesting females prior to 1990 (Turtle Expert Working Group, 2000). The goal of protecting these turtles while in various developmental habitats is extremely complex, par- ticularly when they intermingle with large numbers of sea turtles from a healthy (large) breeding population (SFL), as well as with small (NWFL) and foreign populations (Yucatan). Management measures affecting immature loggerhead sea turtles in the east-central Florida forag- ing habitats need to take into account the contributions from other source populations that may not be as robust as that of the south Florida population. Specific measures will depend largely on the demographic composition of stranded turtles with known sources of mortality. The genetic compositions of juvenile loggerhead sea turtles that are impacted by disease, boat collisions, anthropo- genic debris, channel dredging, and recreational and com- mercial fisheries need to be determined before resource managers can take appropriate action. Unfortunately it is difficult to determine specific causes of mortality from stranded individuals, and researchers are unable to accu- rately determine the activities that affect specific nesting aggregations. The conservative approach would be to regu- late all known sources of mortality to immature coastal loggerhead sea turtles until more data become available through aggressive genetic sampling and necropsies of stranded animals. 630 Fishery Bulletin 100(3) Figure 3 Monthly captures of loggerhead sea turtles less than 85 cm SCL at the St. Lucie Power Plant (1977-98). The coastal foraging habitat near Hutchinson Island is also an important developmental habitat for endangered im- mature green sea turtles, Chelonia mydas (Bass and Witzell, 2000 ). This is a highly cosmopolitan species with contributions to the Florida foraging grounds from Costa Rica (53%), the United States and Mexico (A2%). and Venezuela and Surinam (4%). It may be advisable for managers to designate specific areas, such as Hutchinson Island, as critical developmental habitats for immature sea turtles and to restrict public access. The need for regional and internationally coordinated sea turtle research and management becomes obvious as genetic analysis reveals the complicated mixed stock of foraging turtle aggregations. International organizations, such as the International Union for the Conservation of Nature (lUCN) and the Intergovernmental Oceano- graphic Commission (lOCARIBE) need to become actively involved with the appropriate regional agencies of gov- ernment to ensure the protection of sea turtle stocks in regional as well as international waters. Acknowledgments Funding for this work was provided from National Marine Fisheries Service, National Science Foundation, and the Turner Foundation. We thank Shiao-Mei Chow and Alicia Francisco for their valuable laboratory support and Jeff Schmid for graphics assistance. We thank Brian Bowen for providing laboratory support and expertise and for reviewing an early draft of this manuscript. Two anony- mous reviewers also provided valuable comments on the final draft manuscript. Finally, we would like to extend special thanks to all the biologists who made collections from the nesting locations used in this analysis. Literature cited Amos, B., and A. R. Hoelzel. 1991. Long-term preservation of whale skin for DNA analy- sis. Rep. Int. Whal. Comm. (special issue 13):99-103. Bass, A. L., and W. N. Witzell. 2000. Demographic composition of immature green turtles (Chelonia mydas) from the east central Florida coast: evi- dence from mtDNA markers. Herpetologia 56:357-367. Bass, A. L., C. J. Lagueux, and B. W. Bowen. 1998. Origin of green turtles, Chelonia mydas. at "sleeping rocks" off the northeast coast of Nicaragua. Copeia 1998: 1064-1069. 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 analy- sis. Ecol.Appl. 8:1-7. Bowen, B. W., F. A. Abreu-Grobois, G. H. Balazs, N. Kamezaki. C. J. Limpus. and R. J. Perl. 1995. Trans-Pacific migrations of the loggerhead turtle (Caretta caretta) demonstrated with mitochondrial DNA markers. Proc. Natl. Acad. Sci. U.S.A. 92:3731-3734. Bowen, B. W., J. C. Aviso, J. I. Richardson, A. B. Meylan, D. Margaritoulis, S. R. Hopkins-Murphy. 1993. Population structure of loggerhead turtles {Caretta caretta) in the northwestern Atlantic Ocean and Mediter- ranean Sea. Conserv. Biol. 7:834-844. Bowen, B. W., and W. N. Witzell. 1996. Sea turtle conservation genetics. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFSC-396:l-7. Witzell et al.: Origin of Caretta caretta at Hutchinson Island 631 Broderick, D., C. Moritz, J. D. Millor, M. Guinea. R J F'rinco, and C. J. Limpus. 1994. Genetic studios of the havvksbill turtle KrclDiochclys imhricata: evidence for multiple stocks in Australian wa- ters. Pac. Consei-v. Biol. 1:123-131. Carr, A. F. 1986. Rips. FADS, and little logRcrheads. Bioscience 36: 92-100. 1987. New perspectives on the pelagic stage of sea turtle development. Conserv. Biol. 1:103-121. Carr, A., and S. Stancyk. 1975. Obsei-vations on the ecology and survival outlook of the hawksbill turtle. Biol. Conserv. 8:161-172. Chaloupka, M. Y., and J. A. Musick. 1997. Age, growth, and population dynamics. In The bi- ology of sea turtles (P. L. Lutz and J. A. Musick, eds.), p. 233-276. CRC Press, Boca Raton, FL. Collard, S. B. and L. H. Ogren. 1990. Dispersal scenarios for pelagic post-hatchling sea turtles. Bull. Mar. Sci. 47:233-243. Encalada. S. E., K. A. Bjorndal, A. B. Bolten, J. C. Zurita, B. Schroeder. E. Possardt, C. J. Sears, and B. W. Bowen. 1998. Population structure of loggerhead turtle {Caretta caretta) nesting colonies in the Atlantic and Mediterra- nean as inferred from mitochondrial DNA control region sequences. Mar Biol. 130:567-575. Encalada, S. E., P. N. Lahanas, K. A. Bjorndal, A. B. Bolten, M. M. Miyamoto, and B. W. Bowen. 1996. Phylogeography and population structure of the Atlantic and Mediterranean green turtle Chelonia mydas: a mitochondrial DNA control region sequence assessment. Mol. Ecol. 5:473-483. Epifanio, J. M., P. E. Smouse, C. J. Kobak, and B. L. Brown. 1995. Mitochondrial DNA divergence among populations of American shad [Alosa sapidissima ): how much variation is enough for mixed-stock analysis? Can. J. Fish. Aquat. Sci. 52:1688-1702. Ernest, R. G., R. E. Martin, N. Williams-Walls, and J. R. Wilcox. 1989. Population dynamics of sea turtles utilizing shallow coastal waters off Hutchinson Island. Florida. U.S. Dep. Commer, NOAATech. Memo. NMFS-SEFC-232:57-59. Grant, W. S., G. B. Milner, P. Pvrasnowski, and F. M. Utter 1980. Use of biochemical genetic variants for identification of sockeye salmon (Oncorhynchus nerka) stocks in Cook Inlet, Alaska. Can. J. Fish. Aquat. Sci. 37:1236-1247. Hays, G. C, and R. Marsh. 1997. Estimating the age of juvenile loggerhead sea turtles in the north Atlantic. Can. J. Zool. 75:40-46 Henwood, T.A. 1987. Movements and seasonal changes in loggerhead turtle Caretta caretta aggregations in the vicinity of Cape Canaveral, Florida (1978-84). Biol. Conserv. 40:191-202. HiUis, D. M., B. K. Mable, A. Larson, S. K. Davis, and E. A. Zinimer 1996. Nucleic acids TV: Sequencing and cloning. In Molecu- lar systematics (D. M. Hillis, C. Moritz, and B. K. Mable, eds.), 2nd ed., p. 321-381. Sinauer Associates, Sunderland, MA. Laurent, L.. P. Casale, M. N. Bradai, B. J. Godley, G. Gerosa, A. C. Broderick, W. Schroth, B. Schierwater, A. M. Levy, D. Freggi, E. M. Abd El-Mawla, D. A. Hadoud, H. E. Gomati, M. Domingo, M. Hadjichristophosou, L. Kornaraky, F. Demirayak, and C. H. Gautier 1998. Molecular resolution of marine turtle stock composi- tion in fishery bycatch: a case study in the Mediterranean. Mol. Ecol. 7:1529-1542. Martin, R. E., R. G. Ernst, N. Williams-Walls, and J. R. Wilcox. 1989. Size distribution and seasonal abundance of logger- head and green turtles in nearshore waters off Hutchinson Island, Florida. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFC-226:334-335. Masuda, M., S. Nelson, and J. Pella. 1991. User manual for GIRLSEM, GIRLSYM, AND CONSQRT U.S.Dep.Commer.NOAA-NMFS. (Programs and user manual available from US-Canada Salmon Pro- gram, 11.305 Glacier Hwy, Juneau, Alaska. U.S.A.! Meylan, A., B. Schroeder, and A. Mosier 1995. Sea turtle nesting activity in the state of Florida 1979-1992. Fla. Mar Res. Publ. 52, 51 p. Florida Mar Res. Institute, St. Petersburg, FL. Musick, J. A. and C. J. Limpus. 1997. Habitat utilization and migration in juvenile sea turtles. In The biology of sea turtles ( P. L. Lutz and J. A. Musick, eds. ), p. 137-163. CRC Press, Boca Raton, FL. Norman, J. A., C. Moritz, C. J. Limpus. 1994. Mitochondrial DNA control region polymorphisms: genetic markers for ecological studies of marine turtles. Mol. Ecol. 3:363-373. Norrgard, J. W., and J. E. Graves. 1995. Determination of the natal origin of a juvenile logger- head turtle (Caretta caretta ) population in Chesapeake Bay using mitochondrial DNA analysis. U.S. Dep. Commer, NOAATech. Memo. NMFS-SEFSC-396: 129-136. Pella, J., M. Masuda, C. G. Guthrie, C. Kondzela, A. Gharrett, A. Moles, G. Winans. 1998. Stock composition of some sockeye salmon, Oncorhyn- chus ner/fa, catches in southeast Alaska, based on incidence of allozyme variants, freshwater ages, and a brain-tissue parasite. U.S. Dep. Commer NOAA Tech. Rep. NMFS 132, 23 p. Pella, J., and G. Miller 1987. Use of genetic marks in stock composition analysis. In Population genetics and fishery management (P. L. Lutz and J. A. Musick, eds.), p. 247-276. LIniv. Washington Press. Seattle, WA. Rankin-Baransky, K., C. J. Williams. A. L. Bass, B. W. Bowen, and J. R. Spotila. 2001. Origin of loggerhead turtles {Caretta caretta) strand- ings in the northeastern United States as determined by mitochondrial DNA analysis. J. Herpetol. 35:638-646 Roff, D. A., and R Bentzen. 1989. The statistical analysis of mitochondrial DNA poly- morphisms: chi-square and the problem of small samples. Mol. Biol. Evol. 6:.539-.545. Sears, C. J., B. W. Bowen, R. W. Chapman, S. B. Galloway S. R. Hopkins-Murphy, and CM. Woodley. 1995. Demographic composition of the juvenile loggerhead sea turtle {Caretta caretta) feeding off Charleston, South Carolina: evidence from mitochondrial DNA markers. Mar Biol. 123:869-874. 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 NOA Tech. Memo. NMFS-SEFSC-444, 115 p. Witzell.W. N. (ed.). 1987. The ecology of east Florida sea turtles. U.S. Dep. Commer, NOAA Tech. Rep. NMFS 53, 80 p. Zaykin, D. V., and A. I. Pudovkin. 1993. Two programs to estimate significance of x^ values using pseudoprobability tests. J. Hered. 84:152. 632 Spawning of American shad (Alosa sapidissimd) and striped bass (Morone saxatilis) in the Mattaponi and Pamunkey Rivers, Virginia* Donna Marie Bilkovic John E. OIney Carl H. Hershner Virginia Institute of Manne Science College of William and Mary Rt 1208 Create Road Gloucester Point, Virginia 23062 Email (for D M Bilkovid: donnabeivimsedu In the Atlantic coastal region, Ameri- can shad (Alosa sapidissima ) is highly prized for its flesh and roe. Spawning runs have been heavily fished and since the late 1800s, landings have shown steady declines to the extent that Maryland declared a fishing moratorium in 1980, and Virginia fol- lowed in 1994 for Chesapeake Bay and its tributaries (ASMFC, 1999). Shad restoration projects are underway to restock depleted spawning runs, espe- cially in regions where stream impedi- ments have been or are being removed. Coastal ocean intercept gill-net fisher- ies have remained in place despite criticism and speculation about their impact on populations, particularly those river systems stocks that are depleted. The Atlantic States Marine Fisheries Commission Shad Board (ASMFC, 1999) adopted a fishery man- agement plan for American shad and river herring that included a five-year phase-out of the ocean fishery and that required states to develop an approved fishing or recovery plan for each stock under restoration. In Virginia, this requirement applies to the James and York rivers. Although the roe fishery for Ameri- can shad has historically been impor- tant, there is little information about the specific spawning locations of these broadcast spawners. American shad are anadromous fish native to the Atlantic coast of North America, with a range extending from southeastern Labrador to the St. Johns River, Flor- ida. In Chesapeake Bay tributaries, American shad deposit semidemersal eggs in the freshwater portions of the estuaries in the spring, usually begin- ning in March and ending by early June with peaks in April (Klauda et al, 1991). American shad have histori- cally ascended farther upriver than at present, within tributaries where obstructions to movements upstream now exist. Prior to dam building in the 1800s on the James River, large num- bers of American shad traveled over 335 miles from Chesapeake Bay into the Jackson and Cowpasture rivers (Mansueti and Kolb, 1953). The York River, a coastal plain trib- utary located in the Chesapeake Bay watershed, is formed by the conflu- ence of the Pamunkey and Mattaponi Rivers at West Point (Fig 1). The Pamunkey River has a larger water- shed (3768 km'-) and average spring discharge rate (47.5 m%) than the Mattaponi River (2274 km'^; 27.2 m^/s, respectively). Watershed sizes are based on U.S, Geological Survey digital line graph data (DLG) at 1:100,000. On these unobstructed rivers, annual releases of hatchery-reared American shad approximate two to four million fry through efforts of the Virginia Game and Inland Fisheries (VGIF) and an es- timated 2.5 to 3 million fry are released by the Pamunkey tribal government. In addition there are unknown contribu- tions from the Mattaponi tribal govern- ment (Gunther'). Current monitoring of adult catches indicates that the York River supports the strongest runs of shad in Virginia (Olney and Hoenig^). American shad in the York River are used as the source stock for hatchery ef- forts in the James and Potomac rivers. Thus, the restoration efforts in Virginia are dependent on the productivity of the York River. Within the freshwater tidal portions of the Mattaponi and Pamunkey riv- ers, numerous other species spawn, including striped bass (Morone saxati- lis) (McGovern and Olney, 1988; Grant and Olney, 1991). The Chesapeake Bay stock has rebounded after severe declines in the 1970s and early 1980s as a result of successful management and several years of successful re- production (Olney et al, 1991; Field, 1997). The extent of the spawning area for both American shad and striped bass is in part a function of salinity and temperature. Striped bass spawn from the limit of brackish water to freshwater in the rivers of Chesa- peake Bay from early April through the end of May (Setzler-Hamilton et al., 1981), and American shad spawn in freshwater (Leim, 1924). McGov- ern and Olney (1996) noted that the lower limit of striped bass spawning followed the 1 ppt salinity contour, and Secor and Houde (1995) postulated that the freshwater-saltwater inter- face may act as a down-river barrier to striped bass egg and larval advection. Based on suitable temperature ranges (12-24°C for striped bass [Setzler- Hamilton et al, 1980; Rutherford and Houde, 1995] and 12-25°C for Ameri- * Contribution 2449 of the Virginia Insti- tute of Marine Science, College of William and Mary, Gloucester Point, VA '23062. ' Gunther, T. 2000. Personal commun. Virginia Game and Inland Fisheries, Richmond, Virginia 23230. - Olney, J. E., and J. M. Hoenig, 2000. Monitoring relative abundance of Ameri- can shad in Virginia's Rivers. March 1998-1999. Final report to the U.S. Fish and Wildlife Service and Virginia Marine Resources Commission, 53 p. Contract number F-116-R-1. Virginia Institute of Marine Science, Gloucester Point, Virginia. Manuscript accepted 13 February 2002. Fish. Bull. 100:632-640 (2002). NOTE Bllkovic et al.: Spawning of Alosa sapidissima and Morone saxatilis 633 kilometers 0 40 Figure 1 Extent of ichthyoplankton sampling by bongo net and pushnet in the Mattaponi and Pamunkey rivers (1997-99). Stations are denoted as the number of kilometers from the mouth of the York River. can shad [Leach and Houde, 1999; Walburg and Nichols^] and salinity requirements for the early life stages of these species, the potential for spawning overlap spatially and temporally is high. Species interactions, including preda- tion and competition by both adults and young, may play a role in the spawning and recruitment success of these spe- cies. Similar interactions have been postulated between American shad and other alosines in the Hudson River (Schmidt etal, 1988). Our objectives were to describe the American shad spawning reaches in the Mattaponi and Pamunkey Rivers spatiotemporally, and to determine if striped bass also spawn within the identified spawning habitat of American shad. In year one, we completed an exploratory survey to map the distribution of the American shad spawning ground and the occurrence of striped bass within these reaches. In years two and three, sampling was modified to locate the upper limit of American shad and striped bass spawning within the two rivers. ^ Walburg, C. H., and P R. Nichols. 1967. Biology and man- agement of the American shad and the status of the fisheries, Atlantic coast of the United States, 1960. U.S. Fish and Wild- life Service, Special Scientific Report-Fish .550, 105 p. Fish and Wildlife Service, U.S. Dep. of Interior, Washington, DC 20005. Materials and methods Sampling protocol in 1997 Exploratory sampling in the Mattaponi and Pamunkey rivers for eggs and lai^ae of American shad and striped bass extended from March through April 1997. Sites were chosen on the basis of a prior survey of American shad eggs in the rivers (Massmann, 1952). Sampling protocol included weekly ichthyoplankton collections during day- light hours with stepped oblique tows of a bongo frame fitted with two 333-pm mesh nets (60-cm diameter). Catches from both nets were combined. The same ten stations were sampled weekly on each river within the tidal freshwater reaches. Stations are depicted as river kilometers (rkm) from the mouth of the York River, for example, M68 is a station on the Mattaponi River that is approximately 68 river kilometers from the mouth of the York River The stations were located at approximately 3.2-rkm intervals within the range of 72 to 106 rkm (P72 to P106) on the Pamunkey River and 68 to 102 rkm (M68 to M102) on the Mattaponi River (Fig. 1). Sampling protocol in 1998 and 1999 In 1998 and 1999, station locations were extended upriver to include more shallow stations owing to the low abun- 634 Fishery Bulletin 100(3) dance of American shad eggs in 1997. Bongo nets could not be used, and sampling included pushnet surveys in the upper reaches of the rivers (from 31 March through 20 May 1998 and from 11 April through 7 May 1999). The weekly sampling on each river consisted of pushnet tows at approximately one meter below the surface at each sta- tion. A pushnet frame fitted to the bow of a 14-foot boat (Olney and Boehlert, 1988) accommodated two plankton nets (333 pm, 60 cm). Catches from both nets were com- bined. In 1998, eight stations per river were systemati- cally sampled bracketing M94 to M120 and P109 to P131. In 1999, two stations at M124 and M128 were added on the Mattaponi River; we added six upriver stations (P135-P154) and one downriver station (P104) on the Pamunkey River (each spaced at 3.2-rkm intervals, Fig. 1). Bongo and push nets were fitted with a flow meter for volumetric measurements and tow times were adjusted (three to seven minutes) to meet a lower limit of 50 m^ of water filtered through both nets combined. Laboratory procedures and data analysis Ichthyoplankton samples were preserved in 10'^ phos- phate-buffered formalin. Ichthyoplankton were sorted and larval fish and eggs were identified (Lippson and Moran, 1974; Jones et al.^), enumerated, and removed from the original, whole sample. Densities were reported as number per 100 m'K Relative abundance in both rivers was calculated by average density of each life stage (egg, yolksac larva, and postyolksac) multiplied by total volume of spawning or nursery area sampled. Total volume of spawning or nursery area sampled was determined sepa- rately for each species by including locations within the sampling region where eggs (spawning reaches) or larvae (nursery reaches) were collected. River volumes were calculated by using bathymetric surveys and correspond- ing areal estimates from a digitized record of the mean high-water shoreline position as shown on the 7.5 minute topographic map series of the U.S. Geological Survey completed by Comprehensive Coastal Inventory, Virginia Institute of Marine Science (Bilkovic, 2000). For purposes of comparison, we used data on the abundance of Ameri- can shad and striped bass juveniles in the Pamunkey and Mattaponi rivers. The data were taken from annual surveys of juvenile abundance conducted by the Virginia Institute of Marine Science (Austin et al., 2000; Olney and Hoenig^). in Figures 2 and 3. Average density (total eggs or larvae per total volume filtered) of each species per river is depicted in Figure 4. On the Mattaponi River (1997-99), American shad eggs were collected over a 44-km reach (M81-M124) and the highest densities occurred between M96 and M124. Striped bass spawning occurred over a 27- km reach and the highest densities in the sampled area occurred between M68 and M87, downstream of the pri- mary spawning reaches of American shad (Figs. 2-4). On the Pamunkey River, American shad eggs were collected over a 53-km reach (P98-P150), and the highest densities were found between P104 and P131. Striped bass spawn- ing occurred over a 60-km reach (P72-P131), and the highest densities were found between P72 and P87. There was some spatial overlap in spawning of these species, but the primary spawning reaches were separate. Temporal overlap in spawning of American shad and striped bass occurred throughout the sampled period in both rivers (Figs. 2-3). On the Mattaponi River, American shad larvae (total length, 6.1-19.2 mm) were collected from M68 to M124, and the highest densities were observed between M94 and M102 — a reach that is downstream of the spawning habitat. On the Pamunkey River, American shad larvae (total length, 6.6-12.2 mm) were collected between P76 and P128. Densities were highest at P102, 105, and 124. Larval striped bass were collected from M68 to M94 and from P72 to P109, and peak catches (>l/m') were collected from M68 to M80 and from P72 to P91. In both rivers, we observed overlap in American shad nursery grounds and striped bass spawning reaches. However, the highest densities of larval striped bass were downstream of the primary shad spawning and nursery areas (Fig. 4). Average density of individual life stages of American shad was higher in the Mattaponi River than in the Pamunkey River; the opposite pattern was apparent for striped bass (Table 1). Estimates of the relative numbers of American shad and striped bass (average density x river volume) suggested that abundance of American shad eggs and larvae was higher on the Mattaponi River than on the Pamunkey River by a factor of 5.5 and 4.4, respectively. Relative abundance of striped bass eggs and larvae was higher on the Pamunkey River than on the Mattaponi River by a factor of 29 and 9.9, respectively (Table 2). Discussion Results Trends in density (numbers/100 m'') of eggs for each river and species by date and station are depicted for 1997-99 " Jones, P. W., F. D. Martin, and J. D. Hardy Jr 1978. Devel- opment of fishes of the mid-Atlantic Bight. An atlas of egg, larval, and juvenile stages. Vol. 1, Acipenseridae through Ictaluridae. U.S. Fish and Wildlife Service report FSW/OBS- 78/12. Fish and Wildlife Service, U.S. Dep. Commer., Washmg- ton DC 20005. Over the three years of surveys, eggs and larvae of American shad were rare compared to those of striped bass (Table 1). Despite our successive efforts to relocate sampling stations upstream of known striped bass spawn- ing habitat (Grant and Olney, 1991; Olney et al., 1991), striped bass eggs and larvae were more abundant (-114 times and -38 times, respectively) than those of American shad (Table 2). These differences could be attributed to the relative sizes and egg production of the spawning stocks because the number of mature American shad presently in the York River system is believed to be low in relation to historic run sizes (Nichols and Massmann, 1963; Olney NOTE Bilkovic et al Spawning of Alosa sapidissima and Moione saxatilis 635 Mattaponi 1997: American Shad Eggs 30- Apr 16 Apr ■Apr 29-Mar Mattaponi 1 998; American Shad Eggs _ May May 31-Mar Mattaponi 1999: American Shad Eggs ^"Vo'9 7o^,98 94 18-Apr 11 -Apr Pamunkey 1997: American Shad Eggs Stations (km) 106 102 '29-Apr ;-K-g-ax^l7-Apr ' ^^'JJ'^I 1 -Apr BT ~-'---C~C>^^04-Apr 80 76^^27-Mar'^ Pamunl Mattaponi River egg densities), it is unlikely that NOTE Bilkovic et al : Spawning of Alosa sapidissima and Morone saxattlts 637 Mattaponi River: American Shad 50 T 40 ■■ 30- 0 LJJXtJlJlJlilJuJLaJJUi^ 128 120 113 105 98 91 83 76 68 ■ egg D'a^ae Q Mattaponi River: Striped Bass 600 500 400 300 200 100 0 —aujJjU 128 120 113 105 98 91 83 76 68 I egg p^ larvae Pamunkey River; American Shad 10 T 8 2 154 143 131 124 113 104 94 83 72 ■ egg Dla^^ae Pamunl Floor, Washington. D.C. 20005. Austin, H. M., A. D. Estes, and D. M. Seaver. 2000. Estimation of juvenile striped bass relative abun- dance in the Virginia Portion of Chesapeake Bay. Annual progress report, 32 p. Virginia Institute of Marine Science, Gloucester Point. ,VA. Bilkovic, D. M. 2000. Assessment of spawning and nursery habitat suit- ability for American shad iAlosa sapidissima). Ph.D. diss., 216 p. Virginia Institute of Marine Science. Glouces- ter Point. VA. 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. Crecco, V. A., and M. M. Blake. 1983. Feeding ecology of coexisting larvae of American shad and blueback herring in the Connecticut River Trans. Am. Fish. Soc. 112:498-507. Field, J. D. 1997. Atlantic striped bass management: Where did we go right? Fisheries 22( 7 ):6-8. Gardinier, M. N., and T. B. Hoff. 1982. Diet ofstriped bass in the Hudson River estuary. N.Y. Fish and Game Journal 29:152-165. Grant, G. C, and J. E. Olney 1991. Distribution of striped bass Morone saxatilis (Wal- baum) eggs and larvae in major Virginia Rivers. Fish. Bull. 89:187-193. Johnson, J. H., and D. S. Dropkin. 1997. Food and prey selection of recently released American shad iAlosa sapidissima) larvae. J. Freshwater Ecol. 12: 355-358. Klauda, R. J., S. A. Fischer, L. W. Hall Jr, and J. A Sullivan. 1991. American shad and hickory shad. In Habitat re- quirements for Chesapeake Bay living resources (S. L. Fun- derburk, S. J. Jordan, J. A. Mihursky, and D. Riley, eds.), p. 9-1-9-27. Chesapeake Research Consortium. Inc., Ann- apolis, MD. Leach, S. D., and E. D. Houde. 1999. Effects of environmental factors on survival, growth, and production of American shad larvae. J. Fish Biol. 54: 767-786. Leim, A. H. 1924. The life history of the shad (Alosa sapidissima (Wilson)) with special reference to factors limiting its abundance. Biol . Board of Canada, Contr to Can. Biol. 2(111:161-284. Lippson, A. J., and R. L. Moran., ed. 1974. Manual for identification of early developmental stages of fishes of the Potomac River estuary, 282 p. Mary- land Department of Natural Resources, Baltimore, MD. Manooch, C. S. 1973. Food habits of yearling and adult striped bass, Morone saxatilis (Walbaum), from Albemarle Sound, North Carolina. Ches. Sci. 14:7.3-86. Mansueti, R. J., and H. Kolb. 1953. A historical review of the shad fisheries of North America. Ches. Biol. Lab. Publ. 97: 1-293. Markle, D. F . and G. C. Grant. 1970. The summer food habits of young-of-the-year striped bass in three Virginia rivers. Ches. Sci. ll:.50-54. Massmann, W. H. 1952. Characteristics of spawning areas of shad, Alosa sapidissima (Wilson) in some Virginia streams. Trans. Am. Fish. Soc. 81:78-93. 1963. Summer food of juvenile American shad in Virginia waters. Ches. Sci. 4:167-171. McGovem, J. C, and J. E. Olney. 1988. Potential predation by fish and invertebrates on early life history stages of striped bass in the Pamunkey River, Virginia. Trans. Am. Fish. Soc. 117:152-161. 1996. Factors affecting survival of early life stages and subsequent recruitment of striped bass on the Pamunkey River, Virginia. Can. J. Fish. Aquat. Sci. 53: 1713-1726. Nichols, P. R., and W. H. Massmann. 1963. Abundance, age and fecundity of shad, York Riven, 1953-1959. Fish. Bull. 63:179-187. Olney J. E., and G. Boehlert. 1988. Nearshore ichthyoplankton associated with seagrass beds in the lower Chespaeake Bay Mar Ecol. Prog. Ser. 45:33-43. Olney, J. E., J. D. Field, and J. C. McGovern. 1991. Striped bass egg mortality, production, and female biomass in Virginia Rivers, 1980-1989. Trans. Am. Fish. Soc. 120:354-367. Rinaldo, R. G. 1971. Analysis of Morone saxatilis and Morone americana spawning and nursery areas in the York and Pamunkey Rivers, Virginia. M.S. thesis, 56 p. College of William and Mary, Gloucester Point, VA. Ross, R. M., R. M. Bennett, and J. H. Johnson. 1997. Habitat use and feeding ecology of riverine juvenile American shad. N. Am. J. Fish. Manage. 17:964-974. Rudershausen, P. J., and J. G. Loesch. 2000. Feeding habits of young-of year striped bass, Morone saxatilis, and white perch, Morone americana, in lower James River, VA. Va. J. Sci. 51:23-37. Rutherford, E. S., and E. D. Houde. 1995. The influence of temperature on cohort-specific growth, survival, and recruitment ofstriped bass, Morone saxatilis, larvae in the Chesapeake Bay Fish. Bull. 93:315-332. Schmidt, R. E., R. J. Klauda, and J. M. Bartels. 1988. Distributions and movements of the early life stages of three species of Alosa in the Hudson River, with com- ments on mechanisms to reduce interspecific competition. In Fisheries research in the Hudson River (C. L. Smith, ed.), p. 141-168. The Hudson River Environmental Soci- ety, State University of New York, Albany, NY. Secor, D. H., and E. D. Houde. 1995. Temperature effects on the timing of striped bass egg production, larval viability, and recruitment potential in the Patuxent River (Chesapeake Bay). Estuaries 18: 527-544. Setzler-Hamilton, E. M., W. R. Boynton, J. A. Mihursky, T. T. Polgar, and K. V. Wood. 1981. Spatial and temporal distribution of striped bass* 640 Fishery Bulletin 100(3) eggs, larvae and juveniles in the Potomac estuary. Trans. Am. Fish. See. 110:121^136. Setzler-Hamilton, E. M., W. R. Boynton, K. V. Wood, H. H. Zion, L. Lubber, N. K. Mountford, P. Frere, L. Tucker, and J. A. Milhursky. 1980. Synopsis of biological data on striped bass, Morone saxatUis (Walbaum). U.S. Dep. Commer, NOAA Tech. Rep. NMFS 433. Tresselt, E.F. 1952. Spawning grounds of the striped bass, Roccus saxati- lis (Walbaum), in Virginia. Bull. Bingham Oceanogr. Coll. 14 98-110. 641 Spawning, growth, and overwintering size of searobins (Triglidae: Prionotus carolinus and P. evolansY Richard S. McBride Marine Field Station Institute of Manne and Coastal Sciences Rutgers University 800 Great Bay Blvd Tuckerton, New Jersey 08087 Present address: Florida Manne Research institute 100 Eigtith Avenue SE St. Petersburg, Florida 33701-5095 E-mail address rictiard mcbrideiff fwc state, fl us The northern searobin, Prionotus caro- linus, and striped searobin. P. evolans, are among the most common benthic fishes of continental shelf waters between Cape Cod and Cape Hatteras, and both species have contributed occasionally to landings of the U.S. middle Atlantic states. Although sev- eral life history studies of these species have been published (Marshall, 1946; Wong, 1968; McEachran and Davis, 1970; Richards et al., 1979; McBride et al., 1998), their early life history has been poorly documented until recently (McBride and Able, 1994; Able and Fahay 1998; McBride et al., 2002). Obstacles to early life history studies of these searobins include the follow- ing: 1) eggs and preflexion larvae are difficult to identify; 2) spawning occurs concurrently for several months; 3) slow growth rates confound analysis of size frequencies for determining cohort structure, and 4) conventional sam- pling methods have provided few late- stage larvae and early juveniles. I set out to avoid these problems by collect- ing juvenile specimens from the field and analyzing their otolith microincre- ments to determine spawning period- icity and growth rates. This approach has not previously been applied to any triglidtSecoretal., 1992). This study demonstrates both inter- specific and intraspecific differences in size during the first year. A literature review also suggests that age-0 P. caro- linus are smaller than age-0 P. evolans and that conspecifics at northern latitudes are larger than conspecif- ics at southern latitudes (Table 1). In this study, daily sagittal micro-incre- ments were validated and used to test whether interspecific spawning date differences, growth rate differences, or both, are responsible for size dif- ferences at first winter. Size at first annulus formation was also examined as an independent measure of age and growth rate. The results of this study improve our understanding of the con- tinental shelf as a nursery ground and the geographic variation in life history traits for these two species. Materials and methods Otolith micro-increment validation Otolith ontogeny was examined by using cultured embryos and yolksac larvae. Prionotus carolinus eggs were collected from ripe adults in August 1992 in coastal waters and fertil- ized in the laboratory. Embryos were raised in 10-liter aquaria under a 12: 12 hour light:dark cycle, at 31 ppt salinity, and 26°C. Cultured P. caroli- nus embryos began to hatch after 48 hours and nearly all fish had hatched by 56 hours. Laboratory temperatures unintentionally dropped to 22°C after hatching, but no mortality was ob- served. Embryos and larvae were pre- served in 95'7r ETOH at 8-16 hour intervals for the first three days and 22-26 hours for the following three days. No fish survived beyond six days. No water was exchanged, but constant aeration kept the aquaria well mixed during the experiment. Newly hatched Artemia sp. nauplii were supplied to yolksac larvae, but no feeding was observed. All mortalities appeared to be caused by starvation. Embryos or larvae were placed on a slide under a cover slip or set in immer- sion oil (Secor et al., 1992) to examine otolith ontogeny. Otolith presence, size, and development were noted with both a binocular scope (<50x) and a compound microscope (40-lOOOx) with polarized light. Notochord length (NL) was mea- sured with an ocular micrometer to the nearest 0.1 mm. Otoliths were mea- sured from a digitized video image through a compound microscope to the nearest 0.001 mm. Field-collected P. coro/inus juveniles were chemically marked with tetra- cycline, held in the laboratory under controlled conditions, and the relation- ship between "days captive" and "rings after the tetracycline mark" was tested against a 1:1 ratio. Newly settled P. carolinus (10-15 mm standard length |SL|) were collected near Beach Ha- ven Ridge (New Jersey) by using a 1- or 2-m beam trawl on five different dates in October 1992 (see McBride et al. [2002] for sampling locations). These fish were divided into four rep- licate groups; each group was marked separately with a solution of oxytet- racycline (dihydrate) and ambient seawater (a concentration of 500 mg/L; 26-28 ppt; 20-22°C) for 24 hours. One 40-liter aquarium per repli- cate group was maintained as a flow- through (about 1-10 mL/s) system. Daily water temperature averaged 20.6°C (±2.7 SD), salinity averaged 28.5 ppt (±1.4) and the photoperiod was 12:12 hours of light:dark. Tetra- cycline-marked fish were fed thawed Artemia sp. 2-4 times daily, and a sup- ■ Contribution 2002-15 of the Institute of Marine and Coastal Sciences, Rutgers University, New Brunswick, NJ 08901. Manuscript accepted 20 February 2002. Fish. Bull. 100:641-647 (2002). 642 Fishery Bulletin 100(3) Table 1 Standard length (mm) at age for Prionotus carolinus and P. evolans from two different geographic regions. Fishes older than age 6 are not included because they were typically represented by only individual fishes, nd = no data. Age (yr) 0 1 2 3 4 5 6 ^ Average lengtli values converted from fork length using equations in Wong (1968; Tables 2 and 3). -' Values are medians converted from fork lengths as reported in McEachran and Davis (1970, p. 347-348). ' Values estimated as a median from Richards et al. (1979; Fig. 7i. Chesapeake New England P. carolinus' P. evolans^ P. carolinus'' P. evolans-' nd nd 75 90 126 132 130 160 163 186 195 210 180 207 220 235 197 224 245 260 204 245 255 290 211 270 270 305 plemental feed with vitamins and oils was given at least once each week. Their ration was approximately 10% of their body weight/day; this ration was adjusted 1-3 times weekly to reflect the changes in weight during the experi- ment. Uneaten food was removed daily. Three to seven fish were removed from each replicate aquarium after 16, 26, 36, 46 days (total n=75), measured to the nearest 0.1 mm SL, and preserved in 95% ETOH. Growth rates between the replicate tanks varied from 0.21 to 0.44 mm/d accord- ing to least squares, linear regression of standard length on date of removal. Eleven of 86 fish initially placed in the aquaria died during the experiment and were not used for validating microincrement deposition rate. After preservation, both sagittae were extracted from each fish, mounted on a coded glass slide in nail polish, sanded along the sagittal plane, and swabbed with im- mersion oil. After a fluorescing increment was located in the ocular crosshairs using a ultraviolet light source, the light was turned off and I counted the microincrements from the crosshair to the otolith margin on three separate dates. An average of these three counts was used to esti- mate increment number from the tetracycline mark to the peripheral edge. Daily age and growth Sagittal microincrements were examined from other postsettlement P. carolinus and P. evolans juveniles col- lected during August-October 1991 at locations in Great Bay (New Jersey), Delaware Bay, and continental shelf sites between Chesapeake Bay and Long Island Sound. Fishes were collected with bottom trawls and seine nets and either preserved in 95% ETOH or frozen. Specimens for aging were subsampled in a stratified (1-mm length category) random manner to proportionally represent the complete size range of each independent sample. Otoliths were mounted on coded slides in nail polish and sanded to the core along one side of the sagittal plane or embedded in an epoxy resin and sectioned to the core (Secor et al., 1992). I counted microincrements through a compound microscope, typically at 400x by using polar- ized light, on three separate occasions. If the range of all three counts was >20%' of the mean count, then the speci- men was excluded from further analyses; this criterium resulted in 14 of 137 specimens being rejected. The mean increment count for sagittae was used to estimate "otolith age" in days. Daily age included four additional days — an estimate of the time between hatching and the date of first ring deposition. Hatching dates were calculated by subtracting daily age from date of capture. Annual age and growth Size attained by the end of the first growing season was estimated from P. carolinus and P. evolans juveniles col- lected during later winter and early spring cruises by the National Oceanic and Atmospheric Administration's National Marine Fisheries Service (see also McBride et al., 1998). Fishes were sampled with bottom trawls during two consecutive cruises (February 1993 and March-April 1993), which included a total of 447 stations (Fig. 1). The effective period of sampling for searobins was actually February-March (only one searobin was caught in 160 tows made during April). At sea, only fishes <120 mm SL were saved because larger fishes were presumed to be older than 1 year (see Richards et al., 1979). Fishes were kept frozen to preserve the otolith structure. Specimens were subsampled from each independent sample (i.e. tow) in a stratified (5-mm length intervals), random manner This subsampling ap- proach was chosen again to moderate the resulting sample size while proportionally representing the complete size range of each independent sample; this approach also exaggerates the size range by dampening the modes of re- sulting size-frequency data. Sagittae were mounted in nail polish on a coded glass slide, sanded, and polished along NOTE McBnde: Spawning, growth, and overwinter size of Pnonotus camlinus and P evolans 643 t' P. camlinus oP. evolans DBoth 76° 74° 72° 70° 68° 66° 77° 76° 75° 74° 73° 72° Figure 1 (A) Locations of all sampling during February-April 1993. Locations are indicated where searobins were and were not collected in tows. (B) Sample locations for age-0 searobins used for estimating overwintering size. Locations are for cruise dates during February- March 1993. No age-0 Pnonotus spp. were collected during April 1993, and sampling areas covered during April (north of 40°N) are not shown. the sagittal plane, and viewed under a binocular scope (typically at 20x) to check for the presence (e.g. age-1) or absence (e.g. age-0) of an annulus. Annulus formation occurs in March for both species (Wong, 1968; McEachran and Davis, 1970), and an annulus on the otolith margin was not counted. Results and discussion Otolith microstructure Two pairs of otoliths (the sagittae and lapilli) were pres- ent prior to hatching in cultured embryos, having formed at the same time as the optic vesicles. No asterisci were observed in cultured yolksac larvae and, from personal observations of field-caught specimens, the asterisci form after flexion of the notochord. First ring deposition in sagittae was contemporaneous with eye pigmentation and yolksac absorption, although this process was not observed directly. Instead, it was observed that the diam- eter of the sagittal core (=primordium) steadily increased during yolksac absorption but that no microincrements were evident. I conclude that the laboratory specimens would have laid down their first ring at yolk absorption had they survived because the diameters of their sagittae were similar to the sagittal core diameters of wild fish (i.e. the diameter of the first microincrement). The maximum sagittal diameter of laboratory-cultured P. carolinus larvae was 26.3 pm and the mean sagittal core diameter of field-caught fish was 27.0 or 26.8 pm (n=8 flexion stage and n=6 postflexion specimens, respectively). Yolksac absorption occurred approximately six days af- ter fertilization and four days after hatching at 22-26°C; therefore four days were added to the average number of microincrements counted for both species, although this developmental pattern was observed only for P. carolinus. The rate of larval development observed in the present study is consistent with other reports of P. carolinus cultured at 15-22°C. Yuschak and Lund (1984) observed yolksac absorption by P. carolinus larvae at 3.4 mm NL and first feeding at 3.5-5 days after hatching. Kuntz and Radcliffe ( 1917) reported P. carolinus hatching at about 60 hours (2.5 days) and starvation 5-6 days after hatching. Sagittal microincrement deposition occurred daily in chemically marked P. carolinus juveniles cultured in the laboratory. Data from four replicate aquaria (75 fish) were pooled because an ANCOVA did not indicate a significant difference by either the interaction of treatments (slope- replicate interaction: P=0.48) or the covariate (replicate: P=0.12). The resulting relationship was Y= 1.23 + 0.94 (X), where Y = the number of microincrements after the tetra- cycline mark; and X = the number of days following marking. A test of statistical power (i.e. 1 - beta, as in Dixon and Massey [1951] and Cohen and Cohen [1975]) indicated a 99.9% confidence that microincrement deposition did not deviate from unity by more than 0.02 rings/d. McBride (1994) presented preliminary evidence of daily sagittal microincrements in P. evolans. 644 Fishery Bulletin 100(3) Reproductive seasonality The reproductive period for both species was prolonged and overlapping (Fig. 2). The range of P. carolinus hatching dates was from 19 May to 5 September and P. evolans hatching dates extended from 2 June to 3 September. Median hatching dates for P. carolinus were significantly later than P. evolans i 13 August versus 26 July; Wilcoxon 2-sample test: P=0.011). Both species spawned for at least three months but P. carolinus hatching dates were dramatically skewed to the left (i.e. most individuals hatched late in the season). Previous studies independently demonstrated that both species spawn contemporaneously in the New York Bight: Keirans et al. (1986) calculated egg densities, Wilk et al. (1990) examined gonadosomatic indices, McBride and Able (1994) used length-frequency analysis, and McBride et al. (2002) examined larval densities. All demonstrated an extended spawning season for both species from about May to September; furthermore, a bimodal pattern of spawning output was reported for Prionotus spp. by Keirans et al. (1986) and McBride and Able (1994). My sampling procedure (i.e. a stratified, random design with respect to length intervals) would flatten frequency peaks and emphasize the range of hatching dates, a method not well suited for identifying multiple spawning peaks. Moreover, McBride et al. (2002) identified notable geo- graphic variation in spawning seasonality, which could not be separated out in the present study and should be accounted for in future research. Intra-annual reproduc- tive periodicity of Prionotus should also be evaluated by examining gonads for cyclic, group-synchronous oocyte development, which could be an underlying process lead- ing to a protracted spawning period with regularly spaced peaks in production. I conclude that interspecific size differences are at least partly the result of interspecific differences in reproduc- tive seasonality (Table 1; i.e. the smaller congener was spawned later in the year). However, Able and Fahay (1998) noted that P. evolans eggs are larger than P. caro- linus eggs; thus size differences exist among embryos. In addition, McBride et al. (2002) showed that interspecific size-at-age differences are evident throughout the larval period. Age and growth Age-0 P. evolans were larger at a common age than P. carolinus (Fig. 3). On the basis of all individuals examined, P. evolans did not grow at a significantly higher rate (ANCOVA interaction of slopes; P=0.099), but this species was significantly larger than P. carolinus in general (ANCOVA test of intercepts; P=0.0001). The regression slopes, based on all data, were significantly different from zero (P<0.001 ) for both P caro- linus (SL=6.12-K0.323xage; r2=0.62; n=70) and P evolans (SL=3. 52-1-0. 429xage; r2=0.53; n=53). The intercepts for these linear growth models were biologically meaning- ful, particularly that for P. evolans because the intercept was roughly equal to the known size at hatching (3 mm; Able and Fahay, 1998). Prionotus evolans collected at four different stations in Delaware Bay during October 1991 deviated strongly from the average growth rates for this species (Fig. 3). Such small, slow-growing fish demon- strated the degree of intraspecific variation possible for P evolans growth rates. Age-0 P carolinus were also much smaller on average than P. evolans during winter ( Fig. 4). By February-March, age-0 P. carolinus ranged from 27 to at least 117 mm SL, but this size distribution was strongly bimodal with modes at 42.5 mm and 87.5 mm SL. The age-0 P evolans were all larger than 75 mm SL and appeared to grow even larger than 120 mm SL, which was the cutoff size for collections, so that the interspecific size difference is even greater than that indicated in Figure 4. Apparently all age-1 P. evolans are larger than this cutoff value because none were ob- served in the samples, whereas age-1 P. carolinus ranged from 56 to at least 118 mm. Although age-1 P. carolinus were generally larger than age-0 conspecifics, the sizes of both age classes overlapped in a manner that would con- found length-frequency analyses of this species. The intraspecific size variation of age-0 fishes, of both Prionotus species, was spatially correlated. Individually, Jun 1 Aug 1 Sep 1 Figure 2 Backcalculated hatching dates (date of capture - [otolith age + 4]) for age-0 Prionotus carolinus (open bars; n=70l and P. evolans (filled bars; n=53) collected in estuarine and continental shelf habi- tats during 1991. „ 100 1 E E 80- ■ ■ ■ f 60- ^ 0) ^ 40- ^ ■ ■ftrf r._"3^ 1 20- '^ n - .*.*:s^*^ 0 T ( ) 25 50 75 100 125 Hatching age (days) Figure 3 Length-at-age relationships for Prionotus carolinus (open symbols) and P. evolans (filled symbols). Prionotus evolans from Delaware Bay are depicted separately (triangles) from all other P. evolans (squares). Sample sizes are identical to those | in Figure 2. NOTE McBride: Spawning, growth, and overwinter size of Prionotus carotinus and P evolans 645 depth and latitude explained a significant (r->0.30; P< 0.01) amount of the variability of the mean size of age- 0 P. coroliniis during winter (Fig. 5). Each variable also contributed significantly (P<0.02) to a multiple regression; nearly half (r^=0.47) of the variance of the mean age-0 P. carolinus size between individual trawl tows was ex- plained by the linear, least-squares equation mean SL = -158 + 0.37 x i.depth[m]) + 5.7 x Uatitudei'N]). 36 24 12 0 36- 24- 12; 0- 36- 24- 12- 0- Prionotus carolinus, age-0 (n=1 12) in.n,n,n,n,n,nr,n,n,n,n,n,n p. carolinus, age-1 (n=30) -^ ^A^DA^ ,n,n,n,nn, p. evolans, age-0 (n=28) — r- 30 40 I 60 I 70 T^ I I I I I 50 60 70 80 90 Standard length (mm) 100 110 120 Figure 4 Size frequency of Prionotus carolinus (open bars) and P. evolans (filled bars) based on collections from continen- tal shelf habitats in February-March 1993 (see Figure IB). The upper boundary (120 mm SL) represents the size cutoff used during field collecting and does not nec- essarily define the maximum size attained for any age class, n = number offish in sample. A Y = 47.9-t-0.499*(X)r2 = 0.40 B Y = -296-^9.85(X) r'=0.3Q 120 100 80 60 40 20 0 50 I ■ 100 —I 150 o Taylor's series of W(, with respect to the random variables is obtained by From Taylor's series (mentioned above), the approximate covariance is obtained. Cov(W;"",Wj'.^') = E(Wl"-Wl")(Wir-Wl?') = £k(Cr'-Cr') + Cr'(i' [sr,,(cif'-c;:')+c;f'(EZF,.-iz^,.)] Here C'f and w, , are independent, and both u'k and a'*. are estimated from different samples. Therefore Gov iC[^,'wi,. ) = Coviw^.O^'^) = Cov(Wi,,Wt) = 0, then we get the covariance as only the first term. Erratum 2 Fishery Bulletin 100 (2):2S8. McFee, Wayne E., and Sally R. Hopkins-Murphy Bottlenose dolphin (Tursiops truncatus) strandings in South Carolina, 1992-1996 Last sentence of abstract should read as follows: Incidents of bottlenose dolphin rope entanglements accounted for 16 of these cases. This is estimated by crv(i,y) = i^^i^^^^y,x,-Z)(y-y Fishery Bulletin 100(3) 649 Superintendent of Documents Publications Order Form *5178 I I YEo, please send me the following publications: Subscriptions to Fishery Bulletin for $45.00 per year ($63.00 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.) Charge your order. IT'S EASY! MasterCard 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 RO. 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 English 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 inteUigible to interdiscipUnary 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 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 common.," respectively. Authors are advised wherever jjossible 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 cdnform 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 itahc 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 anatomical parts (line drawings) or variables (graphs) within figures; use Times Roman bold font to label the dif- ferent sections of a figure (e.g. A, B, C). Figure leg- ends should explain all symbols and abbreviations seen within the figure and should be typed in dou- ble-spaced format on a separate page at the end of the manuscript. We advise authors to peruse a recent issue of Fishery Bulletin for standard for- mats. 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. govemipent publication, 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. John V. Merriner Scientific Editor, Fishery Bulletin Center for Coastal Fisheries and Habitat Research, NOS 101 Pivers Island Road Beaufort, NC 28516 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 sepior 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. ■^ caudal fin and the size and shape of its mouth (Fig. 1). Paranfhias furcifer- is dark red in color; darker dorsally, lightening ventrally (Heemstra and Randall, 1993). There is an orange spot at the upper end of the base of the pecto- ral fin and three white spots dorsal to the lateral line. Ber- muda is the northern limit of the distribution of P. fiarifer in the western Atlantic and the species occurs throughout the Bahamas and Antilles and along the American coast from the Gulf of Mexico south to Brazil (Smith, 1971; Smith-Vaniz et al., 1999). In contrast, C. fulva is usually scarlet in color and cov- ered with light blue-green spots, each surrounded by a black ring (Heemstra and Randall. 1993). There are two black spots on the edge of the lower jaw as well as on top of the caudal peduncle. CephalophoUs fulva has rounded caudal, anal, and dorsal fins, similar to fins of other epinepheline serranids (Heemstra and Randall. 1993) (Fig. 1). In the northwestern Atlantic. C. fulva has a distri- bution similar to that of P. furcifer; it occurs as far north as Bermuda, throughout the Bahamas, Antilles, and along the east coast of the Americas from South Carolina to Brazil (Heemstra and Randall. 1993; Smith, 1971; Smith- Vaniz etal., 1999). Morphologically, the putative hybrids are almost e.xactly intermediate to the parent species (Fig. 1). In his 1966 re- view. Smith noted that the hybrids have some characters unique to C. fulva and P. furcifer. For example, the puta- tive hybrid individuals have both a moderately forked tail and blue spots surrounded by a black ring. Paranthias furcifer is the only Atlantic grouper with a forked tail sug- gesting that it is one of the putative parents and C. fulva is the only Atlantic grouper that has blue spots with a black ring. Presence of both traits together in a single individual strongly suggests interbreeding between the two species. The geographic extent of the putative hybrid is not well known; however specimens exist from Cuba and Bermuda (Smith-Vaniz et al.. 1999). and it has been reported from Jamaica (Thompson and Munro. 1978). Hybridization has traditionally been detected by using morphological characters, but increasingly, genetic analy- ses have also been used for this purpose. Allozyme elec- trophoresis provides a rapid and cost-effective method to Bostrom et a\ Hybridization between two serranids, Ccphalopholis fulvo and Paranlhias lurcifcr 653 assess hybridization (Campton, 1987). Through the analy- sis of multiple loci it is possible to identify F, and post-F, hybrids, as well as to detect the introgression of alleles between species by hybrid backcrossing. It became evident however, that analysis of nuclear DNA loci is necessary in studies of hybridization as a means to overcome some of the sampling restrictions of allozyme analysis and to provide a sur\'ey of biparentally inherited genes ( Verspoor and Hammar, 1991). Mitochondrial DNA (mtDNA) has also been used extensively in studies of hybridization. Be- cause mtDNA is maternally inherited in fishes, analysis of the molecule allows one to identify the maternal parent of F, hybrids, as well as the sexual preferences of Fj hybrids and their offspring (Dowling et al., 1996). The key to using molecular markers to identify hybrid- ization is to find multiple independent nuclear loci and a mitochondrial gene region that have unique alleles in each putative parent species (Dowling et al., 1996). An Fj hybrid would be heterozygous at all nuclear loci and have a mitochondrial haplotype identical to one parent species I Campton, 1987; Dowling et al., 1989). A backcrossed indi- vidual would be heterozygous at some diagnostic nuclear loci and homozygous at others. Therefore the power of demonstrating an F, hybrid, as opposed to a backcross or pure parent individual, increases with the number of nuclear loci examined. In this study, genetic information from four diagnos- tic allozyme loci, two diagnostic nuclear DNA loci, and three diagnostic mtDNA gene regions, was used to assess hybridization and introgression between Cephalopholis fulva and Paranthias furcifer in Bermuda waters. Materials and methods A total of 51 Cephalopholis fulva (Linnaeus, 1758), three C. cruentata (Lacepede, 1802), and 37 Paranthias furci- fer (Valenciennes, 1828) were collected from Bermuda with baited handlines or rotenone solution. In addition, six C. fulva and two C. cruentata were sampled from Navassa Island. Fifteen putative hybrids were captured by Bermudian fishermen using handlines or lobster traps. Cephalopholis cruentata was in-cluded in the study as a possible parent species of the putative hybrid and three Epinephelus guttatus (Linnaeus, 1758) specimens from Lee Stocking Island. Bahamas, were used in the preliminary mitochondrial DNA study. Specimens were frozen upon capture and stored at - 20°C or -80°C and transported to the laboratory for analy- sis. For mitochondrial and nuclear DNA analysis, muscle tissues were removed and placed in storage buffer (0.25M EDTA, 20':^ DMSO and saturated with NaCli. For allo- zyme analysis, 1.5-cm'^ pieces of liver and muscle tissue were separately homogenized in 250 pL of chilled (4°C) grinding buffer (0.1 M Tris, 0.9 mM EDTA, and 0.05 mM NADP*, pH 7.2). Samples were centrifuged for 3 min at 16,000 xg and stored at -80°C or analyzed immediately. Genomic DNA was isolated from a l.O-cm'' piece of mus- cle tissue by using the phenol/chloroform protocol of Win- nepenninckx et al. (1993) with the following modifications. Table 1 National Museum of Natural History (USNM) and Vir- ginia Institute of Marine Science (VIMS) catalogue num- | bers for specimens used in the morphological analysis. Standard Species Collection Specimen 1 ength (mm) C. fulva USNM 88717 232 USNM 53134 217 USNM 53134 176 USNM 133689 171 VIMS 10413 222 VIMS 10414 213 VIMS 10415 210 VIMS 10416 170 VIMS 10417 146 USNM 320539 240 P. furcifer VIMS 10410 274 VIMS 10411 259 VIMS 10412 274 USNM 107108 218 USNM 358541 237 USNM 33255 207 USNM 12540 185 USNM 65605-8244 201 USNM 65605-8246 176 USNM 65605-8246 129 Putative hybrid VIMS 10403 190 VIMS 10402 199 VIMS 10401 222 VIMS 10404 237 VIMS 10405 210 VIMS 10406 195 VIMS 10407 172 VIMS 10408 114 VIMS 10409 181 USNM 1240011 230 CTAB (hexadecyltrimethylammonium bromide) was not added to the extraction and phenol was added immediately following incubation of the tissue at 37°C. DNA was pre- cipitated by the addition of 0.04 volume of 5M NaCl and 1.0 volume of isopropanol. DNA was resuspended in 150 pL of sterile O.IX TE (Tris-EDTA) and stored at -20°C. Morphological analyses Specimens of C. fulva and P. furcifer used in the mor- phological analysis were obtained from and measured at the National Museum of Natural History (Table 1). Ten putative hybrids were suitable for morphological analysis. The remaining five samples were not properly preserved and morphological analysis was not possible. Hybrid specimens as well as a small number of the C. fulva and P. furcifer were frozen. Specimens obtained from the National Museum of Natural Historv were fixed in for- 654 Fisherv Bulletin 100(4) malin. Morphometric and meristic characters were exam- ined as described in Smith (1971) by using dial caHpers and a meter stick. The morphometric data were analyzed by using a sheared principal component analysis with a covariance matrix to confine the effect of size to the first principal component (Humphries et al., 1981; Bookstein et al., 1985; Stauffer et al., 1997). The meristic data were analyzed by using a principal component analysis with a correlation matrix. Allozyme analysis Horizontal starch gel electrophoresis followed the pro- tocols described in Murphy et al. (1996). Gels (12'7fw/v; Starch Art Corp., Smithville, TX) were run on one of three buffer systems: Tris-citrate II buffer (TC II) (30 niAmps for 14 hours), lithium hydroxide buffer ( LIOH) (25 niAmps for 14 hours), or Tris borate-EDTA buffer (EBT) (30 mAmps for 14 hours). Histochemical staining followed the proto- cols of Murphy et al. (1996), and locus nomenclature and allelic designations followed Shaklee et al. ( 1990). A preliminary survey of 16 loci in 15 individuals each of C. fidva and P. furcifer (Table 2) was performed to identify those loci for which the alleles were consistently differ- ent among the presumed parent species, C. fulva and P. furcifer. All parent individuals and putative hybrids were then surveyed for all loci that demonstrated differences between the species. Nuclear DNA analysis An actin gene intron and the second intron in the S7 ribosomal protein gene (Chow and Hazama, 1998) were investigated by using restriction fragment length poly- morphism (RFLP) analysis. Amplification primers and reaction conditions are listed in Table 3. The regions were amplified by using the PCR reagent system (GIBCO/BRL Life Technologies®, Bethesda, MD) and a 25-pL reaction cocktail ( IX PCR buffer with MgCl.^, 0.2 mM dNTP, 0.5 pM primer, 2.5 U of Tag DNA polymerase, and 25-50 ng genomic DNA template). Some PCR reactions were per- formed with Platinum® Tag high fidelity (GIBCO/BRL Life Technologies®) with a 25-pL reaction cocktail (IX high fidelity buffer, 2 mM MgSO_,, 0.2 mM dNTR 0.2 pM primer, and 2.5 U of Platinum® Tag DNA polymerase high fidelity). In other cases, 1 pL dimethyl sulfoxide (DMSO, Fisher Scientific BP231-1, Pittsburgh, PA) was added to the reaction to increase sensitivity The amplification products from two individuals of each species for both loci were digested with a panel of restriction enzymes to identify those that exhibited differences between the putative parent species. All samples were subsequently digested with those enzymes that demonstrated differences in the pilot study (Table 3). Digestion reactions ( 1.5-pL lOX buffer, 3 U restriction enzyme, and 4-pL PCR product) were incubated 2 to 18 hours at 37°C. Digestion products were separated on 2.5*7^ agarose gels (1.25% Ultrapure Agarose, GIBCO/BRL Life Technology I R) + 1.25% NuSieve GTG (R) agarose, FMC Biochemical, Rockland, ME), stained with ethidium bromide and visualized under UV light. Table 2 Allozyme analysis: information ncludes loci. buffer sys- terns, and tissues. Locus Buffer Tissue Alcohol dehydrogenase [ADH- 1*1.1.1.1) EBT liver Creatine kinase iCK-B* 2.7.3.2) LIOH liver Creatine kinase (CA'-C*2. 7.3.2) LIOH liver Esterase, (£ST-i* 3.1.1.1) EBT liver Esterase, (EST-2* 3.1.1.1) EBT liver Fumarase,(F//* 4.2.1.2) TCII liver Glucosephosphate isomerase I.GPI-A- 1.1.1.49) EBT liver Isocitrate dehydrogenase (ICDHS* 1.1,1.42) TCII liver Lactate dehydrogenase ILDH-A* 1.1.1.27) TCII muscle Lactate dehydrogenase ILDHB* 1.1.1.27) TCII liver Malate dehydrogenase iMDHA* 1.1.1.37) EBT muscle Malate dehydrogenase (MD//-B' 1.1.1.37) EBT muscle Peptidase-B iPEPl* 3.4.11) LIOH liver Peptidase-S (PEP-2* 3.4.11) LIOH liver Peptidase-C (PEP-3* 3.4.11) LIOH liver Xanthine dehydrogenase (XDH* 1.1.1,204) EBT liver MtDNA analysis The following regions of the mitochondrial genome were surveyed in the three putative parent species and hybrids; the adenosine 5'-triphosphatase subunit 6 (ATPase 6) gene, the 12S/16S ribosomal RNA gene region, and the nicotinamide dehydrogenase subunit 4 (ND4) gene. Ampli- fication primers and reaction conditions are listed in Table 3. Amplified products of two individuals of each putative parent species were screened with restriction enzymes to identify potential differences between C. fulva and P. fur- cifer (Table 3). All individuals were screened at the three regions with those enzymes that revealed differences in the preliminary study. Restriction digestion reactions were performed and visualized as described for nuclear DNA. For each individual, the haplotype designations of each region were combined in sequence creating a composite haplotype. Molecular data analysis Nei's (1978) unbiased genetic distance was calculated from the allozyme data by using the computer program BI0SYS2 (Swofford and Selander, 1989). Mean nucleotide Bostiom el al : Hybridization between two serranids, Ccphalopho/is fulva and Paianthias fu/afei 655 Table 3 PCR primers and conditions used in the RFLP analysis of nitDNA and nuclear intron regions. The forward pri mer is on top, the reverse is on the bottom. Region Primer sequence PCR conditions Citation Enzymes 12S/16S 12SA-L: 5' AAA CTG GGA TTA GAT ACC CCA CTA T 3' 94°C for 1 min., Palumbi et al., BanU.Rsal 16SA-H: 5' ATG TTT TTG ATA AAC AGO CG 3' 45°C for 1 min., 65°C for 3 min. 1991 ATPase 6 H8969: 5' GGG GNC GRA TRA ANA GRC T 3' L8331: 5' TAA GCR NYA GCC TTT TAA G 3' 95°C for 1 min., 45°C for Imin., 65°C for 3 min. Quattro' Ddel NU4 ARG-BL: 5' CAA GAG OCT TGA TTT CGG CTC A 3' 95°C for 1 min.. Bielawski and BstOl.Hpa II, LEU: 5' CCA GAG TTT CAG GCT CCT AAG ACC A 3' 45°C for 1 min., 65-C for 3 min. Gold, 1996 Mbo I, Rsa I S7 ribosomal S7RPEX2F: 5' AGC GCC AAA ATA GTG AAG CC 3' 95°Cfor30sec., Chow and Alu I, Dra I S7RPEX2R: 5' GCC TTC AGG TCA GAG TTC AT 3' eO-C for 1 min., 72°C for 2 min. Hazama, 1998 Actin intron F3: 5' ATG CCT CTG GTC GTA CCA CTG G 3' Rl: 5' CAG GTC CTTACG GAT GTC G 3' 94°C for 1 min., 48''C for 1 min., 65°C for 3 min. Cordes, 2000 Hind ' Quattro, J 1999 Personal commun Department of Biology, Univ. South Carolina, Columbia, SC 29208 sequence divergence was calculated for the mito- chondrial DNA RFLP results by using the equation of Nei and Li ( 1979) for fragment data with weight- ing based on Nei andTajima (1983) as performed by the computer program REAP (McElroy et al., 1992). Genetic distance was not calculated for the nuclear DNA loci owing to an absence of shared restriction fragments between C. fulva and P. furcifer. Results Morphological analysis Of 44 counts and measurements analyzed, mean values of 38 were different between C. fulva and P. furcifer, and the putative hybrids were intermediate in morphology to presumed parents for 33 of these characters (Table 4). The mean value for the puta- tive hybrids exceeded the values of either presumed parent species for four characters (caudal peduncle scales, interorbital width, anal base length, and pectoral length) and was lower than either pre- sumed parent species for two characters (orbit length and depressed dorsal length). In the principal component analy- sis, the first component accounted for 667^ of the variation in the data and the second accounted for 23'^'f . Suborbital width, caudal peduncle length, caudal peduncle to upper fin rays and caudal penduncle to lower fin rays were the char- acters with the highest loadings. In the meristic data, the first principal component accounted for 62% of the varia- tion. Dorsal rays, gill rakers, and transverse scale rows had the highest loadings. A plot of the second principal 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2 & C. fulva P. furcifer Putative Hybrid ^. 4* — I 1 1 I 1 1 0.6 -0.4 -0.2 0 0.2 0.4 0.6 Sheared PC 2 Figure 2 Graph of principal component 2 for the morphological data and factor 1 for the meristic data for ten Cephalopholis fulva. ten Paranthias furcifer, and ten putative hybrids. The arrow indicates putative hybrid H7, an individual identified as a backcross to C. fulva, according to the genetic data. component from the morphological analysis (variation in shape) and the first factor of the correlation matrix of the meristic data shows a discrete difference between the puta- tive parents (Fig. 2). Hybrid individuals were shown to be intermediate between the putative parent species. Allozyme analysis Sixteen allozyme loci were surveyed in 15 C. fulva and 15 P. furcifer to identify those that had different alleles in 656 Fishery Bulletin 100(4) Table 4 Ranges of counts and measurements for ten Pa ranthias furcifer, ten Cephalopholis fulva. and ten putative hybrids. Measurements | are given in millimeters. Raw measurements were divided by the stand 3rd length and multiplied by 1000. Means are given in parentheses, L.L. = lateral line, and the measurement for one hybrid discarded because of a broken third dorsal spine (*). Measurement C. fulva Putative hybrids P. furcifer dorsal rays EK, 15-16(16) IX, 17-18(17) IX, 18-19(19) anal soft rays 8-10(9) 9-10(9) 9-11(10) pectoral rays 32-35(34) 35-37 (36) 36-39 (.38) gill rakers 22-28(25) 24-35(31) 32-39(36) scales above the L.L. 6-8(8) 10-12(11) 10-14(12) scales below the L.L. 22-27(25) 26-30(28) 26-32 (30) transverse scale rows 64-84(71) 75-91 (84) 85-96(91) caudal peduncle scales 40-49 (46) 44-52(47) 43-48(46) head length 374-427(402) 320-3.53(335) 259-.301(280) head width 173-227(200) 150-189(164) 130-162(142) head depth 248-288(266) 212-250(233) 183-241(212) snout length 86-122(107) 79-111 (94) 53-79(67) suborbital width 44-52(47) 29-35(32) 20-24(22) interorbital width 64-78(71) 73-89(82) 76-89(80) orbit length (diameter) 62-76(68) 56-67(62) 52-89(65) postorbital head length 217-243 (232) 173-210(192) 146-171 (1,59) maxillary length 166-194(182) 120-163(141) 96-115(104) lower jaw length 176-200(186) 117-149(138) 98-114(105) snout to angle of preopercle 257-306(282) 213-246(229) 176-200(188) maxillary width 42-56(48) 30-45(41) 28-38(34) tip of lower jaw to gular notch 119-181(145) 90-174(123) 69-121 (89) body width 154-223(178) 155-181 (165) 138-162(1.50) body depth 325-385(357) 302-361 (332) 282-354(316) caudal peduncle depth 128-139(134) 118-145(129) 102-122(111) tip of snout to dorsal origin 386-422 (407) 334-373(354) 321-361 (329) tip of snout to pectoral base 287-415(371) 293-340(319) 265-293(277) tip of lower jaw to pelvic base 398-449(420) 359-420(382) 321-384(351) dorsal base length 523-549(538) 532-561(543) 543-6131.583) depressed dorsal length 609-663 (629) 574-643 (609) 603-658(635) anal base length 166-187(174) 171-192(182) 16.5-204(180) depressed anal length 268-320(294) 248-273(261) 232-278(2.53) end of dorsal to caudal base 129-150(141) 148-165(157) 145-175(162) length of caudal peduncle 166-198(178) 194-217(203) 174-241(219) pectoral length 257-292(274) 253-295 (279) 249-290(274) pelvic length 186-211 (197) 143-192(180) 150-187(173) dorsal spine I length 54-66(62) 43-70(61) 40-64(53) dorsal spine III length 103-138(123) 90-119(106) 86-117(105) dorsal spine IX length 97-150(127) 104-123(111) 76-105 (94) anal spine I length 52-72(62) 35-66(53) 36-48(43) anal spine II length 95-121(106) 91-112(101) 78-98(86) anal spine III length 78-118(108) 97-121(106) 74-101(87) caudal base to tip of upper rays 197-246(216) 228-304(280) 321-373(347) caudal base to tip of middle rays 204-248(222) 164-192(175) 122-143(128) caudal base to tip of lower rays 197-242(217) 278-318(294) 308-346(322) the two species (Table 2). At four loci iCKB*. FH*. LDH- B*. and ICDH-S*) C. fulva and P. furcifer had different alleles. Forty C. fulva, 28 P. furcifer, one C. cruentata, and ten putative hybrids were subsequently screened at the four diagnostic allozyme loci (Table 5). Eight often puta- tive hybrids were heterozygous at all four diagnostic loci. One putative hybrid was heterozygous at all loci except the LDHB* locus, for which it displayed two alleles characteristic of P. furcifer, and another individual was heterozygous at all diagnostic loci, except the FH* locus Bostrom et a\ Hybridization between two serranids, Ccphalopholis futva and Paranthias furafer 657 Allozyme genotypes of CepI loci, CKB*. FH*. ICDHS*. alopholis fit and LDHB Iva, ".n Table 5 Paranthias furcifei; C. cnwr = number offish in a sample lata, and put; itive hybrid individual < at four diagnostic n CKB* FH* ICDHS* LDHB* C. fidva 40 * 100/100 '90/90 *95/95 * 100/100 P. funifer 28 *50/50 * 100/100 ' 100/100 *75/75 C. crucnlala 1 *100/100 *90/90 * 105/105 *75/75 Putative hybrids 8 1 1 *50/100 *50/100 *50/100 *90/100 *90/100 *90/90 * 100/95 * 100/95 * 100/95 *75/100 *75/75 *75/100 Table 6 RFLP genotypes for tht > nuclear intron and mtDNA loci. Some bands (t) were inferred so that fragments would sum | to the total (uncut) size of the amplified gene region. Approximate Locus Enzyme Allele size (bpl actin intron Hin/I A 400. 50' B 450 ST intron Dra I A 575. 550. 75' D 1200 E 575, 525, 75', 25' ATPase 6 Dde I A 600, 50' B 360, 150, 90, 50' C 650 12S/16S Rsa I A 600. 500. 300, 200 B 600, 300, 250, 250, 200 C 450, 375, 300, 275. 100, 100' Ban II A 1600 B 1100, 500 ND4 BstO I A 12.50, 625, 25' B 1250, 450, 200 C 1450, 450 Hpa II A 1400. 500 C 1900 D 1000. 900 Mho I A 700. 500, 400, 300 B 550, 500, 400, 300, 150 C 525, 400, .300. 275, 2.50, 80, 70 Rsa I A 1025, .500. 375 B 610, 500. 400. 390 C 500. 400, 350, 300, 300, 50' D 900, 425. 325, 250 Table 7 Genotypes of Cephalophohs fid •a. Parar thias furcifen C. cruentata. and putative hybrids for the short actin intron 1 and the second intron in the S7 ribosomal protein region. n = number of fish in a sample. Species Actin intron S7 intron n Hind n Dra I Cephalopholis fulva 57 A/A 50 3 4 A/A E/E A/E Paranthias fiircifer 37 B/B 37 D/D C. cruentata 5 B/B 5 D/D Putative hybrid 15 A/B 14 1 A/D E/D for which it displayed two alleles diagnostic of C. fulva. Cephalopholis cruentata was distinguished from C. fulva and P. furcifer at the ICDH-S* locus and was eliminated as a potential parent species. Nuclear intron regions An actin intron appro.ximately 450 base pairs in length was amplified. The region was surveyed with 14 restric- tion enzymes, of which Hint I showed a genetic difference between species. After digestion with Hint I, all C. fulva demonstrated allele A, and all P. furcifer diplayed allele B (Tables 6 and 7). All fifteen putative hybrids were hetero- zygous for both alleles. The second intron region of the S7 ribosomal protein, which was approximately 1200 base pairs in length, was screened with thirty-five enzymes. Two enzymes, Dra I and Alu I, demonstrated differences between P. furcifer and C. fulva. Paranthias furcifer and C. cruentata both exhibited allele D after digestion with Dra I (Tables 6 and 7 ). Cephalopholis fulva was variable at this locus — fifty in- dividuals were homozygous for allele A, three homozygous for allele E, and four heterozygous for alleles A and E. All fifteen putative hybrids were heterozygous at this locus with one of the C. fulva alleles (A or E), and the P. furcifer allele (D). Digestion of the second intron in the S7 region hy Alu I produced a large number of small fragments that were not easily interpreted and the data were not used 658 Fishery Bulletin 100(4) to identify hybridization between C. fulva and P. furcifer. With this enzyme, however, C. cruentata had a unique al- lele and was thus eliminated as a putative parent for the hybrid individuals. Mitochondrial DNA Allelic differences between C. fulva and P. furcifer were found in all three mitochondrial gene regions. The ATPase 6 region was screened with six enzymes, one of which, Dde I, showed differences between C. fulva, P. furcifer, and C. cruentata (Tables 3 and 6). The 12S/16S region was screened with seven enzymes. Two of these. Ban II and Rsa I, demonstrated differences between C. fulva, P. furcifer, and C. cruentata. The ND4 region was screened with nine restriction enzymes, four of which (BstO I, Hpa II, Mbo I, and Rsa I) showed differences between the species; mtDNA composite haplotypes were unique to each species (Table 8). All 15 putative hybrids in the study had a composite haplotype matching the common haplotype of C. fulva, indicating that it was the maternal parent for all hybrid individuals. Three Epinephelus guttatuf! specimens were screened at the ND4 and ATPase 6 regions and showed a unique composite haplotype; therefore this species was not included in the study as a putative parent species. Discussion Morphological analysis In most cases, Fj hybrids should be morphologically intermediate to the parent species and have low variation within characters among themselves. Backcross individu- als, because of random sorting of chromosomes, should have higher variation within intermediate characters and could fall anywhere in the morphological range of the pure parent species (Anderson, 1949). In a principal component analysis plot, a backcross hybrid's score would be expected to be closer to the parent species to which the hybrid Table 8 Composite haplotypes of Cephalopholis fulva. Paranthias furcifer. and C. cruentata for the mitochondrial DNA data. Haplotypes are given for the following sequences of mt- DNA loci and enzymes; 1 ) ATPase 6 Dde I; 2) ND4— B,s/0 I, Hpa II, Mho I, Rsa I; and 3) 12S/16S— fisa I, Ban II. /i = number offish in a sample. Species n Composite haplotype Cephalophtilis fulva 56 1 AAABBAA AAABCAA Paranthias furcifer C. cruentata 36 1 5 BBCADBB BBDADBB CCDCACB Putative hybrids 1.5 AAABBAA backcrossed, whereas an Fj hybrid's characters would be expected to be in the center, closer to an average of the scores of the parent species. A plot of the second principal component of the morpho- logical analysis and the first factor of the meristic analysis (Fig. 2) shows that C. fulva and P. furcifer are well segre- gated according to morphological characters. The putative hybrid individuals were clustered in between the parent species. The post-Fj hybrid detected by using genetic anal- yses, indicated by an arrow in Figure 2, clustered with the putative Fj hybrid individuals. Genetic analyses Results of the allozyme analysis also supported hybridiza- tion between C. fulva and P. furcifer in Bermuda. The puta- tive hybrids were heterozygous at four distinguishing loci, with the exception of two individuals that were homozygous at one diagnostic locus each. One individual was homozy- gous at the LDH-B* locus for the *75 allele. All twenty-eight P. furcifer were homozygous for this same allele, indicating hybrid backcrossing to P. furcifer. Another hybrid individual was homozygous at the FH* locus for the *90 allele. The 40 C. fulva sampled in this study were homozygous for this allele, suggesting hybrid backcrossing to C. fulva. This hybrid individual was among those included in the morpho- logical study and, as shown in the principal component plot (Fig. 2). was morphologically the most similar to C. fulva. It was not possible to distinguish F,, hybrids and backcross hybrid individuals and henceforth the two individuals described above are referred to as post-Fj hybrids. Because all members of the presumed parent samples were homozy- gous at all loci for diagnostic alleles, there was no evidence of introgression between C. fulva and P. furcifer. The nuclear intron data were consistent with the allo- zyme data and supported the hypothesis of hybridization between C. fulva and P. furcifer. Because all hybrid indi- viduals were heterozygous, post-F, hybridization was not evident at these loci. Alleles present at both nuclear DNA loci were unique between parent species and there was no indication of introgression between these species. The mtDNA data clearly showed that C. fulva was the maternal parent for all putative hybrids, including the two post-Fj hybrids. This finding suggests a strong gender bias in hybridization. All C. fulva had composite haplotypes quite distinct from those of P. furcifer, and there was no evidence of mtDNA introgression between the two parent species. Overall, the genetic and morphological analyses sug- gest that all but two of the 15 putative hybrids were F, individuals representing first generation hybridization between a female C. fulva and a male P. furcifer (Table 9). The occurrence of two post-F, hybrids indicates that Fj hybrids are fertile, and the genotypes of the two post-Fj hybrids demonstrate that F, hybrids can backcross with either parent species. Hybridization in Bermuda Hybridization between C. fulva and P. furcifer is known from only certain localities in the tropical Atlantic, despite Bostrom et al : Hybridization between two serranids, Cephalopholis fulva and Paianthias furcifer 659 Table 9 Classification of all putative hybrid individuals as F, and post-F, individuals based on morphological and genetic data. Sam jlelD Classification Data 111 post-F, allozyme, mtDNA, nDNA H2 F, mtDNA, nDNA H3 F, mtDNA, nDNA H4 F, mtDNA, nDNA H5 F, mtDNA, nDNA H6 F, mtDNA, nDNA H7 post-F, morphology, allozyme, mtDNA, nDNA H8 F, morphology, allozyme. mtDNA, nDNA H9 F, morphology, allozyme, mtDNA, nDNA HIO F, morphology, allozyme, mtDNA, nDNA HU F, morphology, allozyme, mtDNA, nDNA H12 F, morphology, allozyme, mtDNA, nDNA H13 F, morphology, allozyme, mtDNA, nDNA H14 F, morphology, allozyme, mtDNA, nDNA H15 F, morphology, allozyme, mtDNA, nDNA USNM 124001 F, morphology broad overlap in the geographical ranges of the two spe- cies. Bermuda exists at the northern range of C fulva and P. furcifer, and the two species have restricted spawning times. Cephalopholis fulva and P. furcifer both spawn in Bermuda from May to early August (Smith, 1958, as cited in Thompson and Munro, 1978), and spawning individuals of the two species have been sampled in the same location (Burnett-Herkes, 1975; B. Luckhurst, personal observ.). Similarly, two hybrids have been reported from Jamaica where the parent species also have overlapping spawning times (Thompson and Munro, 1978). Hybridization between C. fulva and P. furcifer could be the result of directed interspecific interactions, or the chance meeting of gametes spawned at the same time in the same general location. However, because all 13 F, hybrids were the result of C. fulva eggs fertilized with P. furcifer sperm, it appears that there is a gender bias in hybridization. Possible hybridization scenarios include differences in sex ratio between C. fulva and P. furcifer, a biochemical block on fertilization off! furcifer eggs by C. fulva sperm, or "sneaker" P. furcifer males in C. fulva spawning groups. Based on the number of individuals collected in Ber- muda over the last two years, hybridization between C. fulva and P. furcifer seems to be relatively rare. Although the reproductive status of the hybrids is unknown, the oc- currence of two fertile female F, hybrids (one of which was ripe), a spent male F, hybrid, and the presence of post-Fj hybrids, it can be concluded that some F, hybrids are ca- pable of producing viable offspring. However, reproduction of F, hybrids does not appear to be very extensive. Within the limits of the samples analyzed there was no evidence of introgression of alleles between parent species. Intergeneric hybridization Intergeneric hybridization in animals seems to be rela- tively rare. It is believed that the ability to hybridize is an indication of evolutionary relatedness and that divergent taxa should have lost the ability to interbreed through the evolution of reproductive isolating barriers (Sibley, 1957). However, there are several examples of intergeneric hybrid- ization in fishes, most notably in the cyprinids (Hubbs, 1955; Smith, 1973; Aspinwall et al, 1993; Stauffer et al., 1997). Intergeneric hybridization between two such ecologi- cally different species as C. fulva and P. furcifer has also been noted in the lutjanids. Poey (1860) described an in- tergeneric hybrid between Lutjanus synagris and Ocyurus chrysurus. Both Loftus (1992) and Domeier and Clarke (1992) presented reviews of this case of hybridization and Loftus (1992) theorized that the species were capable of interbreeding on the basis of overlap of spawning time and habitat. Both Loftus (1992) and Domeier and Clarke (1992) surmised that O. chrysurug should not constitute its own genus but be included within the genus Lutjanus, owing to its ability to hybridize with a member of that genus. Further support for this revision was provided by Chow and Walsh ( 1992), who reported a high genetic simi- larity between O. chrysurus and several Lutjanus species of the western North Atlantic. They suggested that the species appeared to be misplaced due to its imique mor- phological features. A parallel situation exists with the epinepheline serrands because the genetic distance data suggest that P. furcifer does not belong in a separate genus from Cephalopholis. Recently, Craig et al. (2001) used mitochondrial DNA se- quencing to demonstrate a close phylogenetic relationship 660 Fishen/ Bulletin 100(4) between the genera Paranthias and Cephalopholis. They al- so cited several morphological and ontogenetic similarities between the two genera. In the present study, an analysis of 16 allozyme loci produced a Nei's (1978) unbiased genetic distance of 0.356 between P. furcifer and C. fulva, which is within the range reported between Ocyurus chrysurus and species ofLutjanus (Chow and Walsh, 1992). The mitochon- drial DNA sequence divergence data indicated that P. fur- cifer is closer to, or at least no farther from, C. fulva (0.036) than is C. cruentata (0.042) and that the congeners are most distant, (0.052). Both values are within the range reported between other congeneric serranids (Graves et al., 1990). Taxonomic revision of P. furcifer is not recommended at this time; however, the incidence of hybridization and the small genetic distance between C. fulva and P. furcifer sug- gest that a full-scale phylogenetic analysis of the subfam- ily is warranted. Such a study would have to include more species of Cephalopholis, other members of the subfamily Epinephelinae, and P. colonus, the eastern Pacific gemi- nate species of P furcifer. Acknowledgments Special thanks are extended to J. Parris Jr, J. Parris Sr., W. McCallan, J. Payne, A. Marshall, K. Gregory, L. Mollis, D. Young, L. Outerbridge, and M. Battersbee for donat- ing captured hybrid specimens. T Trott kindly assisted with field dissections in Bermuda, J. R. McDowell pro- vided expert guidance with molecular techniques, and J. Stauffer Jr. is credited with the analysis of morphological data. J. Quattro and J. Cordes provided primer sequences. Financial assistance to MAB was generously provided by a scholarship from the International Women's Fishing Association. A draft of this manuscript was reviewed by T A. Munroe and B. W. Bowen. Literature cited Anderson, E. 1949. Introgressive hybridization, 109 p. J. Wiley, New York, m' Aspinwall, N., D. Carpenter, and J. Bramble. 1993. The ecology of hybrids between the peamouth, A/v/o- cheilus caunnus. and the redside shiner, Richardsonius balteatus. at Stave Lake, Bntish Columbia, Canada. Can. J. Zool. 71:83-90. Bielawski, J. P., and J. R. Gold. 1996. Unequal synonymous substitution rates within and between two protein-coding mitochondrial genes. Molec. Biol. Evol. 13:889-892. Bookstein, F. L., B. Chernoff, R. L. Elder, J. M. Humphries, G. R. Smith, and R. E. Strauss. 1985. Morphometries in evolutionary biology, 177 p. Acad- emy of Natural Sciences of Philadelphia. Special Publica- tion 15, Philadelphia, PA. Burnett-Herkes, J. 1975. Contribution to the biology of the red hind, Epineph- elus guttatus, a commercially important serranid fish from the tropical western Atlantic. Ph. D. diss., 154 p. Univ. Miami, Coral Gables, FL. Campton, D. E. 1987. Natural hybridization and introgression in fishes: meth- ods of detection and genetic interpretations. In Population genetics and fishery management (N. Ryman. and F. Utter, eds. ), p. 161-192. Univ. Washington Press, Seattle, WA. Chow, S., and K. Hazama. 1998. Universal PCR primers for 87 ribosomal protein gene introns in fish. Mol. Ecol. 7:1255-1256. Chow, S., and P J. Walsh. 1992. Biochemical and morphometric analyses for phylo- genetic relationships between seven snapper species (sub- family Lutjaninae) of the western Atlantic. Bull. Mar. Sci. 50<3):508-519. Cordes, J. F A. 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. College of William and Mary, Gloucester Point, VA. Craig, M. T, D. J. Pondella II, J. PC. Franck, J. C. Hafner 2001. On the status of the serranid fish genus Epinephelus: evidence for paraphyly based upon 16S rDNA sequence. Molec. Phylo. Evol. 19i 1 ):121-130. Domeier, M. L., and M. E. Clarke. 1992. A laboratory produced hybrid between Lutjanus syn- agris and Ocyurus chrysurus and a probable hybrid be- tween L. griseus and O. chrysurus (Perciformes: Lutjan- idae). Bull. Mar Sci. 50(3):501-507. Dowling, T. E., C. Moritz, J. D. Palmer, and L. H. Rieseberg. 1996. Nucleic acids III: analysis of fragments and restriction sites. In Molecular systematics (D. M. Hillis, C. Moritz, and B. K. Mable, eds.), p. 249-320. Sinauer Assoc. Inc., Sunderland, MA. Dowling, T E., G. R. Smith, and W. M. Brown. 1989. Reproductive isolation and introgression between Notropis cornutus and Notropis chrysocephalus (family Cyprinidae): comparison of morphology, allozymes, and mitochondrial DNA. Evolution 43(3):620-634. Graves, J. E., M. J. Curtis, P A. Oeth, and R. S. Waples. 1990. Biochemical genetics of southern California basses of the genus Paralabrax: specific identification of fresh and ethanol-preserved individual eggs and early larvae. Fish. Bull. 88:59-66. Heemstra, P. C, and J. E. Randall. 1993. FAG species catalogue, vol. 16. Groupers of the world (family Serranidae, subfamily Epinephelinae). FAO Fish. Synop. 125, vol. 16, 382 p. FAO. Rome. Hubbs, C. L. 1955. Hybridization between fish species in nature. Syst. Zool. 4:1-20. Humphries. J. M., F L. Bookstein. B. ChernofT, G. R. Smith, R. L. Elder, and S. G. Poss. 1981. Multivariate discrimination by shape in relation to size. Syst. Zool. 30:291-308. Loftus, W. F 1992. Lutjanus ambiguus (Poey), a natural intergeneric hybrid of Ocv(/n/,s chrysurus (Bloch) and Lutjanus synagris (Linnaeus). Bull. Mar Sci. 50(3):489-500. McElroy, D., P. Morgan. E. Bermingham, and I. Kornfield. 1992. REAP — the restriction enzyme analysis package. J. Heredity 83:157-158. Murphy, R. W.. J. W. Sites, D. G. Buth, and C. H. Haufler 1996. Proteins: isozyme electrophoresis. //! Molecular sys- tematics ( D. M. Hillis. C. Moritz, and B. K. Mable, eds. ), p. 51- 120. Sinauer Assoc. Inc., Sunderland, MA. Nei, M. 1978. Estimation of average heterozygosity and genetic Bostrom et a\ Hybridization between two serranids, Cephalopholis fulva and Paranthlas furafei 661 distaiici' fniiii a small nunibcr of individuals. Genetics 2:!:;M1-.'!69. Noi.M..andW.-H. Li. 1979. Mathematical mode! for studying genetic variation in terms of restriction endonucleascs. Proc. Natl. Acad. Sci. 76:5269-5273. Nei. M.. and F. Tajinia. 1983. Maximum likelihood estimation of the numlier of nucleotide substitutions from restriction site data. Ge- netics 105:207-217. Palumbi, S, R., A. Martin, S. Romano, W. O. McMillan, L. Stice, and G. Grabowski. 1991. The simple foors guide to PCRver. 2.0, 47 p. [Available from the Dept. of Zoology and Kewalo Marine Laboratoi-y, Univ. Hawaii, Hawaii, HI.l Poey, F. 1860. Memorias sobre la historia natural de la Isla de Cuba. La Habana 2:97-336. 1875. Enumeratio piscium cubensium. An. See. Esp. Hist. Nat. Madrid 4:75-112. 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. Sibley, C. G. 1957. The evolutionary and taxonomic significance of sex- ual dimorphism and hybridization in birds. The Condor 59:166-191. Smith, C. L. 1958. The groupers of Bermuda, /n Bermuda Fisheries Re- search Program Final Report (J. E. Bardach, ed.), p. 37- 59. Bermuda Trade Board, Hamilton, Bermuda. 1966. Menephorus Poey, a serranid genus based on two hy- brids ni Ci'phcilopholifi fulva and I'araiitluas furctfer, with comments on the systematic placement of Paranthlas. Am. Mus. Nov. 2276:1-11. 1971. A revision of the American groupers: Epinephelus and allied genera. Bull. Am. Mus. Nat. Hist. 146:71-225. Smith, G. R. 1973. Analysis of several hybrid cyprinid fishes from west- ern North America. Copeia 1973:395-410. Smith-Vaniz, W. F, B. B. Collette, and B. E. Luckhurst. 1999. Fishes of Bermuda: history, zoogeography, annotated checklist, and identification keys. ASIH (American Soci- ety of Ichthyologists and Herpatologists) Special Publ. 4, 424 p. Stauffer, J. R. Jr., C. H. Hocutt, and R. L. Mayden. 1997. Pararhinichthys, a new monotypic genus of minnows (Teleostsi: Cyprinidae) of hybrid origin from eastern North America. Ichthyl. Explor Freshwaters 7(4):327-336. Swofford, D. L., and R. B. Selander 1989. BIOSYS-2. A computer program for the analysis of allelic variation in population genetics and biochemical systematics, release 1.7. Illinois Natural History Survey, Champaign, Illinois. Thompson, R., and J. L. Munro. 1978. Aspects of the biology and ecology of Caribbean reef fishes: Serranidae (hinds and groupers). J. Fish Biol. 12: 115-146. Verspoor, E., and J. Hammar 1991. Introgressive hybridization in fishes: the biochemical evidence. J. Fish Biol. 39 (suppl. A):309-334. Winnepenninckx, B., T. Backeljau, and R. De Wachter. 1993. Extraction of high molecular weight DNA from molluscs. Trends Genet. 9(12 1:407. 662 Abstract— Commercial hai-vest of red sea urchins began in Washington state in 1971. Harvests peaked in the late 1980s and have since declined substan- tially in Washington and other areas of the U.S. west coast. We studied effects of experimental hai-vest on red sea urchins in San Juan Channel (SJC), a marine reserve in northern Washing- ton. We recorded changes in density and size distribution of sea urchin populations resulting from three levels of experimental harvest: 1) annual size-selective harvest (simulating cur- rent commercial urchin harvest regula- tions), 2) monthly complete (non-size- selective) harvest, and 3 1 no harvest I control I sites. We also examined re- colonization rates of harvested sites. The red sea urchin population in SJC is composed of an accumulation of large, old individuals. Juvenile urchins rep- resent less than 1% of the population. Lower and upper size limits for com- mercial hai"vest protect 5'7r and 4.5'* of the population, respectively. Complete harvest reduced sea urchin densities by SS'i. Annual size-selective harvest significantly decreased sea urchin densities by 67% in the first year and by 47% in the second year. Two years of size-selective harvest significantly altered the size distribution of urchins, decreasing the density of legal-size urchins. Recolonization of harvested sites varied seasonally and occurred primarily through immigj"ation of adults. Selective harvest sites were recolonized to 51% and 38% of original densities, respectively, six months after the first and second annual harvests. Yields declined substantially in the second year of size-selective harvest because of the fishing down of the population and because of low recoloni- zation rates of harvested sites. We recommend that managers consider the potential efficacy of marine hai-vest refuges and reevaluate the existing upper and lower size limits for com- mercial harvest to improve long-term management of the sea urchin fishery in Washington. Effects of experimental harvest on red sea urchins iStrongylocentrotus franciscanus) In northern Washington Sarah K. Carter Glenn R. VanBlaricom Washington Cooperative Fish and Wildlife Research Unit School of Aquatic and Fishery Sciences Box 355020 University ol Washington Seattle, Washington 98195 5020 Present address (for S K. Carter); Wisconsin Department of Natural Resources Bureau of Endangered Resources tOl South Webster St Madison. Wisconsin 53707-7921 E mail address (for S K Carter) carteskm^dnrstate wi us Manuscript accepted 28 March 2002. Fish. Bull. 100:662-673 (2002). Red sea urchins iSti-ongyloccntrotus franciscanus) are the most commonly harvested species of sea urchin on the west coast of North America. Sea urchin harvest in this region occurs primarily in California, Washington, and British Columbia. The commercial sea urchin fishery in Washington began in 1971, and landings were low through the early 1980s (Fig. 1). Landings increased dramatically in the late 1980s, peaking at over 4000 metric tons (t) in 1988. Landings have since declined. Approxi- mately 387 t of sea urchins were har- vested in 2000, valued at $699,052 ( LTlrich' I. Red sea urchins currently con- stitute approximately 60% of landings — the remainder being green sea urchins (S. droebachiensis. UlrichM. The Washington sea urchin fishery currently is managed by using harvest quotas, size limits, license restrictions, limited entry, and mandatory log books (Lai and Bradbury, 1998). Season length is not limited, but haivest occurs pri- marily during the winter when roe qual- ity is highest (Bradbury^). A rotational harvest strategy was practiced from 1977 until 1995, in which each harvest district ( Fig. 2 ) was hai-vested once every third year. In 1995 a U.S. federal court ruling on shellfishery management in Washington's coastal marine waters (Shellfish Subproceedings of United States vs. State of Washington, 873 F. suppl. 1422, 1994, known commonly as the "Rafeedie Decision") allotted one half of all hai-vestable shellfish to native tribes. As a result, rotational harvest was discontinued and replaced by an- nual harvest to ensure that all tribes had equal access to their usual and ac- customed fishing areas each year. Sea urchin harvests declined sub- stantially in the early 1990s along parts of the west coast (Fig. 1). Quotas and season lengths were reduced in Washington because of overharvest- ing concerns (Bradbury-). Quotas were not reduced in California, and catches declined substantially (Kalvass and Hendrix, 1997). Sea urchin densities in some harvested areas in northern California are less than one quarter of those in nearby reserve areas (Kalvass and Hendrix, 1997). In Washington, densities and the proportion of legal- size sea urchins in the population have declined (Pfister and Bradbury, 1996). Harvesters may be maintaining catch per unit of effort at high levels by exploiting new populations, thereby masking stock declines (Pfister and Bradbury, 1996). Experimental hai-vests may indicate potential effects of commercial harvest ' Ulrich.M. 2001. Personal commun. Wash- ington State Department of Fish and Wildlife, 600 Capitol Wav North, Olympia, WA 98.501. - Bradbury, A. 2000. Personal commun. Washington State Department of Fish and Wildlife. Point Whitney Shellfish Labora- tory. 1000 Point Whitnev Road, Brinnon, WA 98320. Carter and VanBlaricom Effects of experimental harverst on Strongylocentrotus franascanus In northiern Washington 663 Table 1 Depth and area of study sites in San Juan Channel, Wash ngton. Site numbers are those shown in Figure 2. Initial .sea urchin densities are from sampling in permanently marked circula • sampling areas in early March 1997. Initial sea urchin Site Abbreviation Treatment D epth ( m ) Area (m-) density (no./m^) 1. Mid San Juan Island Mid SJI complete harvest 6.6 428 1.15 2. South McC'onnell Island South McConnell complete harvest 10.4 405 1.13 3. O'Neal Island O'Neal complete harvest 5.8 430 1.36 4. Point Caution, San Juan Island Point Caution selective harvest 7.5 422 1.95 5. Upper San Juan Island Upper SJI selective harvest 9.1 421 1.27 6. North McConnell Island North McConnell selective harvest 9.1 415 0.37 7. Yellow Island Yellow control 6.9 404 2.05 8. Point George, Shaw Island Point George control 7.5 406 1.96 9. Shady Cove, San Juan Island Shady Cove control 10.2 447 0.94 on sea urchin populations in Washington. We examined changes in density and size distribution of red sea urchin populations resulting from two levels of experimental har- vest. Our study site was San Juan Channel (SJC), one of two areas in Washington where commercial sea urchin harvest has been prohibited since the 1970s (Fig. 2). We compared density and size distribution data for sea urchins from exper- imentally hai-vested sites in SJC with similar data collected from 1) nearby control sites, and 2) 19 sites in the Strait of Juan de Fuca (SJDF), an area commercially harvested since the early 1970s. We also examined recolonization of hai-vested sites through recruitment and immigration. We discuss implications of our results for management of sea urchin harvest effort and the potential use of marine harvest refuges in enhancing Washington's sea urchin fishery. Methods We established nine study sites in SJC from November 1996 to March 1997 (Fig. 2). Sites were 6-10 m in depth and approximately 10 m x 40 m, with the long axis of each site running parallel to the shoreline along the depth contour (Table 1). Eight permanently marked circular sampling areas (each 7.07 m^) were located along the midline of each site along the depth contour Site selec- tion was based on high red sea urchin density (>1.5lm? in preliminary surveys), substrate (primarily large cobble or bedrock) and safety considerations. We applied one of three harvest treatments to each site. "Selective harvest" consisted of annual removal of all legal- size sea urchins (102-140 mm test diameter) each winter (March of 1997 and 1998). Selective harvest simulated the annual commercial harvest of a bed of sea urchins in the San Juan Islands (Pfister and Bradbury, 1996). "Complete harvest" consisted of removal of all sea urchins present in March 1997, and at monthly intervals thereafter through September 1998. The complete harvest treatment did not represent any current management strategy for com- 25,000 « 20.000 o ^ 15,000 2 10,000^ 5,000 • Washington X Northeast Pacific ° Eastern Central Pacific D □ R O-^D S^s, "■■tB««««»' • •• *?****t 0 1975 1980 1985 1990 1995 2000 Year Figure 1 Annual harvest of red sea urchins (Strongylocentrotus spp.l in Washington state and in the larger Northeast Pacific and eastern Central Pacific areas of the west coast of the United States. Northeast Pacific includes the area north of 40''30'N (approximately Oregon-California border). Eastern Central Pacific includes the area south of 40°30'N. (Sources: Hoines, 1994, 1998; FAO, 1980, 1981, 1992, 2001; Bradbury.^) mercial harvest. Rather, the complete harvest treatment represented one extreme (the control treatment being the second) against which effects of selective harvest may be compared. Sea urchin densities were not manipulated in the "control" treatment. Because of logistical constraints, control sites were located in smaller preserve areas within SJC where harvest of all invertebrates and fish is prohib- ited. Harvest treatments were randomly assigned to the remaining six sites. All sea urchins harvested were mea- sured at the surface and released at other locations in SJC well away from the study sites. In SJC, we sampled at large (sites) and small (circular areas within sites) spatial scales. Small-scale sampling was more frequent, allowing detection of potential short- 664 Fishery Bulletin 100(4) Vancouver Island Oaas Island Strail of Juan de Fuca Washinmon Figure 2 (Upper left ) The commercial hai-vesting districts for sea urchins in Washington (after Lai and Bradbury. 1998). Harvest district numbers are shown in the upper left corner of each district. (Right) Enlargement of San Juan Channel, showing locations of study sites. Sites 1-3 are complete harvest sites, sites 4—6 are selective harvest sites, and sites 7-9 are control sites. Dashed lines represent the boundary of the areas on the west coast of San Juan Island and in San Juan Channel closed to commercial sea urchin harvest through fall 1998 The dotted line represents the modified southern boundary of the reserve area in San Juan Channel established in fall 1998. (Lower left) Enlargement of the Strait of Juan de Fuca sampling area. Black circles indicate locations of randomly chosen sampling sites. term effects of harvest; whereas large-scale sampling better indicated how treatments affected the entire bed of sea urchins. We identify the scale of sampling (sites vs. circular areas within sites) for all results. Size distribution In SJC, we measured all sea urchins removed during the initial harvest of complete and selective hai^est sites in March 1997, and all sea urchins removed during the second annual harvest of selective harvest sites in March 1998. In September of 1997 and 1998, we measured a min- imum of 100 sea urchins in situ in all sites in randomly chosen 5 m x 5 m quadrats. We also measured sea urchins in the circular areas within sampling sites in situ in all sites five times per year: early March, mid March, April, June, and September. The sampling periods represented preharvest and 5, 30, 90, and 180 days after harvesting in each year for the selective harvest treatment. Nineteen randomly chosen sites in the western SJDF were sampled in August and September 1997 (Fig. 2). All sites were 7-8 m below mean lower low water. At each site, we measured sea urchins (/) situ on 1-3 transects parallel to shore, each >30 m-. Lower and upper size limits for commercial sea urchin hai^est are 102 and 140 mm test diameter in SJC, and 83 and 114 mm in SJDF (Pfister and Bradbury, 1996). Sea ur- chin measurements in SJC and SJDF were recorded to the nearest 5 mm (e.g. 130 mm=test diameters 130-134 mm), except during the initial sampling periods (March- June 1997), when sea urchins measured in situ only were re- corded to the nearest 10 mm (e.g. 130 mm=test diameters Carter and VanBlancom Effects of experimental harverst on Stiongyloccntiotus fiancisconus in norttiern Wasfiington 665 130-139 mm). Because of the measurement incTement used, we could only estimate proportions of under-, legal-, and over-size sea urchins in SJC and SJDF. Asterisks (' ) indicate estimated size classes. Estimated size classes are as follows: SJC — under'-size: 0-95 mm, legal*-size: 100- 135 mm. and over*-size: 140-175 mm; SJDF — under*- size: 0-80 mm, legal*-size: 85-110 mm, and over*-size: 115-175 mm. Sea urchins less than 50 mm are less than two years old and were classified as juveniles (see Pfister and Bradbury, 1996). Harvest and recolonization Recolonization was determined for three periods: summer 1997 (March 1997 Ipostharvestl-September 1997), winter 1997 (September 1997-March 1998 Ipreharvestl ), and summer 1998 (March 1998 [postharvest] -September 1998). In complete harvest sites, monthly recolonization for each time period was calculated as the sum of the number of sea urchins removed each month from each site during the time period divided by the number of months in the time period. A few sea urchins that divers did not harvest in March 1997 because they could not safely be removed from the substrate were excluded from the cal- culation of recolonization for the first time period. Size of recolonizers was based on sea urchins harvested from May 1997 to September 1998. In selective harvest sites where sea urchins were removed only once each year, we calculated monthly recolonization for each time period as the difference between the number of sea urchins counted in each site at the beginning and end of the time period, divided by the number of months in the time period. At the end of the study (September 1998), divers sam- pled destructively for juvenile sea urchins in one 0.80-m- wedge within each circular area within a sampling site. Prior to this time, divers may not have seen very small sea urchins concealed underneath rocks or large sea urchins because sampling was not destructive to avoid disturbing other experiments. We include results on both red and green juvenile sea urchins sampled because of the very low number of red sea urchins sampled and to provide an indication of the microhabitats inhabited by juvenile sea urchins in general in SJC. Data analysis Size-frequency data were grouped into discrete size classes and compared by using chi-square analysis. Sea urchin data collected at each site were correlated over time. Therefore, we used a paired ^test to analyze the effect of harvest on the total number of sea urchins in sites and on the density of sea urchins in each size class. Similarly, we analyzed sea urchin densities in permanent circular areas over time by using repeated measures analysis of vari- ance. In the analysis, site was treated as a random factor and was nested within treatment, which was treated as a fixed factor. Densities of juvenile sea urchins in Septem- ber 1998 were analyzed by using analysis of variance. Sea urchin densities were log-transformed to improve the variance structure for analysis (Zar, 1984). Mean values 50 -, Test diameter (mm) Figure 3 Size-frequency distribution of red sea urchins sampled in the three control sites combined in San Juan Channel (SJC) in September 1997 (n=5.37l and in the western Strait of Juan de Fuca (SJDF) in August 1997 (n=405). Dashed lines indicate minimum and maximum size limits for commercial harvest in each location. generally are followed by one standard deviation in the text. The level of significance for all tests was a = 0.05. Results Size distribution of sea urchins in SJC and SJDF Size distributions of red sea urchins in SJC were strongly skewed to the right and had a modal size of 140 mm (range 20-175 mm. Fig. 3). 4.8% (±3.1%) of sea urchins in circular areas in all sites in early March were <100 mm (under*- size), 50.0% (±13.0%) were 100-139 mm (legal*-size), and 45.2% (±12.5%) were >140 mm (over*-size). Juveniles rep- resented 0.3% (±0.4%) of the population. The modal size of red sea urchins in SJDF was 100 mm (range 30-175 mm. Fig. 3). 12.1% of the red sea urchin population was under*-size, 35.6% was legal*-size, and 52.4% was over*-size. Juveniles represented 2.5% of the population. Effect of harvest on size distribution Sea urchins harvested in the initial size-selective harvest in 1997 were 90-160 mm in diameter (mode 135 mm. 666 Fishery Bulletin 100(4) Lo in CM in ID LO Ln LO LO (D CO O (M ^ CO ""11998 1 80 - 1^ 60 - j 40 - u 20 - .III 1. LnLOLOLOLOLOlOLn CNj M- (£> 00 o CN 'd- LO CD Test diameter (mm) Figure 4 Size distribution of red sea urchins harvested from the three selective harvest sites com- bined in San Juan Channel (SJC) in March 1997 (fi = 1194) and March 1998 l/!=387). Dashed lines indicate minimum and maxi- mum size limits for commercial harvest in SJC. Fig. 4). The harvest consisted of 0.5% (±0.4%) under*-size sea urchins, 59.5% (±6.6%) legal*-size sea urchins, and 39.9% (±9.2%) over*-size sea urchins. Sea urchins har- vested in the second annual size-selective hai^vest in 1998 were 100-155 mm in diameter (mode 130 mm). The com- position of the harvest was similar: 0.0% (±0.0%), 57.1% (±18.7% ), and 42.9%' (±18.7%) of sea urchins were under*-, legal*-, and over*-size, respectively. The initial size-selective harvest did not significantly al- ter the size distribution of the population (P=0.18, Fig. 5) but did significantly reduce the density of legal*-size sea urchins (P=0. 05, Table 2). The cumulative effect of two an- nual size-selective harvests on the size distribution of the population was highly significant (P<0.0001, Fig. 5) and significantly reduced the density of legal*-size sea urchins (P=0.03, Table 2). The modal size of sea urchins in selec- tive harvest sites increased to 150 mm after the second an- nual harvest (Fig. 5). Six months after the second annual harvest, the size distribution of the population remained significantly different from the original (1997 preharvest) size distribution (P=0.0001). Size distributions of sea urchins in selective harvest sites did not differ between fall 1997 and fall 1998 according to large-scale sampling 45 40 35 30 25 20 15 10 5 0 Complete tiarvest > <-t*» -97 pre-tiarvest - 97 5 d post -97 180 d post ■ ■ -D ■ ■ 98 pre-tiarvest ■ -i ■ 98 5 d post ■ -• ■ ■ 98 180 d posl " Selective harvest E Z "I^O n Control 120 Test diameter (mm) Figure 5 Size distribution of red sea urchins in the complete han'est, selective harvest, and control treatments in San Juan Channel (SJC) in early March (before harvest), mid-March ifive days after harvest), and September (180 days after harvest) of 1997 and 1998. Solid lines represent sampling periods in 1997, and dashed lines represent sampling peri- ods in 1998. Cumulative sample sizes for the three sites within each treatment and for all sampling periods illustrated are n = 149, 518, and 1627 for the complete harvest, selective harvest, and control treatments, respectively. Samples are from urchins measured in situ in sampling circles. One site (mid SJIl was excluded from the 1997 early March sampling period for the complete harvest treatment. (P=0.06), averaging 5.8%, 34.3%-, and 60.0% for under*-, legal*-, and over*-size sea urchins, respectively (:7=756). Size distributions of sea urchins in control sites, based on small-scale sampling, were similar throughout the study (Fig. 5) and did not differ between fall 1997 and fall 1998 according to large-scale sampling (P=0.10). Carter and VanBlaricom: Effects of experimental han/ersl on Strongylocentrotus franascanus in northern Washington 667 Table 2 ChanKos in the density (no./m-) of under*-, legal*- and ovor*-size sea urchins as a result of selective harvests in March of 1997 and 1998. Values are averages (±SI)l of the three selective harvest sites and are based on sea urchins sampled in permanent circular sampling areas within each site. See text for explanation of size categories. 1997 1998 Preharvest 5 days after harvest 180 days after harvest Preharvest 5 days after harvest 180 days after harvest Under*-size 0.06 (±0.06) Legal*-size 0.62 (±0.37) Over*-size 0.52 (±0.39) 0.03 (±0.04) 0.14 (±0.10) 0.20 (±0.13) 0.01 (±0.02) 0.18 (±0.24) 0.37 (±0.18) 0.00 (±0.00) 0.14 (±0.18) 0.32 (±0.06) 0.01 (±0.01) 0.04 (±0.06) 0.19 (±0.08) 0.01 (±0.01) 0.07 (±0.05) 0.15 (±0.10) Table 3 Total number of sea urchins (±SD) in San Juan Channel study sites over time Harvests occurred in early March of 1997 and 1998. 1 "Postharvest" data were collected immediately after the harvest by divers performing the harvest. SJI = San Juan Island. 1997 1998 180 days 180 days Site Preharvest Postharvest after harvest Preharvest Postharvest after harvest Complete harvest O'Neal Island 511 19 Mid SJI 639 40 South McConnell 641 37 Average 597 (±74) 32 (±11) Selective harvest North McConnell 261 100 173 181 113 147 Upper SJI 560 206 229 207 92 139 Point Caution 907 228 421 430 226 290 Average 576 (±323) 178 (±68) 274 (±130) 273 (±137) 144 (±72) 192 (±85) Control Yellow Island 547 646 Point George 823 859 Shady Cove 347 266 Average 572 (±239) 590 (±300) Effect of harvest on density Sea urchin densities in the SJC sites initially averaged 1.35/m2 (±0.55, Table 1). Differences in initial sea urchin density between sites were marginally significant (P=0.05) because of the low density of sea urchins at North McConnell. Because of the difference in initial den- sity, we expressed changes from preharvest to posthar- vest densities in SJC sites as percentages of the original population. Initial harvest in March 1997 decreased the number of sea urchins in complete harvest sites by 94.7% (range; 93.7-96.3%) and in selective harvest sites by 66.6% (range: 61.7-74.9%, Table 3). Reharvest of selective har- vest sites in March 1998 reduced sea urchin numbers 46.9% (range: 37.6-55.6%). All decreases were significant (P<0.04). Yields from selective harvest sites averaged 0.95 sea urchins/m' in 1997, and 0.31 sea urchins/m^ in 1998. The number of sea urchins in control sites did not differ between 1997 and 1998 (P=0.77, average 1.41/m2). Small-scale sampling also demonstrated significant changes in sea urchin density as a result of the experimen- tal harvest treatments (Table 4). Initial harvest reduced sea urchin densities by 97.6% and 69.5% in complete and selective harvest sites, respectively. Mid-March (5 days after harvest) densities averaged 0.03/m- (±0.07/m-) and 0.37/m- (±0.45/m^) in complete and selective harvest sites, respectively. Sea urchin densities in complete harvest sites remained low for the duration of the study (<0.04/m2). Sea urchin densities in selective harvest sites increased slight- 668 Fishery Bulletin 100(4) Table 4 Summary of repeated measure analysis of variance results for red sea urchin density. In cases where the degrees of freedom for the tests of significance for the within-subjects effects were adjusted by the Huynh-Feldt epsilon, the unadjusted degrees of freedom are given in parentheses. Asterisks in column five indicate factors significant at the 0.05 level. Power was estimated by using graphs in Zar ( 1984). ( — ) indicates that power was very low for the effect but could not be computed exactly (Sokal and Rohlf 19951. Source Degrees of freedom Mean square F-ratio P-value Observed power Within-subjects effects Year 1 0.24 1.88 0.22 <0.20 Year x treatment 2 0.11 0.88 0.46 — Year x site (treatment) 6 0.13 Season 3.56(4) 0.28 14.34 * 0.000 >0.99 Season x treatment 7.11(8) 0.055 2.78 * 0.033 0.44 Season x site (treatment) 21.34(24) 0.020 Year x season 3.72(4) 0.12 9.42 * 0.000 0.98 Year x season x treatment 7.45(8) 0.070 5.32 * 0.001 0.86 Year x season x site (treatment) 22.34(24) 0.013 Between-subjects effects Intercept 1 22.11 214.92 0.000 1.00 Treatment 2 6.79 14.99 * 0.005 0.95 Site (treatment) 6 0.45 Table 5 Monthly recolonization ino of sea urchins/site) for complete and sele ctive harvest sites from April 1997 thro jgh September 1998. The summer and winter time periods are April-September and October-March. respectively. SJI = San Juan Island. Treatment Site Summer 1997 Winter 1997 Summer 1998 Complete harvest O'Neal Island Mid SJI South McConnell Average 22.7 29.5 16.3 22.8 11.7 13.8 4.7 10.1 16.0 34.2 6.3 18.8 Selective hai-vest North McConnell Upper SJI Point Caution Average 12.2 3.8 32.2 16.1 1.3 3.7 1.5 0.3 5.7 7.8 10.7 8-1 ly in the summer following each harvest (see next section). Sea urchin densities in control sites changed little over time, averaging 1.61/m- (±1.4/m'-) over the study period. Recolonization of harvested sites Recolonization of complete harvest sites averaged 17.2 sea urchins per month (range: 0-68), or 0.04/m- month. The average size of sea urchins recolonizing complete harvest sites was 129 mm (±19 mm, range: 35-175 mm, ;!=824). Recolonization of complete harvest sites was higher during the summer of each year than during the winter and decreased slightly but not significantly over time (P=0.31,Table5). Recolonization of selective hai^est sites was lower than observed in complete harvest sites (average 7.9 sea urchins/month or 0.02/m- month, Table 5). Recolonization was highest during summer 1997, followed by summer 1998. There was no net recolonization during winter 1997. Sites were recolonized to 51.2% of original densities by September 1997 (range: 40.9-66.3%, Table 3). In Septem- ber 1998, after two annual harvests, sites averaged 37.7% of original densities (range: 24.8-56.3%). Juvenile recruitment Juvenile sea urchins were rare in all SJC study sites in March 1997 (0.4% (±0.6%) of sea urchins sampled in Carter and VanBlaricom: Effects of experimental harverst on Strongylocentrotus franciscanus in northern Washington 669 Table 6 Modal size and proportion of juv jniles in red sea urchin populations along the west coast of North America. Location Modal size (mm ) % of juveniles Source Southeast Alaska, shallow habitats 100-120 -35% <60 mm Carney, 1991 Southeast Alaska, deep habitats 140-160 -8% <60 mm Carney, 1991 Vancouver Island, BC 100-120 <5% <30 mm, ~50'7f <100 mm Watson, 1993 Vancouver Island, BC 100-150* -7% <50 mm* Breen et al., 1978 Vancouver Island, BC 80-150* O-ll'/f <50mm* Breen et al., 1976 Strait of Georgia, BC 100-139* 9.5% <50 mm* Sloan et al., 1987 San Juan Channel, Washington 140-144 O.S'/r <50 mm, 4.8% < 100 mm This study Strait of Juan de Fuca, Washington 100-104 2.5%' <50 mm, 24.7% <100 mm This study Northern California -107 -8-12%'<50mm Rogers-Bennett et al., 1995; Smith etal., 1998 San Nicolas Island. Southern Califorr ia 100-120 -7% < 50 mm Cowen. 1983 Southern California 25 and 110 39% < 30 mm Tegner and Dayton, 1981 ' Coniniercially harvested populations. circular areas). Over the entire study period, juvenile red sea urchin densities in circular areas averaged 0.005/m- in control sites (0.6% of the population), 0.00 1/m- in selec- tive harvest sites (0.9% of the population), and 0.0006/m'- in complete harvest sites. Sampling at the larger spatial scale revealed similar results. Juvenile sea urchin density averaged 0.002/m-' in control sites in September of both 1997 and 1998. Juvenile sea urchin density averaged 0.00 1/m- and 0.005/ni- in selective harvest sites in Sep- tember of 1997 and 1998, respectively. Invasive sampling in September 1998 confirmed that ju- venile sea urchins were rare in SJC sites during the study. One juvenile ( 13 mm) red sea urchin and 45 juvenile (2-12 mm) green sea urchins were found in a search of 6.4 m^ of substrate in each site (57.6 m^ total). Density of juvenile green sea urchins sampled did not differ between treat- ments (P=0.79; average 0.79/m- |±0.86|). Only 4.3% of the juvenile green sea urchins were associated with (found within 8 cm of) an adult red sea urchin. 26.7% of juvenile green sea urchins were found in cryptic microhabitats (algal holdfasts), and the remainder were found on algal blades or on top of cobble, boulders, or bedrock. Discussion Size distribution of sea urchins in SJC and SJDF The modal size of sea urchins in SJC is larger than ob- served in most other locations on the west coast (Table 6). The size distribution of red sea urchins in SJC sites is strongly skewed toward large individuals. The slow growth rate of sea urchins in Washington (Lai and Brad- bury, 1998), combined with the fact that SJC was, at the time of our study, a reserve area that has never been sub- ject to significant harvest, indicates that the sea urchin population in SJC is composed primarily of older individu- als. Fewer juveniles are present in SJC and SJDF than in most other areas on the west coast (Table 6). Large recruitment events appear to be rare for red sea urchins in SJC, evidenced by the lack of peaks in the size distribu- tion at small test diameters. Effect of harvest on size distribution The first annual size-selective harvest only slightly affected the size distribution of sea urchins in SJC. The cumulative effect of two annual size-selective harvests was significant, however, significantly reducing the den- sity of legal*-size sea urchins in the population and in- creasing the modal size of the population. Commercial fisheries elsewhere along the west coast have also affected the size distribution of sea urchin populations. Three com- mercial harvests in the San Juan Islands (outside the SJC reserve) over a nine-year period decreased the proportion of legal-size sea urchins by about 73% (Pfister and Brad- bury, 1996). In southern California, two years of harvest reduced the proportion of legal-size sea urchins in shallow waters by 15% (Tegner and Dayton, 1981). In northern California, the modal size of sea urchins in unharvested reserve sites was 108 mm, whereas the modal size of sea urchins in nearby heavily harvested sites with a 76-mm minimum size limit was 73 mm (Smith et al., 1998). On the east coast of Vancouver Island, however, the modal size of sea urchins remained above the legal size limit after 3-1- years of significant harvest (Sloan et al., 1987). The modal size of the sea urchin population and the upper size limit for legal harvest coincided in SJC. Small errors in the measurement of sea urchins by divers led to an incomplete harvest of legal-size sea urchins and inad- vertent "poaching" of many over-size sea urchins (41% of the catch, likely a slight overestimate because sea urchins 670 Fishery Bulletin 100(4) 140 mm in diameter were incorrectly classified as over- size). Both actions would tend to minimize effects of our experimental size-selective harvest. Commercial hai-vest- ers take a much smaller proportion of over-size sea ur- chins (11.2% of catch) and harvest a higher proportion of legal-size sea urchins (Pfister and Bradbury, 1996). Thus commercial harvest should produce larger differences in size distribution than observed in our study. Effect of harvest on sea urchin density Commercial hai"vest in nonresei-ve habitats of the San Juan Islands for over two decades has significantly de- creased sea urchin densities (Pfister and Bradbury, 1996). Both harvest treatments in our study reduced sea urchin densities dramatically and immediately, as expected. The accumulation of older larger individuals in the SJC popu- lation led to high yields in selective harvest sites in 1997. Yields in selective hai"vest sites declined by more than two thirds in 1998 owing to the fishing down of the population (Hilborn and Walters, 1992), low immigi'ation of legal- size sea urchins, and the presence of few under-size sea urchins in the population with a potential of growing to legal size. Recolonization of harvested sites Recolonization of harvested sites occurred primarily by immigration of adult red sea urchins. Recolonization began within one month of harvest and continued for the next 18 months. In northern California sites at 11 m in depth, red sea urchins also immigrated into harvested sites within a short time period (Rogers-Bennett et al., 1998). Complete and selective han'est sites in California were recolonized to 32% and 86% of their original densi- ties, respectively, within nine days. These recolonization rates are much higher than those observed in SJC, pos- sibly because of high sea urchin movement (up to 10 m/h, Rogers-Bennett et al., 1995, 1998) and smaller study plots (64 m- per site) in the northern California location. Sea urchins move less (~1 m/day) in areas with abundant food, such as in SJC, than in areas with little food (Mattison et al., 1977; Carney, 1991). Seasonally high recolonization rates observed each spring in both harvest treatments may be related to feed- ing activity. Spring and summer are seasons when sea urchins feed more frequently (Vadas, 1968) and densities of algae preferred by sea urchins in SJC (e.g. Laminaria, Nereocystts, Ala/'ia, and Costana; Vadas, 1968, 1977) are highest (Carter, 1999). Both factors may have stimulated movement of sea urchins from deeper waters (with lower algal densities) into our sites. On a monthly basis, selective harvest sites were re- colonized at about half the rate at which complete harvest sites were recolonized. Sea urchins recolonizing complete harvest sites were removed monthly, whereas those recolo- nizing selective harvest sites were not. The presence of sea urchins, however, did not inhibit immigration in another study (Watson, 199.3). Differences in sea urchin densities in adjacent habitats or food availability may have contrib- uted to the observed differences in recolonization rates both within and between treatments. Recolonization of selective harvest sites was insufficient to maintain sea urchin populations at the densities and size distributions observed prior to harvest under an an- nual harvest scenario. A site in Washington was recolo- nized to preharvest density and size distribution after 2.5 years (Bradbury, 1991). Recolonization of SJC sites varied substantially in the first year following harvest (range 1-202 sea urchins per site). One year, after harvest, sites varied between 37% and 69% of their original densities. Such variability in recolonization between sites may be common and should be considered when estimating re- colonization rates for commercially harvested areas. Recolonization rates observed in this study should be considered maximum estimates for recolonization of com- mercially harvested areas. SJC study sites were small, and sea urchins were relatively abundant adjacent to sites. Commercial fishermen harvest entire beds of sea urchins before moving on to the next bed and prefer to harvest at shallow depths (Pfister and Bradbury, 1996; Kalvass and Hendrix, 1997; Bradbury'). Thus sea urchins in adjoining habitats at the same depth would likely be harvested, and only sea urchins inhabiting deeper waters would be available for recolonization. Juvenile recruitment The unimodal size distribution of red sea urchins and the rarity of juvenile red sea urchins in SJC suggest variable and infrequent red sea urcliin recruitment in northern Washing- ton. In central California, red sea urchin recruitment was also rare; no major recruitment events occuiTed during a ten-year period ( Pearse and Hines, 1987 ). Size structures of other sea urchin populations north of Point Conception also suggest low red sea urchin recruitment (Table 6). Commercial harvest may negatively affect juvenile re- cruitment in several ways. Commercial harvest decreases sea urchin densities in northern Washington (Pfister and Bradbury, 1996; our study), potentially decreasing fertil- ization rates and larval supply (Levitan et al., 1992). A reduction in adult sea urchin density might also nega- tively affect juveniles in benthic habitats because adult sea urchins may provide associated juveniles with protec- tion from predators (Duggins, 1981; Breen et al., 1985) and provide an increased food source (Tegner and Dayton, 1977, but see Andrew and Choat, 1985). In southern and northern California, 81% and 73%, respectively, of juve- nile red sea urchins were found under the spine canopy of adults (Tegner and Dayton, 1977; Rogers-Bennett et al., 1995). In southern California, juvenile sea urchins were less abundant in harvested sites than in control sites (Tegner and Dayton, 1977). In northern California, juvenile red sea urchins were rare or absent in completely harvested sites but were present in similar numbers in selectively harvested and control sites (Rogers-Bennett et al., 1998). In British Columbia, 69% of juvenile red sea urchins were associated with adults (Breen et al., 1985). Adults may be less important to juveniles in SJC than in other west coast areas because few (4.3%) juvenile green Carter and VanBlancom: Effects of experimental harverst on Stmngylocentrotus franascanus in nortfiern Washington 671 soa unliiiis were found to be associated with adult red sea urchins (our study), juvenile abundance did not difTer by harvest treatment (our study), refuge habitats (cobble, crevices, kelp holdfasts) are abundant, and fast moving sea urchin predators (e.g. fish, lobster) are rare or absent (Breen et al., 1985; Sloan et al., 1987). Management implications The current hai-vest strategy in Washington, applied to sea urchin beds in SJC, results in low yields in the second year of harvest owing to the fishing down of the popula- tion and slow recolonization of harvested areas. Annual commercial har\'est of a single location is probably not economically viable. Commercial hai-vests in Washington and other areas of the west coast are well below levels observed in the late 1980s, and biomass estimates are not available to evaluate the status of stocks in Wash- ington (Bradbury'-'). In such circumstances, options for managing the fishery to conserve, and possibly increase, stocks over the long term include artificially enhancing stocks, redirecting or reducing harvest, and establishing marine harvest refuges (Tegner, 1989; Quinn et al., 1993; Rogers-Bennett et al., 1995). Stock enhancement efforts often confer limited success (e.g. Tegner, 1989) and may negatively impact behavioral and genetic diversity of wild populations (Rogers-Bennett, 1997). Sea urchin harvest in Washington is currently controlled by using harvest quotas, limited entry, and size limits; sea- sonal closures and rotational harvests were also employed until relatively recently (Pfister and Bradbury, 1996; Lai and Bradbury, 1998). Washington also has two marine harvest refuges (Fig. 2). Both rotational harvests and ma- rine harvest refuges increase the probability of long-term survivorship of populations, particularly when recruitment is low and variable, as it appears to be in SJC (Botsford et al., 1993; Quinn et al., 1993; Pfister and Bradbury, 1996). Rotational harvests were discontinued in Washington in 1995, and the size of one of the two marine harvest refuges was reduced substantially in 1998 (Fig. 2). Additional ma- rine harvest refuges in Washington might be established in areas difficult for harvesters to access (e.g. areas hazardous to navigation and far from port) with little decrease in the actual size of harvested areas (StaiT, 1998). Critical to the ef- fectiveness of marine harvest refuges in enhancing recruit- ment and fishery yields outside of refuge areas is the ability of larvae produced in refuges to recruit to areas outside of the refuge (Carr and Reed, 1992). The extended period that sea urchin larvae spend in the plankton (9-19 weeks, Strathmann, 1978) and the short residence time of water in SJC (Thomson, 1981; Hickey et al., 1991 ) suggest that larval dispersal from this and other refuge areas to areas outside of the refuge is highly likely, probably occurring on a scale of tens to hundreds of kilometers. Research on sea urchin re- cruitment patterns (including larval dispersal, settlement, and juvenile survival) in this region is needed to assess the potential for marine harvest refuges in Washington to con- tribute to recruitment of fished stocks in Washington. The current size limits in SJC protect a substantially smaller proportion of the population than in SJDF (SO'X vs. 63"%, respectively). Because there are very few small sea urchins in SJC, the lower size limit protects less than 5% of the population. The upper size limit, established to protect larger "broodstock," currently protects 45% of the population. Size limits in Washington were originally es- tablished to protect the lower and upper 20% of the popu- lation and to allow some individuals to grow through the legal-size "window" under a three-year rotational harvest policy (Pfister and Bradbury, 1996). Considering the small number of juveniles in the population and the return to an annual harvest in 1995, the existing size limits should be reevaluated for their efficacy at protecting local stocks in the context of other harvest control techniques. Acknowledgments This work was conducted as part of the requirements for a Master of Science degree for S. K. Carter in the School of Fisheries (now the School of Aquatic and Fishery Sci- ences) at the University of Washington. David Duggins, Terrie Klinger, Chris Foote, and Ken Chew provided helpful suggestions and guidance on many aspects of the project. Brian Allen cheerfully provided many hours of expert assistance with diving work. Sam Sublett, Martin Grassley, and many others also assisted with data col- lection. Dennis Willows and the staff and faculty at Friday Harbor Laboratories provided logistical support and use of the facilities and equipment. David Duggins and Terrie Klinger provided helpful comments on ear- lier versions of the manuscript. Funding for the project was provided by the U.S. Geological Survey-Biological Resources Division, the Washington Department of Fish and Wildlife, the North Pacific Universities Marine Mammal Research Consortium, the Washington Coop- erative Fish and Wildlife Research Unit, the H. Mason Keeler Endowment for Excellence, the Egtvedt Endow- ment Scholarship, the John N. Cobb Scholarship, and the School of Fisheries, University of Washington. We offer sincere thanks to all. Literature cited Andrew, N. L., and J. H. Choat. 1985. Habitat related differences in the survivorship and growth of juvenile sea urchins. Mar Ecol. Prog. Ser. 27: 155-161. 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 Proceedings of the international symposium on manage- ment strategies for exploited fish populations, p. 409-428. Alaska Sea Grant College Program AK-SC-93-02. Bradbury. A. 1991. Management and stock assessment of the red .sea urchin {Strongylocentrotus franciscanus) in Washington state: periodic rotation of the fishing grounds. J. Shellfish Res. 10(1):233. Breen, P A., B. E. Adkins, and D. C. Miller 1978. Recovery rate in three exploited sea urchin popula- 672 Fishery Bulletin 100(4) tions from 1972 to 1977. Fish. Mar. Serv. Manuscr. Rep. 1446:1-27. Breen, P. A., W. Carolsfeld, and K. L. Yamanaka. 1985. Social behaviour of juvenile red sea urchins, Stro?igy- locentrotus franciscanus ( Agassiz). J. Exp. Mar. Biol. Ecol. 92:45-61. Breen, P. A., D. C. Miller and B. E. Adkins. 1976. An examination of har\'ested sea urchin populations in the Tofino area. Fish. Res. Board Can. Manuscr Rep. Ser 1401:1-23. Carney, D. 1991. A comparison of densities, size distribution, gonad and total-gut indices, and relative movements, of red sea urchins (Strongylocentrotus franciscanus), in two depth regimes. M.S. thesis, 37 p. Univ. California, Santa Cruz, CA. Carr, M. H., and D. C. Reed. 1992. Conceptual issues relevant to marine harvest ref- uges: examples from temperate reef fishes. Can. J. Fish. Aquat. Sci. 50:2019-2028. Carter, S. K. 1999. Ecosystem effects of simulated sea otter predation and commercial sea urchin harvest on nearshore benthic communities in northern Washington. M.S. thesis, 302 p. Univ. Washington, Seattle, WA, Cowen, R. K. 1983. The effect of sheephead iSemicossyphus pulcher) predation on red sea urchin (Strongylocentrotus francis- canus) populations: an experimental analysis. Oecologia 58:249-255. Duggins. D. O. 1981. Interspecific facilitation in a guild of benthic marine herbivores. Oecologia 48:157-163. FAO (Food and Agriculture Organization of the United Nations). 1980. Yearbook of fishery statistics, catches and landings. 1979. Vol. 48. FAO Fish. Ser. 13, 384 p. FAO, Rome. 1981. Yearbook of fishery statistics, catches and landings. 1978. Vol. 50. FAO Fish. Ser 16, 386 p. FAO, Rome. 1992. FAO yearbook, fishery statistics, catches and landings. 1990. Vol. 70. FAO Fish. Ser 38, 646 p. FAO, Rome. 2001. FAO yearbook, fishery statistics, capture production. 1999. Vol 88/1. FAO Fish" Ser 57, 752 p. FAO, Rome. Hickey. B. M., R. E. Thomson, H. Yih, and P H. LeBlond. 1991. Velocity and temperature fluctuations in a buoyancy- driven current off Vancouver Island. J. Geophys. Res. 96(C6): 10507-10538. Hilborn, R., and C .J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics and uncertainty, 570 p. Chapman and Hall. New York, NY. Hoines, L. 1994. 1991 fisheries statistical report, 80 p. Washington State Department of Fish and Wildlife, Olympia, WA. 1998. 1995 fisheries statistical report, 80 p. Washington State Department of Fish and Wildlife, Olympia, WA. Kalvass, P. E., and J. M. Hendrix. 1997. The California red sea urchin, Strongylocentrotus franciscanus, fishery: catch, effort, and management trends. Mar Fish. Rev 59(21:1-17. Lai, H. L., and A. Bradbury. 1998. A modified catch-at-size analysis model for a red sea urchin (Strongylocentrotus franciscanus) population. In Proceedings of the North Pacific symposium on inverte- brate stock assessment and management (G. S. Jamieson and A. Campbell, eds.), p. 85-96. Can. Spec. Publ. Fish. Aquat. Sci. 125. Levitan, D. R., M. A. Sewell and F. S. Chia. 1992. How distribution and abundance influence fertili- zation success in the sea urchin Strongylocentrotus francis- canus. Ecology 73:248-254. Mattison, J. E., J. D. Trent, A. L. Shanks, T B. Akin, and J. S. Pearse. 1977. Movement and feeding activity of red sea urchins (Strongylocentrotus franciscanus) adjacent to a kelp forest. Mar Biol. 39:25-30. 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. Pfister, C. A., and A. Bradbury. 1996. Harvesting red sea urchins: recent effects and future predictions. Ecol. Appl. 6:298-310. Quinn, J. K, S. R. Wing, and L. W. Botsford. 1993. Harvest refugia in marine invertebrate fisheries: models and applications to the red sea urchin. Strongylo- centrotus franciscanus. Am. Zool. 33:537-550. Rogers-Bennett, L. 1997. Marine protected areas and the red sea urchin fishery. In California and the world ocean '97, proceedings of the conference; March 24-27, 1997, p. 412-423. American So- ciety of Civil Engineers, San Diego, CA. 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. Ecol. Appl. 5:1171-1180. Rogers-Bennett, L., H. C. Fastenau, and C. M. Dewees. 1998. Recovery of red sea urchin beds following experimen- tal harvest. In Echinoderms: San Francisco, Proceedings of the ninth international echinoderm conference, San Francisco, CA, August 5-9, 1996 (R. Mooi and M. Telford, eds.), p. 805-809. A. A. Balkema, Rotterdam, and Brook- field, VE. 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. 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 urchin (Strongylocentrotus franciscanus) as an example. Can. J. Fish. Aquat. Sci. 55:1236-1247. Sokal, R. R., and F J. Rohlf 1995. Biometry: the principles and practice of statistics in biological research. 887 p. W.H. Freeman and Company, New York, NY. Starr, R. M. 1998. Design principles for rockfish reserves on the L^.S. west coast. In Marine harvest refugia for west coast rock- fish: a workshop (M. Yoklavich, ed.t, p. 50-63. U.S. Dep. Commer, NOAA-TM-NMFS-SWFSC 255. Strathmann, R. 1978. Length of pelagic period in echinoderms with feeding larvae from the northeast Pacific. J. Exp. Mar Biol. Ecol. 34:23-27. Tegner, M. J. 1989. The feasibility of enhancing red sea urchin, Strongy- locentrotus franciscanus, stocks in California: an analysis of the options. Mar Fish. Rev 51(2):l-22. Tegner, M. J., and P. K. Dayton. 1977. Sea urchin recruitment patterns and implications of commercial fishing. Science 196:324-326. Carter and VanBlaricom: Effects of experimental fnarverst on Strongylocentiotus Iranascanus in northern Wasfiington 673 1981. Population .structure, rccruilnu'iil and mortality 1977. Preferential feeding: an 0[>timization strategy in sea of two .sea urchins ^Stnmgylocentrolus franciscanun and urchins. Ecol. Monogr. 47:337-371. S. purpuratus) in a kelp forest. Mar. Ecol. Frog. Ser. 5: Watson, J. C. 255-268. 1993. The effects of .sea otter (Enhydra liitris) foraging on Thomson, R. E. shallow rocky communities off northwestern Vancouver 1981. Oceanography of the British Columbia coast. Can. Island, British Columbia. Ph.D. diss., 169 p. Univ. Cali- Spec. Publ. Fish Aquat. Sci. 56, 291 p. fornia, Santa Cruz, CA. Vadas, R. L. Zar, J. H. 1968. The ecology oi Agarum and the kelp bed community. 1984. Biostatistical analysis, second edition, 718 p. Pren- Ph.D. diss., 280 p. Univ. Washington, Seattle, WA. tice Hall, Englewood Cliffs, NJ. 674 Abstract— The northwest Atlantic pop- ulation of smooth dogfish iMustetus canis) ranges from Cape Cod, Massa- chusetts, to South Carolina. Although M. canis is seasonally abundant in this region, very little is known about important aspects of its biology, such as growth and reproductive rates. In the early 1990s, commercial fishery landings of smooth dogfish dramati- cally increased on the east coast of the United States. This study investigated growth rates of the east coast M. canis population through analysis of growth patterns in vertebral centra. Marginal increment analysis, estimates of preci- sion, and patterns in seasonal growth supported the use of vertebrae to age these sharks. Growth bands in verte- bral samples were used to estimate ages for 894 smooth dogfish. Age-length data were used to determine von Ber- talanffy growth parameters for this population; K = 0.292/yr, L, = 123.57 cm. and t^ = -1.94 years for females, and K = 0.440/yr, L„ = 105.17 cm, and t„ = -1.52 years for males. Males matured at two or three years of age and females matured between four and seven years of age. The oldest age estimate for male and female samples was ten and six- teen years, respectively. Age and growth of the smooth dogfish (Mustelus canis) in the northwest Atlantic Ocean Christina L. Conrath Virginia Institute of Marine Science College ol William and Mary 1208 Create Road Gloucester Point, Virginia 23062 E-mail address conrathig'vims edu James Gelsleicliter Mote Manne Laboratory 1600 Ken Thompson Parkway Sarasota, Flonda 34236 Jolin A. Musick Virginia Institute of Marine Science College of William and Mary 1208 Create Road Gloucester Point, Virginia 23062 Manuscript accepted 14 June 2002. Fish. Bull. 100:674-682 (2002). The smooth dogfish, Mustelus canis, is a small shark species found throughout the western Atlantic Ocean from Mas- sachusetts to Florida, and in the north- ern Gulf of Mexico, including Cuba, Jamaica, Barbados, Bermuda, Baha- mas, and southern Brazil to northern Argentina. Smooth dogfish are demer- sal and typically are found in inshore continental shelf and slope waters (Compagno, 1984). Several discrete pop- ulations of smooth dogfish likely exist, separated by large geographic areas; and there appears to be little intermi- gration between the different popula- tions (Bigelow and Schroeder, 1948). The northwest Atlantic population of smooth dogfish ranges from Cape Cod, Massachusetts, to South Carolina and migrates seasonally in response to changing water temperatures (Bigelow and Schroeder, 1948; Castro, 1983). Recently, commercial harvest of smooth dogfish has increased on the east coast of the United States. Annual landings were under 80,000 pounds before 1990, over 300,000 pounds in 1990, and increased to around 1 million pounds from 1998 to 2000. In one year (1995) landings exceeded 2.5 million pounds (Fig. 1) (NMFS, 2002). Smooth dogfish have been landed in significant amounts (i.e. over 50 metric tons) in Massachusetts, New Jersey, Maryland, Virginia, and North Carolina (NMFS, 2002). Sharks are often highly susceptible to overfishing because of life history traits that include slow growth, large adult size, late reproduction, and the production of a few large well-formed young (Hoenig and Gruber, 1990). Be- cause of these characteristics, shark fisheries tend to decline drastically af- ter a short time and take long periods to recover (Holden, 1974). The determi- nation of how increased exploitation will affect a shark population, like that of M. canis in the northwest Atlantic Ocean, requires information on the growth and reproductive rates of the species targeted by the fishery. The purpose of this study was to determine the growth rates of smooth dogfish from the northwest Atlantic Ocean by using age estimates derived from ver- tebral growth-band counts. Materials and methods Smooth dogfish were collected from NMFS groundfish and longline surveys, Virginia Institute of Marine Science (VIMS) longline surveys, Grice Marine Laboratory longline surveys, the Massa- chusetts state trawl survey, and by the Massachusetts Division of Marine Fish- Coniath et al Age and growth of Mustelus cams 675 Year Figure 1 Reported smooth dogfish landings from National Marine Fisheries Service com- mercial catch statistics for the Atlantic and Gulf states, from 1981 to 2000. eries (MDMF). Total length (TL), precaudal length (PCL), and male clasper length (CD were measured, and sex was recorded at the time of collection. A section of the vertebral column containing eight to twelve vertebrae was removed from directly under the first dorsal fin and stored frozen. Reproductive samples were taken from smooth dogfish at this time and maturity state was assessed from these samples. Vertebral samples were cleaned, soaked in 70% ETOH for 24 hours, and air-dried for 24 hours. Dried vertebrae were sagitally sectioned through the focus with an Isomet rotary diamond saw ( Bue- hler. Lake Bluff, IL). Afterwards, vertebral sections were affixed to microscope slides with mounting medium and polished with wet fine-grit sand paper to a thickness of about 0.5 mm. The vertebrae were viewed under a binocular dissecting microscope with transmitted light. Vertebral radius was measured from the focus of the vertebra along the axis of the corpus calcarium to the edge of the vertebra (Fig. 2). Total length (TL) was plotted against vertebral radius (VR) to deter- mine if the growth of the vertebra was proportional to somatic growth of the animal and whether the structure was appropriate for estimating growth rate of the animal. Growth patterns of the vertebrae consisted of wide translucent bands separated by narrow opaque bands that extended from the intermedialia to the corpus calcareum (Fig. 2). An angle change — the result of a change in growth rates at this time — was present in the intermedialia ap- Figure 2 An age-.3■^ Mustelus canis vertebra, VR = vertebral radius, AC = angle change denoting birthmark, 1, 2, 3 = age 1, 2, and 3 growth bands, CC = corpus calcareum, and I = intermedialia. proximately 2 mm from the focus of each vertebra and was considered to be a birthmark. Age was estimated by enumerating the narrow opaque bands, which were con- sidered to form annually owing to a slowing or stopping of 676 Fishery Bulletin 100(4) MW MIR = MW/PBW AC :K Figure 3 An agG-2+ Mustelus canis vertebra showing the calculation of the mar- ginal increment ratio, MW = margin width, PBW = previous band width, MIR = marginal increment ratio, AC = angle change. growth during the winter months. This study followed the criteria found in Casey et al. (1985), who defined an "an- nulus" as a mark that appears as an opaque band in the intermedialia and continues as an opaque band into the corpus calcareuni. A random sample of twenty vertebrae from each of ten 10-cm size classes (33-132 cm TL) was read independent- ly by two readers, and a chi-square test was used to test for systematic differences between the ages. The number of observations above the main diagonal of a contingency table of reader one's and reader two's ages was compared with the number of observations below the main diagonal to determine if this ratio was significantly different from 1:1. The percent agreement (PA), i.e. the percentage of ver- tebrae in each length group that were assigned the same ages by both readers, was determined to test for precision between the two readers. Marginal increment analysis verified the annual nature of the narrow opaque growth bands. The distance from the last opaque band to the edge of the margin was measured and divided by the width of the last growth band on the vertebra to determine the marginal increment ratio (MIR, Fig. 3). The margin width was divided by the distance to the angle change or birthmark for age-1 animals. For age- 0 animals, the distance from the angle change to the edge of the vertebrae was measured and divided by the distance from the focus to the angle change. The mean MIR for each month was plotted for juvenile-size animals to determine if there was a yearly pattern in margin width. The length-at-age data were used to generate a von Ber- talanffy growth curve for males and females by using the computer program SigmaPlot (SPSS Inc., 2000) and the Marquardt-Levenberg algorithm to estimate curve-fitting parameters (Press et al., 1986; Marquardt, 1963; Nash, 1979; Shrager, 1970, 1972). To determine if there was a period of the year when smooth dogfish were growing at a faster rate, the mean to- tal length of age-0 and age-1 smooth dogfish was plotted for each month. For this procedure, we used data from several years and we assumed that every year class followed the same general growth pattern during the first two years of life. The mean monthly length of age-0 animals determined for the 1997 and 1998 cohorts was also calculated. Because we did not collect free-swimming age-0 M. canis during May, the largest estimate of birth length (40 cm) was used to minimize the possibility of creating an artificially large growth increment during the summer months. Results Vertebrae were collected from 918 smooth dogfish ranging in size from 33 to 132 cm TL. The relationship between TL and VR for males and females was not significantly differ- ent (ANCOVA, P<0.05); therefore the data for both sexes were combined. The statistically significant relationship (P<0.001) between TL and VR was positive and curvilin- ear (Fig. 4): TL = -0.477(W?)-2 + 17.06 (VR) + 0.807 (;!=833, r2=0.97,P<0.001]. Of the original 918 vertebral samples, 894 animals were aged and vertebrae from 24 animals were found to be an- Conrath et al.: Age and growth of Mustelus cams 677 140 1 TL = -0 4766(VR) • + 1 7 06(VR) + 0 807 r = 0.97 130 ■ n = 833 p<0001 120- o - ,^v! y 110- ,-, tiip^- 0 -, 100 ■ • S iti^wj* E • §■1 lKlr» ^ ^ 90 ■ * ^j^M T" • • M c ,9^^ ^ 80- ca _Jmt^ * o 1- TO- • GO- ^ ^ »!»• . Females 50 • •1^ 40- 30- • , • Males 6 7 8 Vertebral radius (mm) 10 12 Figure 4 Relationship between vertebral radius (VR) and total length (TL) for Mustelus canis. Table 1 Contingency table of reader one's ages versus reader two's ages, the bold numbers are along the main diagonal (where reader one's | age = = reader two's age . Reader one 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 53 1 19 2 2 1 19 1 3 19 o 4 3 2 1 ^ 5 2 6 1 ■a CO 6 7 8 9 10 11 12 13 14 15 2 11 2 14 4 1 2 1 1 1 1 2 2 1 1 2 1 1 2 2 0 1 0 readable. To test for precision, a second reader read verte- brae from a total of 185 animals (twenty from each 10-cm size group, except for the 33-42 cm size group |«=9|) and four vertebrae were found to be unreadable. A contingency table of reader one's versus reader two's ages was made and a chi-square test resulted in a ;|f ^ = 3. 19, which was less than the critical value of x'nor,.i = 3.84; thus the hypothesis of symmetry was not rejected (Table 1). The overall percent 678 Fishery Bulletin 100(4) 09 n 0 9 0 7 ■ 0) o 0 6 H c 05 H (a E CD b 0 4 -I c 1 03 CO S 02 0 1 0 ♦ Age-0 Age > 0 I I ! } /^ 0? ^^ / #" of ^° 90%) with the excep- tion of the largest size class (123-132 cm). At this length it becomes very difficult to interpret the margin of the vertebrae and to distinguish between real growth bands and growth checks. Therefore, the maximum age may be 680 Fishery Bulletin 100(4) 65 62.5 60 57.5 • 55 • 525 50 • 47 5 • 45 ■ 42 5 ■ 40 37.5 35 32 5 ■ 30 — ♦ — Age 0 n = 150 n=91 ^*~^ ,->' .S^ 80 78 .- 76 ,. 74 > in ■72 South Latitude o p" >v ^ ->" ^ V 0 0 * 3° 1 • ■ % o ^ £-5 "5 o ' ' ' O 1 Metric ton Figure 3 Spatial distribution of the bycatch (t) per school set for the sailfish group from the observers' trips during the Euro- pean Union bigeye tuna program. 25 " tr o ^ 20: 15: % 10 : ^ 5^ 0 : o ; West Longitude 5, , , , 30, ,, ,^5, ,,.?),,, ,15, ,, ,10, ,, ,q East 10 ( ; , . • c o -' X •(; \ □ B □ 1 * e { O 1 Metric ton Figure 4 Spatial distribution of the bycatch (t) per FAD set for the sailfish group from the observers' trips during the Euro- pean Union bigeye tuna program. Table 3 Results of the Monte Carlo simulation for a standard vear with and without a moratorium on FAD sets. Monte Carlo estimates are the average bycatch (t) taken by the European purse seiners in the eastern Atlantic Ocean and the con fidence intervals (CI. in tons). "The ratio of billfish to tuna' is the ratio of billfish bycatch to tuna catch (in tons) and corresponding confidence intervals. Group case Marlins Sailfishes Moratorium Without moratatorium Moratorium Without moratorium Monte Carlo estimates (t) 238.79 396.38 37.52 14.86 Lower CI 0.025 192.86 338.25 23.22 2.89 Upper CI 0.975 283.51 450.54 53.91 33.38 Ratio of billfish to tuna 1.718E-03 2.853E-03 2.700E-04 1.069E-04 Lower CI 0.025 1.388E-03 2.434E-03 1.671E-04 2.076E-05 Upper CI 0.975 2.040E-03 3,242E-03 3.879E-04 2.402E-04 tropical areas. If we compare the spatial distribution of the sailfishes with the spatial distribution of the marlin, we find that the bycatch of sailfishes was associated more with school sets than with FAD sets (Figs. 3 and 4). How- ever, the "sampling scheme" adopted for assessing these incidental catches was constrained by the purse seiners' fishing-effort distribution. Consequently, because purse seiners move seasonally from one area to another, there is a lack of information about the presence of billfishes when the fishing area is temporarily abandoned. In both runs of the Monte Carlo simulation we used the same conditional probability about the presence of each billfish group by fishing mode (Table 1), as well as for the observed distribution of the bycatch per set by fishing mode. Results indicated that the bycatch of marlins de- creased during the moratorium from 396 t to 289 t (Table 3 and Fig. 5). This result is a consequence of the larger as- sociation of marlins with floating objects than with school sets. In looking at Table 1, we can see that the occurrence of marlin was around 35% for FAD fishing operations com- pared to only 4'7f for school sets. Marlin were also present in 25% of the seamount sets but, from the small number of sets made on seamounts, we could not determined any ap- parent effect on the total bycatch for this group. Sailfishes were more commonly observed in school sets than in FAD sets (Table 1 ). Thus it appeared that the bycatch of sailfish increased from about 15 t to 38 t with a moratorium on FADs, as summarized in Table 3 and Figure 5. Gaertner et al Bycatch of biilfishes by the European purse seme fishery 687 Table 4 E.stimate.s of the total bycatch (t) taken by the entire purse-seine fishery in the eastern Monte Carlo simulation applied to the European purse seiners. Because of the moratori age of FAD sets was assumed for the last three years. Atlantic Ocean based on the results of the um on FAD sets, a decrease of the percent- Year Total tuna catch (t) Marlins Sailfishes Estimated bycatch (t) Lower CI 0.025 Upper CI 0.975 Estimated bycatch (t) Lower CI 0.025 Upper CI 0.975 1990 211,882 604.50 515.72 686.92 22.65 4.40 50.89 1991 246,110 702.15 599.03 797.89 26.31 5.11 59.12 1992 204,040 582.13 496.63 661.50 21.81 4.24 49.01 1993 249,086 710.64 606.28 807.54 26.63 5.17 59.83 1994 225,646 643.77 549.22 731.54 24.12 4.68 54.20 1995 218,735 624.05 532.40 709.14 23.38 4.54 52.54 1996 194,173 553.98 472.62 629.51 20.76 4.03 46.64 1997 168,826 290.04 234.33 344.41 45.58 28.21 65.49 1998 172,478 296.32 239.40 351.86 46.57 28.82 66.90 1999 157,933 271.33 219.21 322.18 42.64 26.39 61.26 Introducing some elements of uncertainty in the inputs highlighted the large variability of the bycatch estimates (see the values obtained for the lower and upper CI [confi- dence intervals); Table 4. ). However, even including uncer- tainty in the inputs, these values remained very low com- pared with bycatches reported from other fisheries. Based on these results, the total bycatch of biilfishes taken by the entire purse-seine fleet operating in the eastern Atlantic was tentatively estimated. This calculation is supported by the facts that 1) the European fleet is the main com- ponent of the purse-seine fishery operating in this part of the ocean and 2) it is reasonable to assume that other fleets of purse seiners adopted the same fishing strate- gies as the European fleet. With this approach, the ratio of the billfish bycatch per tons of tunas (obtained from the European fleet; Table 3) was raised to the total tuna catch taken by the entire purse-seine fishery (Table 4). To account for the change in fishing strategies caused by the ban on FAD fishing operations, we performed new billfish ratio estimates for the years 1997, 1998, and 1999. These results give an indication of the bycatch of biilfishes in the eastern Atlantic purse-seine fishery (Table 4). Discussion The ecosystem approach to assessing and managing large coastal marine ecosystems has been developing since the early nineties (Sherman and Duda, 1999). To date, with the exception of the central Pacific Ocean (Kitchell et al., 1999), this approach generally has not been used for monitoring large pelagic fisheries in offshore waters. Our study, in presenting data on bycatch of biilfishes taken by the tuna purse-seine fishery, helps to extend this approach to the monitoring of the eastern Atlantic epipelagic ecosystem. IVIarllns 200 150 100 50 0 150 120 90 60 30 0 -A U- au 175 225 275 325 375 425 475 Sailfish ■ Moratorium n Without moratorium :jliii 4=L 5 15 25 35 45 55 65 Bycatch (t) Figure 5 Histograms of Monte-Carlo-generated total billfish bycatch ( marlin group in the upper histogram and sailfish group in the lower histogram) taken by the European Union tuna purse- seine fishery, taking into account whether a moratorium on FAD fishing was applied or not, in the eastern Atlantic Ocean. A purse-seine fishery cannot be assessed purely in terms of the tuna catch. In the eastern Atlantic Ocean, it can be assumed that the large catches of tunas taken by the purse seiners (around 200,000 t per year in the last decade, Table 4) affect the abundance of biilfishes 1) directly, by generat- ing bycatch and 2) indirectly, by increasing or decreasing 688 Fishery Bulletin 100(4) the abundance of predators or competitors, thereby chang- ing the ratio of predators to prey in the trophic chain. That the bycatch of Istiophoridae represents less than 0.021% of the total tuna catch and less than 10% of the to- tal catches of billfishes currently reported for the eastern Atlantic Ocean (assumed to fluctuate around 7000-8000 t per year) suggests that the direct impact of the purse- seine fishery on these stocks is weak. By comparison, pre- vious research has shown that the discards of small tunas and the total bycatch (billfishes, sharks, other fishes, etc.) generated by this fishery were close to 2% and 1.9%, re- spectively (Ai-iz and Gaertner, 1999). Compared with the longline fishery, the European tuna purse-seine fishery generates less bycatch of billfishes than the longline fish- eries targeting tuna (Matsumoto and Miyabe, 2000; Gon- zalez Ania et al., 2001), swordfish (Mejuto et al., 2000), or both species (Cramer, 2000; Marcano et al, 2000). One of the more general implications of our findings concerns the impact of the ban of FADs by the purse-seine fishery on the bycatch of Istiophoridae. Our analysis sug- gests that this moratorium led to a decrease in inciden- tal catch of marlin from 600-700 t to less than 300 t. In contrast, this trend was reversed for sailfishes, but the corresponding bycatch increased only from 25 t to 45 t. Because in the present study we did not take into account different probabilities (see Table 1 ) for each strata, it could be argued that the Monte Carlo simulations lead to only a partial exploration of the uncertainty in the calculation of the total billfish bycatch. However, it would be interesting to consider this source of uncertainty in the future. Conse- quently, the potential for possible regulations at different spatial and temporal scales needs further exploration. Large bycatches of billfishes could affect the food web of the epipelagic ecosystem inhabited by other apex preda- tors. However, the "zero bycatch solution" propounded by some environmentalist groups could accelerate the change in biomass ratios between the different trophic levels of the ecosystem. In a critique of the conventional risk fac- tors (e.g. biological reference points) used to define the risk of extinction in marine fishes, Musick (1999) introduced other interesting criteria, such as rarity of a species, the small distribution range of a species, endemic species, and specialized habitat requirements. As can be seen in Fig- ures 1-4, the range of spatial distribution of the billfishes is very large. It could be argued that billfishes are rela- tively widespread but occupy very specific habitats within their range, and as a consequence, habitat loss could be ex- amined as a risk factor. However there is no clear evidence to support this hypothesis for billfishes. The difficulty of objectively measuring an ecological risk when it concerns unexploited components of the ecosystem must be stressed because of the vagueness of this concept (Antoine et al., 1998). As a consequence, estimating the ecological risk is reduced to the analysis of the impact of a fishing practice on a limited number of symbolic species (e.g. dolphins, sea turtles; Hall, 1996). Nevertheless, there is no reason to believe that these charismatic species play a larger role in the food web of the epipelagic ecosystem than other targeted or nontargeted species. As shown by Kitchell et al, ( 1999), there was no clear conclusion on the ecological role of apex predators (including billfishes) in foods webs of the central North Pacific. Furthermore, if a decision is made to reduce the ecological impact of the bycatch in a given fishery, management actions cannot be focused only on providing full protection for a single spe- cies (Hall, 1996). Although everybody understands what ■'ecosystem overfishing" means, Murawski (2000) high- lighted the lack of consensus for defining this concept and suggested the need for objective metrics that gauge prop- erties associated with the main features of the ecosystem (e.g. production, diversity, and variability). In addition, decision makers need to evaluate manage- ment options that are both scientifically credible and eco- nomically practical regarding the use of the ecosystems. Because billfishes are sold on the local African fish market (Romany et al., 2000), the effects of the fishing on the ecological processes, as well as on human activities, must be evaluated. Although in the past the regulatory process did not account for sharing of gains nor the social costs associated with fishing practices (Antoine et al., 1998 ), evi- dence suggests that the ecosystem approach for managing marine resources should also include these socioeconomic considerations in a multicriteria analysis (Chesson et al, 1999, Sherman and Duda, 1999). Conclusion This study examined the bycatch taken by the European tuna purse-seine fishei-y in the eastern Atlantic Ocean. Results obtained from this fleet have been extrapolated to the entire purse-seine fleet operating in the same areas of this ocean. The main conclusion of this paper is that the direct impact of the purse-seine fishery on the billfish component of the epipelagic ecosystem is weak. Results of this study provide additional information on the effect of the moratorium on FAD fishing. Although some caution is required at this stage, due to limitations of the spatial and temporal sampling cov- erage, it was found that the ban on FAD fishing operations led to a substantial decrease in marlin bycatch. The present state of knowledge allows us to reach only preliminary conclusions. However, it should be borne in mind that inadequate data can lead to the formation of misguided policies. It is clear that detailed information on bycatch is needed to counter the arguments of those who propose total bans on some fishing practices. Conse- quently tuna commissions must continue to pay attention to the collection of bycatch statistics and must encourage fishermen to report incidental catches in their logbooks, at least by large taxa (e.g. billfishes, sharks, etc.). In order to use accurate information for management purposes, we recommend that data from regularly conducted observer programs be used as part of any future research. Acknowledgments This research was funded in part by the European Com- mission (DG XIV) research project n° 96/028: "A study of the causes of the increase in the catches of bigeye tuna Gaertner et al : Bycatch of billfishes by the European purse-seine fishery 689 by the European purse-seiner tuna fleets in the Atlantic Ocean." This study would not have been possible without the help of the Spanish and French tuna companies and the skippers and crews who accepted observers aboard. Literature cited Antoine, L., O. Guyader, and M. Goujon. 1998. Un changement de technique de peche estil compatible avec une peche responsable? L'exemple de la peche au germon iThunnus alalunga) au filct derivant en Atlantique Nord. In ICCAT tuna symposium I J. S. Beckett, ed.), p. 651-660. ICCAT (International Commission for the Conservation of Atlantic Tunasl Coll. Vol. Sci. Papers 50(2). Ariz, J., and D. Gaertner 1999. A study of the causes of the increase in the catches of bigeye tuna by the European purse seiner tuna fleets in the Atlantic Ocean, 70 p. + appendices. Program n° 96/028, European Union, DG XIV, Bruxelles, (Belgium). Chesson, J., H. Clayton, and B. Whitworth. 1999. Evaluation of fisheries-management systems with respect to sustainable development. ICES J. Mar Sci. 56: 980-984. Cramer, J. 2000. Pelagic longline bycatch. ICCAT Coll. Vol. Sci. Pap. 51:1895-1930. Fonteneau, A. 1991. Monts sous-marins et thons dans I'Atlantique tropical est. Aquat. Living Resour 4:13-25. Gaertner, D., M. Pagavino, and J. Marcano. 1996. Utilisation de modeles lineaires generalises pour eval- uer les strategies de peche thoniere a la senne en presence despeces associees dans I'Atlantique Quest. Aquat. Living Resour. 9:305-323. Gonzalez Ania, L. V., C. A. Brown, and E. Cortes. 2001. Standardized catch rates for yellowfin tuna ( Thunnus albacares) in the 1992-1999 Gulf of Mexico longline fishery based upon observer program from Mexico and the LInited States. ICCAT Coll. Vol. Sci. Pap. 52:222-237. Hall, MA. 1996. On bycatches. Rev Fish. Biol. Fish. 6:319-352. Kitchell. J. K., C. H. Boggs, Xi. Le, and C. J. Walters. 1999. Keystone predators in the Central Pacific. In Eco- system approaches for fisheries management, p. 665-683. Univ. Alaska Sea Grant, Fairbanks AK. Klimley A. F, S. B. Butler, D. R. Nelson, and A. T Stull. 1988. Diel movements of scalloped hammerhead sharks, Sphyrna lewini Griffith and Smith, to and from a seamont in the Gulf of California. J. Fish Biol. 33:751-761. Longhurst, A. 1998. Ecological geography of the sea, 398 p. Academic Press, San Diego, CA. Marcano, L., F Arocha, J. Marcano, and A. Larez. 2000. Actividades desarolladas en el programa expandido de IC(^AT para pcces de pico en Venezuela. Periodo : 1998- 99. ICCAT Coll. Vol. Sci. Pap. 51:981-993. Matsumoto, T, and N. Miyabc. 2000. Report of 1999 observer program for Japanese tuna longline fishery in the Atlantic Ocean. ICCAT Coll. Vol. Sci. Pap. 51:729-750. Mejuto, J, B. Garcia-Cortes, and J. M. De La Sema. 2000. Estimaciones cientificas preliminares de desembar- cos de peces de pico capturados en el O. Atlantico y Mar Mediterraneo por la flota espanola de palangre de super- ficie de pez espada, durante el periodo 1988-1998. ICCAT Coll. Vol. Sci. Pap. 51:976-980. Murawski, S. A. 2000. Definitions of overfishing from an ecosystem per- spective. ICES J. Mar Sci. 57:649-658. Musick, J. A. 1999. Criteria to define extinction risk in marine fishes. The American Fisheries Society initiative. Fishery 24 (12):6-14 Pallares, P.. and C. Petit. 1998. Tropical tunas: new sampling and data processing strategy for estimating the composition of catches by spe- cies and size. ICCAT Coll. Vol. Sci. Pap. 48:230-246. Perkins, P C, and E. F Edwards. 1996. A mixture model for estimating (discarded bycatch from data with many zero observations: tuna discards in the east- ern tropical Pacific Ocean. Fish. Bull. 94:330-340. Romany, B., F. Menard, P. Dewals, D. Gaertner, and N. N'Goran. 2000. Le "faux-poisson" dAbidjan et la peche sous DCP derivants dans lAtlantique tropical Est: circuit de commercialisation et role socio-economique. In Peche thoniere et dispositifs de concentration de poisons (J. Y. Le Gall, P Cayre, and M. Taquet, eds. ), p. 634-652. IFREMER (Institut Fran^ais de recherche pour I'exploitation de la mer) Actes de Colloques 28. Shelton, P. A., W. G. Warren, G. B. Stenson, and J. W Lawson. 1997. Quantifying some of the major sources of uncertainty associated with estimates of harp seal prey consumption. Part II: LIncertainty in consumption. Estimates associated with population size, residency, energy requirement and diet. J. Northwest Atl. Fish. Soc. 22:303-315. Sherman, K., and A. M. Duda. 1999. An ecosystem approach to global assessment and man- agement of costal waters. Mar Ecol. Prog. Ser 190:271- 287. 690 Abstract— Snoek [Thyrsites atun) is a valuable commercial species and an important predator of small pelagic fishes in the Benguela ecosystem. The South African population attains 50'7f sexual maturity at a fork length of ca.73.0 cm (3 years). Spawning occurs offshore during winter-spring, along the shelf break (150-400 m) of the western Agulhas Bank and the South African west coast. Prevailing currents transport eggs and larvae to a primary nursery ground north of Cape Colum- bine and to a secondary nursery area to the east of Danger Point; both shal- lower than 150 m. Juveniles remain on the nursery grounds until maturity, growing to between 33 and 44 cm in the first year (3.25 cm/month i. Onshore- offshore distribution (between 5- and 150-m isobathsi of juveniles is deter- mined largely by prey availability and includes a seasonal inshore migration in autumn in response to clupeoid recruitment. Adults are found through- out the distribution range of the spe- cies, and although they move offshore to spawn — there is some southward dispersion as the spawning season pro- gresses— longshore movement is appar- ently random and without a seasonal basis. Relative condition of both sexes declined dramatically with the onset of spawning. Mesenteric fat loss was, however, higher in females, despite a greater rate of prey consumption. Spa- tial differences in sex ratios and indi- ces of prey consumption suggest that females on the west coast move inshore to feed between spawning events, but that those found farther south along the western Agulhas Bank remain on the spawning ground throughout the spawning season. This regional differ- ence in female behavior is attributed to higher offshore abundance of clupeid prey on the western Agulhas Bank, as determined from both diet and rates of prey consumption. Life history of South African snoek, Thyrsites atun (Pisces: Gempylidae): a pelagic predator of the Benguela ecosystem Marc H. Griffiths Department of Environmental Affairs and Tourism Branch Marine and Coastal Management Pnvate Bag X2 Roggebaai 8012 Cape Town, South Afnca E-mail address, mgriffitia'sfriwcapegovza Manuscript accepted 27 Mav 2002. Fish. Bull. 100:690-710 (2002). Snoek (Thyrsites atun) is a medium- size, pelagic predator (max. size 9 kg; Nepgen, 1979a) inhabiting the coastal waters of the temperate Southern Hemisphere; it is found from the sur- face to the seabed, to depths of 550 m (Kailola et al., 1993). Occurring off southern Africa, Australia, New Zea- land, the east and west coasts of south- ern South America, Tristan da Cunha, and the islands of Amsterdam and St. Paul (Nakamura and Parin, 1993), snoek have successfully colonized envi- ronments as diverse as oceanic island, west coast upwelling, and subtropical convergence ecosystems. It is an impor- tant food fish throughout much of its distribution, supporting moderate fish- eries (<1000 metric tons ItJ/yr) off southern Australia, Chile, and Tristan de Chuna, and substantial fisheries 010,000 t/yr) off New Zealand and Southern Africa (Andrew et al.. 1995; FAO. 1997). Southern African snoek have been recorded from northern Angola to Algoa Bay on the South African east coast but are mostly found between the Cunene River and Cape Agulhas, i.e. in the Benguela ecosystem. Thyrsites atun has been an important commercial species in this system since the early 1800s, caught initially with hand lines but al- so trawled after 1960 (Crawford, 1995). Total catch peaked at about 81,000 t in 1978 but dropped substantially with the exclusion of foreign trawlers from the Namibian fishing grounds in 1991 (FAO, 1978, 1981, 1990, 1997). Current annual-catch ranges between 14,437 and 22,920 t (1991-95), and 93% of it is made in South African waters (FAO, 1995). Thyrsites atun is far the most important species caught by the South African commercial line fishery^ (comprising 39^?^ of the 1986-97 catch (National Marine Linefish System-] ); it is also targeted by recreational an- glers, for which catch statistics are not available. Around 40% of the South Af- rican catch (1990-96) is made by com- mercial handline fishermen and 60% by trawlers (Demersal Commercial Data Base^). However, the commercial line-catch may be under-reported by as much as 75% (Sauer et al., 1997); therefore the total catch could be substantially larger. In addition to its fishery significance, T atun is a major predator of anchovy (Engraulis japoni- cus) and sardine (Sardinops sagax) in the southern Benguela ecosystem (Wickens et al., 1992) and has been implicated in top down effects on both prey and consequently zooplankton populations (Verheye et al.. 1998). Life-history information is funda- mental to identifying and assessing fish stocks, as well as to formulating management strategies for their sus- tainable use. Whereas snoek stocks off Australia (Blackburn and Gartner, 1954; Blackburn, 1957; Grant et al., 1978) and New Zealand (Mehl, 1971; ' The South African commercial linefishery consists of about 2500 vessels (5.5-15 m long) that operate on the continental shelf using handline or rod-and-reel. - National Marine Linefish System. 2000. Unpubl. data. Linefish Section, Marine and Coastal Management. Private Bag X2 Roggebaai 8012. Cape Town. South Africa. ^ Demersal Commercial Data Base. 2000. Unpubl. data. Demersal Section. Marine and Coastal Management. Private Bag X2 Roggebaai 8012, Cape Town, South Africa. Griffiths: Life history of Jhyrsitcs ntun 691 29° ■gi-ia^ tV^Js Orange River ^ ^PortNollotti 30" 31 32° 33° 34' 35' 36' Region 0 ■> Hondel500 g), cut open, and sex was determined. Gonads were removed, assigned a macroscopic index of maturity (see Table 1), and weighed to the nearest 0.1 g. Random samples (/?=.5-30) of each macroscopic ovarian stage were fixed in Bouins solution for 48 hours and then stored in 70% ethanol for microscopic verification. They were routinely embedded in paraffin wax, sectioned to 3-7 pm, and stained with haemotoxylin and eosin. Stomach contents were analyzed fresh, and prey items were identified to the lowest possible taxon and weighed (wet) to the nearest 0.1 g. Bait was rec- ognized easily and discarded. The size at 50% maturity (Lr,,,* for males and females was estimated by PROBIT analysis (SPSS, 1992) of the fractions of mature fish (gonad stage 3-i-) per 5-cm length class (midpoint), sampled during the breeding season. Upper and lower 95% confidence limits were calculated by the software package by using maximum-likelihood estimation. Seasonal patterns of reproduction were established by calculating gonadosomatic indices (GSIs) and the monthly percent frequency of each maturity stage, for fish >Lc^q: GSI = gonad weight I fish weight - gonad weight + stomach content weight xlOO. The extent of the spawning area was determined by com- puting the percent frequency of each maturity stage for females i>Lc,Q) that were sampled in each region during peak spawning (June-October). Snoek spawn on the trawl grounds (see below); thus spawning grounds were further delineated by mapping commercial trawl CPUE (stan- dardized by using general linear modelling to account for vessel size ) during June-October ( 1986-97 ) with a 20 x 20 mile grid system. Commercial trawl CPUE was also used to establish the depth distribution of snoek on the trawl grounds, and hence the depth of the spawning area. Sex ratios were tested statistically for significant deviations from equality with chi-square tests. Nursery areas were delineated by comparing the length-frequency distributions of snoek caught in each region 1 ) during pelagic and demersal biomass surveys, 2) by the line fishery, and iii) by Cape gannets, Morus capensis. Because trawling activities during PBSs were aimed at small pelagic clupeoids, it was not possible to use CPUE data from these cruises to analyze spatial pat- terns in juvenile snoek abundance. However, gannets from Lambert's Bay and Malgas Island (Fig. 1) have feeding distributions that, although large, are separated at Cape Columbine and do not overlap to any large extent (Berruti, 1987). Previous studies (Crawford et al., 1992; Berruti et al., 1993; Crawford, 1998) have demonstrated that gan- net diet adequately reflects temporal and spatial patterns in prey abundance. Relative abundance of early-juvenile snoek north and south of Cape Columbine was therefore estimated by comparing rates of snoek consumption by the two colonies. Stomach contents of gannets at the two colonies have been sampled on a monthly basis since 1978 (see Berruti et al., 1993, for methods). Lengths of snoek prey were obtained from undigested specimens, and snoek consumption by the two colonies was compared annually (1978-97) using the mean mass per stomach containing food and percentage frequency of occurrence (i.e. percent- age of stomachs with food that contained snoek). Dietary importance of snoek prey was assessed by per- centage frequency of occurrence (%F), which provides an indication of how often a particular item is selected within a population (Hynes, 1950), and by percentage by mass (%M) (Windell and Bowen, 1978), a measure of the energy contribution of that item (Macdonald and Green, 1983). An index of relative importance ilRI) was calculated for each prey category / as the product of %M, and %F,. To facilitate comparisons of prey importance between analyses (e.g. sep- arated spatially, temporally or according to predator size), this was expressed as a percentage (Cortes, 1997): Griffiths Life fiislory of Thyisitcs atun 693 Table 1 Doscription ofponad maturity stages of snoek (7\rs(7(>s aliin ) in Soutli African waters. Microscopic descriptions are given only for leniaU's. Stage Macroscopic appearance Microscopic appearance 1 Immature and Ovaries appear as clear, pinkish, or translucent orange Some ovaries in this stage consist entirely of pri- resting tubes. Eggs are not visible to the naked eye. Testes mary growth oocytes and others contain oocytes thread-like and clear, to ribbon-like and pinkish white to the early cortical alveoli stage, a atresia of in color. unyolked and /3 atresia of yolked oocytes are also observed. 2 Active 3 Ripe 4 Ripe or running 5 Spent Eggs discernible to the naked eye as yellow granules that do not occupy all available space in ovary. There is very little increase in the diameter of the ovary. Testes are wider, triangular in cross-section, and beige or cream in color. Sperm is present if the gonad is cut and gently squeezed. Ovaries completely opaque and orange to yellow in color They are larger in diameter and eggs occupy all available space. Testes still larger in cross-section and softer in te.xture. They become creamier in color due to considerable quantities of sperm. Ovaries considerably larger in diameter, amber in color with a substantial proportion of hydrated eggs. Sperm is freely extruded when pressure is applied to the abdomen of the whole fish. Ovaries are reduced in size, similar in appearance to stage- 1 ovaries, and have a few yolked oocytes remaining. These yolked oocytes are generally aspher- ical and appear to be undergoing resorption. Testes are shrivelled in appearance and mottled beige and cream in color. A little viscous semen may still ooze from the genital pore when pressure is applied to the abdomen. Primary growth to early yolked oocytes evident. Those at the end of the spawning season show a high degree of « atresia. Primary growth to late yolk-stage oocytes pres- ent. Atresia and postovulatory follicles evident in some. All stages from primary growth to hydrated oo- cytes present. Primary growth to advanced yolk-stage oocj^tes present, but a much lower proportion of yolked oocytes than in previous stages; major atresia of yolked oocytes also present. %IRI, =10077?/,/^//?/,, where n = the total number of food categories considered at a given taxonomic level. Numerical percentage contribution (Pillay, 1952) was excluded because T. atun has a diverse diet and this method would bias the results towards small crustaceans, e.g. euphausiids and amphipods, which are not individu- ally selected. Inshore samples for dietary analysis were collected from the west coast (regions 1-3) while offshore samples were collected from both the west coast (regions 1-3) and western Agulhas Bank (regions 4 and 5). Proportion of stomachs containing food and the mean mass of stomach contents (including fish with empty stom- achs) were used as indices of the rate of prey consumption for adult (>75 cm) snoek. Sexual and spatial differences were tested for statistical significance by using chi-square tests (2x2 contingency tables) for frequency data (i.e. num- bers with and without food ), and a two-way ANOVA with single observations for mean prey mass data. Monthly relative condition (Kn) was calculated for adult snoek (>75 cm) as follows; 'aFn Kn, = ■ where Wy = the gonad and stomach content free weight (g) of the i^^ individual in thej"" month; FL, = the fork length (mm) of the ;"^ individual; a (0.000018) and b (2.80) are constants from the length-weight (gonad and stomach content free) relationship derived from data collected during the present study (all months combined); and n = the number offish sampled in the/'^ month. Snoek accumulate fat as three longitudinal mesenteric deposits along the outer walls of the stomach. Monthly 694 Fishery Bulletin 100(4) 29< 30° 31 > 32° 33° 34° 35° 36° '^-V -^Orange River ::'^ X ■•:■■' ■■■ Port Nolloth •I'THondeklip Bay SOUTH AFRICA •t ■-•'■ St Helena Bay ^Cape Columbine SUMMER (January - February) KEY = 0 • < 10 • 10-50 • >50 t s : i',l ,\ 29° = 9 >i •••• .;. -% Orange River ■ -Oy Port Nolloth 1 ,' . . - - -• H y- i .1 . A 30° r •^ : H '- ' . ■•••ii \: . *^V Hondeklip Bay Z v ■ • ■ • • • ^\:*':.A - N ■ ■ - •s-*'-A = '\' -• ■ • . -A 31° _ • ■ ■ \ ;: \\ SOUTH AFRICA 1 ' , ■ n: '•• v I ^ ■ ': • ..;-;^^ Z N . .:;/ . ■ ■ :-.,\ 32° \ I ' ■-■ .; *..•'') ■.'"»' '•^I'r\. St Helena Bay 33° \ \ i;r,Cape Columbine ^ . ' * ^ Saldanha ^ ■ ■ ■ ■ ■■ t ; >-.? ;,', 34° '•,.\ |r CAPE TOWN WINTER ; ■ •'••iAn S^P^r '- (July - August) : KEY 35° - =0 '•".'*.*■. >^^^ ; • < 10 \ ■ ^ ■ ■ : •10-50 • >50 36° \ A \ T : 15° 16° 18° 19° E20° 15° 16° 17° 18° 19° E20° Figure 2A Mean number ofThyrsites atun per trawl per grid block (5x5 nmi) by season for (Al the west coast (July 1985-Jan 1991; n=l& surveys and 1624 trawls) and (B) the south coast (September 1987-April 1996; n=18 surveys and 1554 trawls) demersal biomass surveys off South Africa. proportions of stomachs with no fat were calculated sepa- rately for adult (>75 cm) males and females. Results Migration Spatial analysis of CPUE from fishery-independent DBSs indicated that T. atun are distributed farther off- shore and farther southeast in winter and spring than in summer and autumn (Fig. 2). Fishery-dependent data support this pattern; commercial catches from the trawl grounds (i.e. >150 m) were highest from June to October, and those in the southern regions (3-6) lagged behind catches in the north (regions 0-2) by approximately one month (Fig. 3). Analysis of trawl CPUE by depth, area, and season (Fig. 4) further confirmed offshore movement during the spawning period (June-October) and further- more revealed that T. atun were most abundant between bottom depths of 150 m and 350 m while on the trawl grounds. Snoek found on the trawl grounds were gener- ally >65 cm FL (Fig. 5A). Line-based catch and CPUE in regions 0-2 was dis- tinctly seasonal; most of the catch was made from April to June (Fig. 3) and a slight southwards progression in peak catch and CPUE occurred within this period (Fig. 3). Line-caught snoek from regions 1 and 2 were substan- tially smaller (50-75 cm) and younger ( 1-3 years; author, unpubl. data) than those from region 3 (80-95 cm; 3-7 years) (Fig. 5B). Line-caught snoek in region 3 were mostly >L5q (Fig. 5B). Monthly catch and CPUE statistics from this region de- picted no trend, indicating that adult snoek are available to line fishermen throughout the year (Fig. 3). Catch rates in region 4 were highest between July and October, which is consistent with the winter-spring south-eastward dispersal of adult snoek evident from trawl data (above). Handline catch and CPUE in regions 5 and 6 were highest during the first half of the year but declined dramatically in winter. Size at maturity The ratio of active (stage-2) to ripe (stage-3) ovaries during the spawning season decreased with fish length Griffiths: Life history of Thyrsites atun 695 33° S 34' 35' 36° SOUTH AFRICA c? ^ Cape St Francis Port Alfred c \ PORT ^S" , . \ ELIZABETl MOSSEL BAY Knysna ^%Jsitsikamma /.. '/■ \^' -*---— i" ' ^T W' ^ : \ -* ■ • "■■/ / ^ '. ' . \ . 1 . ■ -m- 20° 21 22° 23° t AUTUMN (Apnl - May) KEY . =0 • < 10 • 10-50 • >50 _ Untrawlable ground I I 1 1 I ' I ' ' ^ ' i I ' I I ' 24° 25° 26° 33< S 36' SOUTH AFRICA Cape St Francis Port Alfred i Cape Seal MOSSEL BAY Knysna Tsitsikamma PORT ELIZABETH I« ™ •• • ■ ' ^ , s " - yv ;*. t WINTER/SPRING (June and September) KEY = 0 . < 10 • 10-50 • >50 Untrawlable ground hj_L lJ_ 1 I I t I i 1 ' I I 1 1 1 I I I I I i 20° 21° 22° 23° 24° Figure 2B 25° 26° (Fig. 6A), suggesting (as in other teleosts; Griffiths and Hecht, 1995) that substantial proportions of snoek in the smaller size classes undergo partial gonad development during the season prior to maturity. Because these fish are unlikely to spawn, only specimens with gonads developed to at least stage 3 (ripe) were regarded as mature. Given that snoek spawn offshore (see below) and that almost all fish on the trawl grounds (regardless of size) were mature (Fig. 6B). L,,, maturity calculations were limited to fish sampled inshore (<150 m)^: 72.0 cm for males and 73.4 cm for females (Fig. 7). Upper and lower 95% confidence limits were 69.3 and 74.3 cm for males, and 71.3 and 75.4 cm for females. Lr^^ for combined male and female data was 73.0 cm with 95% confidence intervals of 70.0 and 75.2 cm. In all cases the L^g values corresponded with an age of 3 years (author, unpubl. data). In situations where fish migrate to spawn, all fish on the spawn- ing ground are mature. As a result, even if W/i of a particular size class (at-.v cm) in the population are mature, the sampling of that IWc once it had migrated onto the spawning grounds would suggest that IOC'S of that size class are mature. Because mature snoek move back and forth between the spawning and feeding grounds lofTshore and inshore! in regions 0-3 during the spawning season, and could readily be distinguished from immature fish, inshore samples were used to determine size at maturity. 696 Fishery Bulletin 100(4) 40 30 20 10 40 30 20 10 20 Region 0 iso 160 "•-— 400 Region 2 10 bti 300 30 200 20 100 10 O c Region 6 i „. ^ 1 25 Region 3 30 20 10 I .— fl ° '^ ^ * a " JFMAMJJASOND Month : \ji JFMAMJJASOND 2^° Month 200 150 100 50 20 15 10 5 Figure 3 Catches of Thyrsites atun made by South African commercial trawlers (bars I and line fishermen (black circles) summed on a monthly basis in each region and expressed as percentages of the total catch for the period 1985-1997. Mean monthly catch per unit of effort (kg/boat/day) of T! atun caught by South African commercial line fishermen in each region (open circles) is also given (1985-97). Because the bulk (70 ^t) of the line catch in region 5 was made on offshore pinnacles (i.e. over a 72-mile bank), the CPUE trend for this region was based on data from this area. The offshore pinnacles (the 7'2-mile bank being the deepest) are, however, located well inshore of the trawl grounds in that region (see Fig. 1). Spawning Gonadosomatic indices (Fig. 8) and gonad maturity indi- ces (Fig. 9) showed that South African snoek spawn from May to November and that peak spawning occurs from June to October (winter-spring). Monthly male GSIs were similar inshore and offshore, but female values were higher offshore during the spawning season. In addition, a substantial proportion (20-i-%) of offshore females in regions 2 to 5 had hydrated oocytes (stage 4), whereas this gonad stage was rarely observed inshore. Spatial patterns in trawl CPUE from both fishery-independent (Fig. 2) and fishery-dependent (Fig. 10) data collected during winter- spring suggest an extensive spawning gi'ound that encom- Griffiths: Life history of Thymtes atun 697 LU Q. o Region 4 Region 5 Region 3 1— «. ■-- -■—»—«- «. >^« i 100 200 300 400 500 600 Depth class (50 m) June - October November - May 100 200 300 400 500 600 Depth class (50 m) Figure 4 Catch per unit of effort (CPUE; kg/h) per 50-m depth category for Thyrsites atun commercially bottom trawled in each of the seven regions during winter-spring and summer-autumn, 1986-97. passed the western edge of the Agulhas Bank and most of the South African west coast, to a point just north of Hon- dekhp Bay. The predominance of ripe (stage-3) females (both inshore and offshore) throughout the protracted spawning season (Fig. 9) is indicative of multiple spawning (Griffiths, 1997). Moreover, the simultaneous occurrence of postovulatory follicles (POFs) with both primary growth and advanced yolked-stage oocytes (Fig. 11) confirmed that snoek are indeterminate, serial spawners. Sex ratio Sex ratios in regions 0-3 were skewed towards females inshore and males offshore (Table 2), a pattern most pro- nounced during the spawning season (winter-spring). Inshore, this pattern was most evident in the adult size class (>75 cm). Adult sex ratios in regions 0-3 during the spawning season were 2.9F:1M inshore and 1F:2M offshore. Inshore and offshore sex ratios of adult snoek in regions 4 and 5 (Table 2) revealed no clear pattern; although moderately more females were sampled on the trawl grounds of these regions during the spawning season ( 1.6F:1M), the inshore ratio ( 1M:1.2F) did not devi- ated significantly from unity. Nursery areas Monthly length-frequency distributions comprised fairly discrete modes in spite of variation in collection method and sampling period and clearly depicted early juvenile growth (Fig. 12). Young-of-the-year snoek first appeared 698 Fishery Bulletin 100(4) 40 30 - 20 10 Region 0 ^Jhl n = 42 n = 549 go 1 Region 1 50 40 30 20 10 30 20 10 30 20 10 .^A k n = 206 n=968 Region 2 140 :1 597 ii Region 4 ■ n=0 ^ n= 1 017 Region 5 ■ n = 411 E3 n = 1 405 Region 6 ■ n = 82 m n = 47 mil «'.ia,^. Region 3 ■ n = 37 E n = 1 785 _J__Jj .? -? <§^ 'P c? Fork length class (5 cm) Demersal trawl <150m Dennersal trawl >150m M Fork length! class (5 cm) Figure SA Fork-length distributions of Thyrsites atun caught by (A) demersal trawl and (B) handline and midwater trawl in each region. Handline and midwater catches were made shorewards of the 150-m isobath and demersal catches were divided into those made deeper and shallower than this depth. The dotted line indicates length at maturity ^L^,-,). in the diets of gannets and in pelagic trawls during spring (Oct-Nov) at lengths 7-12 cm and grew rapidly to 33-44 cm by the spawning season (winter) of the following year (±3.25 cm/month). Early juveniles (<1 year old) were sampled (with fishing gear) in each of the seven regions (Fig. 5) but formed a larger proportion (by mass and frequency of occurrence) of the diet of gannets at Bird Island (Lamberts Bay) than in gannets of Malgas Island (Fig. 13). Snoek smaller than 70 cm comprised a far higher proportion of the handline and pelagic trawl catches of regions 0-2 than those from region 3 (Fig. 5B). Length frequencies of pelagic trawl, demersal trawl (<150 m), and line-caught snoek suggested that juveniles were also found in reasonable numbers to the east of Danger Point in regions 4 and 5. Thyrsites atun collected in demersal trawls deeper than 150 m were mostly >65 cm, whereas those from shallower bottom trawls included substantial proportions smaller than this length (Fig. 5A). Juveniles <30 cm, although present in pelagic trawls (Fig. 5B), were notably absent from demersal trawls, including those shallower than 150 m (Fig. 5A). Diet Snoek prey on a wide variety of demersal and pelagic organisms, including teleosts, crustaceans, and cephalo- pods (Tables 3 and 4) and show ontogenetic shifts. The diet of T atun sampled inshore of the 150-m isobath along the west coast (regions 0-3) consisted predominantly of pelagic fishes, and crustaceans comprised a smaller but Griffiths Life tiistory of Tliyisitcs attin 699 £ 30 Region 0 M Pelagic trawl <150m E3 Line < 150m Fork length class (5 cm) Figure SB Table 2 Inshore and offshore sex (M:F) ratios of Thyrsites atun during summer-aut umn and winter-spring spawning season) for the period 1994-97. Levels of significance—* (P<0.05), ** (P<0.01), and *** (P<0.001) — were determined by using a chi-square test | (Yates' correction factor was apphec asdf = l). Inshore (<150 m ) Offshore ( >150m) Summer-Aut umn Winter-Spring Summer-Autumn Winter-Spr ing Fork length (cm) M:F n X' M:F n x' M:F n x'^ M:F n X- Regions 0-3 20-49 1:1 203 0.01 1:1.3 1 0 50-74 1:1.2 647 4.5* 1:1.1 524 0.6 1.2:1 105 1.0 2.4:1 228 20.3'*' >75 1:1.8 523 42.0*** 1:2.9 1131 263.0*** 1:1.2 134 1.3 2.0:1 704 61.3*** All sizes 1:1.4 1373 30.5*** 1:2.0 1662 192.3*** 1:1 241 0 2.1:1 933 82.3*** Regions 4—5 20^9 1.3:1 7 0 50-74 1:1.1 53 0 1:1 24 0 No data 1:1 440 0 >75 1:2.0 6 0.1 1:1.2 46 0.2 1:1.6 586 31.1*** All sizes 1:1.2 59 0.4 1:1.1 70 0.1 1:1.3 1033 18.4*** 700 Fishery Bulletin 100(4) A Inshore n = 1 039 c Boffshoren=509 80- _^ r~ t I^H "~~ ^f 60 40 J 1 20- T— 1 m m m ^ • ^ <§^ <^ (§*<§' 'e '^ c? c^ 75 cm). In offshore waters snoek diet consisted almost exclu- sively of teleosts. including pelagic and demersal taxa. The most important prey items were sardine, round herring, and hake (Merluccius spp.) for snoek <75 cm; and hake, sardine and horse-mackerel (Trachurus trachurus cap- ensis) for larger specimens. The main difference between the two offshore diets consisted in the greater importance of sardine (IRI=44.4 vs. 17.2 for snoek >75 cm) and the r •G Male n = 502 1UU /.50 = 72.0 cm 9**^' * • Female n = 1 039 /' 80 >> 2 60 r /.50 = 73,5 cmf 40 d7 20 1 1 .1. .l..WlBlQlniB.i>i*1* 1 1 1 1 i 1 1 10 30 50 70 90 110 Fork-length class (5 cm) Figure 7 Percentage of mature (gonad stage 3-^) Thyrsites atun by 5-cm fork-length class sampled during the spawning season (June-October), 1994-97. Symbols repre- sent the observed data and the lines the fitted models. L^fi= length-at-maturity (cm fork length); n = sample size. lesser importance of hake (IRI=31.8 vs. 60.8 for snoek >75 cm) off the western Agulhas Bank than off the west coast (Table 4). In both areas, large prey species, such as hake and horse mackerel, were more important in the larger (>75 cm) size class. Prey consumption Both indices of prey consumption — proportions of stom- achs containing food and mean mass of stomach con- tents— indicated that within each area and sampling season, adult females consumed more prey than adult males (Table 5); spatial and season trends were also evi- dent. Two-way ANOVA with single observations revealed that differences in the mean mass of stomach contents between sexes (F= 65.9, df=l, F,,., = 10.1) and between areas (F=215.5, df=3, F,.r,(,(a/=9-3* were both highly sig- nificant (P<0.001). Chi-square tests revealed that dif- ferences between the proportions of males and females with stomach contents were highly significant inshore on the west coast (WC) and offshore along the western Agulhas Bank (WAB) in winter-spring, but were not significant inshore on the WC in summer-autumn, or offshore on the WC in winter-spring (Table 5). Female mean stomach content mass was lowest (22.1 g) inshore on the WC in summer-autumn and highest (60.1 g) off- shore on the WAB in winter-spring (spawning season). Female proportion with stomach contents was also lowest (52.5%) inshore on the WC in summer-autumn but high- est (81.2%) in the same area during winter Differences in female proportions with stomach contents between sea- sons inshore on the WC (j^^-gg 9 df=l), between inshore and offshore areas of the WC in winter-spring (^^=130, df=l) and between the offshore WAB and inshore WC during winter-spring (^-=26.3, df=l) were highly sig- Griffiths Life f:istory of Thyrsttes atun 701 10 10 Females Inshore Offshore 39 25 1 215 ^130 186 113 I 13 63 86 B Males 65 72 49 39 ><, Y J J A S O N D Month Figure 8 Mean monthly gonadosomatic indices (GSI; ±2 standard errors) for mature (>75 cm) (A) female and (B) male Thyrsites atun sampled both inshore (cir- cles) and offshore (squares), 1994-97. Inshore and offshore sample sizes are given above and below the error bars, respectively. nificant (P<0.01 ), but differences between the two offshore areas — WC and WAB — in winter-spring were not. Mean stomach content mass was, nevertheless, substantially higher (60.1 g vs. 44.6 g.) offshore on the WAB than off- shore of the WC during winter-spring (Table 5). Condition Relative condition (Kn) of adult snoek depicted a clearly seasonal cycle: highest between March and May, declin- ing steeply through June to October (spawning season), increasing again from November through to March (Fig. 14). The proportion of stomachs without any fat was inversely related to Kn (Fig. 14): it was lowest from March to May but increased dramatically during the spawning season. The rate of fat decline was, however, higher in females than males; by September 90^7^ of females and 58% of males had no mesenteric fat reserves. Discussion Crawford and De Villiers (1985) postulated that the snoek of the Benguela ecosystem comprise a single stock that undergoes a seasonal longshore migration — moving southwards into South African waters to spawn in winter and returning north, as far as southern Angola, in spring- summer. Although this theory has become widely accepted (Crawford et al., 1987; Crawford, 1995), this study showed that adult snoek are available to South African line fisher- 702 Fishery Bulletin 100(4) J FMAMJ JASOND Month Q Inactive (1) : Ripe/running (4) Q Active (2) ■ Spent (5) □ Ripe (3) Figure 9 Monthly percentage of macroscopic ovar- ian stages for adult Thyrsites atun (>75 cm) sampled (A) offshore (f!=369) and (B) inshore (n = 1172). men throughout the year, and that the seasonal availabil- ity of adults on the trawl grounds results from an offshore spawning migration rather than southward movement from Namibian waters. Based on these results, niter alia. separate nursery areas and egg and lai-val distributions north and south of the cold upwelling cell (25-27°S) off southern Namibia, Griffiths (in press) concluded that Ben- guela snoek exist as two subpopulations, and have limited exchange. The life history of South African T. atun is summarized by the conceptual model presented in Figure 15. Movement patterns were inferred largely from spatiotemporal trends in fishery-dependant catch and effort data. Considering that snoek is the most important line fish on the South African western seaboard and that it is a bycatch in the trawl fishery, catch trends are unlikely to have been biased by switches in targeting. Moreover, even though line-fish catches are under-reported (Sauer et al., 1997), they have been demonstrated to accurately reflect seasonal trends in the abundance of migratory species (Griffiths and Hecht, 1995). Given the offshore movement of adult fish during the spawning season and the associated ovarian condition (greater GSls and the presence of hydrated oocytes), it is concluded that South African snoek spawn offshore between the 150- and 400-m isobaths. Although most of the inshore catch was line-caught, the absence of stage-4 fe- males from trawls made shallower than 150 m during the spawning season (n=46, including 32 ripe |stage-31 speci- mens) indicates that this observation is not an artifact of gear selectivity. Spatial patterns in trawl CPUE from both fishery-independent (Fig. 2) and fishery-dependent (Fig. 11) data, collected during winter-spring, suggest an extensive spawning ground that encompasses the western edge of the Agulhas Bank and most of the South African west coast, to a point just north of Hondeklip Bay. Presence of snoek eggs and larvae (off South Africa) from the southern tip of the Agulhas Bank to Hondeklip Bay and their great abundance between the 200-m and 600-m isobaths (Olivar and Fortuiio, 1991; Olivar and Shelton, 1993) corroborate offshore spawning and an extensive spawning ground. The occurrence of snoek preflexion larvae in region 6 (Wood, 1998 ) indicates that spawning does occur farther to the east, but the relatively low abundance of adult fish and spawning products suggests it is not an important spawning area. The sex ratio of adults (>75 cm) on the west coast (regions 0-3) was skewed towards males on the spawning grounds (2M;1F), and towards females further inshore (2.9F:1M), particularly during the spawning season. Based on the differential loss of intestinal fat, it is evident that the energetic demands of spawning are higher in females than in males. The higher rate of prey consumption by females during winter-spring (i.e. spawning season) than during summer-autumn and the greater prey consumption by females than by males during the spawning season indicate that they are able to enhance their spawning effort with exogenous energy. It is therefore postulated that because snoek are indeterminate serial spawners, females on the west coast move inshore between spawning events, where their principal prey — Sardinops sajax and Engraulis japonicus — are more abundant. Females spawning off the western Agulhas Bank do not appear to move inshore to feed between spawning events, according to near equal sex ratios and low inshore catch rates in winter-spring (region 5). Although line catches in region 4 were highest during the spawning season, which at first glance may appear to be contradictory, CPUE was an order of magnitude lower than in region 3 or region 5, indicating relatively low inshore abundance at that time of the year High female mean stomach content mass (60.1 g) and the significantly higher proportion of females than males with food (shared only with the inshore feeding ground of the WC ) suggest that the offshore WAB functions as both spawning and feeding ground during winter-spring. In addition, PBSs have revealed that clupeids — adult sardine and round herring — are more abundant near the shelf edge of the WAB in winter than along the WC (Coetzee'^), and dietary comparison (this study) showed that clupeids indeed represent a larger component of the offshore diet Coetzee, J. 2000. Personnal commun. Pelagic Section, Marine & Coastal Management, Private Bag X2, Roggebaai, 8012, Cape Town. Griffiths: Life hiistory of Thyivtes atun 703 S 29" 30" 31" 32° 33"'b- 34' 35"! 36° ^ m % $ Orange River -? • • ^ Port Nolloth : \ s • '.S i : • •" • • • \ : • • • '•■ _• • \ Hondeklip Bay : .* ■f • • • •' k : • "! • • • •- \ 1 F- • '• • • *^ \ s •' • • • •,' A h ^ • • *\ SOUTH AFRICA • •> • • ' St Helena Bav •, -'J • • • 1 ;ape Columbine cape I^Saldanha St Francis Port ; • \ • 1 • '• • T>, " MOSSELBAY cape Seal \ PORT '^"^^ • w T^^CAPE TOWN Jl^,_X Knysna.Z_Tsitsika^l^^^ -^ J- • •'.' \\ mOape nanqKiip ^' i oici vai i\»,^r^. Vt> ^ ^^ Point „?-■ 1 • 'T ^-Ji '-■!-■' y-^- -- _, • ^* • ■"• •v> L r!^ P_j; ^ -'/ -; -.- _," O" 'f "''' ■ * ^« '.-' ?^ P^ f-~- ' .— _.' ■ .- '- 'I' ■. • • ' ■? • •' V ■ ^ -'. \ -.- ; ■ 0-0.09 • 0.1 - 0 9 2 « • ^9 • • *, • ;. ' .' • ■i <• • • • • ,'.' •r • 1-10 A • « ,f • • • •' •,- • • • 11-20 1 • • ^; • • ^ •, '• Tl 1 1 1 1 1 1 1 1 1 1 1 1 1 II 1 1 1 1 1 < 1 1 1 1 III 1 1 1 1 1 1 1 1 1 II 1 II 1 Ml II 1 II 1 -ft 1 1 1 1 1 1 1 1 1 1 1 1 II 1 1 1 1 II 1 II 1 1 1 1 \ II 1 1 1 1 1 1 1 1 1 il 1 1 1 1 1 1 1 1 1 1 MM 15" 16° 17" 18° 19" 20° 21° 22° 23° 24° 25° 26° Figure 10 Catch per unit of effort (kg/h) by 20-niile block ( 1986-97) for Thyrsites atun caught by commercial trawls in the southern Benguela system during the spawning season. of the WAB than offshore along the WC, where hake was dominant. Sex ratios and feeding patterns of adult snoek in the present study were confirmed as persistent biological features of the western Agulhas Bank by experimental trawl surveys on the spawning ground of region 4 during August 1999 and August 2000. The principal prey taxon for 1875 snoek with a sex ratio of 1M:1.2F (P<0.001) was Clupeidae (67% by weight); mean stomach content masses were 58.2 g for females and 27.6 g for males. Even though young-of-the-year snoek were sampled (with fishing gear) in each of the seven regions, they formed a larger proportion (by mass and frequency of occurrence) of the diet of gannets at Bird Island (Lamberts Bay) than in gannets at Malgas Island. Either early juveniles are more abundant north of Cape Columbine than between Cape Columbine and Cape Infanta or they are selected at a higher frequency by gannets in the former area. Higher propoi ions of snoek smaller than 70 cm in the handline and pelagic-trawl catches of regions 0-2 than in catches from region 3 support the first suggestion, confirming that the area north of Cape Columbine has an im- portant nursery function. Length frequencies of pelagical- ly trawled, demersally trawled (<150 m), and line-caught Figure 11 Photomicrograph of transverse section of Thyrsites atun ovary stained with haemu.oxylin and eosin. showing co-occurrence of postovulatory follicles (POF) together with primary growth (PGl and advanced yolk- stage oocytes (YO). snoek, nevertheless suggest a second nursery area to the east of Danger Point. But because handline catches in 704 Fishery Bulletin 100(4) 50 30 10 20 10 20 -10 I i 1 r I 20 10 20 10 20 10 October n = 21 April November 0 n = 44 ^ n = 277 May December n = 46 June January r7 = 58 iBftuHMnttmilaHMmi , I iHa February n n = 39 ^ n=72 ^JJ^JJ-UJJJJ^^/l I August n= 103 n= 120 n = 36 n = 27 March 4 8 12 16 20 24 28 32 36 40 44 48 Fork iengtti class (1 cm) 0 Gannet —^ Pelagic trawl • Demersal trawl 4 8 12 16 20 24 28 32 36 40 44 48 Fork length class (1 cm) Figure 12 Monthly fork-length distributions of Thyrsites atun <50 cm caught by midwater trawls (1985-971. demersal trawls (1985-97), and Cape gannets ( 1978-97) in the southern Benguela system. ;! = sample size. Dotted line at 25 cm is included as a reference point. regions 4 and 5 amounted to only 10% of those in regions land 2 (1985-97) and given that juvenile snoek made a substantially lower contribution to the diets of gannets foraging in this area, the area north of Cape Columbine is regarded as considerably more important to juvenile snoek. Thyrsites atun in demersal trawls made deeper than 150 m were mostly >65 cm, whereas those from shallower bottom trawls included substantial proportions smaller than this length. It is therefore concluded that the primary nursery ground is situated on the west coast, north of Cape Columbine and in water shallower than 150 m; a secondary nursery exists to the east of Danger Point. The absence of specimens <30 cm from demersal trawls, particularly those made shallower than 150 m, and their presence in pelagic trawls, indicates that 0-year-old snoek are largely epipelagic but become pelagic with growth (at ca. 8 months old). The northward flowing Benguela Current provides a mechanism for transporting epipelagic snoek eggs and larvae (De Jager, 1955; Olivar and Fortuno, 1991) from the spawning ground to the proposed nursery areas. The Benguela current is found offshore (generally between the 150- and 400-m isobaths) and flows along the western edge of the Agulhas Bank and up the South African west coast; rate of flow is fastest (typically 25-80 cm/s) between Cape Point and Cape Columbine, where it develops into a shelf edge jet (Shannon and Nelson, 1996). Although water movement is primarily parallel to the shelf, inshore advec- tion occurs to some degree due south of Danger Point and to a greater extent between Cape Columbine and Hon- Griffiths: Life fiistory of Thy/site'; atun 705 Figure 13 Annual frequency of occurrence (%F; dotted line) and mean mass (solid line) of Thvrsites atun in the stomachs of Cape gannets from Bird Island (circles) and Malgas Island (squares), 1978-97. n=number of gannet stomachs examined. 1.15 1.05 m 0.95 Figure 14 Mean monthly relative condition (Kn) and the proportion of adult Thyrsites atun (>75 cm) without intestinal fat reserves, in the southern Benguela Eco- system (1994-97). deklip Bay (Shelton and Hutchings, 1990; Shannon and Nelson, 1996). Snoek hatch about 2 days (50 hours) after fertilization (De Jager, 1955), are phytoplanktivorous from first feed- ing (3.5 mm and 3-4 days after hatching) until 8 mm long (standard length), after which they prey largely on the larvae of other fishes (Haigh, 1972). Total abundance of fish larvae in the southern Benguela is highest in spring and summer (Shelton, 1986); snoek larvae spawned dur- ing the winter and spring months are therefore assured of an abundant supply of food. Following northward advec- tion in the jet current, young-of-the-year anchovy, round herring, and sardine, which share the northern nursery area with snoek, move shoreward across the shelf, with a concomitant increase in size (Armstrong et al., 1987; Roel and Armstrong 1991; Hampton, 1992; Roel et al., 1994; Barange et al., 1999; Van der Lingen and Merkle, 1999). They arrive at the coast from February through through June at total lengths of 5-11 cm, and, unlike snoek, which are found on the nursery grounds for 2-3 years, immedi- ately begin a southwards migration back onto the Agulhas Bank, where they were originally spawned. Coupled with earlier spawning, rapid growth ensures that 0-year-old snoek are large enough (ca. 35 cm TL) to partake with the older juveniles in this seasonally abundant food source. Substantial increases in line catches during the period of clupeoid recruitment suggest that juvenile snoek follow their prey inshore. Differences in the size and age com- position of catches north and south of Cape Columbine, however, indicate that they do not move southwards with them and onto the Agulhas Bank. The large contribution of lanternfish to the diets of snoek <50 cm is attributed to the fact that most of these fish were sampled during the second half of the year, i.e. when juvenile clupeoids (i.e. 706 Fishery Bulletin 100(4) Table 3 Stomach contents of snoek ( Thyrsites atun ) sampled in shore of the 150-m isobath along South African west coast (1994-97 .M = percent mass, F = percent frequency of occurrence, IRI = percent index of relative prey importance at the species and at a higher | (bold) taxonomic level. ^ = pel agic and ° = demersal species. 5-24 cm FL 25-49 cm FL 50-74 cm FL >74 cm FL Taxon and (/! = 11) (n =2121 (;! =540) ( 1 = 1,069) prey item M F IRI M F IRI M F IRI M F IRI Annelida <0.1 0.2 <0.1 0 0 0 Polychaeta <0.1 0.2 <0.1 Crustacea 27.7 54.5 24.7 19.4 45.5 12.6 22.0 39.5 14.3 7.4 5.9 0.5 Amphipoda Themisto gaudichaudi^' 20.9 9.1 14.2 17.8 42.0 32.4 1.42 10.9 1.0 Brachyura Megalopa larvae^ <0.1 0.2 <0.1 Euphausiacea Euphausia lucens^ 6.8 45.5 23.0 1.3 3.6 0.2 15.0 27.4 26.0 0.3 1.6 <0.1 Isopoda Paridotea ungulata^ <0.1 <0.1 <0.1 <0.1 0.1 <0.1 Macrura Lobster pueruliP <0.1 0.2 <0.1 <0.1 0.3 <0.1 Pasiphaea spp.^ <0,1 0.5 <0.1 0 0 0 MysidaceaP <0.1 0.9 <0.1 0.3 1.0 <0.1 <0.1 0,1 <0.1 Stomatopoda Pterygosquilla armata^ 5.0 10.4 3.3 7.1 4.5 1.1 Unidentified crustaceans 0.3 0.9 <0.1 0.1 0.6 <0.1 <0.1 0.3 <0.1 MoUusca Cephalopoda 0.3 2.2 <0.1 0.2 16 <0.1 Inioteuthis <0.1 0.2 <0.1 0 0 0 Loligo vulgaris reynaudi 0 0 0 <0.1 0.3 <0.1 Lolliguncula mercatoris? 0.2 1.5 <0.1 <0.1 1.1 <0.1 Lycoteuthis? 0 0 0 <0.1 0.1 <0.1 Sepia australis 0.1 0.4 <0.1 <0.1 0.1 <0.1 Todaropsis eblanae 0 0 0 <0.1 0.1 <0.1 Unidentified squid <0.1 0.2 <0.1 0 0 0 Teleostei 72.3 63.6 75.3 80.6 76.0 87.4 77.8 66.9 85.7 92.4 95.7 99.5 Engra ill is japon iciis^ 6.1 10.7 2.8 30.9 19.6 38.3 26.9 44.0 38.6 Sardinops sagax'' 8.9 6.3 2.4 18.9 6.8 11.7 43.1 33.8 47.4 Etrumeus ivhiteheadi^ 5.1 1.8 0.4 6.5 3.2 1.3 2.3 3.5 0.3 Lampanyctodes hectoris^ 68.4 9.1 46.5 37.2 21.4 34.4 0.8 1.9 0.1 0.4 1.3 <0.1 Maurolicus muelleri^ 1.0 2.7 0.1 0.8 1.5 0.1 <0.1 <0.1 <0.1 Trachurus t. capensis^ 4.1 2.4 0.6 3.9 6.8 0.9 Scomber japonicus^ 0 0 0 0.5 0.1 <0.1 Lepidopus caudatus^ 0.3 0.4 <0.1 0.1 0.5 <0.1 Thyrsites atun'' 0 0 0 <0.1 0.1 <0.1 Sufflogobius bibartus'' 1.3 1.0 0.1 0.1 0.2 <0.1 Gonorhynchus gonorhynch is 0 0 0 0.2 1.1 <0.1 Gnathophis capensis <0.1 0.2 <0.1 0.3 2.9 <0.1 Merluccius spp.'' 5.0 1.8 0.4 5.7 5.2 1.9 3.1 2.6 0.3 ClinidaeD <0.1 0.2 <0.1 <0.1 0.1 <0.1 Emmelichthys nitidusr^ 0 0 0 0.5 0.1 <0.1 Spondyliosoma emarginatum^ 0 0 0 0.3 0.2 <0.1 Unidentified Teleostei 4 54.5 16.3 17.4 35.7 28.8 8.5 29.3 15.7 10.7 33.0 11.5 Griffiths Life fiislory of Jhyisitcs atun 707 Table 4 Stomach contonts of snock iThyrsites atun) sampled oH'shore of tlio 150-m isobath along thu west coast (regions 1-3) and the western Agulhas Bank (regions 4 and 5) (1994-97). M = percent mass, F = percent frequency of occurrence, IRI = percent index of relative prey importance at the species and at a higher (bold) taxonomic levels. '' = pelagic and '^ = demersal species. West coast Western Agulhas Bank 50 -74 cm FL >75 cm FL 50 - 74 cm FL >75 cmFL n = 147) (n =447) tn =336) (n = 367) Taxon and prey item M F IRI M F IRI M F IRI M F IRI Annelida <0.1 <0.1 <0.1 Polychaeta <0.1 <0.1 <0.1 ' Crustacea 1.3 4.8 0.1 1.5 6.9 0.1 Amphipoda Themisto gaudichaudi^ 0.1 0.2 <0.1 Brachyura Parapagurus dimorphus^ 0.1 0.7 <0.1 0.1 0.4 <0.1 Euphausiacea Euphausia lucens^ 1.1 3.4 0.2 1.3 5.3 0.4 Macrura Funchalia?^ <0.1 0.7 <,0.1 <0.1 0.7 <0.1 Stomatopoda Pterygosquilla armata^ 0.1 0.4 <0.1 MoUusca Cephalopoda 1.4 2.7 <0.1 1.7 1.9 <0.1 Lycoteuthisl 0.1 0.2 <0.1 Sepia australis 0.1 0.4 <0.1 Todaropsis eblanae 1.2 1.8 0.1 1.7 1.4 0.1 Unidentified squid <0.1 0.2 <0.1 <0.1 0.5 <0.1 Teleostei 98.7 97.3 99.9 97.1 96.4 99.8 100 100 100 98.3 99.4 99.9 Engraulis japonicus^ 0.1 0.7 <0.1 <0.1 0.2 <0.1 0.2 0.6 <0.1 0.1 0.3 <0.1 Sardinops sagax^ 25.7 23.8 36.3 17.0 19.2 17.2 61.4 45.2 79.4 25.6 28.3 44.4 Etrumeus whiteheadi^ 18.5 13.6 14.9 4.9 5.4 1.4 19.6 22.9 12.9 6.3 11.2 4.3 Lampanyctodes hectoris^ 4.5 12.2 3.2 6.0 18.1 5.7 1.0 4.1 0.3 Maurolicus muelleri^ 2.5 7.5 1.1 1.4 13.0 1.0 0.4 5.4 0 1 0.2 3.5 <0.1 Trachurus t. capensis^ 13.2 12.8 8.8 1.4 0.6 <0.1 14.2 10.1 8.8 Scomber japonicus^ 5.3 0.7 0.2 3.0 1.6 0.1 Lepidopus caudatus^ 2.6 2.7 0.4 1.2 0.9 0.1 9.3 6.0 3.4 Thyrsites atun^ 0.8 0.3 <0.1 Gonorhynchus gonorhynchus^ 2.6 2.7 0.2 4.7 6.8 2.0 Emmelichthys nitidus^ <0.1 0.3 <0.1 Diaphus hudsoni^ <0.1 0.2 <0.1 <0.1 0.3 <0.1 Scomberesox saurus '' 1.9 0.9 0.1 0.2 0.3 <0.1 Merluccius spp. ^ ' 31.4 15.0 27.8 47.6 24.4 60.8 4.4 3.3 0.4 28.0 18.5 31.8 Caelorinvhus simorhynchus D 0.1 0.7 <0.1 Sebastes capensis^ 0.5 0.7 <0.1 0.4 0.4 <0.1 0.2 0.5 <0.1 Zeus capensis^ 0.7 0.2 <0.1 Paracallionymus costatus^ 0.6 0.7 <0.1 1.8 3.6 0.3 1.5 1.6 0.2 Unidentified Teleostei 6.9 37.4 15.2 2.8 24.8 3.7 8.1 30.4 7.0 3.2 23.4 4.6 Two species of hake occur in South African waters— M. paradoxus and M. capensis-but owing to difficulty differentiating partially digested speci- mens, they were not analyzed separately. 708 Fishery Bulletin 100(4) S 29' so- ar 32< 33- 34' 35° 36" NAMIBIA ""% "vOrange River .:, V- a Spawning grounds (June -October) "'\ o Eggs and larvae m\ ■ Juveniles ~"vo«iV, Egg and larval drift V W% k ■a SOUTH AFRICA N f r .^ 3CAPE TOWN c^ ■■>/ -^ '75 cm) ma e(M) and female iF) 7^ atun sampled on the South African west coast (WC; | regions 1-3) and western Agulhas Bank iWAB ; regions 4 and 5). x- values were Dased on the freq uencies of males 1 and females with and without stomach contents (2x2 | contingency tables) in each area (Yates' correction factor | was applied as df = = 1). Significance level s are given by * iP<0.05), ** (P<0.01), and *** (P<0.001). Mean Area Sex ;! '"r X- mass (gl Summer-autumn wc,„,„„. M 197 .50.2 0.18 13.3 F 383 52.5 22.1 Winter-spring wc,„,,.,„ M 130 52.4 6.6* 34.0 F 729 81.2 43.0 ^'-'(IfTshore M 374 61.8 2.5 39.0 F 193 69.0 44.6 WAB„,T^„„. M 212 49.5 15.2*- 38.4 F .336 67 60.1 Literature cited Andrew, T. G., T. Hecht, P. C. Heemstra, and J. R. E. Lutjeharms. 1995. Fishes of the Tristan da Cunha group and Gough Island, South Atlantic. Ichthyol. Bull. J. L. B. Smith Inst. Ichthyol.63, 43p. Armstrong, M. J., A. Berruti, and J. Colclough. 1987. Pilchard distribution in South African waters, 1983- 1985. S. Afr. J. Mar Sci. 5:871-886. Badenhorst, A., and M. J. Smale. 1991. The distribution and abundance of seven commercial trawl fish from the Cape south coast of South Africa, 1986- 1990. S. Afr. J. Mar. Sci. 11:377-393. Barange, M., I. Hampton, and B. A. Roel, 1999. Trends in the abundance and distribution of anchovy and sardine on the South African continental shelf in the 1990s, deduced from acoustic surveys. S. Afr. J. Mar Sci. 21:367-391 Berruti, A. 1987. The use of Cape gannets in management of the purse- seine fishery of the Western Cape. Ph.D. diss., 340 p. Uni- versity of Natal, Pietermaritzburg, South Africa. Berruti, A., L. G. Underbill. P. A. Shelton, C. Moloney and R. J. M. Crawford. 1993. Seasonal and interannual variation on the diet of two colonies of the Cape gannet (Moras capensis) between 1977-78 and 1989. Colonial Waterbirds 16:158-175. Blackburn, M. 1957. The relation between the food of the Australian barra- Griffiths: Life fiistory of Thyrsilcs ati/n 709 couta. Thyrsiles atiin (Euphrasen) and recent fluctuations in the fisheries. Aust. J. Mar. Freshwater Res. 8:29-54. Blackburn, M., and P. E. Gartner 19r)4. Populations of barracouta Tliryailca aliin ( Euphrasen) in Australian waters. Aust. J. Mar. Freshwater Res. 5:411- 468. Cortes, E. 1997. A critical review of methods of studying fish feeding based on analysis of stomach contents: application to elas- mobranch fishes. Can. J. Fish. Aquat. Sci. 54:726-738. Crawford, R. J. M. 1995. Snoek and chub mackerel. In Oceans of life off Southern Africa. 2"'' edition (A. I. L. Payne and R. J. M Crawford, eds.l, p. 177-187. Vlaeburg, Cape Town. 1998. Responses of African penguins to regime changes of sardine and anchovy in the Benguela system. S. Afr. J. Mar. Sci. 19:355-364. Crawford, R. J. M., and G. De Villiers. 1985. Snoek and their prey — interrelationships in the Ben- guela upwelling system. S. Afr. J. Sci. 81:91-97. Crawford, R. J. M., L. V. Shannon, and D. E. Pollock. 1987. The Benguela ecosystem. Part IV. The major fish and invertebrate resources. Oceanogr. Mar Biol. Ann. Rev. 25: 353-505. Crawford, R. J. M., L. G. Underbill, C. Raubenheimer, B. M. Dyer, and J. Martin. 1992. Top predators in the Benguela ecosystem — implica- tions for their trophic position. S. Afr J. Mar. Sci. 12: 675-687. Davies, D. H. 1954. The South African pilchard (Sardinops ocellata). Development, occurrence and distribution of eggs and larvae, 1950-51. Investl Rep. Div. Fish. S. Afr. 15:1-28. De Jager. B. v. D. 1955. The South African pilchard {Sardinops ocellata ). The development of the snoek (Thyrsites atun ) a fish predator of the pilchard. Investl Rep. Div. Fish. S. Afr 19:1-11. Dudley S. F. J. 1987. Snoek Thyrsites atun in South African waters: aspects of its biology, distribution and fishery. M.Sc. thesis, 116 p. Univ. Cape Town, Cape Town, South Africa. FAO. 1978. Yearbookoffishery statistics (vol. 46). FAO Statistic series 25, 128 p. FAO, Rome. 1981. Yearbookoffishery statistics (vol. 52). FAO Statistic series 45. 156 p. FAO, Rome. 1990. Yearbook of fishery statistics ( vol . 70 ). FAO Statistic series 105, 325 p. FAO, Rome. 1995. Yearbookoffishery statistics (vol. 80). FAO Statistic series 134, 351 p. FAO, Rome. 1997. Yearbookoffishery statistics (vol. 841. FAO Statistic series 147, 343 p. FAO, Rome. Grant. C .J., T R. Cowper, and D. D. Reid. 1978. Age and growth of snoek, Leionura atun (Euphrasen), in south-eastern Australian waters. Aust. J. Mar. Fresh- water Res. 29:435-444. Griffiths, C. L., and M. J. Blaine. 1988. Distribution, population structure and biology of sto- matopod Crustacea off the west coast of South Africa. S. Afr J. Mar Sci. 7:45-50. Griffiths, M. H. 1997. The life history and stock separation of silver kob, Argyrosomus inodorus, in South African waters. Fish. Bull. 95:47-57. In press. Stock structure of snoek Thyrsites atun in the Benguela: a new hypothesis. S. Afr J. Mar Sci. 25. Griffiths, M. H., and T. Hecht. 1995. On the life-history oCAtractoscion aequidens, a migra- tory sciaenid off the east coast of southern Africa. J. Fish Biol. 47:962-985. Haigh, E. H. 1972. Larval development of three species of economically important South Alrican fi.shes. Ann. S. Air. Mus. 59:47- 70. Hampton, I. 1992. The role of acoustic surveys in the assessment of pelagic fish resources on the South African continental shelf S. Afr J. Mar Sci. 12:1031-10,50. Hulley P A., and R. M. Prosch. 1987. Mesopelagic fish derivatives in the .southern Ben- guela upwelling region. S. Afr J. Mar Sci. 5:597-611. Hurst, R. J. 1988. The barracouta, Thyrsites atun, fishery around New Zealand: historical trends to 1984. New Zealand Fish. Tech. Rep. 5, 43 p. Hurst, R. J., and N. W. Bagely 1987. Results of a trawl survey of barracouta and associated finfish near the Chatham Islands, New Zealand, December 1984. New Zealand Fish. Tech. Rep. 3, 44 p. 1989. Movements and possible stock relationships of the New Zealand barracouta. Thyrsites atun, from tag returns. New Zealand J. Mar Freshwater Res. 23:105-111. Hutchings, L., Pillar, S. C. and H. M. Verheye. 1991. Estimates of standing stock, production and con- sumption of meso- and macrozooplankton in the Benguela ecosystem. S. Afr J. Mar Sci. 11:499-512. Hynes, H. B. N. 1950. The food of fresh-water sticklebacks {Gasterosteus aculeatus and Pygosteus pungitnisl with a review of meth- ods used in studies of the food of fishes. J. Anim. Ecol. 19: 36-58. Kailola, P J.. M .J. Williams, P C. Stewart. R. E Reichelt, A. Mcnee. and C. Grieve. 1993. Australian fisheries resources, 422 p. Bureau of Re- source Sciences and the Fisheries Research and Devel- opment Corporation, Canberra, Australia. MacDonald, J. S.. and R. H. Green. 1983. Redundancy of variables used to describe the impor- tance of prey species in fish diets. Can. J. Fish. Aquat. Sci. 39:651-659. Mehl, J. A. 1971. Spawning and length-weight of barracouta (Teleostei: Gempylidae) from eastern Cook Straight. New Zealand J. Mar. Freshwater Res. 5:300-317. Nakamura, I., and N. V. Parin. 1993. Snake mackerels and cutlassfishes of the world (Families Gempylidae and Trichiuridae). FAO species catalogue 15, 136 p. FAO, Rome. Nepgen, C. S., de V. 1979a. Trends in the line fishery for snoek Thyrsites atun off the south-western Cape, and in size composition, length- weight relationship and condition. Fish. Bull. S. Afr 12: 35-43. 1979b. The food of the snoek T/!yrs(7es a/u/(. Fish. Bull. S. Afr 11:39-42. 1982. Diet of predatory fish in False Bay and possible effects of pelagic purse-seining on their food supply. Fish. Bull. S. Afr 16:75-93. Olivar, M. P., and J. M. Fortuno. 1991. Guide to ichthyoplankton of the southeast Atlantic (Benguela current region). Scientia Marina 55:1-383. 710 Fishery Bulletin 100(4) Olivar, M. P., and P. A. Shelton. 1993. Larval fish assemblages of the Benguela current. Bull. Mar Sci. 53:450-474. Pillar, S. C. 1986. Temporal and spatial variations in copepod and euphausid biomass offf the southern and south-western coasts of South Africa in 1977/78. S. Air. J. Mar. Sci. 4: 219-229. Pillay, T. V. R. 1952. A critique for the methods of study of food of fishes. J. Zool. Soc, India 4:185-200. Roel, B. A., and M. J. Armstrong. 1991. The round herring Etrumeiis whiteheadi. and abun- dant, underexploited clupeoid species off the coast of south- ern Africa. S. Afr. J. Mar Sci. 11:267-287. Roel, B. A., J. Hewitson, S. Kerstan, and I. Hampton. 1994. The role of the Agulhas Bank in the life cycle of pelagic fish. S. Afr. J. Sci. 90: 185-196. Sauer, W. H. H., A. J. Penney, C. Erasmus, B. Q. Mann, S. L. Brouwer, S.J. Lamberth, and T. J. Stewart. 1997. An evaluation of attitudes and responses to monitor- ing and management measures for the South African boat- based linefishery S. Afr. J. Mar Sci. 18:147-163. Shannon, L. V., and G. Nelson. 1996. The Benguela: large scale features and processes and system variability. In The South Atlantic, present and past circulation (G. Wefer, W. H. Berger, G. Siedler. and D. J. Webb, eds. ), p. 163-210 Springer- Verlag, Berlin. Shelton, P A. 1986. Fish spawning strategies in the variable southern Benguela Current region. PhD diss., 327 p. Univ. Cape Town, South Africa. Shelton, P. A., and L. Hutchings. 1990. Ocean stability and anchovy spawning in the south- ern Benguela Current region. Fish. Bull. 88:323-338. SPSS Inc. 1992. SPSS for Windows™': advanced statistics, release 5.0, p. 233-244. SPSS Inc. Chicago, IL. Van der Lingen, C, and D. Merkle. 1999. Predicting anchovy and sardine recruitment from pelagic pre-recruit surveys. South African shipping news and fishing industry review. Cape Town, South Africa. March/April 1999:24-27. Verheye, H. M., A. J. Richardson, L. Hutchings, G. Marska, and D. Gianokouras. 1998. Long-term trends in the abundance and community structure of coastal zooplankton in the southern Benguela system, 1951-1996. S. Afr. J. Mar Sci. 19:317-332. Wickens, R A., D. W. Japp, P Shelton, F Kriel, P C. Goosen, B. Rose, C. J. Augustyn, C. A. R. Bross, A. J. Penney, and R. G. Krohn. 1992. Seals and fisheries in South Africa — competition and conflict. S. Afr J. Mar. Sci. 12:773-789. Windell, J. T, and S. H. Bowen 1978. Methods for the study of fish diets based on analysis of stomach contents. In Methods for assessment of fish production in fresh waters, 3rd edition (T B. Bagenal, ed.), p. 219-226. London, Blackwell. Wood, A. D. 1998. A contribution towards the taxonomy of the ich- thyoplankton species community and understanding of its dynamics along the south east coast of South Africa. Ph.D. diss., 338 p. Rhodes Univ., Grahamstown, South Africa. 711 Abstract— The problem of bias in IV- male petrale sole a^e and lenglh-at- maturity relationships caused by sam- pling; from spawninfj aggregations was investigated. Samples were collected prior to aggregation, and histological methods were used to determine matu- rity status. Mature and immature fish were classified by inspecting oocytes for the presence of yolk in September, when substantial divergence in yolked and unyolked oocyte diameters had been observed. Comparison of macro- scopic and microscopic assessment of maturity showed that maturity status cannot be determined accurately by using macroscopic inspection during the summer. Female petrale sole from the central Oregon coast were 50'^ mature at 33 cm and 5 years of age. Comparison of data fi-om our study with data used in recent petrale sole stock assessments showed that both sampling bias and the use of samples fi-om sea- sons when status cannot be accurately determined have likely caused eiTors in fitted maturity relationships. Length and age at maturity of female petrale sole (Eopsetta jordani) determined from samples collected prior to spawning aggregation Robert W. Hannah Steven J. Parker Oregon Department of Fish and Wildlife Marine Resources Program 2040 Marine Science Drive Newport, Oregon 97365 Email address (for Robert W Hannah) Bob HannahigJhmsc.orsl.edu Erica L. Fruh National Marine Fishenes Service Northwest Science Center Fishenes Research and Monitoring Division 2030 Manne Science Dnve Newport, Oregon, 97365 Manuscript accepted 21 March 2002. Fish. Bull. 100:711-719 (2002). Petrale sole (Eopsetta jordani) have been the target of a valuable com- mercial trawl fishery off the U.S. West Coast since before World War II (Samp- son and Lee' ). Management and assess- ment of petrale sole is complicated by the fact that modern fishing activity targets winter deepwater aggregations of spawning fish (Fig. 1). Because most of the available female maturity data are derived from the winter commer- cial fishery, they are potentially biased with respect to age and length at matu- rity (e.g. small mature fish are more likely to be sampled than small imma- ture fish). Some summer samples are available for analysis, including those from the National Marine Fisheries Service triennial shelf survey (e.g. Zim- mermann et al., 1994). However, in the summer, when mature and immature fish are well mixed, maturity status is not easily determined visually because of the similarity in appearance between ovaries of immature fish and those of fish in the "resting" state (Ketchen and Forrester, 1966). The age and length at maturity for female fish is an important param- eter in many stock assessment models. Clark (1991), Lunsford (1999) and Al- Jufaily (1996) have demonstrated that variation in the age at 50% maturity, especially in relation to the median age of recruitment to the fishery, can have a major influence on the calculation of a target fishing rate. Target fishing rates are currently used by the Pacific Fishery Management Council (PFMC) to manage most groundfish stocks (Clark, 1991: PFMC'-). An example of a target fishing rate would be F^q'/,' the fishing mortality rate that would reduce spawning stock biomass per recruit to 40% of the unexploited level. The potential influence of errors in es- timating the median age of female ma- turity can perhaps best be illustrated with an example. In Lunsford's (1999) study of maturity of Pacific ocean perch ' Sampson, D. B., and Y. W. Lee. 1999. An assessment of the stocks of petrale sole off Washington. Oregon and Northern California in 1998. In Pacific Fishery Management Council. 1999. Appendix to the status of the Pacific coast groundfish fishery through 1999 and recommended acceptable catches for 2000: stock assess- ment and fishery evaluation. Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, Oregon, 97201. - Pacific Fishery Management Council. 2000. Status of the Pacific Coast ground- fish fishery through 2000 and recom- mended biological catches for 2001: stock assessment and fishery evaluation. (Docu- ment prepared for the council and its advisory entities.) Pacific Fishery Man- agement Council 2130 SW Fifth Avenue, Suite 224, Portland, Oregon 97201. 712 Fishery Bulletin 100(4) 180 m Washington 47 — Northern Oregon 45 — Newport \180m Pacific Ocean 44" Charleston 43 Southern Oregon Brookings 42 125' 124° I Figure 1 Sampling area for petrale sole (bold-outline polygon) off Newport, Oregon, and areas where winter trawl fishing for petrale sole is concentrated (gray polygons). Also shown is the 180-m depth contour. (Sehastes alutus), a shift in the estimated median age of maturity for female fish from 7.5 to 10.5 years decreased the Fji,,. value from 0.110 to 0.076. a decrea.se of 31'/f. Ac- cordingly, the use of potentially biased maturity data from the winter commercial fishery for petrale sole adds to the uncertainty in stock assessments. The accuracy of matu- rity information used in recent petrale sole stock assess- ments is also uncertain because of two additional factors: outmoded data and the blending of samples across space and time (Sampson and Lee.'; Turnock et al.M. Maturity samples are obtained from various locations and times, and are often combined to produce a general relationship for stock assessment purposes. The combined relationship may poorly account for geogi'aphical differences, long-term ■'' Turnock, J., M. Wilkins, M. Saelcns, and t'. Wood. 199,3. Sta- tus of West Coast petrale sole in 1993. Appendix G. In Status of the Pacific Coast groundfish fishery through 1993 and recom- mended biological catches for 1994: stock assessment and fish- ery evaluation. Pacific Fishery Managemenmt Council, 2130 SW Fifth Ave., Suite 224, Portland, OR 97201. changes in maturity, or for inadequacies in some samples (e.g. lack of immature fish) (Sampson and Al-Jufaily, 1999) and can result in poorly determined curves that bias and degrade the fit of the combined maturity relationship. The principal objective of our study was to collect maturity data for female petrale sole from the late summer to early fall period and compare them with the maturity data used in recent stock assessments (Sampson and Lee'; Turnock et al.'). Although sampling in late summer and early fall is probably optimal for assessing female maturity in the petrale sole population, some ovaries may be difficult to classify accurately as mature or immature by macroscopic inspection. A second objective of our study was to evaluate female petrale sole maturity by using microscopic exami- nation of stained thin sections and compare these results with those obtained from simple visual inspection. The petrale sole maturity data used in recent stock as- sessments have been obtained from the Oregon Depart- ment of Fish and Wildlife's commercial fishery sampling program (Sampson and Lee'; Turnock et al.-^). Potential Hannah el al Length and age at maturity for Eopsctta pidani 713 bias in some of these data, arising from spawning aggrega- tioii, argues that a difPerent sampling approach niiglit he usetiil I'or trying to generate unbiased maturity data from samples of the commercial catch. A third objective of our study was to evaluate problems with available maturity data from fishery samples in light of histological evalua- tions of maturity stages and to reconmiend an alternative approach for sampling the commercial catch for age and length at maturity. Materials and methods Petrale sole were collected by trawl during August and September 2000 by using a chartered commercial fishing vessel from Newport, Oregon. Sampling was conducted over a very limited geographic range to minimize any spatial variation in the maturity ogive. Individual sam- pling sites were selected to attempt to sample across the available size range of petrale sole, based on the skipper's experience and expertise. Fish were sorted at-sea accord- ing to total length (±0.5 cm), and fifteen fish from each 1-cm size interval were retained for maturity sampling. Upon returning to port, the fish were measured again and sex was determined by dissection. An ovary was removed and assigned a macroscopic maturity stage following the criteria of Hagerman (1952). Only experienced samplers were used to assign macroscopic maturity stages to mini- mize any errors in visual staging. Otoliths were removed from all female fish for subsequent age determination, and one ovary was preserved for histological examination to determine maturity. Ages were determined by using the break-and-burn technique for sagittal otoliths (Chilton and Beamish, 1982). Consistency of age determination was evaluated by using a blind double reading. Ovaries were preserved in WA buffered formalin and later transferred to 70% ethanol for storage. Tissue sam- ples from the midsection of the ovary were then embedded in paraffin, sectioned at 5 pm, and stained with Harris's hematoxylin and eosin V. The samples were examined and classified as mature or immature based on the presence or absence of dark-staining yolk globules (vitellogenin). The diameter (pm) of the largest spherical nonatretic oocyte in the most advanced oocyte stage from each of five microscope fields was also measured and used to calcu- late a mean maximum oocyte diameter (MMOD) for each sectioned ovary (West, 1990; Nichol and Pikitch, 1994). MMOD was compared for mature and immature fish. A divergence in MMOD between mature and immature fish was taken as evidence that the presence or absence of dark-staining yolk globules was an accurate indicator of maturity status. Samples from August and September were compared by using this approach. The accuracy of visual examination of ovaries was then evaluated by com- paring the maturity status from visual examination with status determined after sectioning, staining, and micro- scopic examination. Logistic regression was used to fit sigmoid length-ma- turity and age-maturity curves. The model fitted had the general form P = gl'iu + ''l»l V( 1 -I- {'"'() + ''l«ll) •>i ' where p = the probability that a fish is mature in a given length (cm) or age category x-^; and 6,1 and 6, are parameters that define the shape and loca- tion of the fitted curve. The predicted length or age at 50% maturity was calcu- lated as L (or A)r,|, = -hjhy Maturity data collected by the Oregon Department of Fish and Wildlife (ODFW) from the 1986-2000 commercial trawl fisheries were also analyzed and compared with the data collected in our study. The most recent petrale sole stock assessment (Sampson and LeeM relied on maturity data used for the 1993 assessment (Turnock et al.'*), which in turn were derived from combined ODFW port sampling data for the years 1986-91. We fitted a maturity-length relationship to these data for comparison with the matu- rity data from our study. We also summarized the ODFW data from 1986-91 and 1992-2000, by month and maturity status, to compare with our data and assess their adequacy for determining length at maturity for female petrale sole. All of these samples were classified by a simple visual evaluation of maturity stage. To examine the effect that a lack of immature fish may have had on estimated maturity curves from these two time periods, we split the data into groups, according to sea- son. Samples from the months of November through Febru- ary were classified as "fall-winter" samples. These samples represented the seasonal time period when the maturity status of petrale sole was most readily determined visually, but also the time period when samples were most likely to he from spawning aggregations. Samples from April through September were classified as "spring-summer" and represented the seasons in which fish were not aggre- gated for spawning, but the time when maturity status was most difficult to determine by inspection of the ovaries. Results Comparison of visual maturity determinations with results obtained from microscopic evaluation showed that maturity of petrale sole cannot be reliably determined by simple visual inspection in August or September (Tables 1 and 2). For ovaries classified as immature by inspection in August, 16.7% proved to be mature upon microscopic evaluation. Of those classified initially as mature, 6.5% showed no microscopic evidence of vitellogenesis. By September, all of the ovaries classified by inspection as mature, did prove to be vitellogenic. However, 27.9% of those classified as immature, proved to be vitellogenic after viewing the stained sections with a microscope. MMOD increased markedly in vitellogenic ovaries between August and September, but in nonvitellogenic ovaries the increase was minimal (Fig. 2). The resulting divergence in MMOD and the absence of fish with MMOD 714 Fishery Bulletin 100(4) Table 1 Details of petrale sole collected by trawl aboard the FV Olympic. Date Location Females Males Size range (cm) 8 Aug 2000 4428''N 12434''W 140 103 24-51 23 Aug 2000 4446°N 12432°W 24 46 27-49 26 Sep 2000 4428''N 12435°W 117 123 26-50 Total 281 272 24-51 Comparison of macroscopic and microscopic Table 2 determination of maturity m female petrale sole collected by trawl. Sample month Macroscopic classification Microscopic classification Condition Number Nu offish mber immature i'-'r) Number mature lOi ) August September Total immature mature immature mature 102 62 61 56 281 85(83.3) 4(6.5) 44(72.1) 0(0.0) 17(16.7) 58 (93.5) 17(27.9) 56(100) o O 15-, 10 5- 58 ■ Non-vitellogenic n Vitellogenic August Samples 58 118 178 238 298 358 418 478 538 September Samples 538 118 178 238 298 358 418 478 Mean maximum oocyte diameter (nm) Figure 2 Comparison of mean maximum oocyte diameters (pm, see text) for petrale sole with vitellogenic and nonvitel- logenic oocytes, for samples collected m August and September 2000. around 180 \im suggested that by September, the micro- scopic classification of ovaries provided an accurate identi- fication of maturity status of individual fish. Immature fish were well represented in the September collections, indi- cating the fish were still well mixed on the feeding grounds (Table 2). Therefore, the September maturity data gener- ated from microscopic inspection were used to fit age- and length-at-maturity relationships for female petrale sole (Fig. 3). The resulting curves fitted the data well and chi- square tests on the residuals indicated no problems with lack of fit caused by overdispersion (Pearson chi-square of 5.714 Iwith 20 degrees of freedom] for length, and 0.643 [with 11 degrees of freedom] for age, P>0.999). The fitted relationships indicated that female petrale sole off the cen- tral Oregon coast were 50% mature at about 33 cm and at about 5 years of age. Complete maturity was obtained at about 40 cm and about 9 years of age (Table 3). Examination of the monthly samples that, as an ag- gregate, were used to fit the 1986-91 maturity curve showed that numerous samples were included from the IVIarch through August period, especially from the port of Charleston, Oregon (Table 4). The inclusion of these summer samples explains some of the flatness in the fitted maturity curve from Turnock et al.'^ (Fig. 4, L5Q=30.6, slope=0.29). A maturity curve fitted to these da- ta, but excluding spring and summer samples, was much steeper (Fig. 4, Lr,Q=33.56, slope=0.50). Inspection of the 1986-91 and 1992-2000 maturity data from the winter months, when errors in determin- ing maturity should be minimized (Tables 4 and 5, Fig. 5), showed some evidence of the type of bias that can result from sampling spawning aggregations. First, very few im- Hannah et a\ Length and age at maturity foi Eopsctta /oidani 715 Table 3 Results of logistic regression analysis of histologically detcrmnK •d maturity status of pctrale sole versus ength (cm) and age (Si'plemlx'r samples oi ly). Independent variable Coefficients Standard error Chi-square P-value ^50 "'' -"50 95% confidence Length 33.10 cm 32.13-33.93 constant -24.593 4.572 28.936 0.0001 length 0.743 0.136 29.759 0.0001 Age 5.15 yr 4.68-5.54 constant -6.358 1.417 20.122 0.0001 age 1.234 0.261 22.349 0.0001 mature fish were sampled, even though the commercial fishery samples showed a similar lower size range to the fish collected in our study (Fig. 6). The lack of immature fish was most evident in the 1986-91 November-February samples from Charleston. Oregon, where 170 fish were sampled for maturity, and only one fish was reported as immature. The 1992-2000 Charleston data also showed low numbers of immature fish. These samples also showed groups of very small fish, and most or all of them were classified as mature (Fig. 5). This kind of bias, where small fish encountered are more likely to be reported as mature than similar-size fish sampled from a well-mixed popula- tion, would also act to flatten the resultant maturity curve. The Astoria winter data included more immature fish, and accordingly produced steeper maturity curves. However it was unclear to what degree bias from sampling spawning aggregations may have influenced these curves. Discussion Although it is known that late fall and winter is a better time for visually assessing female maturity status for petrale sole, our data showed clearly that late summer and early fall maturity samples, for which there was a visual determination of maturity, should not be used for estimat- mg maturity in this species. Although our study did not spe- cifically examine ovaries from spring collections, the general timing of ovarian development in this species suggests that this caution can be extended to all late spring and summer samples. Our results are similar to the findings of Ramsay and Witthames ( 1996), who suggested that visual maturity evaluations of the common sole iSolea solea) are generally only reliable during a limited portion of the year. The maturity data developed in our study (Figs. 3 and 4) suggest a larger size at 50'7f maturity (33.1 cm vs. 30.7 cm) and a much steeper maturity curve for petrale sole than indicated by the 1986-91 sample data used in the last two petrale sole stock assessments (Turnock et al.^). Although the curves appeared to be quite different, the length at SOVc maturity increased by only 2.4 cm. The most recent stock assessment model indicates an annual fishing mor- tality rate, for fully exploited size groups, of 0.285 (Samp- son and Lee' ). Incorporating a revised maturity ogive from our study decreased the equilibrium fishing rate to 0.268, 20 23 26 29 32 35 38 41 44 47 50 53 56 59 Length 1,25-, jmr"~^ 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Age Figure 3 Proportion mature for female petrale sole, by age and length (FL,cni), from September samples. Maturity status was determined by histological examination by using pres- ence and absence of vitellogenesis to determine maturity. Fitted curve is also shown (Table 3). a change of about 6% (Sampson'*). This modest effect on harvest rate seems odd, given the large apparent differ- ence between the curves in Figure 3. A possible explana- tion for this discrepancy is that the results depend heav- ^ Sampson, D. B. 2000. Personal commun. Coastal Oregon Marine E.xperiment Station and Department of Fisheries and Wildlife, Oregon State University, Hatfield Marine Science Center, Newport, OR 97365. 716 Fishery Bulletin 100(4) Table 4 Summary of the numbers of mature and immature petrale sole s ampled, from the Oregon bottom trawl fishery, 1986-91. Port Month Immature Mature Total Astoria November 20 130 150 Astoria December 0 167 167 Astoria January 2 135 137 Astoria February 32 10 42 Winter total 54 442 496 Astoria March 18 61 79 Astoria October 7 26 33 Astoria total 79 529 608 Charleston November 0 37 37 Charleston December 1 68 69 Charleston January 0 24 24 Charleston February 0 40 40 Winter total 1 169 170 Charleston March 16 64 80 Charleston April 0 37 37 Charleston June 14 115 129 Charleston July 27 123 150 Charleston August 71 162 233 Charleston October 2 78 80 Charleston total 131 748 879 TotaKall ports) 210 1277 1487 Table 5 Summary of numbers of mature and immature petrale sol ? sampled by season and port from the Oregon bottom trawl fishery. 1992-2000. Port Season Immature Mature Total Astoria Winter (Nov-Febl 19 242 261 Spring-fall (Mar-Octi 4 89 93 Charleston Winter (Nov-Feb) 20 347 367 Spring-fall (Mar-Oct) 31 179 210 Brookmgs Winter (Nov-Feb) 0 56 56 Spring- fall (Mar-Oct) 0 142 142 Total 84 995 1079 ily on assumptions about fishery selectivity and discard used in the latest assessment model-parameters that are poorly known for most West Coast fish stocks. The modest effect on harvest rate found in our study also does not pre- clude larger effects for other, later maturing stocks. In the example given earlier from Lunsford's (1999) simulation work with Pacific ocean perch, the drop in target harvest rate was much larger. In that instance, a shift of 3 years in the median age of female maturity resulted in a 31'% drop in the target harvest rate. In that example, spawning stock biomass per recruit also fell to 31'7f, well under the target ofF,,,,,.. The maturity data developed in our study differ from those developed by Turnock et al.-^ primarily in the steep- ness of the maturity curve. It could be argued that the earlier curve is flatter simply because it incorporates data from a wider geographic range. A latitudinal cline in length at 50% maturity, as has been postulated for many flatfish stocks (Ketchen and Forrester, 1966; Castillo, 1995; Brodziak and Mikus, 2000), could cause flattening Hannah et al : Length and age at maturity for Eopsetta joi dcni 717 1 25 1 00 e 075 0 50 025 0 00 20 23 26 29 32 35 38 41 44 47 50 53 56 59 Length (cm) Figure 4 Comparison of fitted pctralc sole female maturity cui"ves from September samples (our study) and 1986-91 com- mercial fishery data used in the last two stock assess- ments (Turnock et al.'l. Bars show 1986-91 maturity data used to fit the curve from Turnock et al.' Dashed line shows the maturity curve fitted to the 1986-91 sam- ples by using only data from the fall-winter period. in a combined maturity curve. However, errors in deter- mining maturity status, caused by including samples from time periods when maturity status cannot be accurately assessed visually, would have the same effect. This can be seen in a comparison of the maturity data generated in our study from macroscopic and microscopic evaluations. The macroscopic curve (bars and solid line in Figure 7) is flattened in relation to the curve according to microscopic evaluation (dashed line). The curve is also shifted to the right because, in this case, more large fish were incorrectly classified than small fish. At other times of the year, errors could be more common with smaller fish and the cui-ve would shift the other way. In either case, however, the curve should, in theory, be flatter than the true maturity curve. This flattening of the maturity curve is caused by the way the most likely errors change along a maturity ogive. In small or young fish, which are mostly immature, most of the errors encountered will be immature fish mis- takenly considered mature. At larger sizes or older ages, the reverse will be true. Near the inflection point, the two types of errors offset each other and have little impact on the slope because the population is split roughly 50/50 with respect to maturity. These effects, in combination, tend to flatten the resulting curve. Compared to a length at 50% maturity of 33 cm for the central Oregon coast (our study), the Astoria data from the commercial fishery are consistent with an increase in the length at 50% maturity with latitude, as suggested by Ketchen and Forrester (1966) and Castillo (1995). The methods developed in our study could be applied to petrale sole samples collected over a wider geographic range to shed more light on how length at maturity varies with latitude. Best ( 1961) reported a length at 50% maturity of 35.5 cm for female petrale sole from California waters, and Harry^ re- ported 40.0 cm for the Columbia River area. In comparison with the recent data from Astoria fishery samples (Fig. 5) 1,25 1,00 075 0.50 0,25 000 1 00 075- o 0,50 Q. O qI 0,25 0,00 1 00- 0,75 0,50 0,25 0,00 Astona Samples 1986-1991 Jl B Astoria Samples l'i'):'-;?0()() Charleston Samples 1992-2000 20 23 26 29 32 35 38 41 44 47 50 53 56 59 62 Length (cm) Figure 5 Proportion of female petrale sole identified as mature from November-February samples of trawl landings in Astoria and Charleston. Oregon, 1986-1991 and 1992-2000. and from the central Oregon coast (Fig. 3), the data from Harry'' suggest a very large decrease in the length at 50% maturity for female petrale sole since the 1960s. If this decrease is a biological reponse to exploitation, it could ex- plain how petrale sole stocks have held up quite well under heavy commercial harvest (Sampson and Lee'). The data presented in our study suggest that the collec- tion of maturity data by sampling on spawning aggrega- tions does cause bias in estimates of female age and length at maturity for petrale sole. Use of spring and summer maturity samples in data sets used to estimate age and length at maturity has also caused errors. Our analysis shows that the overall effect on estimates of age and length '' Harry, G. Y. 1959. Time of spawning, length at maturity, and fecundity of the English, petrale and Dover soles iPar- ophrys vetulus, Eopsetta jordani, and Microstomus pacificus, respectively). Fish. Comm. Oregon, Research Briefs 7(1):5-13. 718 Fishery Bulletin 100(4) 1986-91 Samples From Commercial Catch 37 42 47 Length (cm) Figure 6 Companson of female petrale sole length frequency from the 1986-91 commercial fishery samples and the September samples collected in our study The 1986-91 samples were also used to fit the maturity curve from Turnock et al •' 1.25 1.00 075 0 50 o dI 0.25 0,00 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 Length (cm) Figure 7 Comparison of maturity cui-ves fitted to the data from our study. Bars show proportion offish mature by length from the macroscopic evaluations of the August and September samples. The solid line is the fitted curve for those data. The dashed line is the fitted curve from microscopic evaluations of the September samples only. at 50% maturity has been small. Collection of early fall samples from the commercial catch, followed by histologi- cal preparation and classification based on presence or ab- sence of vitellogenin, should yield more accurate maturity information for female petrale sole than winter sampling on spawning aggregations or maturity sampling where only a visual assessment of ovary condition is made. Acknowledgments We'd like to thank the skipper and crew of the FV Olympic. Staff froin ODFW's Marine Program and from the Pacific States Marine Fisheries Commission helped with sample preparation. Jennifer Menkel aged the otoliths and Margo Whipple of Oregon State University prepared the histo- logical sections. Jim Golden and David Sampson provided helpful reviews of the draft manuscript. Literature cited Al-Jufaily, and M. S. 1996. Variation in the maturity schedule of english sole and its influence on calculating the F35% target fishing rate. M.S. thesis, 85 p. Oregon State Univ., Corvallis OR. Best, E. A. 1961. Savings gear studies on Pacific coast flatfish. Bull. Pac. Mar Fish. Comm. 5:25-47. Brodziak, J., and R. Mikus. 2000. Variation in life history parameters of Dover sole, Microstumus pacificiis, off the coasts of Washington, Ore- gon, and northern California. Fish. Bull. 98:661-673. Castillo, Gonzalo C. 1995. Latitudinal patterns in reproductive life history traits of northeast Pacific flatfish. In Proceedings of the international symposium on north Pacific flatfish. Alaska Sea Grant College Report. AK-SG-95-04, 1995, p. 51-72. Alaska Sea Grant Progi-am. Fairbanks, AK. Chilton, D. E., and R. J. Beamish. 1982. Age determination methods for fishes studied by the Groundfish Program at the Pacific Biological Station. Can. J. Fish. Aquat. Sci. Spec. Publ. 60:2-18. Clark, W. G. 1991. Groundfish exploitation rates based on life his- tory parameters. Can. J. Fish. Aquat. Sci. 48:734-750. Hagerman, F B. 1952. The biology of the Dover sole. California Dep. Fish and Game, Fish Bull. 85. 48 p. Ketchen, K. S., and C. R. Forrester 1966. Population dynamics of the petrale sole. Fish. Res. Board Can. Bull. 153, 195 p. Lunsford, C. R. 1999. Distribution patterns and reproductive aspects of Pacific ocean perch iSebastes alutus) in the Gulf of Alaska. M.S. thesis, 154 p. Univ Alaska Fair- banks, AK. Nichol, D. G., and E. K. Pikitch. 1994. Reproduction of darkblotched rockfish off the Oregon coast. Trans. Am. Fish. Soc. 123:469-481. Ramsay, K., and P. Witthames. 1996. Using oocyte size to assess seasonal ovarian development in Solea solea. J. Sea Res. 36(3/4): 275-283. Hannah et a\ Length and age at matunty for Eopsctta lordani 719 Sampson, D. B., and S. M. Al-Jufaily. 1999. Geographic variation in the maturity and growth schedules of Enghsh sole along the U.S. west coast. J. Fish Biol. 54:1-17 West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar. Freshwater Res. 41:199-222. Zimmermann, M., M. K. Wilkins, R. R. Lauth. and K. L. Weinberg. 1994. The 1992 Pacific west coast bottom trawl survey of groundfish resources: estimates of distribution, abundance and length composition. U.S. Dep. ("ommer, NOAA Tech. Memo. NMFS-AFSC-42, 110 p. plus appendices. 720 Abstract— We have formulated a model for analyzing the measurement error in marine survey abundance estimates by using data from parallel surveys (trawl haul or acoustic mea- surement). The measurement error is defined as the component of the var- iability that cannot be explained by covariates such as temperature, depth, bottom type, etc. The method presented is general, but we concentrate on bottom trawl catches of cod [Gadus morhua). Catches of cod from 10 parallel trawling experiments in the Barents Sea with a total of 130 paired hauls were used to estimate the measurement error in trawl hauls. Based on the experimental data, the measurement error is fairly constant in size on the logarithmic scale and is independent of location, time, and fish density. Compared with the total variability of the winter and autumn surveys in the Barents Sea, the measurement error is small (approximately 2-5%, on the log scale, in terms of variance of catch per towed distance). Thus, the cod catch rate is a fairly precise measure offish density at a given site at a given time. The measurement error in marine survey catches: the bottom trawl case Vidar Hjellvik Olav Rune Gode Institute of Marine Research Nordnesgt 33 PO Box 1870, Nordnes N-5817 Bergen, Norway E-mail address (for V Hiellvik) vidarfi'a'imrno Dag Tjastheim Department of Mathematics University of Bergen, Johs Brunsgt 12 N-5008 Bergen, Norway Manuscript accepted 20 February 2002. Fish. Bull. 100:720-726 (2002). Surveys are vital for estimating the size and composition of marine pop- ulations. In the data collection pro- cess, trawl samples or acoustics (or both) are usually employed. It is well known that the resulting estimates are subject to substantial variations, and it is important to quantify and explain, as much as is possible, the variability in terms of relevant explanatory variables or covariates. Typically these variables will depend on the sampling tool used, but for bottom trawl catches important explanatory variables can be depth and location of the haul (Pola- check and Volstad, 199.3), time of the day (Korsbrekke and Nakken, 1999), season, strength of the year classes involved, etc. Generally, as the number of covar- iates increases, and the model becomes more complex, the residual variation (or remaining uncertainty) not explained by the model decreases. But no matter how refined the model is. there will always be an unexplained random component that cannot be attributed to any observed variable. This residual variation is caused by the interactions between the fish, the measurement device, and the environment (see e.g. Engas, 1994). The purpose of this study was to define and quantify this residual source of random variation. This was done by analyzing meastu-ements from parallel tows of multiple vessels, which is the closest one can come to a controlled statistical experiment in this context. The importance of quantifying this type of fluctuation lies in the fact that it is a benchmark uncertainty, which is inherent in the survey process itself, and in this sense it may be termed a measurement error If the measurement error can be assessed from field data and is consistent over time and space, we improve our understanding and quantification of other causal factors behind the uncertainty associated with survey estimates. In this study we looked at bottom trawl catches of cod, but we would like to stress that the concepts and techniques developed can in principle be applied to acoustic survey estimates or indeed to any type of measurements collected simultaneously by two or more independent parallel sampling devices. It should be noted that there is a growing related literature on comparative survey analysis. We refer to the review paper by Pelletier (1998) and references therein. Materials and methods Parallel trawling experiments During the annual combined bottom trawl and acoustic survey of demersal fish in the Barents Sea during winter and autumn conducted by the Institute Hjellvik et al : Measurement error in marine survey catches 721 Table 1 Summary statistics for the parallel trawling experiments; time, vessel, position (latitude, longitude), the number of hauls in) and the average per cod weight w (in kg) within each group. The vessels involved were Johan Hjort (.JM),Anny Krsemer (LIZY), G. O. Sars (GS), Jan Mayen (JM), and Michael Sars (MS). Group Year Date Vessels Lat. Long. n w 1 1991 3-5 Mar LIZY, JH 71.3 26.2 10 2.01 2 1994 10 Feb LIZY JH 71.2 36.0 5 0.81 3 1994 22-23 Feb LIZY, GS 71.3 26.3 8 1.98 4 1995 22-23 Feb JM,JH 71.3 25.4 12 1.05 5 1995 23-25 Feb JH,GS 71.3 25.4 23 0.94 6 1995 15-17 Aug MS,JH 74.3 17.3 29 0.72 7 1996 17 Feb JM,GS 70.4 36.5 4 0.03 8 1996 24-25 Feb JM,GS 71.8 23.8 10 0.19 9 1997 8-10 Feb JH.GS 71.3 27 17 0.72 10 1997 2-3 Aug MS,JH 72-73 27-30 12 0.21 of Marine Research, Bergen (IMR), (Jakobsen et al.M, parallel trawling experiments were used to compare the efficiency of the participating vessels with gear types as given in Table 1 and Figure 1. During a parallel haul the vessels operated about 500 meters apart and used radio contact to assure proper coordination during hauls. We analyzed ten parallel trawl experiments performed by the IMR during the last decade (Table 1). The data from 1991 are described in Michalsen et al. (1996). Two hauls with unstable bottom contact were excluded from the 1991 data (Michalsen et al., 1996). Similarly, two hauls from 1995 were excluded — one where trawl geometry measurements indicated problems with the doors and another with highly different recorded towed distances (0.7 and 2.2 nautical miles [nmi]). Let d, denote the towed distance for haul ; andvesselj;j=l, ... ,n;y=l,2 wheren is the total number of hauls. The average recorded distance is d ={2n ) ' X -'X '=■ d, , = 1.33 nautical mile (average duration in time is 27 minutes), with 0.8 < d,.^ < 1.8 in 98% of the cases. The abso- lute values |d, j-c/, 9I ofthe differences in towed distance for the two vessels in the same haul, are 0, 0.1, 0.2, 0.3, 0.4, and 0.6 in 43, 45, 31, 6, 4, and 1 cases, respectively. The data were subdivided into 10 groups so that the same two vessels performed all hauls within a group, within a period of one to three days, and usually in a small geographical area. Group 10 is an exception where the trawl stations are evenly spread over about 60 nmi both in the east-west and in the north-south directions. The statistical model Any study of uncertainty depends on the stochastic model adopted. Two different statistical models may yield quite ' Jakobsen, T., K. Korsbrekke, 8. Mehl, and O. Nakken. 1997. Norwegian combined acoustic and bottom trawl surveys for demersal fish in the Barents Sea during winter ICES CM 1997A':17:l-26. Institute of Marine Research, P.O.Box 1870 Nordnes, 5817 Bergen, Norway. different uncertainty estimates. A general model for a series of survey measurements [y^,i=l, ... ,11} is given by >',=/'^i' .^,p' + f,. i = 1, ... ,n. where /■ = a deterministic function, which in general is unknown; and X = the zth measurements of p explanatory variables, such as geographical location, depth, or time. If all of the relevant explanatory variables were included, f, would represent the residual uncertainty. In practice, all conceivable explanatory factors will not be observed, and often /'is assumed to be linear. A difficulty in assessing the uncertainty of fish abundance estimates is that we cannot carry out a controlled experiment, where the setting of each experiment is identical. In such an idealized series of experiments the explanatory variables x,i, ... ,x,p would be fixed, and e would be the only source of random variation so that for a series of A'^ experiments. y^=f(x.„...,x) + e^. 1, ... ,N. The standard error of f^, could then be estimated directly from the observations {y^l as 1 N- \l <.>'*->') and the correctness of the model could be tested by a new series of experiments for a new set of fixed values for the explanatory variables. The closest we can come to such an idealized experiment is that of parallel trawling described above. The values of the explanatory variables, such as geographical location and depth, will vary somewhat from one vessel to another, 722 Fishery Bulletin 100(4) 00 S< X X >««< X ^ X "^x X X ^ r ''°'' X X xX ■ . X X X ^ X . + x + X ^ -^ + ■ ^ X X . X X XX x X ' X V r + X x>^ ^ X X >?*x X X x X ■ X X X X X X X X X.. X - J + X X x" X X X JH JH GS JH GS JH GsJ L X GS X GS JH 1 11 16 24 36 59 88 102 119 Haul Figure 1 I A) Catches v, ^ from the parallel trawling experiments in Table 1. The groups are separated by vertical lines and the group numbers corresponding to Table 1 are given at the bottom of the figure. The symbols represent vessels (see Table 1), and the cumulative number of hauls are given at the horizontal axis. (B) Corresponding differences z, =,v, i - v, ■>■ The symbols indicate gear type; 3236: Campelen 1800 shrimp trawl with 35-mm mesh size, 40 m sweeps and rockhopper gear; 3270: same as 3236, but 20-mm mesh size; 3271: same as 3270 but with strapping. 3270/1 indicate that one vessel uses 3270, the other 3271. To the right, 95'?r confidence intervals for the group means are given. The first and second vessel in each group are given at the top and bottom, respectively, for each group. but as an approximation they will be considered identical. However, we will allow for an additive individual vessel effect a,, j=l,2, which permits differences in equipment and efficiency for the two vessels. This leads to the fol- lowing model for the observations [v,^, i=\, ... , /!;./=l,2) where j=l, 2 corresponds to the first and second vessel in Table 1: .V,;l=A-^,l -V' + ^l+^^l .>':,2 = /'^i xJ + o-1^^,.2- (1) All the factors affecting jointly tow performance are supposed to be in the function /", and therefore the residuals |e, , i=\ n;j=l,2] are assumed to be independent zero- mean identically distributed random variables and a^ = sd (£,-j) is the measurement error. We can now eliminate /' and the explanatory variables by taking differences, i.e. z, = y, «l-«2 + ^,;l-^,;2- The expected difference between the two vessels is then given by E(2, ) = a, - Oj and because of the independence off, j and e^.^, o\ = var(2,) = var(f, J -f,_2' = 2(7^, and the standard error CT can be estimated as - 1 - 1 1 V, (2) Hjellvik et al.: Measurement error in marine survey catches 723 whereas 5 = a, - «., is estimated hy S = z. It should be noted that in the actual computation of a^ and S, the data v,., are log transformed, i.e. y,.j = ]ogU, j/d,.j), 1=1, ,/=1.2. (3) where n^ and d^^ denote the catch in numbers and the towed distance, respectively, for vessel j at the iih haul. Log-transformed data are used to reduce the heterogene- ity of the variance. Tests for differences between experimental groups If our hypothesis, that parallel trawling experiments can be used to quantify a measurement error inherent in the cod catching process itself, is correct, we expect that this error, as estimated by ct,., should be the same for all 10 experimental groups. If the z/s originate from a Gaussian distribution, the null hypothesis of equal variance can be tested by Bartlett's test Icf. all groups tested simultane- ously, Bickel and Doksum, 1977, p. 304) and if needed, followed by a series of F-tests where the groups are tested against each other in pairs. The possible differences in efficiency caused by different vessels or fishing gears ( or both ) can be tested by an AN OVA test followed by a series of f-tests if the AN OVA test leads to rejection. Again, normally distributed observations are a prerequisite for such tests. Our first task was therefore to check whether the z,-data followed a Gaussian distribution. It seemed plausible to assume that observations from different groups followed the same distribution, but possibly with differences in mean and variance. Therefore, when checking for normality, we considered the standardized variables where, k=k{i), k=l. ... , 10, denotes the group that haul ;, ; = 1 71, belongs to; and 2^ and s^. are the average and the estimated standard deviation of the z-values in group k, respectively. Deviations from normality of |.v,l can be checked visually by inspecting a normal plot, and formally by e.g. the Kolmogorov-Smirnov test (Bickel and Doksum, 1977, ch. 9.6). Results The log-catches |y, ,) and the corresponding differences (z,! are presented in Figure l.The catches range from approxi- mately 6^=20 to e*=3000, but on the log-scale the differ- ence in catch between the vessels does not seem to depend on the size of the catches (see formal test at the end of this section). A normal plot of the standardized observations {x={z-z^)/s^] appears linear (Fig. 2) and the Kolmogorov- Smirnov test does not reject the null hypothesis of normal- ity at a 10% level. Testing each group separately (except group 7 where the sample size is too small) yields the OJ - ^/"^^ >< o y^ T ^^y^ CM -2-1012 Quantiles of standard normal Figure 2 Norn a) probability plot of |.v, = (z, -z^)lR,y same result, i.e. normality is not rejected at a 10% level for any group, thus justifying the use of Bartlett, F- and t- tests. Bartlett's test for testing equality of variances yields a P-value of 0.81. In view of this, it is not really necessary to test the groups in pairs for equality in variance using an F-test, but as a source of additional information we have carried out the tests obtaining the lowest P-value of 0.078 for groups 4 and 5. Thus, based on these data, the hypoth- esis of a uniform measurement error independent of geo- graphical location, time, depth etc., could not be rejected. To investigate possible differences in efficiency for the participating vessels we did an ANOVA test. We found a P-value less than 10"^, indicating significant differences. This finding was consistent with earlier findings in calibration experiments (Pelletier, 1998). Thus, because £(Z,) cannot be considered equal for all groups (also the confidence intervals in Fig. IB), a pooled variance could be used for estimating d\. Alternatively, e, in Equation 2 could be replaced by the variables z\=z-Zj,, which are adjusted for group means and are identical to the residuals from the ANOVA fit. The resulting estimates with the last approach are a;^=0.069 and o;.=0.263. The bootstrapped standard errors of (t| and a^ are 0.0077 and 0.0147, respectively (1000 bootstrap replicates were used). Some caution should be exercised in interpreting these numbers (see e.g. Srivastava and Chan, 1989). Compared with the total variability of the survey, the measurement error of a single haul is relatively small. For the last 5 years ( 1996-2000 ), var(.v, ) for the nonzero catches varied between 1.38 and 2.06 for the winter survey and between 2. .53 and 3.92 for the autumn survey. Thus, a\ is about 2-5% of the total variation. This is the percentage of the variation that we cannot expect to be able to explain by explanatory variables. One should carefully note that these numbers are on the log scale. If antilogs of the catch rate were to be used, the additive model (Eq. 1 ) would have to be replaced by a multiplicative model, and the relative magnitude of variances would be changed. The results for length-stratified data are shown in Table 2, as well as the results obtained by measuring the catches by weight instead of by numbers, i.e. by replacing n,^ in (Eq. 3) by, m,^, where w,^ is the weight of the catch in kilograms. Only hauls where both vessels 724 Fishery Bulletin 100(4) Table 2 Estimates of a'\ (with bootstrapped means and standard errors) for catches measured in kg (first column), total number offish caught (second column), and catch offish in various length-stratified gi'oups measured in numbers (last four columns . The number of hauls where both vessels caught at least 10 cod is given in the fourth row, the groups excluded due to less than four remaining hauls are given in the fifth row, the average catch of the remaining hauls in the sixth row and P-values from Bartlett's test for equal variances, in the seventh row. The means and standard errors were estimated by using 1000 bootstrap replicates. Total no kg offish caught Catch offish in length stratified gi-oups >0 cm >0 cm <30 cm 30-59 cm >60 cm >30cm (t2 0.0744 0.0690 0.1124 0.0801 0.0855 0.0643 boot, mean of o-^ 0.0740 0.0684 0.1113 0.0788 0.0844 0.0637 boot.SEofCT2 0.0096 0.0077 0.0140 0.0114 0.0122 0.0083 No.ofhaulswithatleast lOcod 130 130 113 117 103 118 groups excluded none none 2,3 10 2,7,10 10 V 4.77 5.28 4.61 4,06 3.61 4.52 P- value, Bartlett 0.11 0.81 0.31 0.87 0.60 0.97 collected 10 specimen or more were included, and only groups with at least four such hauls. The null hypothesis of equal variance was not rejected for any of the length groups (P-values from Bartlett's test are giyen in Table 2). It is seen that of is highest for small fish, and the bootstrapped standard errors indicate that the difference is statistically significant. Indeed, let D = o'^,„,„ii - ^f /nr/;<> denote the difference in measurement error between small (<30 cm) and large (>30 cm) fish. If (Tf ,.,„„// and o-/^^^,^ are independent and normally distributed (their bootstrap distributions are approximately normal) with standard errors as given in Table 2, it follows that D-NiO. 0.0140- + 0,0083^) under the null hypothesis of equal measurement errors. The corresponding one-sided P-value for the observed D is 0.0016. The hauls differed in towed distance, with the two most frequently recorded values being 1 and 1.5 nmi which correspond to 20 and 30 minutes tow duration. Stratifying on tow duration (which is recorded more precisely than towed distance), with hauls of less than 25 minutes duration (33%) in group A and the remaining hauls (67%) in group B, and estimating the measurement error for each group separately, we get ct^=0.0707 and 0.0656 for group A and B, respectively. Thus, there is no significant difference due to tow duration. For fish less than 30 cm, the corresponding estimates are 0.115 and 0,107 for group A and B, respectively. No significant relationship was found between the mag- nitude of the catches and their differences, A regression analysis was performed with the absolute value of the mean-adjusted differences in catch rate, \2',\, as the de- pendent variable and the average catch y, = (y, i + .v, 2)/2 as the independent variable. The regression equation was |2' I =0.20-1-0. 019v, and the P-value under the null hypothesis of no relationship was 0,31, However, the residuals from the analysis were skewed with a long right-hand tail; therefore a bootstrap test was also done, resulting in an empirical P-value of 0.137, and again the null hypothesis of no relationship was not rejected at a 5% level. Discussion We have estimated the measurement error (a^) of a trawl haul by using data from parallel trawling experiments, including 130 parallel hauls from 10 groups of experi- ments. No significant differences in o'j among the groups were found. Thus, o'j seems to be independent of year, time of the year, and geographical position at which the haul was taken. It also seems to be independent of the catch size on a logarithmic scale. The magnitude of o'j is small ( = 2-5%) compared with the total variability in the survey trawl catches. The results are preliminary in that they are based on a limited set of hauls and more extensive experi- ments would be of interest to check their consistency. In another investigation, Pelletier (1998) examined the vessel effect between two research vessels. These data could possibly be used to test the general pattern revealed by analyses of data in the current study. Stromme and lilende (2001) examined a total of 365 paired hauls from intercalibration experiments off Namibia in 1998 and 1999 between the research vessel Dr. Fndtjof Nansen and commercial trawlers. These data of Namibian hake (catch in kg/h) were kindly made available to us, and an estimate of 6'j=QAd was obtained for the measurement error. The variance of y, for all the 365 hauls was 2,0; the measure- ment error for this study on the log scale was about 10% of the survey variance. Actually o'j may be an overestimation of the measure- ment error because all the explanatory variables are not exactly the same for the two sets of measurements in Equation 1, For example the geographical location is not the same and the fish densities may differ from one vessel to the other because of the distance between them. However, because the towed distance is typically 5-10 Hiellvik et al : Measurement error in marine survey catches 725 times the distance between the vessels, we believe this factor to be of minor importance. Another problem is the determination of the towed distance. The uncertainty connected with subjective judgments and inaccuracies in the GPS should be included in the measurement error because these factors are also present at a standard survey haul. However, it is not obvious to what extent the differences in the recorded towed distances are due to differences in subjective judgments or to differences in actual towed distances. In our calculations, we used the recorded values from both vessels for d,. . If there is no real difference in the towed distances within a comparison, d~ is expected to decrease by setting^, j=(i, 2 f"*" ^1' hauls, thus eliminating one factor of uncertainty. At the other extreme, if the subjective judgments are perfect, and the recorded differences in towed distance are due to real differences, one would expect (T^ to increase by setting (/, ,=c?, .2, because an extra error then is added. Actually, by using the values from vessel 2 only and by setting c^, i=(^, ■, the resulting estimate of&l is a'i=0.061, which is a reduction by about 119< . Even though this value is statistically insignificant, it indicates that uncertainty connected to the measurement of towed distance constitutes a part of the measurement error (see also God0 et al., 1990). In the "Results" section, the null hypothesis of equal efficiency for all the participating vessels was rejected. By joining data from the groups where the same pair of vessels participates, the ability to detect differences in efficiency between the vessels increases through an increased sample size and a smaller number of simultaneous tests (with N tests and a nominal level a, the null hypothesis of equal efficiency is rejected for P-values smaller than the Bonferroni corrected level a/N). For the N=6 tests, we obtained P-values 0.0005 for group 4, 0.009 for groups 1 and 2, 0.011 for groups 6 and 10, 0.034 for groups 5 and 9, 0.107 for groups 7 and 8, and 0.123 for group 3. With a level a=0.05 we have a/6=0.0083, and for group 4 the difference is clearly statistically significant. The higher efficiency of Jan Mayen (JM) in this group was probably due to her heavier trawl doors. At a 10*7^ level, the vessel Army Kraemer (LIZY) was significantly more efficient than Johan Hjort (JH) in groups 1 and 2, and Michael Sars (MS) was significantly more efficient than JH in groups 6 and 10. The differences between G.O. Sars (OS) and JH (groups 5 and 9), OS and JM (groups 7 and 8) and GS and LIZY (group 3) were not significant. However, excluding group 4, and ignoring statistical significance, the vessels can, in fact, be ranged consistently after increasing efficiency as higher for small fish. One explanation may be that the interaction between small fish and the trawl gear is more variable (God0 and Walsh, 1992); another possibility is that small fish operate more in patches than do large fish. If the last assumption is correct, a reduction in (T^ could be expected with increasing tow length. However, for the .set of tows considered, we found no significant difference in a^ due to tow distance. Consistent with the length dependency of the measure- ment error is the length dependency of the total variability of the surveys. The average varly, ) for the winter surveys 1996-2000 for fish <31cm and >64cm, was 2.49 and 0.98, respectively; whereas for the unstratified data it was 1.66. The corresponding numbers for the autumn surveys 1996- 2000 were 3.03, 1.47. and 2.98 for small, large, and unstrati- fied fish, respectively. All numbers are for nonzero catches. Trawl catches have been considered highly variable (see e.g. Gulland, 1964; Doubleday and Rivard, 1981) and as a result the reliability of trawl survey estimates have been questioned. Abrupt changes in catch size and composition over a short time in a limited area have demonstrated the difficulties in using the information as a relative estimate of density without an understanding of the nature and causes of the variability (Godo, 1994). Unexpected an- nual changes in survey indices may also be a problem for a reliable evaluation of fish stocks and can be attributed to a variable bias (changes in catchability) among years (Pennington and Godo, 1995). Our analysis demonstrates that for the bottom trawl survey in the Barents Sea, catch rates and composition from the applied survey trawl are repeatable up to a relatively small and constant measure- ment error and are hence expected to give a reliable pic- ture of the relative fish density at a given site and time. Further, the measurement error of this sampling gear is small compared with the total observed variability. For a particular survey it appears that most of the sui-vey vari- ance is caused by station-to-station differences in catches rather than local conditions at a station. This may be taken as an indication that shorter and more frequent tows may be more efficient for monitoring this cod stock. Moreover, when controlling trawl geometry (Godo and Engas, 1989) and towed distance (Godo et al., 1990), it should be possible to establish explanatory factors to be included in the sur- vey assessment procedure. To the degree that one is able to establish models to determine fish densities at any station, the comparability of density measures throughout the dis- tribution area will improve. The consequences will thus not only be more reliable survey estimates, but we also expect a better understanding of distributional patterns in rela- tion to the physical and biological environment. JM- -^GS- 15,9,31 -^JH- ' LIZY /MS. (4) Acknowledgments The numbers in parentheses refer to experiment groups, and for each group one vessel to the left, and one to the right of the corresponding arrow, are involved, the one to the right being always the most efficient one. There also seems to be a significant difference in the measurement error for small and large fish, it being We are grateful to Atle Totland for help with handling the data and to Michael Pennington for comments on the paper. We are also indebted to the scientific editor and three anonymous referees for a number of useful comments that improved the paper. The work was financially supported by the Norwegian Research Council (127198/120). 726 Fishery Bulletin 100(4) Literature cited Bickel, P. J., and K. A. Doksura. 1977. Mathematical statistics: basic ideas and selected topics. Holden-Day, Oakland. CA, 493 p. Doubleday, W. G., and D. Rivard. 1981. Bottom trawl surveys. Can. Spec. Publ. Fish. Aquat. Sci. 58:1-273. Engas, A. 1994. The effects of trawl performance and fish behaviour on the catching efficiency of demersal trawls. In Marine fish behaviour related to capture and abundance estima- tion (A. Ferno and S. Olsen, eds.), p. 45-68. Fishing New Books, Oxford. God0, O. R. 1994. Factors affecting reliability of abundance estimates of ground fish from bottom trawl surveys. In Marine fish be- haviour related to capture and abundance estimation (A. Feme and S. Olsen, eds.), p. 166-199. Fishing New Books, Oxford. Godo, O. R., and A. Engas. 1989. Swept area variation with depth and its influence on abundance indices from trawl sun'eys. J. Northwest Atl. Fish. Sci. 9:133-139. God0, OR, and S.Walsh. 1992. Escapement of fish during trawl sampling— implica- tion of resource assessment. Fish. Res. 13:281-292. Gode, O. R., M. Pennington, and J. H. Volstad. 1990. Effect of tow duration on length composition of trawl catches. Fish. Res. 9:165-179. Gulland, J.A. 1964. Catch per unit effort as a measure of abundance. Rapp. P-V. Reun. Cons. Int. Explor Mer 155:8-14. Korsbrekke, K., and O. Nakken. 1999. Length and species dependent diurnal variation in catch rates in the Norwegian Barents Sea bottom trawl surveys. ICES J. Mar Sci. 56:284-291. Michalsen, K., O.R. Godo, and A. Ferno. 1996. Diel variation in the catchability of gadoids and its influence on the reliability of abundance indices. ICES J. Mar Sci. 53:389-395. Pelletier, D. 1998. Intercalibration of research survey vessels in fish- eries: a review and an application. Can. J. Fish. Aquat. Sci. 55:2672-2690. Pennington, M., and O. R. Godo. 1995. Measuring the effect of changes in catchability on the variation of marine survey abundance indices. Fish. Res. 23:301-310. Polacheck, T., and J. H. Volstad. 1993. Analysis of spatial variability of Georges Bank had- dock iMelanogrammus aegleftnus) from trawl survey data using a linear regression model with spatial interaction. ICESJ. Mar Sci. 50:1-8. Srivastava, M. S., and Y. M. Chan. 1989. A comparison of bootstrap method and edgeworth expansion in appro.ximating the distribution of sample variance — one sample and two sample cases. Commun- ication in statistics — simulation 18:339-361. Stromme, T, and T. lilende. 2001. Precision in systematic trawl surveys as assessed from replicate sampling by parallel trawling off Namibia. S. Afr J. Mar Sci. 23:385-396. 727 Abstract— Reproductive organs from 393 male and 382 female porbeagles (Lamna nasus), caught in the western North Atlantic Ocean, were examined to determine size at maturity and reproductive cycle. Males ranged in size from 86 to 246 em fork length (FL) and females ranged from 94 to 288 cm FL. Maturity in males was best described by an inflection in the rela- tionship of dasper length to fork length when combined with clasper calcifica- tion. Males matured between 162 and 185 cm FL and 50% were mature at 174 cm FL. In females, all reproduc- tive organ measurements related to body length showed a strong inflection around the size of maturity. Females matured between 210 and 230 cm FL and 50% were mature at 218 cm FL. After a protracted fall mating period (September-November), females give birth to an average of 4.0 young in spring (April-June). As in other 1am- nids, young are nourished through oophagy. Evidence from this study indi- cated a one-year reproductive cycle and gestation period lasting 8-9 months. The reproductive biology of the porbeagle shark (Lamna nasus) in the western North Atlantic Ocean Christopher F. Jensen North Carolina Department of Environment and Natural Resources Division of Marine Fishenes PO Box 769 Morehead City. North Carolina 28557-0769 E-mail address Caltimuscdta'aol com Lisa J. Natanson Harold L. Pratt Jr. Nancy E. Kohler National Marine Fisheries Service, NOAA 28 Tarzwell Dr Narragansett, Rhode Island 02882 Steven E. Campana Marine Fish Division Bedford Institute of Oceanography PO Box 1006 Dartmouth, Nova Scotia Canada B2Y 4A2 Manuscript accepted 29 May 2002. Fish. Bull. 100:727-738 (2002). The porbeagle {Lamna nasus), a pe- lagic shark in the family Lamnidae, inhabits the cold temperate waters of the North and South Atlantic, South Pacific, and southern Indian Oceans, as well as the subantarctic region of the Southern Ocean (Svetlov, 1978; Compagno, 1984). In the western North (NW) Atlantic Ocean, the porbeagle ranges from the Flemish Cap and the Grand Banks off southern Newfoundland, Canada, to the Gulf of Maine and (rarely) south to New Jersey (Templeman, 1963; Compagno, 1984). The porbeagle is most commonly encountered from the Gulf of Maine to the Grand Banks, where it has been the subject of a commercial fishery since 1961 (O'Boyle et al.i; Campana et al^^). Seasonal abundance is related to north-south migrations (Aasen, 1963; Campana et al.-^; Aasen''). Lamnid sharks are ovoviviparous and nourish their embryos by oophagy (Lohberger, 1910). Early descriptions of porbeagle embryos exhibiting ooph- agy were documented by Swenander (1907) and Shann (1911, 1923), and more recently by Francis and Stevens (2000). Litter size has been variously reported as one to five pups (Dunlop, 1897; Swenander, 1907; Shann, 1911, 1923; Gauld, 1989; Francis and Stevens, 2000). Birth size has been reported as 1 O'Boyle, R. N., G. M. Fowler, P. C. F. Huriey, M. A. Showell, W. T. Stobo, and C. Jones. 1996. Observations on porbeagle (Lamna nasus) in the north Atlantic. DFO (De- partment of Fisheries and Oceans) Atl. Fish. Res. Doc. 96/24, 29 p. Marine Fish Division, Bedford Institute of Oceanog- raphy, RO. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2. ' Campana, S., L. Marks, W. Joyce, P. Hurley, M. Showell, and D.Kulka. 1999. An ana- lytical assessment of the porbeagle shark {Lamna «a.s(/s) population in the Northwest Atlantic. CSAC (Canadian Stock Assessment Secretarate) Res. Doc. 99/158, 57 p. Ma- rine Fish Division. Bedford Institute of Oceanography, P.O. Bo.x 1006, Dartmouth. Nova Scotia, Canada B2Y 4A2 ' Campana, S., W. Joyce, L. Marks, P. Hurley, L. J. Natanson, N. E. Kohler, C.F. Jensen, J. J. Mello, and H. L. Pratt Jr 2000. The rise and fall (again) of the porbeagle shark population in the Northwest Atlantic. Unpubl. manuscr Marine Fish Division. Bedford Institute of Oceanography, P.O. Box 1006, Dartmouth. Nova Scotia, Ca- nada B2Y 4A2 ^ Aasen, O. 1961. Some observations on the biolog)' of the porbeagle shark (Lamna nasus L.i ICES CM. Copenhagen 1961, Near Northern Seas Committee 109:1-7. 728 Fishery Bulletin 100(4) 60-80 cm total length (TLl (54-72 cm fork length |FL]) (Shann, 1923; Compagno 1984, Francis and Stevens, 2000). Bigelow and Schroeder (1948) speculated that females are gi-avid by 152 cm TL (136 cm FL), whereas Aasen (1963) concluded that females mature between 200 and 250 cm TL (193-240 FL) and males mature between 150 and 200 cm TL (146-193 FL). In a more comprehensive study, Francis and Stevens (2000) found that females mature between 165 and 180 cm FL in the South Pacific. In this study, we present the results of a comprehensive examination of porbeagle reproduction in the NW Atlantic Ocean. We define the sizes and stages of maturity for both sexes and provide insights into the reproductive cycle, as well as embryo sex ratios and litter sizes. These parameters will be useful for the refinement of fishery management plans for the porbeagle shark in the NW Atlantic Ocean. Materials and methods Porbeagles were collected with pelagic longline onboard both U.S. and Canadian commercial fishing vessels and U.S. research vessels fishing in U.S. and Canadian waters from the Gulf of Maine and Georges Bank ( northeastern U.S.) to the Grand Banks off southern Newfoundland. Sev- eral specimens were obtained at a sport-fishing tourna- ment held on Stellwagen Bank, Massachusetts. Sampling took place between 1979 and 1999, but most of the data were obtained during 1993, 1994, and 1999. Morphometries For each shark, five lengths were measured over the curve of the body to the nearest half centimeter (cm): interdor- sal length (posterior dorsal fin base to origin of second dorsal fin; IDL); dorsal length (origin of first dorsal fin to precaudal pit; DL); precaudal length (snout to precaudal pit; PCD; fork length (snout to fork of tail; FL); and total length (snout to a perpendicular line from the tip of the upper caudal fin in a natural position; TL). Total length of embryos were measured along a straight line with the tip of the tail fully extended (TL,) because of the difficulty in obtaining over-the-body or FL measurements (or both) in embryos less than 6 cm TL. FL and PCL were taken when possible. FL are reported for all sharks in our study; TL values are presented for embryos with calculated or measured FL in parentheses. TL can be converted to FL by using the following regression (Campana et al.^): TLj can be converted to FL for embryos by using the regression FL = 0.832(rL )-0.19 FL=0.885(rL)-i-0.99 [/•-=0.99,;!=361]. All other length and weight conversions can be found in Campana et al.' Aasen's ( 1963) TL values were converted to FL by using the approximate formula: FL = 0.947*(Aast'nrL) + 3.64 (Campana^). [7-2=0.99, ?! = 131] ^ Campana, S.E. 2000. Unpubl. data. Department of Fisher- ies and Oceans, Bedford Institute of Oceanography, P.O. Box 1006, Dartmouth. Nova Scotia, Canada, B2Y 4A2. Whole weight in kilograms (kg) was taken when possible. Literature values of TL were converted to FL for compari- son and the conversions are presented throughout the text in parentheses. The converted values from the literature should be considered good estimates only because of the variation in measurement techniques between studies. Maturity Indicators A number of measurements, weights, and conditions were taken on reproductive organs in both sexes to develop indi- ces of maturity following Pratt (1979, 1993, 1996). Most specimens were measured fresh; however, some frozen reproductive tracts were also measured. Reproductive tracts were measured on the right side of the specimen. Organ terminology follows Pratt (1979) (Fig. 1, A and B). For more detail, uteri were divided visually into anterior and posterior segments. The anterior segment is defined as the portion between the origin of the uterus at the isthmus and the point where both uteri join. The posterior segment is the portion from the junction to the posterior constriction of the uterus (Fig. IB). A length measurement was taken for the anterior uterus, and width was taken midway along this portion of the uterus. Active ovulation was defined as occurring if 1) ova were entering the ostium or were present inside the upper oviduct, 2) ova and capsules were present in the oviducal gland, 3) encapsulated ova were present in the isthmus, and 4) encapsulated ova were present in the uterus. Prior mating activity was assessed according to the pres- ence or absence of a vaginal membrane (hymen), deter- mined by passing a probe through the posterior end of the uterus into the cloaca. The presence of vaginal mating scars was used to determine if recent mating had taken place. Scars were identified as either healed or recent. Maturity status of both sexes was assigned to each shark based on all reproductive organ characteristics (Pratt, 1979). The size at SO'/r maturity was determined by fitting a logistic regression to a plot of percent maturity versus FL. Embryos were either frozen or preserved in 10% buff- ered formalin. Litter size and embryo sex were deter- mined in the field or in the laboratory under a dissecting microscope. Embryo growth was related to month. Linear regressions were fitted to the relationship between mean embryo length and month for samples from the NW At- lantic and southwest Pacific Oceans and compared with analysis of covariance (ANCOVA). Results Male length at maturity Reproductive data were obtained from 393 male porbeagle sharks ranging in size from 86 to 246 cm FL, of which Jensen et al : The reproductive biology of Lamna nasus in the western North Atlantic Ocean 729 Clasper Ampulla B Junction Vagina Epigonal Epididymis Oviducal Gland X_^ Head of Epididymis Upper Oviduct Osteum Cloaca Lower Uterus Ovary Efferent Pore Anterior Uterus Figure 1 Reproductive systems of the porbeagle shark: (A) male reproductive tract with closeup inserts of the head of the epididy- mis and the ampulla epididymis and (B) female reproductive tract with a closeup of the upper oviduct leading into the ostium. 730 Fishery Bulletin 100(4) 160 140 120 e 100 60 40 20 - N =253 o Adult A Juvenile o ^ o c£) c OO Oo OO o o ®o n° 0,0*0 o o x^^« 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Fork length (cm) Figure 2 Relationship of right testis length to fork length. 250 200 ^ ? E £ 150 A. A A A ^ ^ ^ Uncalcified claspers, no spermatophores A ■^ A . M^ AA^ "^ A X Uncalcified claspers with spermatophores A A ^^'J^ A^AAiA ^ O Calcified claspers. no speramtophores J^S^^%^ •Calcified claspers with spermatophores Clasper en o 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Fork length] (cm) Figure 3 Relationship of right outer clasper length to fork length. Relative amount of clasper calcification and presence or absence of spermatophores are also indicated. reproductive organs were measured for 267. Development of the siphon sacs, as well as internal male reproductive organs — testes (diameter and length), epididymis, and ampulla epididymis — is gradual in relation to FL (Fig. 2). The development of the claspers. however, shows a dis- tinct inflection (Fig. 3). Claspers begin to elongate when the males approach 120 cm FL, and rapid clasper growth begins at 135 cm FL and slows by about 170 cm FL when they have reached their maximum adult size of 19-24 cm. The largest shark with uncalcified claspers but containing spermatophores in the ampulla epididymis was 176 cm FL, and the smallest shark with calcified claspers, but with no spermatophores, was 166 cm FL. Claspers rotate freely at all sizes in the porbeagle; therefore clasper rotation is not a good measure of maturity. Males of 135 to 184.5 cm FL are in a transitional phase leading to maturity. At this stage, claspers are lengthening and the head of the clasper (rhipidion) is beginning to develop, whereas at 149 cm FL Jensen et a\ The reproductive biology of Lamna nasus in the western Noith Atlantic Ocean 731 1 A y- r 0.8 - / / IVlale / Female 0.6 / 50% \ 0.4 173 7 cm / 217 5cm 0.2- 1 / n J 90 110 130 150 170 190 210 230 250 270 290 Fork length! (cm) 0) 1 0.8 - B 0.6 0.4 0.2 .' Male 50% 8 1 years 13.1 years 0 2 10 12 14 16 18 20 Age (years) Figure 4 Maturity ogives for male and female porbeagle sharks: (A) ogives based on length, (B) ogives based on age. Fifty percent maturity is indicated. and greater, spermatophores appear in the ampulla epidid- ymis. By 162 cm FL fully calcified claspers were observed. Individual variation exists in body length and the order in which these developments take place. All males observed were mature by 185 cm FL based on clasper calcification and the ability of the rhipidion to open. The presence of spermatophores may be used in addition; however, absence of spermatophores can be attributed to season as well as to maturity. Each clasper bears a sharp, conical, calcified spur near the tip which folds out of the distal surface as the rhipidion opens. In adults, the sharp spur tip was either covered by an epidermal membranous sheath, or naked. A naked spur is an indicator of mating activity because mating ruptures the sheath. The FL at which 50% of the male population is mature, based primarily on clasper con- dition, was estimated as 174 cm FL (Fig. 4A). Both testis size and the amount of spermatophores in the ampulla epididymis showed distinct fluctuations related to the mating season. From November through mid December, the testis lobes were noticeably reduced in diameter, the testes were smaller, and the epigonal organ had infiltrated the space previously occupied by the 732 Fishery Bulletin 100(4) bUU o o N =271 o 500 o o § E F 4nn S oO°oocP * Juvenile o o 03 300 o Adult ^8 «^2§ ° o O 200 A oo oo oo OOo 100 • » A ^' 0 **«#.*« *t'>«»irf»**4»V*. *»f^' fVA^A^ 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 Fork length (cm) Figure 5 Relationship between ovary length (mm) and fork length (cm) in the porbeagle shark. Juveniles and adults are indicated. 60 N = 281 E E 50 E 40 * Juvenile 0 Adult 30 •S 20 - O ;g O 10 J ^»>o° ^ 2- 0 oo* A 1 - A A A A "A ° A n 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 Forl< length (cm) Figure 7 Relationship between maximum ovum? diameter (mm) and fork length (cm) for the | porbeagle shark. as flaccid uteri with trophonemata and reduced ovaries containing few maturing or mature oocytes. Juvenile females have small undeveloped ovaries with white to clear oocytes; the uterus appears narrow and constricted, and the oviducal gland is seen as a barely perceptible widening of the oviduct. The juvenile and mature virgin sharks that we examined had a membrane separating the vagina from the cloaca, whereas mature (reproductively active) sharks had no membrane. In nongravid females, maturity is based on the presence of mature oocytes in a developed ovary, an expanded uterus, a well-developed oviducal gland, internal and external mating scars, and the absence of a vaginal membrane. The smallest mature female in our sample was 210 cm FL based on the absence of a vaginal membrane and the condition of the internal reproductive organs, and the largest immature female was 230 cm FL based on the presence of a vaginal membrane and internal organ condition. Fifty percent of the female porbeagle population was mature at 218 cm FL (Fig. 4A). Mating injuries Fresh mating scars, containing at least small areas of unhealed dermal lacerations, were observed from late September through mid-December, during which time most adult females appeared to have recently mated or were in the early stages of pregnancy. Scars and cuts were often observed on either or both pectoral fins as single or multiple jaw outlines, and the trailing edge of the fin was often shredded. Tooth scrape marks, puncture wounds, and gouges, some as open, penetrating subdermal lacerations, were observed mainly along the posterior lat- eral, dorsal, or ventral body surface. None of the wounds appeared to have penetrated the abdominal wall nor were they debilitating. No infected or necrotic tissue was seen in these wounds. Most mature female porbeagles had dis- tinct fresh or healed mating scars. Internal mating scars were observed on the vaginal walls from late September through mid-December. The scars were generally small, round to ovoid, hematose marks on an otherwise white to light yellow vaginal wall. These were probably caused by the spur that holds the clasper in place during copulation. Gravid females, embryo growth, and nutrition Data from 80 litters and 309 embryos were obtained from gravid females sampled from mid-September through April. In addition to embryos, these females had several different types of egg capsules in their oviducal gland, isthmus, and uteri. These were empty capsules, capsules containing one large ovum without a visible embryo (single ovum capsules), capsules with one developing embryo and attached yolk sac, capsules with 22-45 blastodisc-stage ova, and capsules containing 8-100 nonblastodisc-stage ova (nutritive capsules), some appearing atretic and in which individual ova were difficult to detect. The number of capsules present in each uterus ranged from none to 63 (Fig. 8). The proportion of mature nongravid females with recent mating scars, but no obvious fertilized ova, decreased from September through December, whereas the proportion of gravid females increased (Fig. 9). All mature females ex- amined in December were gravid. During October, several mature females contained either parts of, or entire, sper- matophores in the uteri, but no ova capsules, indicating 734 Fishery Bulletin 100(4) that mating had taken place but that ova were not yet fertilized. On the basis of the occurrence of early- stage gravid females, the ripe condition of the males, and the presence of recent mating scars, we concluded that most mating occurs from late September through November. Mature males with copious amounts of spermatophores and mature females were captured on the same gear numerous times during the fall. The right uterus of the one gravid female observed during late September had one single ovum capsule with no visible embryos, along with several blastodisc or nutritive capsules, whereas the contents of other uterus appeared to have been aborted. This was the only recently mated female observed in September. Some gravid females in October and November con- tained capsules with a single ovum, 6-7 mm diameter, in addition to varying numbers of empty, blastodisc, or nutritive capsules. These single ovum blastodisc cap- sules (usually two per uteri) were considered to be the annual developing embryos, their single ovum larger than those of the aggregate blastodisc or nutritive capsules. Gravid females examined during October had either single ovum capsules or embryos 0.9-6.8 cm TL^ (0.6-5.3 cm FL), and those seen in November had a de- crease in occurrence of single-ovum capsules, and embryos ranged from 1.1 to 37.8 cm TL^ (0.7-31.2 FL). No gravid females with single ovum capsules were seen during De- cember, and embryos ranged from 2.3 to 43.1 cm FL (Fig. 10). Only one nearterm, gravid porbeagle, caught in mid- April, was examined in our study. This shark had three embryos ranging in size from 59 to 72 cm TL^ (50-59 cm FL). The considerable variation in embryo length among litters during early gestation indicates a protracted mat- ing season. Postpartum females were observed from the first week of May to the first week of June. Our data from these fe- males suggested that parturition extends from early April through early June. The presence of postpartum females in May-June after the September-November mating in- dicates a gestation period of about 8-9 months. Although numerous gravid females are present from late September through December on the Scotian shelf and Grand Banks region, mature gravid or nongravid fe- male sharks are seldom seen from January through June in the Canadian fishery. Gravid females were caught be- tween Georges Bank and the Grand Banks (Fig. 11). Fecundity Embryos were found to be at the same developmental stage in each female, although embryo size varied as much as 14.6 cm in one litter Runts were encountered in five females and although clearly developing, were consider- ably smaller than the coexisting embryos. Fecundity was calculated by using all embryos. The average number of embryos per porbeagle was 4.0 (304 embryos from 76 lit- ters) although litters ranged from 3 to 6. In 66 litters there were two embryos per uterus. The sex ratio of 202 embryos (99 male and 103 female) was not significantly different from one (;t:2=0.08, P>0.1). Figure 8 Photograph of an ovary and open uterus of a gravid porbeagle showing a developing embryo amid ova capsules. In the uteri of six females, fertilized ova were found in the same ova capsule with one to three late developing yolksac embryos. Some of these fertilized ova appeared to be decomposing. These capsules were in the same uteri as a larger, developing litter, but the size of the embryos associated with the capsules (averaging 4-20% of the aver- age size of the developing litter) were smaller than runts (runts averaged 24-43% of the average size of siblings) or the main litter. Because of their small size in relation to the developing litter, and their appearance of decomposition, it was presumed that they were ova that had been fertilized late by remaining sperm and that had probably not contin- ued to develop. However, this is not to rule out the possibil- ity that some of these embryos could have developed into runts or be consumed when littermates initiated feeding on egg capsules. Further research is being conducted on this subject and will be reported at a later date. Embryonic development and growth Newly fertilized ova, without visible embryos, were 6-7 mm in diameter As the embryo developed, their yolk sacs decreased in size, and by the time the embryos were between 4.2 and 9.2 cm TLs (3.3-4.5 cm FL), the yolk sacs were tiny, white, ovoid structures attached to the embryo by a short stalk. Embryos were found in capsules up to 4-^.2 cm TLj, (3.1-3.3 cm FL). The smallest posthatch embryos were 3.2^.2 cm TL, (2.5-3.3 cm FL). The abdomens of posthatch embryos between 4.7 and 5.1 cm TL^, (3.7 and 4.0 cm FL) were slightly swollen. Functional embryonic denti- tion appeared in embryos at 12 cm TL^ (9.8 cm FL). As the embryo continues to grow, the yolk stomach increases in size as a result of in utero consumption of ova capsules by the embryo. The stomach becomes grossly distended by the time the embryo reaches 27-13 cm TL J 23-36 cm FL). The large amounts of yolk material in the stomachs of these embryos was characteristic of oophagy. Frequently, parts of ova cap- sules, parts of, or whole, ova, and white flocculent material were observed in the clear intrauterine fluid surround- Jensen et al. The reproductive biology of Lanma nasus in the western North Atlantic Ocean 735 W=4 N=19 N=60 W=27 W = 1 W=5 /V=9 N-2 W = 1 100 80 60 40 20 I I Sept Oct Nov Dec Jan Feb March April IVIay June July August Month D%non-gravid ■ % gravid Figure 9 Relationship of gravid to nongravid adult female porbeagle sharks by month from 1979 to 1999. ing the embryos, suggesting recent embryo feeding. One embryo was seen with a whole, undamaged ova capsule protruding from its mouth. The stomach contents of embryos from several litters did not show parts of other embryos, suggesting that adelphophagy (com- petitive embryonic cannibalism), found in the sandtiger shark (Gilmore et al, 1983) does not occur in the porbeagle. The existence of devel- oping runts in several litters and consistently two embryos per uterus reinforces these obser- vations. A slight modification to our knowledge of oophagy would be the possible consumption of the late-fertilized ova capsules that had small undeveloped and decomposing embryos along with unfertilized ova, as mentioned previously. Most gravid porbeagles examined during this study were ovulating and, there- fore, nourishing their embryos. The times at which ovulation ceases, ovary size decreases, and the large amount of yolk in the stomach is consumed, need further definition. The mean growth rate of embryos from this study was 11.4 cm per month; however, this is most likely inflated by the lack of larg- er embryos. Combining data from the pres- ent study with previous data from the North Atlantic, which includes later term embryos (Francis and Stevens, 2000), leads to an estimate of 8.15 cm per month. The regression of growth per month for the North Atlantic embryos was significantly different from that of the South Pacific population in intercept (time of year) but not in slope (growth rate) (ANOVA P>0.10 slopes, P<0.1 intercept) (Fig. 10). (NH) FL=22 7 r-=0 68 Sep Mar Apr May Aug IVlonth Southern Hemisphere Northern Hemisphere NW Atlantic. This study Figure 10 Monthly variation in mean lengths of embryo litters from both hemi- spheres. Data indicated by open circles and closed triangles are from Francis and Stevens (2000) and Francis,'' respectively. Neonates and young of the year The smallest free-swimming porbeagle examined during this study was 77.5 cm FL (87 cm TL) on 24 May Other records of free-swimming individuals measured in April- June, ranged from 55 to 79 cm FL (mean=71 cm FL, /2=9) 736 Fishery Bulletin 100(4) Grand Banks ♦ February ▲ Apnl ■^ September -> October • November ® December ^ Map of the capture locations of al Grand Banks are indicated. Figure 11 I gravid female porbeagles by month of capture. Approximate locations of Georges Bank and the (Kohler''). The largest embryos in our study from one litter in April were 49-60 cm FL (59-72 cm TL). Discussion In the NW Atlantic, male porbeagles mature between 166 and 184 cm FL, and 50'^^^f maturity was at 174 cm FL, which corresponds to an age of 8 years (Fig. 4B; Natanson et al., 2002). The most accurate means of determining male maturity is clasper length and calcification. Because of the distinct seasonal variability, the presence of seminal products alone, is not a good indicator of maturity. Clasper rotation occurred at all sizes, eliminating it as a maturity indicator. Our estimate of length at maturity substantially refines Aasen's* estimate of 150 to 200 cm TL (146-193 cm FL) for this population. Aasen^ based his estimates on clasper length in relation to dorsal length. Female NW Atlantic porbeagle sharks mature between 210 and 230 cm FL and 50':{ maturity was at 218 cm FL, at an age of 13 years (Fig. 4B; Natanson et al., 2002). The sizes of all female reproductive organs measured showed a definitive inflection at maturity in relation to body size and are good indicators of maturity. Aasen^ estimated fe- male maturity between 2.0 and 2.5 m TL ( 193-240 cm FL) based on uterus length. Our estimates compare well to the upper part of his range; however our data do not support his lower size at maturity. The seasonality of spermatophore production, observa- tions of females with fresh mating scars, and observations of males and females on the same longline indicated that « Kohler, N. K. 2000. Unpubl. data. NMFS Apex Predators Program, 28Tarzwell Dr. Narragansett, RI 02882. porbeagles mate in the fall, primarily between September and November. In an earlier study, sperm in the shell gland of a female, and the presence of a male caught on the same longline set in October, also suggested a fall mating period (Pratt, 1993). Aasen (1963) proposed a fall mating season (September-October) based on an increase in the amount of sperm present in the males "towards the end of August." This period agrees with the sugges- tion of a long mating season in the SW Pacific porbeagle population (Francis and Stevens, 2000). During this time, females have fresh external mating scars, internal vaginal scars, and spermatophores in the uterus. Gauld (1989) suggested that mating occurs during December-January in the northeast Atlantic on the basis of fresh mating scars on the pectoral fins. Mating injuries have been documented for many spe- cies of sharks. Bite marks on females during mating have been documented in the blue (Pnonace glaitca), nurse iGinglymostoma cirratum), sandtiger, (Carcharias taii- rus), blacknose iCarcharhinits acronotus) finetooth iC. isodon ), blacktip (C. limbatus), sandbar (C plumbeus), and Atlantic sharpnose {Rhizoprumodon terraenovae) sharks (Stevens, 1974; Gilmore et al., 1983; Schwartz, 1984; Cas- tro, 1993, 1996, 2000). Fresh bite marks may coincide with insemination and ovulation, marking the approximate beginning of the gestation period (Parsons, 1983; Castro, 1996). Matthews ( 1950) noted lacerations in the vagina of the basking shark iCetorhinus maximus) from the clasper spur. Pratt ( 1979) also observed abrasions from claspers in the vagina of female blue sharks. Gravid females typically carry single-ovum capsules from September to November, and developing embryos from October to April. Aasen (1963) reported no gravid porbeagles from July to September 1961, in the NW At- Jensen et a\ The reproductive biology of Lamna nasus in the western North Atlantic Ocean 737 lantic. Bigelow and Schroeder (1948) described gravid porbeagles in the (iulf of Maine during November, Janu- ary, and August, althougii the latter month eonfiicts with l)oth the present findings and those of'Aasen ( 1963). Aasen (1963) reported that pregnant females carrying large embryos were observed in late May at the Flemish Cap. Templeman ( 1963) reported three gravid females from the SW Grand Banks during January and February 1953-56. Gauld ( 1989) reported the presence of gravid females from December to June in the NE Atlantic ocean, whereas Francis and Stevens (2000) reported gravid females from March to July in the SW Pacific Ocean. Length variability within and between litters has been found in most porbeagle studies (Fig. 10). We found a dif- ference of up to 14.6 cm in the lengths of individuals in one litter. Gauld (1989) found length differences of up to 11 cm within individual litters, and Shann (1923) found a 12-15 cm difference in embryo length within individual litters. Runts were observed in pregnant porbeagles in our study as well as by Francis and Stevens (2000). As with other lamnids, female porbeagle sharks nurture their young through oophagy. Upon hatching from the sin- gle-ovum capsules and after absorption of the external yolk sac, embryos begin orally feeding on yolk-filled nutritive ova capsules. As a result, the internal yolk stomach of the embryos expands to the large size characteristic of lamnid embryos at this stage in their development. Evidence from this study suggests that adelphophagy does not occur in the porbeagle. The ova capsules observed in our study were similar to those found in the sandtiger and other lamnid sharks described by Gubanov (1972), Fujita (1981), Gruber and Compagno (1981), Gilmore et al. (1983), Francis and Stevens (2000), and Mollet et al. (2000). Embryonic growth was estimated at 8.15 cm per month in our study (Fig. 10). Francis and Stevens (2000) esti- mated embryonic growth of approximately 7.48 cm per month for embryos in the South Pacific. New data from the South Pacific (Francis'') refined this estimate to 8.47 cm per month. Examination of more late-term embryos will further refine this estimate for the NW Atlantic. Our results indicated a one year reproductive cycle, with gestation lasting 8-9 months. Mating occurs between Sep- tember and December Based on the one late-term female and several postpartum females it appears that parturi- tion occurs from early April though June. Aasen (1963) estimated that parturition occurs from late May to early June in the NW Atlantic and suggested about an 8-month gestation period based on the lack of gravid female por- beagles from June to September. All females we examined in December were gravid. Although a nongravid mature portion of the female population could reside elsewhere, we have no data to support this and thus assume that porbeagle females reproduce annually. Parturition time and location and late embryonic development and gnwth rates need further investigation. Francis, M. 2000. Personal commun. National Institute of Water and Atmospheric Research P.O. Box 14-901. Kilbirnie, Wellington, New Zealand. Mean litter size in the NW Atlantic was found to be 4.0 — usually two embryos per uteri. Mean litter size from the SW Pacific Ocean was 3.8 from 138 embryos representing 43 litters (Francis and Stevens, 2000). Mean litter size from the NE Atlantic Ocean was 3.7 based on 12 litters examined by Gauld (1989). The sex ratio of the embryos examined in our study was 1:1, which agrees with the findings of Francis and Stevens (2000) for the SW Pa- cific Ocean. The smallest free-swimming porbeagles in the NMFS historical tagging database range in size from 55 to 79 cm FL (mean 71 cm FL)^ from April to June; this length range, along with that of the one late-term litter (50-59 cm FL), suggests a birth size similar to Aasen's"* ( 1963) prediction of 67 cm FL and to Francis and Steven's (2000) estimate of 58-67 cm FL for the SW Pacific. How- ever, length at birth needs further analysis because few neonates were observed during the parturition period. Acknowledgments We thank Clearwater Fine Foods, Karlsen Shipping, the Atlantic Shark Association, and Stephanie Jane, Inc. for providing access to their fishing vessels. We thank Cap- tain Steve James of the Boston Big Game Fishing Club for access to tournament-landed porbeagles from Stellwagan Bank. We also thank Warren Joyce and Andy Kingman for sample collection. Gregg Skomal of the Massachusetts Division of Marine Fisheries coordinated our participation in the porbeagle tournament. Malcolm Francis kindly provided data on embryo growth from the Southern Hemi- sphere, and corresponding data that he compiled from the Northern Hemisphere, and reviewed an early draft. We also thank Jose Castro for reviewing an early draft. Literature cited Aasen, O. 1963. Length and growth of the porbeagle {Lamna nasus, Bonnaterre) in the North West Atlantic. Fisk. Skrift. Ser. Havund. 13(6):20-37. Bigelow, H. B., and W. C. Schroeder. 1948. Fishes of the Western North Atlantic. Part 1. Lance- lets, cyclostomes, sharks. Mem. Sears Found. Mar. Res., 576 p. Castro, J. I. 1993. The biology of the finetooth shark, Carcharhinus isodon. Environ. Biol. Fish. 36:219-232. 1996. The biology of the blacktip shark, Carcharhinus lim- batus, off the Southeastern United States. Bull. Mar. Sci. 59(31:508-522. 2000. The biology of the nurse shark, Ginglymostoina cir- ratum. off the Florida east coast and the Bahama Islands. Environ. Biol. Fish. 58:1-28. Compagno, L. J. V. 1984. FAO species catalogues. Vol. 4. Sharks of the world. An annotated and illustrated catalogue of the shark spe- cies known to date, parts 1 and 2. FAO Fish. Synopsis 125, vol. 4, parts 1 and 2, 655 p. FAO, Rome. Dunlop. J. 1897. The habits and anatomical structure of the porbeagle shark {Lamna cornubica,L, Cuv. i. Proceedings and Trans- 738 Fishery Bulletin 100(4) actions of the Natural History Society of Glasgow 4:136- 137. Francis, M. P., and J. D. Stevens. 2000. Reproduction, embryonic development, and growth of the porbeagle shark, Lamna nasus, in the southwest Pacific Ocean. Fish Bull. 98:41-63. Fujita, K. 1981. Oviphageous embryos of the pseudocarchariid shark, Pseudocarcharias kamoharai. from the central Pacific. Jap. J. Ichthyol. 28:37-44. Gauld, J. A. 1989. Records of porbeagles landed in Scotland, with obser- vations on the biology, distribution and exploitation of the species. Scot. Fish. Res. Rep. 45, 15 p. Gilmore, R. G., J. W. Dodrill, and P A. Linley 1983. Reproductive and embryonic development of the sandtiger shark, Odontaspis taurus (Rafinesque). Fish. Bull. 81(2):201-225. Gruber, S. H., and L. J. V, Compagno. 1981. Taxonomic status and biology of the bigeye thresher, Alopias superciliosus. Fish. Bull. 79(41:617-640. Gubanov, Y. P 1972. On the biology of the thresher shark. Alopias cul- pinus (Bonnaterre), in the northwest Indian Ocean. J. Ichthyol. 12(41:591-600. Lohberger, J. 1910. Uber zwei riesige Embryonen von Lamna. Abhand. Math.-Phys. Klasse Konig Bayer. Akad. Wissen. Munchen suppl. Bd. 4 no. 2 Abhandlung, 45 p. Matthews, L. H. 1950. Reproduction of the basking shark, Cetorhinus ma.xi- nius (Gunnerus). Phil. Trans. R. Soc. Lond., Ser B., Biol. Sci. 234:247-316. Mollet, H. F, G. Cliff, H. L. Pratt Jr., and J. D. Stevens. 2000. Reproductive biology of the female shortfin mako, Is- uriis oxyrinchus. Rafinesque, 1810 with comments on the em- bryonic development of lamnoids. Fish. Bull. 98:299-318. 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 Ocean. Fish. Bull. 100:266-278. Parsons, G. R. 1983. The reproductive biology of the Atlantic sharpnose shark, Rhizoprionodon terraenovae (Richardson). Fish. Bull. 81:61-73. Pratt, H. L., Jr. 1979. Reproduction in the blue shark, Prionace glauca. Fish. Bull. 77:445-470. 1993. The storage of spermatozoa in the oviducal glands of western North Atlantic sharks. Environ. Biol. Fish. 38: 139-149. 1996. Reproduction in the male white shark. In Great white sharks, the biology of Carcharodon carcharias. Proceedings of the symposium on the biology of the white shark. Bodega Marine Lab., Bodega, CA, 4 March, 1993 (A. P Klimley and D. G. Ainley, eds.), p. 131-138). Acad. Press, San Diego, CA. Schwartz, F J. 1984. OccuiTence, abundance, and biology of the blacknose shark, Carcharhinus acronotus in North Carolina. NE Gulf Sci. 7(11:29-47. Shann, E.W. 1911. A description of the advanced embryonic stage of Lamna cornubica. Ann. Rep. Fish. Board Scotland 28(3): 73-79. 1923. The embryonic development of the porbeagle shark, Lamna cornubica. Proc. Zool. Soc. Lond. 11:161-171. Stevens, J. D. 1974. The occurrence and significance of tooth cuts on the blue shark iPrionace glauca L.) from British waters. J. Mar Biol. Assoc. U.K. 54:373-378. Svetlov, M. F 1978. The porbeagle, Lamna nasus, in Antarctic waters. J. Icthyol. 18(51:850-851. Swenander, G. 1907. Uber die Ernahrung des Embryos der Lamna cornu- bica. Zool. Stud. Tullberg, Uppsala 1907:283-288. Templeman, W. 1963. Distribution of sharks in the Canadian Atlantic (with special reference to Newfoundland waters). Bull. Fish. Res. Board Can. 140, 77 p. 739 Abstract— In the face of dramatic do- clines in groundfish populations and a lack of sufficient stock assessment information, a need has arisen for new methods of assessing groundfish popu- lations. We describe the integration of seafloor transect data gathered by a manned submersible with high-reso- lution sonar imagery to produce a ha- bitat-based stock assessment system for groundfish. The data sets used in this study were collected from Heceta Bank, Oregon, and were derived from 42 submersible dives (1988-90) and a multibeam sonar survey (1998). The submersible habitat survey in- vestigated seafloor topography and groundfish abundance along 30-minute transects over six predetermined sta- tions and found a statistical relation- ship between habitat variability and groundfish distribution and abundance. These transects were analyzed in a geographic information system (GIS) by using dynamic segmentation to dis- play changes in habitat along the tran- sects. We used the submersible data to extrapolate fish abundance within uni- form habitat patches over broad areas of the bank by means of a habitat classi- fication based on the sonar imagery. Alter applying a navigation correction to the submersible-based habitat seg- ments, a good correlation with major boundaries on the backscatter and topographic boundaries on the imagery were apparent. Extrapolation of the extent of uniform habitats was made in the vicinity of the dive stations and a preliminary stock assessment of sev- eral species of demersal fish was calcu- lated. Such a habitat-based approach will allow researchers to characterize marine communities over large areas of the seafloor. Integration of submersible transect data and high-resolution multibeam sonar imagery for a habitat-based groundfish assessment of Heceta Bank, Oregon* Nicole M. Nasby-Lucas Pfleger Insdtute of Environmental Research 901 -B Pier View Way Oceanside, Calilornia 92054 E mail address: nnasbyig yahoo, com Bob W. Embley Pacific Marine Environmental Laboratory, NOAA Hatfield Manne Science Center Newport, Oregon 97365 Mark A. Hixon Department of Zoology Oregon State University Corvallis, Oregon 97331 Susan G. Merle Cooperative Institute for Manne Resources Studies Hatfield Manne Science Center Newport, Oregon 97365 Brian N. Tissot Program in Environmental Science and Regional Planning Washington State University Vancouver, Washington 98686 Dawn J. Wright Department of Geosciences Oregon State University Corvallis, OR 97331 Manuscript accepted 29 May 2002. Fish. Bull. 100:739-751 (2002). Dramatic declines in several grouncifish populations have occurred along the U.S. West Coast during the last decade (Ralston, 1998: PFMC^; Sampson,^ Bloeser'). One problem exacerbating these declines is that current stock assessments are not sufficiently precise or accurate to effect empirically based management. This is especially true for commercially important species of rockfish (Scorpaenidae, Sebastes). which comprise major groundfish fish- eries along the Pacific Coast. Although evidence has accumulated for sub- stantial declines in the abundance of several species of rockfish, the overall picture is unclear because 42 of 54 rockfish species ( 78'7r ) have never been assessed (Ralston, 1998; NMFS, 1999; Bloeser^). Of the 12 species that have been assessed by the National Marine Fisheries Service, five were listed as "overfished" and one species was listed as "approaching overfished condition" (NMFS, 1999). A possible alternative to single-spe- cies stock assessments of demersal fishes is a habitat-based community assessment, which serves to estimate groundfish population sizes by recog- nizing that species are not randomly distributed among varying habitats. It is known that the diversity, quality, and extent of bottom habitats are impor- tant in determining the distribution, abundance, and diversity of rockfishes (Carlson and Straty, 1981; Pearcy et al., 1989; Carr, 1991; Stein et al., 1992; O'Connell and Carlile, 1993). It has been previously demonstrated, within local study areas, that species richness and composition correlate with seafloor texture (Hallacher and Roberts, 1985; Richards, 1986; Love et al., 1991; Stein et al., 1992; Krieger, 1993; Yoklavich et * Contribution 2477 of the Pacific Marine Laboratory, Newport, Oregon 97365. 1 PFMC (Pacific Fishery Management Council). 1999. Status of the Pacific Coast ground- fish gishery through 1999 and recom- mended acceptable biological catches for 2000: stock assessment and fishery evalu- ation, 75 p. Pacific Fishery Management Council, 7700 NE Ambassador Place, Suite 200, Portland, OR 97220. - Sampson, D. B. 1997. Effective fishing effort in the Oregon groundfish trawl fish- ery. Final report to the Oregon Trawl Com- mission, 80 p. Oregon Trawl Commission, PO. Box 569, Astoria, OR 97103. ■^ Bloeser, J. A 1999. Diminishing returns: the status of west coast rockfish. Pacific Marine Conservation Council, P.O. Box 59, Astoria, OR 97103. 740 Fishery Bulletin 100(4) Heceta Bank, Oregon l^'^^^^'^'l "x -. i:S Od'W I24'40'W 124°00'W r 44 4(1'N 44"2(I'N 44 = IH)'N **■ >^'- v/ -mem C^ ' o I- I |:i 2i] ; .tcplhs in niclcrs! .i Newport Waldport '^ Washington 6.5 and <25.5 cm), boulder (B, >25.5 cm), continuous flat rock (F, low vertical relief), and diagonal rock ridge (R, high vertical relief). The latitude and lon- gitude positions of each transect were determined by using Loran-C with a Trackpoint II ultrashort baseline tracking system and by positioning the vessel directly above the submersible every 10 to 15 minutes. At least three position points were made per transect and the lo- cations of bottom type and biological data were interpo- lated between these points. The absolute accuracy of the submersible's position, obtained by using Loran-C, was within about 150 to 500 m (Melton, 1986). Multibeam sonar The acoustic survey of Heceta Bank was conducted in May of 1998 with a Simrad EM300 (30 kHz) multibeam sonar system on the RV Ocean Alert (Merle et al.*). This survey provided a highly detailed, precisely navigated seafloor map of bathymetry and seafloor texture ( Figs. 2 and 3). The data were processed with Swathed software (Hughes-Clarke et al., 1996). The data processing steps used in Swathed were the following: navigation and sound- ing editing, roll bias correction, tide correction, refraction correction, map sheet setup, gridding, and mosaicing which resulted in a composite map made up of acoustic backscatter imagery. The survey consisted of 47 overlap- ping north-south swaths up to 45 km long, which provided images of approximately 725 km^ of the seafloor and nearly 100% coverage of high-resolution bathymetry and backscatter amplitude. These data were displayed in grids with a resolution of less than 5 meters on the shallowest portions of the bank from depths of 70 to 150 meters, and of about 5 to 10 meters at depths down to about 500 meters. Data integration and habitat assessment The sonar and submersible transect data were combined by using ArcView and Arclnfo geographic informa- tion system (GIS) software. The sonar data used were Nasby-Lucas et al.: Use of submersible transect data and multibeam sonar imagery for fiabitat assessment 743 bathymetry and backscatter and the submersible transect data included bottom type characteristics and fish density data. In order to represent the dive transects in GIS as linear features displaying changes in habitat and fish density, a dynamic segmentation data structure was used (ESRL 1994). Bottom type and fish density data to be displayed using dynamic segmentation were derived from transect observa- tion data. Transects were divided into segments by uniform bottom type. Fish density was calculated along each seg- ment of habitat tjT)e by using the data for the most common species observed, accounting for 90"^ of the total, plus a few rare species of commercial importance (i.e. lingcod, sablefish, Dover sole and rex sole). This complex of species consisted of a mixture of demersal and benthopelagic species. The follow- ing species were assessed: juvenile Sebastes sp. (unknown juvenile rockfish), Sebastes chlorostictiis (greenstriped rock- fish), Sebastes wilsojii (pygmy rockfish), Sebastes helvomac- ulatus (rosethorn rockfish), Sebastes zacentrus (sharpchin rockfish), Sebastes flavidus (yellowtail rockfish), Ophiodon elongatus (lingcod), Sebastolobus alascanus (shortspine thomyhead), A?iop/opo/nn fimbria (sablefish). Microstomus paclficus (Dover sole), and Errex zachirus (rex sole). Tlie density of fish (number per hectare) was calculated by taking the number of fish sighted in that habitat segment, dividing by the area of the habitat segment in meters, and multiplying by 10,000 square meters per hectare. The use of dynamic segmentation data structure al- lowed for the display of changes in bottom type and fish density data within the transect lines. This was done by creating a "route" system in Arclnfo from the dive transect data and associating it with an "event table." The event table consisted of bottom type and fish density data, and their corresponding locations along the transect, and a route-identifier number to link the information to the corresponding transects in the route system. For visual display, bottom type segments were combined into three major habitat groups: 1) mud, which consisted of "MM" observations, 2) rock ridge, which consisted of "RR" obser- vations, and 3) mixed substrate habitat, which consisted of combinations of all other bottom type observations. In order to combine the sonar and submersible data sets, all segmented dive transect data were then re- projected with a 500-meter offset to the east. This was determined to be the best correction for discrepancies between the transect position data which were acquired by Loran-C and the sonar data which came from GPS posi- tions. It was determined that this offset was necessary by comparing the two data sets and matching depth contours and borders of well-defined habitat, specifically interfaces between the mud and rock features of the bank. There did not appear to be a significant north-south offset, al- though this effect was more difficult to determine because the submersible transects did not cross any well-defined north-south boundaries. Transects segmented by both bottom type and fish density were overlaid on the sonar data in AxcView (Fig. 4, A-C). Assessments of fish abundance within large habitat ar- eas were performed by selecting patches of relative habi- tat homogeneity on the sonar map around the location of each submersible transect. These patches were chosen by examining both patterns in the backscatter values and topographic features indicated by the backscatter and bathymetric data. In areas of mud off the bank, borders were chosen by maintaining constant depth as well as equal distance from the bank. In selecting patches to represent areas of similar habitat, the boundaries were relatively well defined in areas of rock and mud, but for mixtures of sand, cobble, pebble, and boulder, it was more difficult to distinguish distinct boundaries and therefore these patch borders were drawn conservatively. Using the observational data from the transects from all three years, we were able to characterize each habitat patch by percent bottom type, density of fish, and esti- mated abundance of fish as extrapolated from the dive transect data contained within that patch. The grand mean density and standard error for each species was de- termined by using a weighted density for each habitat seg- ment based on a proportion of the length of that segment to the overall transect distance within that habitat patch. The grand mean density was calculated as * = X'^'^" (1) where x = x; d= density offish within a segment of continuous bottom type; and p = bottom type segment length/total transect length within the patch. This calculation used the associations of the fish species with substrate type and weighted its contribution to the overall density by the comparative length of that segment. Total fish abundance for each habitat patch was deter- mined by multiplying the area of the patch and the grand mean density and standard error of each species. The total abundance for each species for all habitat patches was determined by adding the abundance for that species for all eight habitat patches. The standard error for the total abundance for each species was determined by calculating a grand mean standard error weighted by using the stan- dard error of each habitat patch multiplied by the propor- tion of the abundance of that species for that habitat patch to the overall abundance for that species. Total abundance standard error was calculated by using Equation 1 where X = SE; d = standard error offish abundance within a habi- tat patch; and p = abundance within that habitat patch/ total abundance in all eight patches. Results Comparing the submersible data with the sonar data, we found that there was high correlation between the direct observations of bottom type and the habitat type indicated by the sonar data. Side-lit bathymetry revealed areas of outcropping substrata and the backscatter data 744 Fishery Bulletin 100(4) Bottom Type Mud /\/ Mixed Dover Sole fish/hectare 0 0-200 /\/ 200 • 300 /\/300 ■ 4.000 --'/' £^ ^ %^^ Rosethom Rockflsh fish/hectare 0 0- 200 /\/ 200- "WO /\J^a ■ 17.000 i^£»jij. w- *' ■f:e Figure 4 Submersible transects in the middle portion of Heceta bank segmented by obsei-vations of I A) bottom-type habitat, categorized by mud. mixed substrate, and rock ridge (B) den- sity of Dover sole iMicrustonius pacificus ) and IC) density of rosethorn rockfish tScbastcs helromaciilatusl. showed changes in seafloor texture. Borders of rock ridge and mud observed in the submersible survey matched the bathymetric data, indicating areas of high reHef for rocky patches and low relief for areas of mud (Figs. 4A and 5). Changes in backscatter also corresponded with changes observed in bottom type from the submersible data (Fig. 5). It was noted consistently that high-backseatter areas were associated with mixed substrate habitat, comprising a combination of pebble, cobble, and boulder, which due to their size and geologic composition have a relatively high reflectivity. Mid-backscatter values were found to be associated with rock ridge features. These are typically Nasby Lucas et al Use of submersible Iransect data and multibeam sonar imagery for hiabital assessment 745 Figure 5 Habitat patch boundaries and transects segmented by bottom type overlaid on both bathymetric and baclt- scatter sonar images. Habitat patches are labeled A-H. The bottom-type observations were grouped into three predominant habitats: mud. mixed substrate, and rock ridge. large-scale rolling ridges with less small-scale relief. Low- backscatter values were associated with mud bottoms, which have characteristically low reflectivity. Eight regional patch boundaries were drawn based on areas of relative homogeneous seafloor sonic character- istics around the six stations surveyed by submersible (Fig. 5). These patches (labeled A-H) ranged from 1.8 km- (patch B) to 38.9 km-^ (patch C) (Table 1). Patch boundaries for A, C, and F were drawn around areas of mid-backscat- ter reflectivity and high relief indicated by the bathymet- ric data, and were located in relatively shallow areas of the bank (75-125 m). Patch boundaries for D and H were drawn around mud areas located off the edge of the bank, which exhibited low-backscatter values, low relief on the bathymetric data, and were located in deeper waters ( 150- 300 m). Patch boundaries for B, E, and G were drawn around areas of high backscatter and were present in intermediate depths ( 100-150 m). 746 Fishery Bulletin 100(4) 0^ CO ■>* c Oi ^ in C lO ^ in m 00 '^ t^ t^ CO 00 t> r- Tl- o Ol cd' *« ■^ C^l CO CM in CM 00 co [~- CO e2 oa +1 -H +1 +1 +1 •H ■H +1 +1 +1 +1 +1 rH o 03 T— 1 CO CM t~ CO CO ■cP in o 00 lO ^ ■* in ^ ^ CO 00 '^ Tt 05 QJ t:^ o CD lO tr- ^H ^ CM Tf Tf in co' in CO c^ CM rH cd' o CO in 0) T-* CD CO in CM CJi lO Tt r- T-H Cfi "2 o o o o O o o C^I C-J o o o o o o -t in CO Tf CM CO 1 — 1 1 — 1 ^ o ^1 ^ ^ Tf +1 +1 o o +1 o +1 o +1 +1 o o +1 +1 in CO +1 +1 o o +1 +1 o o +1 +1 o o +1 +1 C^i +1 CD +1 CD +1 +1 t^ o +1 +1 as +1 O in ^ o> r-H ^ CD CM CM CD Tf in ^H CM CM 02 Xi C ,-1 ,-H i> in CO ^ o 00 T3 (M 00 CO in 00 o CO CO O 00 CO 1 — 1 in Oi 1 — I 1 — 1 CO c^ CO in c^ 00 rt in r^ o o CO t r- CM CO c~ ^ CO in C^ +1 +1 O CO +1 in ■H +1 +1 CD CO +1 +1 in CM +1 +1 CO Oi +1 +1 o o +1 fl 03 t +1 03 tl +1 CO +1 CO +1 +1 tr- CM +1 CO +1 in +1 T-H CO ■* CO O CO •-I O^ CM 03 o CM CO CO i-H 00 >_ O) o in oi OJ in CO ^ C^I CO lO ^ Tf t-h' 'cfi T-H 0) Q T3 in 03 CO —1 in O O 00 CO 00 lO -H CO CO d CO CD in r- CM rt CM f-f o in 00 m C'} Tf CM CO ^ ^ CD CJ o Tf Ol l> CO o o o o in CO o o T-H a> +1 +1 a> o +1 o +1 +1 +1 •H -H CO Ol ■H +1 CM .-H +1 +1 00 t~ +1 +1 CO CM +1 o +1 o +1 o +1 o +1 +1 00 ^ +1 o +1 o +1 CO o C<1 O 00 rt CM C- CO I> CJ3 CO ^ CD t^ CO -^ C pq Ol CO ■* c3 r- CO CO r^ CM ^ CD CD -u T3 rf CO CO CO ^ CM .-H CO ^ CM ■* 03 in f 0; Oj Cn CM ^ O) ■* ^ CM C^l 't ^ CM ^ CO CM CO o o 6 -H C^l [r^ T-H 0) W .2 oi 00 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 ffi § -" CD lO O CM CO o a> 00 >o 1> O CO CO CO CD o o tr- t-H o> 00 CO CD ^1 (N CTi Ol CO CO CO o CO o CM in 6 r^ in CM ^ T-H X CM Tf CM in CM ^ [^ rf ^H CO o CM -* t^ en -4J _a; rt a CO CM CD O , 1 o o LO CO Ol 00 t CO a 03 -o ^ -H CO ' ' CO o o l-H -^ o o o o -r CM CO 00 •+ CM T3 IS 03 Q ^ -o ■H +1 CO CD +1 00 +1 CD +1 +1 o o ■H -H O Tf +1 +1 03 CO +1 +1 o o +1 +1 o o +1 00 +1 in +1 CM +1 T-H +1 +1 CM CD +1 CO +1 Ol +1 02 ^ C i-H -^ ^ CM CM CM t- in CO r-l 00 Tl' in ^ ■2 ^ i5 1 ^ o j: ,- 'C o? CJ CO ^ CM CO CM CO »— 1 l-H in CO a a. in in o Ol CT) lO .-1 O 03 03 CO CO c^^ CO •-I ^ in CM o o •-1 CO in --H o o o o o o o o CM +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 CM [^ lO CO CO CD O) CJ3 CO o o C^ CO t^ CO o o o o O O o o l> M O CTl CO CD t^ t^ 00 rt CM O CO ■^ lO lO in Oi CO 'J' CM ^ ^H CM lo' O)' " 2 r-( iH co' CM O) r-i x i-H >, J3 CM ~ yi to Oi in 00 ^ cn o T3 00 00 Ol c^ CO o Tf o o in C<1 o C^ CO a C Oj c-^ CO CD CM in CO CO in rH CO in CM --t> in ^ o o o o .-H CO CD -* CJ a- ffl .2 ^ C^ ■H +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 tl +1 +1 +1 +1 +1 +1 +1 +1 T3 a -a S CO r-l 00 C^ Tf lO 00 O CO C<1 .-H Oi t -t ^ o o o o T-H [-- t^ CM O [> CO CM CO .-1 ^ a> CO «:> CO ^ ^ CO Qj o -- -r O) •n CD C-] CD -H 05 ^ CM ^ ^^ - OJ Vj So CD lO CD in in Tj- 00 +1 £ T3 CO CO CO in CO CO CD ,-H tP CO CO O C^I 00 ^ -c ^ <■ _^ JO lO CO CM m CO in in r- 't in CO in CO o o o o CM ^ o o CM Oj o ^ -H +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 o 1 CO O -H 00 00 in o t^ f— t ^ CO CO CM CM i-H 00 in •a c 3 o CM ^ CO QJ <1> QJ Oi o QJ o dj Oj Cj Oj Oj o T3 ^ o CJ CJ o CJ CJ CJ CJ CJ CJ CJ C 3 c C C c c^ C C C C C C C ra c >. ta >, a >-. M >. CO >. CO >. CO >. CO > to >•. to >. CO > to to $ II 03 — c '« -a c S c S c S c ■" -a 'S c 'tfi T3 C 'tfi -a c S C tn -a c -a c C Oj c ■ti =^ C 3 c 3 C 3 C 3 C 3 C 3 C 3 c 3 c 3 C 3 C 3 3 -a c o 01 OJ ^ OJ _D Q, X OJ _D Oj _D OJ _Q 01 J2 0) JH QJ -Q OJ -Q CJ -D XI -a rt -a ra -a CO -O CO T3 CO T3 CO T3 to T3 to T3 to -ri to -a to CO -c B "rt en C o o o en g d QJ CO ^ -£! a> C3 > -o J3 cC CC -M OJ e -a S 6 X Si CO m u OJ S to X a IS So C/3 i cC c -2 O 3 tH CJ C 8 2 o ■B £ QJ ^ CD -C s « .S 2 So g s — rS C/3 3 0 o c: ^ o -a CO 2 cS CO C o Cj to "5 to 0) 3 .a S QJ O to X to > tj a f^j 3 C ^c^ O CO ca CO x; CO CO >- CO ►J O j= CO CO CO CO •t a^ OJ Ki Nasby Lucas et al : Use of submersible transect data and multibeam sonar imagery for habitat assessment 747 0.9 B 0.6- 0.3 • n 1 II . 1 ■ ■ 1 RR BB BC CB CC CM MB MC MP MtVI RR BB BC CB CC CM MB MC MP MM Bottom type Figure 6 Percent cover of bottom types calculated from observations of all transects contained within each of the eight habitat patches labeled A-H. Bottom types are listed from left to right by decreasing relief and particle size, where the first letter is the dominant substratum and the second letter is the second most prevalent substratum. Substrate categories shown comprise only the 10 most dominant bottom types, consisting of com- binations of R = rock ridge, B = boulder, C = cobble, M = mud. and P = pebble. Combining the sonar-derived habitat patches and sub- mersible observations gave an indication of bottom-type compositions for the defined patches. Of the eight habitat patches analyzed, three of the habitat patches were pre- dominantly rock ridge (patch A—63Vc RR and 14% RC/RB; patch C— 55% RR and 3% RB/RC; and patch F— 66% RR), two were predominantly mud (patch D — 98% MM; and patch H— 60% MM and 36% MC/MP), and three were a mixture of boulder, cobble, pebble, and mud (patch B — 75% BC/CB/MB/MC/MM/CM/MP/BM/BB; patch E— 83% MC/MM/MP/CM/MB/CP/BM/BC/CB/PM/CC/BB/BP/PB; and patch G— 100% BC/CB/CM/CC/MP/MC/MM/PM/BB) (Fig. 6). There were differences in fish density within the patch- es between the three major habitat classification types, as well as differences among patches of similar bottom types (Fig. 7, Table 1). Species with the highest association with rock-ridge habitat patches were yellowtail rockfish, juvenile rockfish, and lingcod. Those primarily associated with mud habitats were Dover sole, rex sole, and short- spine thornyheads. Those associated with mixed substrate patches were sharpchin rockfish, rosethorn rockfish, green- 748 Fishery Bulletin 100(4) JR GR PR RR SR YR LC ST SF DS RS JR GR PR RR SR YR LC ST SF DS RS Fish species Figure 7 Fish densities (mean ±SE) for each habitat patch calculated from all transects con- tained within each patch. JR = juvenile rockfish iSebastes species); GR = greenstriped rockfish tSehastes elongatus); PR= pygmy rockfish iSebastes wilsoni); RR = rosethorn rockfish iSebastes helvomaculatus); SR = sharpchin rockfish iSebastes zacentriis);YVi - yellowtail rockfish iSebastes flavidus); LC = lingcod iOphiodon elongatus); ST = short- spine thornyhead iSebastolobus alascanus): SB = sablefish iAnoplnpoina fimbria ); DS = Dover sole iMicrostomus pacifictis). and RS = rex sole iE. zachinis}. striped rockfish, and pygmy rockfish. Juvenile rockfish were found predominantly in rock-ridge habitats, and found in substantially high density (25,577 |±65211 fisli/ hectare) in patch C as compared with the other rock ridge patches (patch A 1140 |±536| fish/hectare and patch F 4709 [±5309] fish/hectare). Both Dover sole and rex sole were associated with mud habitat, and both were found in higher densities in patch D (Dover sole — 312 |±33| fish/hectare, rex sole — 143 |±18| fish/hectare) than patch H (Dover sol— 217 |±29| fish/hectare, rex sole— 24 |±141 fish/hectare). Pygmy rockfish were found in high density in mixed substrate patches B and F (12,435 |±3392] fish/ hectare and 12,119 [±3531] fish/hectare) but at a substan- tially lower density in patch G (3236 |±81| fish/hectare). Among all eight patches, the species found in the highest abundance overall were juvenile rockfish and pygmy rock- fish, and those in the lowest abundance were lingcod and sablefish (Table 1). The total area of all habitats assessed was 124 km-, which is approximately 17'7f of the total area of the sonar survey, and the total number of estimated fish and standard error for that area was 156,598,000 ±16,854,000. The coefficient of variation was relatively low Nasby Lucas et al.: Use of submersible transect data and multibeam sonar imagery for habitat assessment 749 for p-eenstriped rockfish, rosethorn rockfish, sliarpchin rockfish, shortspino tliornyhoad. and Dover solo (between 3.7'r and 6'; ) and slightly higher for pygmy rockfish, yel- lowtail rockfish, lingcod, sablefish, and rex sole (between 7.8' r and 11.9%). Discussion A primary finding of our study was that distinct bottom types found on Heceta Bank are distinguishable through the use of sonar data and that these interpreted habitats correlate with direct submersible observations of bottom type. The determination of habitat information from sonar data is significant in that it provides a broad-scale view of the seafloor habitat, previously unavailable, and allows a habitat-based groundfish assessment. Although seafloor imaging and GIS techniques have previously been used in the study of marine habitats (e.g. Meaden, 1999; McRea et al., 1999;Sherin, 1999; Yoklavich et al., 2000), ours is one of the first published studies where GIS technique was used to combine a detailed analysis offish and habitat transect data with broad-area high-resolution sonar seafloor imag- ery and where total fish abundances were calculated for large areas of the seafloor (see also O'Connell et al.^). Habitat type could not be determined by bathymetry or backscatter data alone, but the information provided by both data sets, in addition to groundtruthing by direct sub- mersible observation, gives a clearer picture of the overall habitat environment. The use of the backscatter data combined with the bathymetric data has the advantage of providing an indication of substrate type, which is clearly important in fish-habitat associations. In general, however, backscatter provides a better indication of habitat for fish association purposes. Bathymetric data can provide geo- logical structure on the resolution of five to ten meters, whereas backscatter data give an indication of structural variation on a smaller scale, which is of ecological impor- tance in influencing the distribution of groundfish. For ex- ample, the physical properties of a substrate influence the types of invertebrates that colonize the seafloor, and local relief can provide microhabitats for some fish. The extent to which a groundfish habitat can be ef- fectively mapped by remote sensing is determined by the resolution of the system used. In general, sonars are optimized for specific operational depth ranges. A sys- tem designed for very shallow water can have sufficient resolution to provide contours of features or objects that deeper water (lower frequency and longer range) systems will only "see" as backscatter changes. As more sites are studied by combining visual seafloor transects, high-reso- lution sonar, and GIS techniques, it is likely the geologic indices most relevant to groundfish habitats will become s O'Connell, V., and C. Carlile, and C. Brylinsky. 2001. De- mersal shelf rockfish stock assessment and fishery evaluation report for 2002. Regional Information Report lJOl-35, 43 p. Southeast Region, Division of Commercial Fisheries, Alaska Department of Fish and Game, P.O. Box 25526. Juneau. AK 99802. apparent. These methods should lead to a more coherent approach to habitat-based stock assessments. One of the limitations of this habitat-based approach to stock assessments has to do with strong reliance upon the fish-substratum association. The distribution and abundance of groundfish has been shown to be strongly correlated with substrate type (see introduction), but fish distributions and densities may vary with other factors as well, such as depth, currents, nutrients, and food avail- ability. In this study there was an attempt to address this potential problem of over emphasizing the fish substra- tum relationship through the grouping of habitats into patches. The designation of habitat patches allowed the grouping of areas of potentially similar biotic and physi- cal characteristics. Thus, the use of patches as areas for fish density estimates allowed for increased accuracy in forming abundance estimates from the habitat-groundfish association information. For example, this advantage was apparent by high variance in fish density estimates among patches of similar bottom type, such as the high density of juveniles in one of the three rock ridge patches. Variations in density in similar patch types of our study were also ob- served for Dover sole, rex sole, and pygmy rockfish. These patterns may be due to differences in depth, food avail- ability or variations in percent composition of substrate type in separate areas of the bank. The other benefit of using habitat patches was that it allowed the testing of new groundfish assessment methods without making pre- dictions for areas of high uncertainty where submersible transects were not performed. Hixon et al.'s study* is one of only a few comprehensive habitat-groundfish studies available and has provided a foundation for testing this new approach. Hixon et als dataset provided invaluable habitat information, but had several shortcomings. One problem was that of the incon- sistency in positional data because of the use of Loran-C ( GPS was not yet available ). Another problem, characteris- tic of all submersible studies, was the overall limited spa- tial sampling provided by the survey. The stations for the study were chosen as representative habitats for Heceta Bank from exploratory submersible dives conducted by Pearcy et al. (1989) in 1987. However, not all of the rep- resentative habitat areas were sampled because maps of high-resolution bathymetry and backscatter were not available at that time. Lack of complete habitat data made it difficult to extrapolate bottom-type and fish-density data to the entire bank. The use of observational data from submersibles for de- termination of fish density, and extrapolation from these data to total abundance within regional habitat patches, were based on several assumptions. First, we assumed that the areas sampled were representative of the entire regional patch to which each transect belonged. Whatever error was associated with this assumption, our approach is certainly more accurate that any method that ignores habitat variation. Second, we assumed that all fish within each submersible transect were accurately identified and counted. Identification of similar species of rockfish can be problematic, and counts clearly become estimates when dense schools are encountered. Third, we assumed 750 Fishery Bulletin 100(4) that avoidance or attraction of fish in response to the submersible was minimal, and thus did not affect counts substantially. This assumption was tested by Hixon et al.'' by observing the local distribution and abundance of fish around the submersible just before and after the 10-15 minute "quiet periods'" on the bottom, during which all motors and lights were turned-off Quiet periods were con- ducted for all transect dives, and there was no indication that fish behavior was altered by the presence of the sub- mersible. However, midwater schools of yellowtail rock- fish sometimes circled the submersible during transects, which could have affected counts if fish were counted more than once. The final assumption was that transect width, which varied as a function of the altitude of the submers- ible above the seafloor, was constant. Certainly, there was some variation in altitude, yet the error introduced was presumably insubstantial except perhaps in areas of ex- tremely heterogeneous vertical relief This study provides an expanded view of the groundfish habitat of Heceta Bank in areas adjacent to the histori- cal Delta transects. In order to perform a full assessment of the bank it will be important to groundtruth habitat types for the entire extent of the bank. The new sonar da- ta have indicated areas on the bank that contain unique habitats that have not been identified. The best study plan would have been to gather the sonar data first, then use the detailed imagery to define patches of uniform bottom type for planning subsequent stratified random sampling and groundtruthing using submersibles. The next phase of our project was initiated in June 2000 and involved operations with the manned submersible used in the original study and an advanced remotely operated vehicle (ROV) to conduct transects on unsurveyed areas, as well as to repeat the original historical transects. Not only will this approach optimize the techniques developed in our project but may also provide the opportunity to es- timate changes in fish density on Heceta Bank over the past decade. Conclusions In this exploratory project we have demonstrated how sonar and submersible data can be combined to allow habitat- based stock assessments of multiple species of groundfish. Despite its limitations, this method provides the possibility for a detailed look at fish abundance using habitat associa- tions. This approach could be used to address the problem of the high number of groundfish species that are currently unassessed. It offers the prospect of examining multiple species of fish and may provide a better indication of fish abundance estimates, particularly for multiple species, than is possible using current methods. The method presented in our study provides an alter- native for assessing these ten groundfish species and ad- ditional groundfish species that are currently unassessed, as well as for monitoring species that are considered to be overfished. Of the groundfish species examined in this study using the 1988-90 submersible observations, the status of all, but the pygmy, rockfish has been reported recently (NMFS, 1999). In the NMFS report on the sta- tus of fisheries in the United States, only the lingcod was reported as "overfished." Yellowtail rockfish, shortspine thornyhead, sablefish, and Dover sole were listed as "not approaching overfished," and the status of rex sole, greenstriped rockfish, rosethorn rockfish and sharpchin rockfish was reported as "unknown" (NMFS, 1999). In addition to the ten species that were examined here, the method described in our study would be important for assessing species of rockfish found on Heceta Bank that are considered overfished, such as the canary rockfish, Pa- cific ocean perch, and darkblotched rockfish. It should be stressed, however, that this method would be useful only for those species that are closely associated with seafloor habitats and spend most of their time near the bottom. For example, yellowtail rockfish are found as high as 25-35 m off the seafloor, so that transect data collected on the seafloor may not accurately reflect their true abundance (Pearcy, 1992). The preliminary work in this study is a step toward creating a model approach for characterizing and quanti- fying groundfish and their habitat associations on a scale meaningful to the stock assessment of commercial species and the conservation of benthic communities. Traditional stock assessment methods for groundfish have been inad- equate. Our study is the first step in the development of a new quantitative method of assessing groundfish stocks that is independent of traditional trawl surveys. Overall, this habitat-based approach to stock assessment has par- ticular recommendation for defining and mapping essen- tial fish habitat, as well as providing important data for designing and managing marine reserves and protected areas. Acknowledgments We would like to thank all who were involved in the col- lection and analysis of both the Delta submersible data and the Simrad EM.300 multibeam sonar data. We espe- cially thank D. Stein, B. Barss, R. Methot, B. Malouf and J. Auyuong. W. Wakefield provided advice and support with this project. The high-resolution multibeam survey of Heceta Bank would not have been possible without the efforts of the captain and crew of the MV Ocean Alert, and the personnel from C&C Technologies. Also, D. Clague and Monterey Bay Aquarium Research Institute pro- vided valuable support for the EM300 survey with both the onboard processing system and in allowing us to use some of their survey time to complete the Heceta Bank survey. J. Reynolds generously stayed on board Ocean Alert for the Heceta Bank survey and provided invaluable expertise in data collection and processing. We also thank N. Maher for valuable technical advice on EM300 pro- cessing. This research was funded by Oregon Sea Grant (NOAA), NOAA's National Undersea Research Program through NURP Headquarters and the West Coast and Polar Regions Undersea Research Center. The submers- ible and sonar portions of the work were funded through the U.S. Department of the Interior Minerals Manage- Nasby-Lucas et al Use of submersible transect data and nuillibeam sonar imagery for habitat assessment 751 ment Service, the National Undersea Research Program, Northwest Fisheries Science Center (NOAA/NMFS), and Oregon Sea Grant. Literature cited Able, K. W., D. C. Twitchell. C. G. Grimes, and R. S. Jones. 1987. Sidescan sonar as a tool for detecting demersal fish habitats. Fish. Bull. 85:725-744. Carlson, H. K.. and R. R. Straty. 1981. Habitat and nursery grounds of Pacific rockfish. Si'bastcs spp., in rocky coastal areas of Southeastern Alaska. Mar Fish. Rev. 43:13-19. Carr, M. H. 1991. Habitat selection and recruitment of an assemblage of temperate marine reef fishes. J. Exp. Mar Biol. Ecol. 146:113-137. ESRI 1994. Dynamic segmentation. In Network analysis, 263 p. Environmental Systems Research Institute, Inc., Red- lands, CA. Greene, H. G., M. M. Yoklavich, D. Sullivan, and G. Cailliet. 1995. A geophysical approach to classifying marine benthic habitats: Monterey Bay as a model. Alaska Dep. Fish and Game Special Publ. 9:15-30. Hallachen L. E ,and D. A. Roberts. 1985. Differential utilization of space and food by the inshore rockfishes (Scorpanidae.- Sehastes) of Carmel Bay, Cahfornia. Environ. Biol. Fish. 12:91-110. Hughes Clarke. J. E., L. A. Mayer, and D. E. Wells. 1996. Shallow-water imaging multibeam sonars: a new tool for investigating seafloor processes in the coastal zone and on the continental shelf. Mar Geophys. Res. 18:607-629. Krieger, K. J. 1993. Distribution and abundance of rockfish determined from a submersible and by bottom trawling. Fish. Bull. 91:87-96. Love, M. S., M. H. Carr. and L. J. Haldorson. 1991. The ecology of substrate-associated juveniles of the genus Sehastes. Environ. Biol. Fish. 30:225-243. McRea, J. E.. H. G. Greene. V. M. O'Connel. and W. W. Wakefield. 1999. Mapping marine habitats with high resolution sides- can sonar Oceanol. Acta 22:679-686. Meaden. G. J. 1999. Applications of GIS to fisheries management. In Marine and coastal geographical information systems (D. J. Wright and I). J. Bartlett, eds.J, p. 205-226. Taylor & Francis, London. Melton, L. 1986. The complete Loran-C handbook, 221 p. Interna- tional Marine Publishing Company, Camden, ME. NMB'S (National Marine Fisheries Service). 1999. Report to Congress: status of fisheries of the United States, p. 35-39. NOAA Fisheries, 1315 East West High- way SSMC3, Silver Spring, MD 20910. O'Connell, V. M., and C. W. Cariile. 1993. Habitat-specific density of adult yelloweye rockfish Sehastes ruberrimus in the eastern Gulf of Alaska. Fish. Bull. 91:304-309. Pearcy W. G. 1992. Movements of acoustically-tagged yellowtail rockfish Sehastes flavidus on Heceta Bank, Oregon. Fish. Bull. 90; 726-735. Pearcy W G.. D L. Stein, MA. Hixon. E. K. Pikitch, W. H. Barss, R. M. Starr. 1989. Submersible observations of deep-reef fishes of Heceta Bank. Oregon. Fish. Bull. 87:955-965. Ralston. S. 1998. The status of federally managed rockfish on the U.S. West Coast. In Marine harvest refugia for West Coast rockfish: a workshop (M. Yoklavich, ed.), p 6-16. U.S. Dep. Commer. NOAA Tech. Memo.. NOAA-TMNMFS-SWFSC- 255. Richards, L .J. 1986. Depth and habitat distributions of three species of rock- fish [Sehastes) in British Columbia: observations of deep- reef fishes of Heceta Bank, Oregon. Fish. Bull. 87:955-965. Sherin.A. G. 1999. Linear reference data models and dynamic segmenta- tion: apphcation to coastal and marine data. In Marine and coastal geographical information systems (D. J. Wright and D. J. Bartlett, eds. ), p. 95- 1 16. Taylor & Francis. London. 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. Yoklavich. M. M.. G. M. Cailliet. G. Greene, and D. Sullivan. 1995. Interpretation of sidescan sonar records for rockfish habitat analysis: examples from Monterey Bay. Alaska Dep. Fish and Game Special Publ. 9:11-14. Yoklavich. M. M.. H. G. Greene, G. M. Caillet, D. E. Sullivan, R. N. Lee, 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. 752 Abstract— We used allozyme. mic- rosatellite, and mitochondrial DNA (nitDNA) data to test for spatial and interannual genetic diversity in wall- eye pollock [Theragra chalcogramma) from six spawning aggregations rep- resenting three geographic regions: Gulf of Alaska, eastern Bering Sea, and eastern Kamchatka. Interpopulation genetic diversity was evident primar- ily from the mtDNA and two allozyme loci (SOD'2*, MPI*). Permutation tests indicated that F^j. values for most allo- zyme and microsatellite loci were not significantly greater than zero. The microsatellite results suggested that high locus polymorphism may not be a reliable indicator of power for detecting population differentiation m walleye pollock. The fact that mtDNA revealed population structure and most nuclear loci did not suggests that the effective size of most walleye pollock popula- tions is large (genetic drift is weak I and migration is a relatively strong homogenizing force. The allozymes and mtDNA provided mostly concordant estimates of patterns of spatial genetic variation. These data showed signifi- cant genetic variation between North American and Asian populations. In addition, two spawning aggregations in the Gulf of Alaska, in Prince Wil- liam Sound, and off Middleton Island, appeared genetically distinct from walleye pollock spawning in the She- likof Strait and may merit manage- ment as a distinct stock. Finally, we found evidence of interannual genetic variation in two of three North Ameri- can spawning aggregations, similar in magnitude to the spatial variation among North American walleye pol- lock. We suggest that interannual genetic variation in walleye pollock may be indicative of one or more of the following factors: highly variable reproductive success, adult philopatry, source-sink metapopulation structure, and intraannual variation (days! in spawning timing among genetically distinct but spatially identical spawn- ing aggregates. An examination of spatial and temporal genetic variation in walleye pollock (Theragra chalcogrammd) using allozyme, mitochondrial DNA, and microsatellite data* Jeffrey B. Olsen Susan E. Merkouris James E. Seeb Gene Conservation Laboratory Alaska Department of Fish and Game 333 Raspberry Road Anchorage Alaska 99518-1599 E-mail address (lor James E Seeb, contact author! |im_seebmifishgame stale ak us Manuscript accepted 2.5 March 2002. Fish. Bull. 100:7,52-764 (20021. Detecting spatial structure in the ge- netic variation of some marine fishes is challenging because populations are often closely related due to high gene flow, and the relationships be- tween populations may change over years (Hedgecock, 1994; Shaklee and Bentzen, 1998; Waples, 1998). For these species, independent population studies of genetic variation may result in conflicting evidence on the extent and complexity of population structure (McQuinn, 1997; Shaklee and Bentzen, 1998). Interpreting these results may be further complicated if the loci and classes of genetic markers differ by study and year. In addition, the discor- dant patterns of population structure may reflect complex population biology and ecology or merely differences in the resolving power of loci and marker classes to detect spatial structure. The walleye pollock iTheragra chalco- gramma) is a marine fish that typifies this situation. Walleye pollock inhabit basin, slope, and shelf waters of four major seas in the North Pacific Ocean: Sea of Japan, Sea of Okhotsk, Bering Sea, and Gulf of Alaska (Bailey et al., 1997). Their abundance is greatest in the eastern Bering Sea and the Sea of Okhotsk, but large aggregations are also found in the Sea of Japan, western Bering Sea, and the Gulf of Alaska. Smaller ag- gregations occur along coastal waters in bays and fjords such as in Prince William Sound, Alaska, and Puget Sound in Washington State. Popula- tion boundaries for walleye pollock may be correlated with the margins of spatially distinct spawning aggre- gates that occur throughout the species range in predictable locations during late winter and early spring (Bailey et al., 1999). Many of these spawning aggregates can be distinguished by spawning time and habitat, and by meristic and morphometric characters (e.g. Iwata, 1975b; Hinckley, 1987; Mul- ligan et a!., 1989). Nevertheless, genet- ic support for discrete populations is equivocal, and the extent of population structuring in walleye pollock remains a controversial and unresolved issue for management and consei-vation in U.S. and international waters (Bailey etal., 1999). Independent attempts to define ge- netic population structure in walleye pollock on a regional scale have shown mixed results. Some allozyme loci show genetic differentiation between walleye pollock from contiguous sea basins (e.g, eastern Bering Sea and Gulf of Alaska; Grant and Utter, 1980), and a single locus (SOD) appears to reveal a major east-west division between Asian and North American populations (Iwata, 1975a, 1975b; Grant and Utter, 1980). This major division is supported by a ' Contribution PP-217 of the Alaska Depart- ment of Fish and Game, Division of Com- mercial Fisheries, Juneau, AK 99802-5526. Olsen et al : An examination of spatial and temporal genetic vanation in Theragm chalcogiamma 753 Gulf of Alaska \ If ,/^ X T" T T Figure 1 Map of the Gulf of Alaska and Bering Sea showing the location of walleye pollock spawning aggregations sampled for this study. single study of mitochondrial DNA (mtDNA) sequence variation (Shields and Gust, 1995), but it is not evident in an unpublished study of microsatellite variation at two loci (reviewed in Bailey et al., 1999). Conversely, genetic difTerentiation among eastern Bering Sea and Gulf of Alaska populations is supported by the microsatellite study but is not supported by a study of mtDNA-RFLP (restriction fragment length polymorphism) variation (Mulligan et al., 1992). These discordant results may reflect attributes of walleye pollock population biology, the unique evolutionary properties of each marker class, or both. Nevertheless, an interpretation of the collective data is not possible because the studies were conducted as much as 22 years apart, and many samples were not from spawning aggregations. Nonrepresentative sampling may bias attempts to define genetic population structure in walleye pollock, particularly on a fine spatial scale (within sea basins). For example, some studies using allozymes (Grant and Utter, 1980) and mtDNA (Shields and Gust, 1995) have included population samples taken during summer and fall, even though walleye pollock aggregate for spawning in late winter and early spring. These samples may obscure ge- netic differentiation because walleye pollock populations are believed to mingle in summer and fall during feeding migrations (Bailey et al., 1999). Also, the sample size used in some studies may be inadequate to detect population structure. The sample size in walleye pollock studies thus far has been typically less than 50, and the minimum sample size has been 22 (e.g. Grant and Utter, 1980; Mul- ligan et al., 1992). Sample sizes in this range (22-50) are generally insufficient to detect statistically significant differences in allele frequencies between populations that exhibit weak population structure (e.g. Fgj <0.02; Goudet et al., 1996; Ruzzante, 1998; Waples, 1998). These studies suggest that a sample size of 50 should be considered an absolute minimum for high gene flow species and that sample sizes of 100 or greater may be necessary when al- lele frequencies differ by 5% or less. In our empirical study, we accounted for the sources of sample bias described above and addressed two important questions: 1) Do allozymes, mtDNA, and microsatellites provide concordant estimates of inter- and intraregional population structure in walleye pollock? 2) Is the genetic variation in walleye pollock populations stable from year to year? We tested for spatial patterns of genetic variation among six population samples from three regions: Gulf of Alaska, eastern Bering Sea, and eastern Kamchatka. We also tested for interannual stability of the genetic signal in replicate samples from three of the North American populations. Our results showed that two of 24 polymor- phic allozyme loci (SOD-2*. MPI*) reveal significant spa- tial heterogeneity at the inter- and intraregional level. In general, the mtDNA data were concordant with these two allozymes, revealing significant genetic variation between North American and Asian (eastern Kamchatka) popula- tions, as well as evidence of intraregional genetic varia- tion, particularly among the Gulf of Alaska populations. The microsatellites, although highly polymorphic, showed little spatial variation. The allozyme and mtDNA data provided evidence of interannual genetic variation in two North American populations. F^j values showed this in- terannual variation is of similar magnitude to the spatial variation in these populations. Materials and methods Sample collection Tissue samples for allozyme and DNA analysis were obtained from six spawning aggregations of walleye pollock representing three major geographic regions: Gulf of Alaska, eastern Bering Sea, eastern Kamchatka (Fig. 1, Table 1). 754 Fishery Bulletin 100(4) Table 1 Location, sample data, and spawning habitat of walleye pollock examined in this study. Spawning Major region and population habitat Date n NLat. W Long. North America-Gulf of Alaska Shelikof Strait shelf 1 Mar 97 100 57°55' 154°35' 19 Mar 98 100 57°55' 154°40' Prince William Sound fjord 6 Feb 97 100 60°05' 148°20' 6 Feb 98 100 60°05' 148°19' Middleton Island shelf 5 Mar 98 100 59°26' 145°45' North America-Bering Sea Bogoslof Island basin 1 Mar 97 100 53n5' 169°00' 7 Mar 98 100 53^08 ' 169-11' Unimak Pass shelf 1 1 Apr 98 40 54°33' 165-38' Asia-eastern Kamchatka Kronotsky Bay shelf 1 Feb 99 96 52°00' 161°00E In total, nine samples were collected from 1977 through 1999: six from each of the major spawning aggregations shown in Figure 1, and three samples that were interan- nual replicates from Prince William Sound (PWS), Shelikof Strait (SHEL), and Bogoslof Island (BOG). Other samples included Unimak Pass (UNI) and Middleton Island (MID) in 1998 and Kronotsky Bay (KRON) in 1999. Tissue sam- ples from heart, liver, muscle, and eye were taken from 100 individuals per population with the exception of Unimak Pass (n=40) and Kronotsky Bay (7?=96). Only muscle tissue was sampled from Bogoslof Island walleye pollock in 1997. Samples were stored at -80°C until analyzed. Allozyme analysis AUozyme alleles were resolved by using horizontal starch- gel electrophoresis and enzyme-specific histochemical staining procedures described by Aebersold et al. (1987). Thirty six loci were screened for polymorphism (Table 2): sAAT-l*,sAAT-2*,mAAT-l*.ADA-l*\ADA-2*.AH-l*.AH- 2*,ALAT*, CK-A'\ FH*, GAPDHl*. G3PDH-1*, G3PDH- 2*. G3PDH-3', G3PDH-4\ GPIl*. GPL2\ IDHPl*. IDHP-2*. IDHP-3\ LDH-2'\ LDH-3'. MDH-A*. MEP-l*, MEP-2*. MPI'. PEPA\ PEPB*, PEPD*. PEPLT*, PGDN*, PGM-P, PGM-2*, SOD-1*. SOD-2*. and TPI*. Twenty- four loci were polymorphic in at least one of nine popula- tions, and these loci were used in the tests of spatial and temporal genetic variation (Table 3). DNA analysis Total genomic DNA was isolated from 20-30 mg of heart tissue by using a Centra Systems^'^ (Minneapolis, MN) Puregene DNA isolation kit. Precipitated DNA was hydrated in 50-100 pL tris-ethylenediaminetetraacetic acid (EDTA) buffer (10 mM Tris, 0.1 mM EDTA, pH 8.0) and heated at 55°C for approximately 12 h. Approximately 1 pL of hydrated DNA was used from each sample for poly- merase chain reaction (PCR) amplification of mtDNA and microsatellite fragments. mtDNA For an initial screen, six segments of the walleye pollock mitochondrial genome were examined for restric- tion fragment length polymorphism in a sample of 12 wall- eye pollock. The mtDNA segments examined were ND5/6 (-2400 base pairs /bp/; Cronin et al., 1993), cytochrome b 1-1200 bp and~800 bp; Bickham et al., 1995), cytochrome oxidase I (-700 bp; PowersM, D-loop (-1400 bp; Cronin et al., 1993), and 16S (-600 bp; Palumbi et al., 1991). We used the following restriction enzymes: A/m I.Apa I, Ase l,Ava I, Ava II, BamH I, Bel I, Bgl 1, Bgl II. BstE II. BstU I, Dpn II, EcoR I, EcoR V. Hae II, Hae III, Hha I, Hinfl, Hind III, Kpn I, Mse I, Msp I, Nci I, Pst I, Rsa I, Sac I, Sac II, Sau96 I. Sea I, Stii I. Tag I. Xba I. and Xho I. PCRs were performed in 25-100 pL volumes containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2.0-3.5 niM MgCl.^, 0.8 mM dNTPs, 0.05 units/pL Tag polymerase, 0.3-1.2 pM primer, and about 250 ng DNA template. The PCR profile was 92°C (2 min) -^ 30-40 cycles of (92°C (30 sec) -hX°C (30 sec) -I- 72°C (140 sec)) -t- 72°C (5 min). where the anneal- ing temperature X varied among primer pairs. Restriction digests followed manufacturer's specifications (New Eng- land Biolabs. Beverly, Maine). The mtDNA fragment and enzyme combinations ND5/ 6 (-2400 bp) — Ase I, Ava I, Msp I, Rsa I; cytochrome b (-1200 bp) — Alu I. Hae III, Mse I; and cytochrome oxidase (-700 bp) — Alu I revealed polymorphism and were used to test for spatial and temporal genetic variation. The PCR annealing temperatures for these fragments were the following: (for ND5/6) 50°C; (for cytochrome b) 54°C; and (for cytochrome oxidase) 50°C. Restriction fragments ' Powers. D. A. 1997. The use of molecular techniques to dis- sect the genetic architecture of pollock populations. Unpubl. rep. to the National Marine Fisheries Service, lip. Hopkins Marine Station. Stanford University, Pacific Grove, CA, 93950. Olsen et a\ An examination of spatial and temporal genetic variation in Jlieiagia chalcogiamma 755 Table 2 BiifTors and tissues used to resolve enzyme-coding loci in walleye pollock. Enzyme nomenclature and Enzyme Commission (EC I number follow lUBNC ( 1984); locus nomenclature follows Shaklee et al., ( 1990). Enzyme or protein EC number Locus Tissue' (buffer^) Aspartate aminotransferase Adenosine deaminase Aconitatc hydratase Alanine aminotransferase Creatine kinase Fumarate hydratase 2.6.1.1 3.5.4.4 4.2.1.3 2.6.1.2 2.7.3.2 4.2.1.2 sAATl* L(AC6.5) sAAT-2* H(AC6.1), M{AC6.1) niAAT-l* M(AC6.1) ADAl* L(AC6.5) ADA-2* L(AC6.5) AH-1* L(AC6.5) AH-2* H(AC6.5), M(AC6.1, AC6.5I, L(AC6.5) ALAT* M(TBE) CK-A* M(TBE) FH* M(ACEN7.0) Glyceraldehyde-3-phosphate dehydrogenase 1.2.1.12 GAPDHl* M(AC6.9) Glycerol-3-phosphate dehydrogenase 1.1.1.8 G3PDH-1^ G3PDH-2* G3PDH-3* G3PDH-4* L(AC6.5) M(AC6.5), L(AC6.5) H(AC6.5), L(AC6.5) L(AC6.5) Glucose-6-phosphate isomerase 5.3.1.9 GPI-1* GPI-2* H(TBCL) H(TBCL) Isocitrate dehydrogenase (NADP+) 1.1.1.42 IDHP-1* IDHP-2* IDHP-3* L(ACE7.0), L(TC7.0) M(ACEN7.0) LIACE7.0) L-Lactate dehydrogenase 1.1.1.27 LDH-2* LDH-3* H(TBCL). M(TBE) H(TBCL), M(TBE) Malate dehydrogenase 1.1.1.37 MDH-A* H(AC6.1), M(AC6.1, AC6.9) Malic enzyme {NADP+) 1.1.1.40 MEP-1* MEP-2* M( AC6.5, AC6.9. ACEN7.0) M(AC6.5, AC6.9, ACEN7.0) Mannose-6-phosphate isomerase 5.3.1.8 MPI* H(TBCL), M(TBE) Dipeptidase 3.4.-.- PEP A* H(AC6.5) Tripeptide aminopeptidase 3.4.-.- PEPB* M(AC6.9) Proline dipeptidase 3.4.13.9 PEPD* M(AC6.1) Peptidase-LT 3.4.-.- PEPLT* H(AC6.1) Phosphogluconate dehydrogenase 1.1.1.44 PGDH* M(AC6.9,ACEN7.0) Phosphoglucomutase 5.4.2.2 PGM-1* PGM -2* H(TBCL), M(TBE), L(ACE7.0) M(TBE) Superoxide dismutase 1.15.1.1 SOD-1* SOD-2* M(ACEN7.0,TBE) H(ACEN7.0) Triose-phosphate isomerase 5.3.1.1 TPI* H(AC6.1), M(AC6.1) ' H = heart; E = eye; M = muscle; L = liver. - Buffers: AC. ACN. ACE, ACEN: amine-citric aci 19871. values indicate pH; TBCL: tris-citric acid tris-boric acid-EDTA buffer, pH 8.7 (Hover et al. d buffer ( gel, pH i 196.3). Clayton and Tretialc, 19721 modified .7 and lithium hydroxide-boric acid e with EDTA (El, NAD (N), or both (Aebersold et al.. lectrode buffer, pH 8.0 (Ridgway et al., 19701; TBE: were size fractionated on 3% agarose gels, stained with ethidium bromide, and visualized and photographed un- der ultraviolet light (312 nm). Fragment sizes as small as 50 bp were estimated by comparison with 100 bp and 250 bp ladders on each gel, A composite haplotype was generated for each individual by recording the presence or absence of restriction sites for all restriction enzymes and mtDNA segments (Lansman et al,, 1981), The length of 756 Fishery Bulletin 100(4) Table 3 Single locus statistics: number of individuals Ui ); number of allel iPJ\ total heterozygosity (Hj,); analogs of Wright's F-statistics F,^, 0,7., is computed for mtDNA. Values in bold type with an asterisk for k tests, where k is the number of loci (allozvmes, 24; microsate es(A are s Ihtes ; frequency of the most common allele over all populations and Fj^lf, 6, F). Note: The analog ofF^j- for haplotype data, tatistically significant. The ff-level (0.05) was adjusted ia/k) 3). Locus n A P, H^ /' e F Allozyme sAAT-1* 692 3 0.996 0.008 -0.001 -0.001 -0.003 .■iAAT-2* 793 4 0.992 0.016 -0.011 0.006 -0.005 mAAT-1* 803 2 0.998 0.004 0.001 -0.002 -0.001 AH-l" 657 4 0.958 0.081 0.060 0.001 0.061 ALAr* 811 3 0.898 0.188 -0.019 0.002 -0.017 FH* 814 2 0.999 0.002 0.001 -0.001 -0.001 a3PDH-2* 811 4 0.994 0.012 -0.002 -0.002 -0.004 G3PDH-3* 715 4 0.992 0.016 -0.004 -0.002 -0.007 GPir* 815 3 0.992 0.016 -0.008 0.002 -0.006 GPI-2* 815 4 0.990 0.020 -0.010 0.003 -0.007 IDHP-1* 714 4 0.976 0.047 -0.018 -0.001 -0.020 IDHP-2* 814 3 0.996 0.008 -0.002 -0.001 -0.003 LDH-2* 815 2 0.999 0.002 0.001 -0.001 -0.001 LDH-3* 814 2 0.999 0.002 0.000 0.000 0.000 MDH-A* 815 3 0.996 0.008 -0.002 0.000 -0.002 MEP-1* 815 4 0.994 0.012 -0.004 -0.001 -0.004 MPI* 810 5 0.788 0.337 0.016 0.017* 0.032 PEPB* 802 4 0.962 0.074 -0.001 0.003 0.003 PEPD* 815 3 0.985 0.030 0.158* -0.001 0.157 PGDH* 814 9 0.544 0.590 0.043 -0.002 0.041 PGM-1* 815 4 0.993 0.014 -0.004 0.000 -0.004 SOD-1* 815 2 0.999 0.002 0.000 -0.001 0.000 SOD-2* 711 3 0.916 0.154 -0.014 0.088* 0.075 TPI* 815 3 0.997 0.006 -0.002 0.000 -0.002 Mean 788 4 0.956 0.069 0.024 0.010* 0.033 SE 1 0.100 0.136 0.014 0.011 0.010 Microsatellite Teh 10 741 32 0.395 0.811 0.069 0.001 0.070 Teh 12 671 12 0.305 0.782 0.035 0.000 0.035 Tch22 637 15 0.367 0.713 0.039 0.005* 0.043 Mean 683 20 0.356 0.769 0.048* 0.002* 0.050 SE 11 0.046 0.050 0.011 0.001 0.011 mtDNA ND5/6-CB-CO 684 66 0.371 0.837 0.022* the recognition sequence (bp) and the number of recogni- tion sites identified by each enzyme were the following: for ND5/6— Ase I (6, 5},Ava I (6, 4), Msp I (4, 4),Rsa 1(4, 5); for cytochrome b—Alu I (4, 2), Hae III (4, 2), Mse I (4, 7); and for cytochrome oxidase — Alu I (4, 4). A total of 150 nucleo- tides were surveyed for polymorphism or approximately 3.5'7f of the combined 4300-bp region. Microsatellites Thirteen microsatellite loci were screened for polymorphism, amplification quality, and null alleles in walleye pollock by using primer pairs derived from Atlan- tic cod, Gadus morhua, (Gmol, Gmo2, Gmo9, GmolO, Gmol23, Gmol32, and Gmol45; Brooker et al., 1994) and walleye pollock (Tch.5, TchlO, Tchll. Tchl2. Tchl8, and Tch22; O'Reilly et al., 2000). PCRs were performed in 10 pL volumes containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCU, 0.8 mM dNTPs, 0.05 units/pL Taq polymerase, 0.3-1.2 pM primer, and about 250 ng DNA template. The following PCR profile was used: 92°C (2 min) -I- 30 cycles of (92°C (30 sec) + X°C (30 sec) + 72°C (140 sec)) + 72°C (5 min) where the annealing tempera- ture X varied among microsatellites. Microsatellites were size fractionated by using an Ap- plied Biosystems Inc. (ABI, Foster City, (ilA) 377-96 au- tomated DNA sequencer operated in GeneScan^'^' mode (ABI, 1996a). Data were analyzed by using the internal Olsen et al : An examination of spatial and temporal genetic variation in Themgm chalcogiamma 757 lane sizing standard and local Southern sizing algorithm in the GeneScan 672 software vers. 1.1 (ABI, 1996a). Al- leles for each locus were scored and data were tabulated for importing into statistical software with Genotyper software, vers. 2.0 (ABI, 1996b). Of the thirteen loci screened, Teh 10, Teh 12, and Tch22 were used to test for spatial and temporal genetic varia- tion. The PCR annealing temperatures for these loci were the following: (for TchlO) 54°C; (for Tchl2) 47°C; and (for Tch22) 52°C. The loci GnwlO, Gmol23, Gmol45, Tch5, and Teh 18 did not amplify consistently by the methods above, and Gmol. Grno2, Gmo9, Gmol32, and Teh 1 1 appeared to possess null alleles as revealed by significant heterozygote deficits (data not shown). nificance of estimates for each Fg^ analog (9, R^j., and 0,57.) was determined by using a permutation test option in the respective computer programs. In each case the data set was permuted 1000 times (alleles or haplotypes among population samples). Second, allele and haplotype fre- quency homogeneity was tested for the same population comparisons by using a G-test (allozymes and microsatel- lites) and probability test (mtDNA). The threshold for sta- tistical significance (a=0.05) for multiple comparisons was determined by using the sequential Bonferroni method (Rice, 1989). Results Statistical analyses Estimates of allele and haplotype frequency were calcu- lated for each locus and population sample. Heterozygos- ity estimates (W) were calculated by using Equation 8.4 of Nei (1987) and haplotype diversity estimates (/;) were computed by using the program ARLEQUIN vers. 1.1 (Schneider et al., 2000). A permutation test of the statistic f (Weir and Cockerham, 1984) was used to assess confor- mity to Hardy-Weinberg equilibrium (HWE) for each locus and over all loci for each population with the computer program FSTAT vers. 2.8 (Goudet 2000). The data set was permuted 1000 times (alleles were permuted among indi- viduals, within population samples I, and the threshold for statistical significance (a=0.05) was corrected for simul- taneous tests by using the sequential Bonferroni method (Rice, 1989). Spatial and temporal genetic variation were quantified by estimating analogs of Wright's Fg-j. for each marker class: 6 (allozymes and microsatellites. Weir and Cock- erham, 1984), Rgj, (microsatellites, Slatkin, 1995), cPgj, (mtDNA, Excoffier et al., 1992). Estimates of 6, Rgj,, and 0gj^ (9, Rgj^ and gj,) were computed for all nine popula- tion samples, for all contemporaneous population samples ( 1997, 1998), and for the temporal replicates. The hap- lotype frequency estimates were used to compute gj. varied among loci and failed to ex- pose a single marker class, regardless of overall variabil- ity, as most informative for detecting population structure in walleye pollock (Table 3). Permutation tests revealed two allozyme loci [MPT\ SOD-2*) and one microsatellite (Tch22) with 9 values significantly greater than zero over all population samples (Table 3, Fig. 2). The 9 of 0.088, for SOD-2* was exceptionally high. The 0gj. for mtDNA was within the range of 6* for allozymes and microsatellites and was significantly greater than zero (P<0.001). Intrapopulation genetic variation and Hardy-Weinberg equilibrium Estimators of intrapopulation diversity (alleles per locus, Z/^. and h) differed among marker types within popula- tions but generally showed little variation between popu- lations (Table 4). As expected, the values for the diversity estimators were usually lowest in the population with the smallest sample sizes (e.g. UNI, «=40). This was most notable in haplotype number (range 12-27) because most 758 Fishery Bulletin 100(4) SOD-:* MP I* I Allele 1 D Allele 2 D Allele 3 mtDNA I Allele 1 a Allele 2 D Allele 3 Teh:: I Hap 1 a Hap 2 D Hap 3 D Hap 4 ■ Hap 5 ■ Hap 6-66 I Allele 1 D Allele 2 D Allele 3 ■ Allele 4-15 Figure 2 Frequency histograms for SOD-2*, MPI*, and Tch22 alleles, and mtDNA haplotypes. Abbreviations: Kronotsky Bay (KRON); Bogo-slof Island. 1997 (BOG97); Bogoslof Island, 1998 (BOG98); Unimak Pass (UNI); Shelikof Strait, 1997 1SHEL97); Shelikof Strait, 1998 (SHEL98); Prince William Sound, 1997 (PWS97); Prince William Sound, 1998 (PWS98I; Middleton Island c;j^0.028, P<0.001) but were not distin- guishable from each other (Table 5). These results were confirmed by goodness-of-fit tests of haplotype homoge- neity. Spatial genetic structure among the 1997 samples was not evident from values of 6 over all allozymes and microsatellites (Table 5). 1998 Gulf of Alaska and Bering Sea, and 1999 eastern Kamchatka Indices of population structure for allozymes (0=0.017, P<0.001) and mtDNA (0sy=O.O21, P<0.001) were similar over all population samples and larger than for microsatellites (0=0.002. P>0.050; ^,,.^=0.001, P>0.050; Table 5). Because the values of 9 and R^j over all microsatellites and pop-ulations were not significant, analyses of population structure were conducted with only the allozyme and mtDNA data. Significant genetic differ- ences were revealed between pooled samples from North America and Asia with allozymes (0=0.030. P<0.001) and mtDNA «i',,.7^0.035, P<0.001; Table 5), between samples from the Gulf of Alaska and Bering Sea with allozymes (0=0.005, P=0.006), and within the Gulf of Alaska with allozymes (0=0.009, P<0.001) and mtDNA ((P^.j^O.OlS, P=0.009). Values for 0 were considerably larger for SOD-2* than for the two other informative markers MPI*. and mtDNA (Table 6, Fig. 2). Estimates of P,;^. from mtDNA and MPI* for each of the regional comparisons were similar but the most informative microsatellite locus, Tch22. did not vary significantly among regions. Based on permutation test results from the hierarchical analyses, estimates of genetic differentiation were computed for all population pairs by using both allozyme and mtDNA data, except for Gulf of Alaska versus Bering Sea popula- tions (allozyme data only). Values of 0 (allozymes) and 0,.^ for most pairwise comparisons from North America and Asia were relatively large and highly significant (Table 5). In contrast, values of 0 (allozymes) for most population pairs from the Gulf of Alaska and Bering Sea were rela- tively small, and only the UNI x SHEL and UNI x PWS pairs were significant (0=0.021, P<0.003; 0=0.017, P<0.006). Paii-wise comparisons within the Gulf of Alaska were sig- nificant for SHEL X PWS (allozymes, 0=0.017, P<0.002) and for SHEL X MID (mtDNA, *^..;,=0.025, P<0.007; Table 5). In goodness-of-fit tests, allele and haplotype frequency homogeneity were concordant with the statistically sig- nificant values of 0 (allozymes) and (t>gj for all populations and pooled samples within and between regions (Table 5). However, results from the two types of tests conflicted for some population pairs. For example, G-tests of allozyme data revealed significant genetic variation for two popula- tion pairs (BOG x PWS; BOG x MID; Table 5) for which 0 was not significant. In contrast, 0 was significant for two population pairs, KRON x PWS and UNI x SHEL, which were not genetically different based on G-test results. One population pair, PWS x MID, was genetically different in the probability test of mtDNA data, but i>gj. for this pair was not significant (Table 5). Temporal change In genetic variation: Bogoslof Island, Shelikof Strait, and Prince William Sound Significant temporal change in mtDNA variation was detected in one population, BOG, by using the probability test of haplotype homogeneity (P<0.001) and 0gj. (P=0.003, Table 5). The ' N F,T analog tests e SOD-2* e MPI* mtDNA 1997 populations 3 N/A 0.001 0.017* 1998 populations, Asia 6 0.095* 0.028* 0.021* Between NA and Asia 6 0.155* 0.059* 0.035* Between GOA and BS 5 0.043* 0.002 0.004 Within BS 2 0.001 0.000 0.000 Within GOA 3 0.051* 0.013* 0.018* InterannuaK 1997-98) Bogoslof Island 2 N/A 0.011 0.023* Prince William Sound 2 0.047* 0.000 0.012 Shelikof Strait 2 0.000 0.011 -0.003 an(i 6). Second, two allozyme loci [SOD-2*; MPI*) provided the strongest evidence of regional genetic variation among tlie 1998 samples (Table 6). In fact, the locus SOD-2* had exceptionally high values of 6 for most regional com- parisons in contrast to the other nuclear loci and mtDNA. Diversifying selection may be acting on this locus (e.g. Hudson et al., 1997); however, the general concordance between SOD-2*. MPI*, and mtDNA data suggests that the values of d most likely reflect the influence of genetic drift(Table6. Fig. 2). Spatial structure in genetic variation North America to Asia The signature of a major east- west division between Asian and North American walleye pollock was evident in our allozyme and mtDNA data but not in our microsatellite data. The results of our F^^ analog tests and goodness-of-fit tests confirm the findings of ear- lier studies (e.g. Iwata, 1975a, 1975b; Grant and Utter, 1980; Shields and Gust, 1995, Bailey et al., 1999) and add some new insight regarding genetic heterogeneity in wall- eye pollock from different geographic regions. For example, we confirm the significant SOD-2* variation observed in the past but we also reveal that a second allozyme locus, MPI*, exhibits substantial spatial variation at this geo- graphic scale (Table 6, Fig. 2). Nevertheless, only these two loci and the mtDNA reveal genetic variation between these two regional groups although some of the other 25 nuclear loci screened were equally or more polymorphic (e.g. ALAT*, PGDN*, TchIO, Tchll, and Tch22). These results suggest that, despite broad spatial separation, Asian and North American walleye pollock populations are remarkably genetically similar. This outcome supports the notion that the effective population size in walleye pollock is large, the rate of genetic drift is very low, and migration despite physical distance is a strong homogenizing force. North America-Bering Sea and Gulf of Alaska The allo- zyme and mtDNA provided discordant results with re- spect to inter- and intraregional population structure with- in North American walleye pollock. The most noteworthy instance was the contradictory evidence of genetic varia- tion between populations from the Bering Sea and the Gulf of Alaska. That is, significant genetic differentiation between populations from these two regions was evident from both the allozymes and mtDNA, but in different years (Table 5). We believe that the intraannual differ- ences between marker classes is an artifact due primarily to the lack of SOD-2* data for the Bogoslof Island sample in 1997. The values of 0 and .4'1f5ii*M» 100 - \ 150 200 ■ 1 / 250 I ; 300 - 350 - Vj w f April 27. 2000 ■21 ■ 15 14 13 ■13 12 12 Figure 2 Depth and temperature records for a bigeye tuna carrying an archival tag (tag 99-889), exhibiting unassociated type-1 behavior. Approximate area. 1°N 97°W. is based on geolocation estimates from the archival tag. CTZ = central time zone. /?=9), even though the larger fish (mean=16.2'7(, n=12) still had an obvious affinity for floating objects. Deep-diving behavior Deep-diving behavior was defined as dives in excess of 500 m (Fig. 7). The mean duration of deep-diving behaviors for individual fish ranged from 0.3 to 2.3 h (grand mean=1.2 h, 95% CI=0.3 h) (Table 2). The distributions of the time of day, duration, and maximum depth of the 115 deep-diving events are shown in Figure 8. The majority of the deep dives occurred during daylight hours (mean=ll:56 central time zone |CTZ| ), although some deep dives were made at night. The prominent mode in the maximum depth distribution was about 650 to 850 m. The estimated maximum depth and minimum temperature reached by three bigeye tuna (99- 801, 99-803, and 99-889) were 1500 m and 3°C, respectively Reliability of the geolocation estimates The estimated mean accuracy and precision in the geolo- cation estimates of latitude were 2.04 and 0.79 degrees, respectively (Table 3). The associated uncertainty in those estimates averaged 2.57 degrees (95% CI=0.85) for the south error and 3.40 degrees (95% CI=1.00) for the north error The estimated mean accuracy and precision of the geo- location estimates for longitude was 0.46 and 0.26 degrees, respectively (Table 3). The associated uncertainty in those estimates averaged 0.33 degrees (959; CI=0.13) for the west error and 0.17 degrees (95% CI=0.13) for the east error Additional information on the accuracy and precision of the estimates from the processed archival tag light-level data was obtained by comparing the differences between estimates and actual latitudes and longitudes of recapture for tags 99-864 and 99-874 (Table 3). The two bigeye tuna 772 Fishery Bulletin 100(4) 90 -, 80 A 70 - n = 240 60 - 50 - 40 ' 30 1 20 ■ ll. 10 Ill . 0 4 8 12 16 20 24 28 32 36 40 44 48 52 130 1 120- 110 - B 100 - f7 = 245 90 - Frequency o o o o o 30 - 20 - 1 10 - A 1 ■ H . ■ ■ 0 4 8 12 16 20 24 28 32 36 40 60 -| 55 - 50 - c 45 - 40 - n = 140 35 - 30 - 25 - ■ 20 - 1 15 - 1 10 - 5 - 0 -1- 1 . 0 2 4 6 8 10 12 14 16 18 Days Figure 3 Distributions of the durations of all unassociated type-1 lAl and unassociated type-2 (B) events and events associated with floating objects (C) observed and classified for bigeye tuna listed in Table 2. The total number of events are given in the upper right corner of each panel. carrying these tags were recaptured at the same location in the same purse-seine set. The respective differences between estimates and actual values were 0.20 and 0.30 for latitude and 0.23 and 0.06 for longitude. We also noted that the differences between estimated and actual lati- tudes and longitudes were 1.95 and 0.29, respectively, for the fish with tag 99-812, and 2.67 and 0.28, respectively for the fish with tag 99-865 (Table 3). These two fish car- rying these tags were recaptured by longline vessels, and scientific observers were not aboard to verify recapture positions. Aside from the large differences between the estimates and actual recapture latitudes in Table 3 for tag numbers 99-801, 99-847, and 99-891, the majority of the estimates were within 2 degrees of the actual latitudes. Tags 99-792, 99-835, 99-861, and 99-877 were the only tags that provid- ed differences more than 1 degree between the estimates and actual recapture longitudes. Fish with tags 99-792, 99-793, 99-804, and 99-826 were at liberty during the autumn equinox. The geolocation estimates for longitude were unaffected by this event, but the estimates for latitude were unreliable for a few weeks Schaefer and Fuller Movements, behavior, and habitat selection of Thunnus obesus 773 400 Time (CTZ) 18 00 -99-862 (93 cm) -^99-883 (116 cm) 28 18 15 H h14 g M3 I •13 O •13 •12 12 B 18 00 Time (CTZ) 18 00 Figure 4 Simultaneous depth and temperature records for two bigeye tuna carrying archival tags, exhibiting unassociated type-1 behavior. 23-25 April 2000. Approximate area. 0°N 96°W. is based on geolocation estimates from the archival tags. (A) Depth and ambient temperature records (B) delta T's (dif- ferences between peritoneal cavity and ambient temperature) for the same time period as in A above. CTZ = central time zone. surrounding the autumn equinox because there was very little variation in day length near the spring and autumn equinoxes (Hill and Braun, 2001). Movements The movements of 8 bigeye tuna at liberty for 76 to 340 days are plotted in Figure 9. A-D. No bigeye tuna went farther west than about 110°W and most spent the major- ity of time between QCW to 105°W and 5°N to 5°S. Data for the estimated movement paths of 22 fish, derived from the filtered geolocation estimates, are given in Table 4. The estimated mean speeds ranged from 76 to 165 km/d (grand mean=116.6 km/d. 95^7^ CI=10.0 km/d). The hypothesis that the observed movement path is random was rejected for 17 of the 22 fish (Table 4). A significant positive cor- relation (r=0.61, P<0.05) was found between the number of days at liberty and their corresponding 95*7^ utilization distributions (Tables 1 and 4). The areas encompassed by the 95^f probability ellipses for the geolocation estimates of unassociated behavior and behavior associated with floating objects (Fig. 10), for the 22 bigeye tuna at liberty for 30 d or more (Table 4). were 3.9 and 3.1 x 10*' km-, respectively. The difference between 774 Fishery Bulletin 100(4) Time (CTZ) 12 00 12 00 12 00 12 00 12 00 12 00 12 00 12 ^ 22 00 02 04 06 08 10 12 14 16 18 20 22 00 02 26 O Figure 5 Depth and temperature records for a bigeye tuna carrying an archival tag (99-8691, exhibiting unassociated type-2 behavior Approximate area, 2°N 107°W, is based on geolocation estimates from the archival tag. CTZ = central time zone. the two spatial distributions was not statistically signifi- cant (f =0.313, P=0. 524). The fish with tag 99-793 was at liberty 340 d and ex- hibited behavior associated with floating objects for 9.2% of those days. It traveled initially to the east, remained for a considerable period in a relatively restricted area north of the Galapagos Islands, displayed some range in latitudinal movements between about .5°N and 5°S, and was recaptured only 83 nmi north of its release location (Fig. 9A, Tables 1, 2, and 4i. The fish with tag 99-812 was released at a different FAD and was at liberty for a shorter period (159 d). It exhibited behavior associated with float- ing objects for 15.4% of the days at liberty. In contrast to the fish with tag 99-793, the fish with tag 99-812 traveled westward, displaying a fairly directed movement path, and was subsequently recaptured 821 nmi west of its re- lease location (Fig. 9A, Tables 1, 2, and 4). The fish with tag 99-826 was at liberty 272 d and exhib- ited behavior associated with floating objects for 7.5% of those days. It traveled the greatest total distance, display- ing extensive longitudinal and latitudinal movements, and yet was recaptured only 279 nmi west of its release location (Fig. 9B, Tables 1, 2, and 4). The fish with tag 99- 804 was released at the same FAD as the fish with tag 99-826 and was at liberty for a comparable period (250 d). It exhibited behavior associatede with floating objects for 12.4% of the days at liberty. In contrast to the fish with tag 99-826, the fish with tag 99-804 displayed a much more constrained movement path. This fish started moving to the northwest during its last few months at liberty and was subsequently recaptured 544 nmi west of its release location (Fig. 9B. Tables 1, 2, and 4). The fish with tag 99-792 was at liberty for 183 d and ex- hibited behavior associated with floating objects for 37.5% Schaefer and Fuller Movements, behavior, and habitat selection of Thiinnus obesus 775 Time (CTZ) 12 00 12 00 12 00 12 00 12 00 12 00 12 0 I ■ ■ — ' — ' — ' — ^^ — ■ — ■— ' — ' — ' — ' — ' — ' — ' — ■-^ — ' — ' — ■ — ' — ■ — ' — 129 oT 22 00 02 04 06 08 10 12 14 16 18 20 22 00 02 ^ Figure 6 Depth and temperature records for a bigeye tuna carrying an archival tag (99-792), exhibiting behavior associated with a floating object. Approximate area, 1°N 97°W, is based on geolocation estimates from the archival tag. CTZ = central time zone. of time at liberty. The fish traveled a considerable distance west and southwest to about 5°S and 110°W, before mov- ing back toward the east. It was recaptured only 82 nmi east of its release location (Fig. 9C, Tables 1, 2, and 4). The fish with tag 99-787 was released at the same FAD where the fish with tag 99-792 was also released. It was at liberty for 129 d and exhibited behavior associated with floating objects for 12.8'^^f of time at liberty. In contrast to the fish with tag 99-792, the fish with tag 99-787 traveled in the opposite direction, toward the east, moving fairly rapidly to around 82°W, and then moving north and south between about 3°N and 3°S. This fish was recaptured 843 nmi east of its release location (Fig. 9C, Tables 1, 2, and 4). The fish with tag 99-889 was at liberty 76 d and exhib- ited behavior associated with floating objects for 10.0% of time at liberty. This fish traveled a considerable distance south to about 7°S, before moving back northward, and was subsequently recaptured only 100 nmi southwest of its release location (Fig. 9D, Tables 1, 2, and 4). The fish with tag 99-869 that was released at the same FAD where fish 99-889 was released, was at liberty 112 d and exhib- ited behavior associated with floating objects for 19.2% of time at liberty. The fish initially traveled in a similar direction as that of fish 99-889 before moving westward. It showed extensive latitudinal movements from about 5°S to 5°N, and was recaptured 430 nmi northwest of its release location (Fig. 9D, Tables 1, 2, and 4). Habitat selection The habitat selected by bigeye tuna is presented by month in Figure 11, for days with unassociated type-1 behavior only. The nighttime and daytime depth distributions were very similar for April, May, and June. Most of the time 776 Fishery Bulletin 100(4) Time (CTZ) 12 00 12 00 12 00 12 00 12 00 12 00 12 00 12 00 22 00 02 04 06 08 10 12 14 16 18 20 22 00 02 0 100 200 300 400 500 600 700 800 900 1000 -.^yUrAM^Jrt-U'lj-jj^.jiiiUuJ^ -^-^fljvi;ju May 16,2000 15 - 13 - 12 - 10 25 Figure 7 Depth and temperature records for a bigeye tuna carrying an archival tag (99-792). exhibiting deep diving behavior Approximate area, 1°S 97°W, is based on geolocation estimates from the archival tag. CTZ = central time zone. was spent above 50 m at night and between 250 to 300 m during the day. From April to June, the average depths of the 20°C and 15°C isotherms were 50 and 100 m, respec- tively. The habitat selected by the fish during July, August, and September was different and more variable during the day than that for the previous three months. Most of the time was spent above 50 m during the night. During the day the primary mode in depth in July was similar to the previous three months, but in August the primary mode shifted to about 200-275 m. In July the average depths of the 20°C and 15°C isotherms were 17 and 90 m, respectively, whereas in August and September the aver- age depths of the 20°C and 15°C isotherms were 10 and 75 m, respectively. In September the depth distribution during the day was more uniform from about 100-350 m, with a slight mode between 175 and 250 m. The daily vertical behavior of the fish during September was highly unusual and erratic in relation to previous months and to October; TSVe of the days for five fish were classified as unassociated type-2 behavior. The habitat selected by the fish during October, November, and December was different during the night, than that for the previous six months. Most of the time was spent above 25 m during the night. During the day the primary modes in depth during October, November, and December were not as distinct as in previous months, excluding September, but were primarily between 150 and 275 m. The average depths of the 20°C and 15°C isotherms for October and December were 15 and 75 m respectively, whereas in November the average depths of the 20°C and 15°C isotherms were 25 and 50 meters, respectively. The habitat selected by the fish during the day in January, February, and March was similar, with a distinct mode in depth between about 200 and 300 m. During the night, there was a transition Schaefer and Fuller Movements, behavior, and habitat selection of Thunnus obesus 777 0 2 00 4 00 6 00 8 00 10 00 12 00 14 00 16 00 18.00 20 00 22 00 24 00 Time of day (CTZ) x=1:18 2:00 3 00 4:00 5:00 Duration (h) 6:00 x=872 500 600 700 800 900 1000 1100 1200 Maximum depth (m) 1300 1400 1500 Figure 8 Distributions of the time of day, duration, and maximum depth for 115 deep-diving events observed and classified for bigeye tuna listed in Table 2. The average value is given in the upper right corner of each panel. CTZ = central time zone. during these three months from most of the time being spent above 25 m in January to a greater amount of time spent between 25 to 50 m at night in March. This transi- tion corresponded with a downward shift in the depths of the 20°C and 15°C isotherms from an average of 30 and 80 m in January and February to 50 and 100 m in March, respectively. The percentages of time that 13 bigeye tuna (112-126 cm in length at release) spent within 25-m depth intervals from the surface to 400 m, during each hour of each day at liberty, while exhibiting unassociated type-1 behavior, are given in Table 5. The average daytime temperatures and depths for these fish throughout their times at liberty, when exhibiting type-1 behavior, indicated that the high- est concentration (about 507r ) were between 13° and 14°C and 200 and 300 m depth. About 85% of the data were distributed between 13° and 16°C and 150 and 300 m in depth. Twelve of the twenty-two bigeye tuna showed signifi- cant correlations (r=0.61 to 0.75, P<0.05) between the vis- ible disk area of the moon and their average nighttime depth distributions. There was a significant correlation (/•=0.81, P<0.05) between the average nighttime depth for all 22 fish and the visible disk area of the moon (Fig. 12A). Fish occupied significantly greater depths for the 7-day period surrounding the full moon (29.0 m), in contrast to the other 22 days of the lunar cycle (21.2 m) (Fig. 12A). The average nighttime light levels indicated that the change in depth did not totally compensate for the greater light intensity during the full moon phase (Fig. 12A). Eight of the twenty-two fish showed significant correlations (r=0.54 to 0.75, P<0.05) between average daytime depth 778 Fishery Bulletin 100(4) Table 3 The accuracy, precision, and uncertainty in estimated latitudes and longitudes derived from archival tag light-level data from 21 bigeye tuna, processed with the geolocation programs of Wildlife Computers (Hill and Braun, 2001). The latitudes and longitudes are given as decimal degrees (dd). The error estimates are the uncertainty about the individual estimates of latitude or longitude. The differences are those between the actual latitude and longitude of recapture and the corresponding geolocation estimates. Tag no. Recapture latitude (dd) Difference Recapture longitude (dd) Difference Actual Estimate South error North error Actual Estimate West en'or East error 99-787 1.72 N 1.0 N 2.5 2.5 0.72 83.08 W 82.97 W 0.5 0.0 0.11 99-792 1.35 N 3.0 N 3.0 4.0 1.65 95.82 W 93.93 W 0.5 0.0 1.89 99-801 1.47 N 5.0 S 1.5 4.5 6.47 90.72 W 90.62 W 0.5 0.0 0.10 99-803 2.20 N 1.0 N 2.5 2.0 1.20 102.72 W 102.12 W 0.5 0.0 0.60 99-810 1.65 S 2.5 S 1.5 2.0 0.85 93.17 W 93.21 W 0.0 0.5 0.04 99-812 0.55 N 2.5 N 0.0 2.0 1.95 109.00 W 109.29 W 0.0 0,5 0.29 99-816 1.78 N 1.5 N 3.0 3.5 0.28 97.60 W 97.40 W 0.0 0.5 0.20 99-817 1.12S 2.0 S 3.5 5.5 0.88 95.87 W 95.47 W 1.0 0.0 0.40 99-835 1.20 N 5.0 N 2.5 2.5 3.80 99.68 W 98.48 W 0.5 0.0 1.20 99-839 2.33 S 0.0 N 6.5 0.5 2.33 95.75 W 95.54 W 0.5 0.0 0.21 99-847 1.77 N 3.0 S 2.5 5.0 4.77 97.23 W 97.17 W 0.0 0.5 0.06 99-860 2.18 S 2.0 S 3.0 2.0 0.18 97.87 W 97.20 W 0.5 0.0 0.67 99-861 2.23 S 1.5 N 6.0 6.5 3.73 96.67 W 95.56 W 0.0 0.0 1.11 99-864 2.70 N 2.5 N 1.0 3.0 0.20 97.87 W 98. low 0.5 0.5 0.23 99-865 2.67 S 0.0 N 1.5 6.5 2.67 99.05 W 99.33 W 0.5 0.0 0.28 99-869 4.85 N 4.0 N 2.5 2.0 0.85 103.18 W 103.35 W 0.5 0.0 0.17 99-874 2.70 N 3.0 N 1,0 2.0 0.30 97.87 W 97.81 W 0.5 0.0 0.06 99-877 2.77 S 2.0 S 7.0 3.0 0.77 99.20 W 97.37 W 0.0 1.0 1.83 99-883 1.60 S 0.5 N 0.5 0.5 2.10 95.55 W 95.67 W 0.0 0,0 0.12 99-889 0.37 N 2.5 S 1.5 2.5 2.87 98.58 W 98.57 W 0.0 0,0 0.01 99-891 0.25 S 4.5 S 1.0 9.5 4.25 92.88 W 92.74 W 0.5 0,0 0.14 Mean 2.57 3.40 2.04 0.33 0,17 0.46 95-^; CI 0.85 1.00 0.79 0.13 0,13 0.26 (distributions and visible (disk area of the moon. The pat- terns in average daytime depths for 5 of the 8 fish showed a significantly shallower depth distribution for a 3-day period surrounding the full moon (190 m) compared to the other 26 days of the lunar cycle (221 m). There was no apparent pattern, and the correlation coefficient (7-=0.15, P>0.05) was not significant for the average daytime depth for all 22 bigeye tuna in relation to the visible disk area of the moon (Fig. 12B). Discussion The results obtained in our study are useful for evaluat- ing fine- to large-scale horizontal and vertical movements, behavioral patterns, and habitat characteristics on spatial and temporal scales previously undocumented for bigeye tuna. Movement paths, residence times at FADs, and habitat selection are essential for understanding the ecol- ogy of this species, and should be incorporated into stock assessment models to evaluate its vulnerability to various modes of fishing. In addition to the 96 bigeye tuna released with archival tags, we released 101 bigeye tuna, in the same area and time period, with conventional plastic dart tags only. To date, 29 (30%) of the fish with archival tags and 22 (23%) of these with conventional tags have been confirmed as recaptured. The difference between these percentages was not significant (P>0.05), indicating that tagging mortality was probably no greater for the fish with archival tags than the fish with conventional tags. Several of the archival tags were removed from recap- tured bigeye tuna by members of the scientific staff of the Inter- American Tropical Tuna Commission (lATTC). All but two of the tags were situated in the peritoneal cavity, in the general area where they were implanted, and were apparently encapsulated by fibrous connective tissue. Of the two other tags, one had apparently been invaginated into the lumen of the stomach and the other into the lu- men of the intestine. Apparently the fish were attempting to expel these foreign bodies from their peritoneal cavities. Transintestinal expulsion of surgically implanted trans- mitters by fish was previously considered an exceptional phenomenon, except in the case of catfish (Marty and Schaefer and Fuller: Movements, behavior, and habitat selection of Thunnus obcsus 779 10°N 5°N CN 5°S ICS 10°N 0°N 5°S 10°S no'w 100°W 90°W 80°W 110°W 100°W gcw 80°W Figure 9 Movement paths of bigeye tuna derived from filtered geolocation estimates. The open squares are the release positions at FADs. The open circles and solid squares are geolocation estimates for unassociated behavior and behavior associated with a floating object, respectively, and the crosshairs are the recapture positions. The color code for each fish, in each map, corresponds with the following tag number ( Al blue: 99-812. The dashed line, between two geolocation estimates on 16 August and 27 September 2000, represents our inability to obtain reliable estimates of latitude from light level data near the equinox, red: 99-793. The dashed line, preceding the recapture position, represents our inability to obtain reliable estimates of latitude from light level data near the equinox and the failure of the tag to log data on 3 October 2000. (B( blue: 99-804. The dashed line, preceding the recapture position, represents the failure of the tag to log data on 9 October 2000. red: 99-826. The dashed line, between two geolocation estimates on 20 August and 4 October 2000. represents our inability to obtain reliable estimates of latitude from light level data near the equinox. The dashed line, preceding the recapture position, reflects the failure of the tag to log data on 2.5 December 2000 iC I blue: 99-792. The dashed line, begin- ning on 31 August 2000 preceding the recapture position, represents our inability to obtain reliable estimates of latitude from light level data near the equinox, red: 99-787. (Di blue: 99-889. red: 99-869. Summerfelt, 1986; Baras and Westerloppe. 1999). In most specimens, the ventral region of the body wall, where the stalk protruded, did not appear to be completely healed. A small, dark, circular crater was seen at the base of the stalk — obviously a mark of irritation from the movement of the stalk even after the tag body had been encapsulated by flesh. Unless it is necessary to collect internal tempera- ture data, the dorsal musculature is a potentially better location for implanting archival tags, of appropriate size and shape, for long-term deployment (Brill et al.-'). The long-term performance of the archival tags used in our experiment was questionable. The bigeye tuna that we studied pushed these tags to the limits of their design specifications by undergoing regular daily vertical forays with fairly dramatic temperature and pressure fluctua- tions, in addition to making the unexpected deep diving events exceeding 1000 m. Of the 27 archival tags recovered to date, four of them failed to collect light data because of apparent problems ' Brill, R. W, K. Cousins, and P. Kleiber. 1997. Test of the feasibility and effects of long-term intramuscular implantation of archival tags in pelagic fishes using scale model tags and captive juvenile yellowfin tuna {Thunnus albacares). NMFS Admin. Rep. H-97-11.12 p. Southwest Fisheries Center Hono- lulu Laboratory, National Marine Fisheries Service, NOAA, Honolulu, HI 96822-2396. 780 Fishery Bulletin 100(4) Table 4 Spatial statistics for 22 bigeye tuna at liberty for 30 days or longer, based on filtered estimates of the geographic locations derived from the archival tag light level data, n is the total number of geographic locations in the data set. Distance is the total distance traveled per data set. Linearity is the ratio of the distance between the data set endpoints and the total distance traveled. MSD is the mean squared distance from the center of activity, p is the proportion of MSD values, from a Monte Carlo simulation, higher than the MSD value from the observed data. UD is the utilization distribution, for the 95'* and 50'7r probability levels, reported as area in km^. NA stands for non applicable because those estimates are only valid if the movement path indicates site fidelity. Tag no. n Distance (km) .V bearing .V speed (km/d) Linearity MSD (km 109) Site fidelity (p) 95':; UD (km-) 50';i UD (km-l 99-787 75 13596 11 105 0.12 408.017 59.6 NA NA 99-792 94 20017 217 142 0.01 195.310 99.7 1,182,310 131,437 99-793 52 11996 30 86 0.06 111.726 99.9 793,555 103,429 99-801 32 7513 171 94 0.09 227.857 78.9 NA NA 99-803 72 11721 186 123 0.08 134.198 95.3 802,037 100,354 99-804 75 17044 219 115 0.06 218.924 98.4 1,169,153 104,391 99-810 33 6975 51 122 0.05 79.566 99.9 4.30,586 83,410 99-812 78 14603 182 92 0.11 262.036 84.2 NA NA 99-814 70 17671 138 143 0.02 109.497 99.9 836,649 158,936 99-816 27 4752 309 128 0.08 62.609 96.7 393,121 49,882 99-817 36 7970 158 129 0.03 83.159 99.9 478,858 50,205 99-826 108 23,085 219 113 0.03 494.010 78.9 NA NA 99-835 37 6785 183 119 0.09 111.445 91.9 NA NA 99-839 24 6143 267 110 0.03 90.090 99.4 501,057 107,094 99-847 11 3188 126 76 0.04 113.948 96.3 516,561 145,177 99-860 38 8893 324 165 0.02 65.013 99.9 389,426 51,713 99-865 29 6572 258 111 0.06 69.959 99.8 508,706 94,889 99-869 81 15811 335 140 0.05 221.813 95.5 1,295,394 205,956 99-874 17 4395 96 146 0.03 123.812 95.4 526,085 84,107 99-883 29 7282 58 121 0.03 83.922 99.9 564,556 103,.384 99-884 149 32501 115 98 0.04 174.997 99.9 1,164,992 341,468 99-889 31 6694 252 87 0.03 85.269 99.3 609,874 154,383 with their stalks. The batteries in four of the six tags, which were in fish at liberty for 175 days or more, failed and these tags stopped collecting data. Fortunately, how- ever, previously collected data were preserved in the non- volatile memories of the tags. Perhaps the most important feature of archival tags is their ability to collect data on the movement of tagged fish at frequent intervals from re- lease until recapture (Hunter et al., 1986; Gunn and Block, 2001). There are, however, several factors that can affect the accuracy of the geoposition estimates. These include, but are not limited to, latitude, equinoxes, resolution of the light sensor, light attenuation, and behavior of the fish (Gunn and Block, 2001; Musyl et al., in press). Gunn et al.'* previously reported the accuracy of geolo- cation estimates from the light data from archival tags Gunn, J. S., T. W. Polacheck, T. L. O. 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 technology to groundfish research. In Proceedings of ICES mini-symposium on fish migration, 23 p. International Council for the Exploration of the Sea, Palaegade 2-4, DK-1261 Copenhagen K, Denmark. attached to southern bluefin tuna iThinuius maccoyii) held in cages in the Indian Ocean to be about 0.5° in lon- gitude and 1.5° in latitude. Welch and Eveson (1999) and Musyl et al. (2001) estimated the accuracy of geolocation estimates from the light-level data recorded by archival tags by comparing the known and estimated locations of tags that were attached to oceanographic buoys in the north Pacific. The reported accuracy by Welch and Eveson (1999) was ±0.9° in longitude and ±1.2° in latitude. The reported accuracy by Musyl et al. (2001) ranged from 0,2° to 0.3° in longitude and from 1.5° to 4.4° in latitude. In our study, we estimated the accuracy of geolocation estimates (longitude: 0.5°, latitude: 2.0°) by comparing the known and estimated locations of 21 bigeye tuna on their days of recapture (Table 3). We used our estimates of accuracy and precision as criteria for filtering the daily geolocation estimates. In other studies where archival tags were used to provide estimated movement paths, geolocation estimates derived from light data have been verified or adjusted by compar- ing recorded temperatures from archival tags with maps of estimated sea-surface temperatures from satellite data (Gunn and Block, 2001). In the equatorial EPO, sea-sur- Schaefer and Fuller Movements, behavior, and habitat selection of Thunnus obesus 781 10°N sn CN- 5^- ID'S- 110°W 105°W lOCW 95°W 90°W 85'W 80°W Figure 10 Geolocation estimates for 22 bigeye tuna at liberty for 30 d or longer (Table 4) classified as unas- sociated (blue dotst or associated with floating objects (red triangles). The yellow solid squares are the release locations and the green crosshairs are the recapture locations. The larger blue and red dots, axes, and elipses are the arithmetic means, major and minor axes, and SS'r probability ellipses, respectively, for the spatial distributions of the geolocation estimates for unassociated behavior and behavior associated with floating objects. face temperatures vary little over extremely large areas (Fiedler, 1992); therefore this technique is much less use- ful than in temperate regions. The movement paths shown for bigeye tuna in our study (Figs. 9 and 10) derived from the filtered archival tag light data indicated that the area was restricted to the equatorial EPO. No fish traveled further west than about 110°W, and most movements were constrained be- tween about 95° and 100°W and 3°N and 5°S. However, the value of the archival tags in providing fisheries-inde- pendent information on dispersion and movement paths is apparent, especially considering the fact that 16 of the fish were recaptured within 300 nmi of where they were released. Furthermore, the minimum convex polygon for the filtered archival tag data is approximately four times the area of the minimum convex polygon surrounding the release and recapture positions. Movements of big- eye tuna inferred from large-scale conventional tagging programs in the western Pacific (Hampton and Gunn, 1998; Hampton et al.''; Kaltongga''! and Hawaii (Itano and Holland, 2000) indicate that, although there are some long-distance movements, most recoveries are near their points of release. Those data appear to indicate, as do those of the present study, regional fidelity for bigeye tuna, and that the expected degree of mixing is quite low between the EPO and the central and western Pacific Ocean (CWPO). The estimated mean velocity of 117 km/d or 2.6 knots (Table 4) is comparable to the estimate of 130 km/d for Pacific bluefin tuna {Thunnus orientalis) from archival tag data (Tsuji et al.. 1999). Although this estimate should not be interpreted as actual swimming speed through the water, considering the imprecision of the movement paths and the fact that daily vertical movements were not Hampton. J., K. Bigelow, and M. Labelle. 1998. A summary of current information on the biolog>', fisheries and stock assess- ment of bigeye tuna [Thunnus obesus) in the Pacific Ocean, with recommendations for data requirements and future research. Secretariat of the Pacific Community. Oceanic Fisheries Pro- gramme. Technical Report 36,46 p. Oceanic Fisheries Pro- gramme. SPC. B.P. D5. 98848 Noumea Cedex. New Caledonia. Kaltongga. B. 1998. Regional tuna tagging project: data summary. Oceanic Fish. Prog. Tech. Rept. 35. 70 p. Secretar- iat of the Pacific Community. Noumea, New Caledonia. Oce- anic Fisheries Programme, SPC. B.P D5. 98848 Noumea Cedex. New Caledonia. 782 Fishery Bulletin 100(4) 45 90 90 Percent time 45 0 45 90 90 45 10 15 20 25 30 15 20 25 Temperature (C) 10 15 45 90 20 25 30 Figure 11 Depth frequencies of bigeye tuna during the night (solid bars) and day lopen bars), along with the vertical thermal profiles by month and year (unassociated type-1 behavior days only). Data from depths greater than 400 m are excluded. The number of fish represented in each month is given. included in the calculations of velocity. Nevertheless, this estimate should be useful for incorporation into a spatially stratified movement model that is designed to evaluate dispersion and mixing rates between large regions (see Sibert and Fournier, 2001). Sonic tracking studies have shown that the diel verti- cal migrations of bigeye tuna are closely associated with vertical movements of organisms of the deep scattering layer (DSL). Bigeye tuna probably forage on squids and other mesopelagic organisms within the DSL throughout the day and night (Josse et al., 1998; Dagorn et al., 2000). As reported by Blunt (1960), squid are very important in the diet of subsurface bigeye tuna, in the eastern tropical Pacific (ETP) — 10% of the stomachs examined contained squid equivalent to 60'7( of the total food volume. Fiedler et al. ( 1998) reported the depths of the DSL in the ETP as 300-400 m during the day and 0-100 m at night. Other studies have documented that bigeye tuna have evolved anatomical and physiological adaptations to enable them to exploit organisms of the DSL during the daytime in a dark, cold, and oxygen-poor environment (Kawamura et al., 1981; Holland et al., 1992; Brill, 1994; Holland and Si- bert, 1994; Schaefer, 1999; Lowe et al.. 2000; Graham and Dickson, 2001). The depth distributions of bigeye tuna not associated with FADs near Hawaii and in the Coral Sea are signifi- cantly greater than those for bigeye tuna in the equatorial EPO. It seems possible that the greater daytime depths exhibited by bigeye tuna in the CWPO are related to the greater daytime DSL depths (>400 m) in that region (May- nard et al., 1975;Tont, 1976; Davies, 1977; Kuznetsov et al., 1982; Fiedler et al., 1998; Josse et al., 1998). For a 112-cm Schaefer and Fuller Movements, behavior, and habitat selection of Thunnus obesus 783 Percent Time 1( )0 75 50 25 0 25 5 25 50 3 75 1 100 : 125 =1 150 =1 175 =1 200 =1 225 Z3 250 z: 275 n 300 =1 325 1 350 375 7 October 2000 - 400 / n=4 - 1 I I 10 15 20 25 30 50 30 100 75 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 Percent Time 50 25 0 25 50 :j* J» ^^ 3 ^ D 4 D i D i D i n i ] =: i Zl i =3 J Zl i Zl * D i 2 I November 2000 " i J n=3 -^ 1 1 : 30 50 10 15 20 25 Temperature (C) Figure 11 (continued) 30 Percent Time 100 75 50 25 0 25 50 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 400 Zl ^'^ 'A 1 1 /^ 1 I 3 i Zl i =1 i =] i =1] i =1 I =1 f 1 December 2000- l n = 3- 1 T ' I 10 10 15 20 100 75 50 25 25 30 0 25 50 15 20 25 30 bigeye tuna in the equatorial EPO not associated with a FAD (Fig. 2), the baseline daytime depth was 250 to 300 m at temperatures of 12° to 13°C. For a 131-cm bigeye tuna exhibiting similar behavior in Hawaiian waters, the base- line daytime depth was 400 to 500 m at temperatures of 7° to 10°C (Musyl et al., in press). Gunn and Block (2001) reported that bigeye tuna with archival tags in the Coral Sea showed that the mean depth of the fish at night was 50 m and that during the day they were at depths of 450 to 500 m and at temperatures of 7° to 9°C. The average light level experienced by bigeye tuna at night (77 Wildlife Computer's light level [well at 24 m). is below that experienced during the day ( 126 wcl at 242 m). Therefore, bigeye tuna do not occupy an isolume as has been suggested for other vertically migrating organisms (Widder and Frank. 2001). Bigeye tuna are also able to adapt to much higher light levels ( 195 wcl I for prolonged periods when remaining at shallow depths during day- light hours when they are associated with FADs. Unassociated type-2 behavior observed in bigeye tuna (Fig. 5 and Table 2) may be attributed to a shift in the ver- tical distribution of prey items. Monospecific 100- to 200- ton schools of bigeye tuna (>100 cm length) were observed feeding at the surface on the mesopelagic fish Vincigueria lucetia during daylight hours on 15 December 1978 in the equatorial EPO (K. Schaefer, unpubl. data). Vincigueria lucetia is normally distributed at depths of 500 m or more during the day and is common in the 0-90 m layer at night (Blackburn, 1968). During the 1971-91 period, previous to the development of the drifting FAD fishery in the EPO, many purse-seine sets made during daylight hours were successful in capturing bigeye tuna schools not associ- ated with drifting objects (Calkins et al., 1993). Atypical behavior of large schools of Vincigueria nimba?-i. present in large concentrations within the mixed layer during the day, has also been observed in the equatorial Atlantic Ocean (Marchal and Lebourges, 1996). This type of behav- ior has also been observed in the Coral Sea, where bigeye 784 Fishery Bulletin 100(4) • c 123456789 10 11 Moon phase o a • 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 5 - II III Ml) 10 - 15 \ MO 70 60 20 ■ ' 1 50 40 25 ■ 30 30 - A 20 ■ 10 ,- E £ • © g- 123456789 10 1 r-i 0 O 3 • 1 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 ~ — 50 ■ 100 - 150 - 200 - 0 B Figure 12 Average (A) nighttime and (Bi daytime depths (solid bars) and light levels (open circles) for 22 bigeye tuna, by moon phase (unassociated type-1 behavior days only). tuna are caught throughout the day near the surface by both handhne and longhne fisheries between October and December (Hisada, 1973). A second explanation for some of the unassociated type-2 behavior is that the fish were still possibly associated with a FAD but were making af- ternoon excursions into the DSL, foraging for food. Occasionally bigeye tuna make dives in excess of 500 m (Fig. 7 and Table 2). The durations of the deep diving events are not correlated (r=0.0008, P>0.05) with body size. Moreover, there appear to be two types of deep dives (Table 2). One may be for predator avoidance. The fish descend and then ascend rapidly back to the depth at which it had been previous to the dive. The second type may be a foraging behavior, where prolonged time is spent at greater depths, up to 1000 m. Deep dives to 1000 m or more have also been reported from archival tag data for bigeye tuna in the Coral Sea (Gunn and Block, 2001) and for Atlantic bluefin tuna (Block et al., 2001a; Block et al., 2001b). Bigeye tuna exhibiting behavior associated with floating objects ( Fig. 6 and Table 2) generally remain above the ther- mocline, but they still show a diel shift in depth distribu- tion. They remain at about 6 m, on average, deeper during the day than at night (Fig. 6). Stomach content analysis of FAD-associated bigeye tuna in the eastern Atlantic Ocean by Menard et al. (2000) showed that 82.7'7r of the stomachs were empty, where as only 25% of the stomachs of bigeye tuna unassociated with floating objects were empty. They concluded that FADs do not have a trophic function. How- ever, the observed excursions to depths of about 300 m for a few hours at about 1800 h for four consecutive days by a FAD-associated bigeye tuna (Fig. IB) may be related to foraging for prey in association with the DSL. Parin and Fedoryako (1999) stated that tunas associate with floating objects only temporarily because there are in- sufficient food resources in the vicinity of these devices. We found that residence times and total times spent at FADs are limited (Table 2). Our observations do not support the Schaefer and Fuller: Movements, behavior, and habitat selection of Jhunnus obesus 785 hypothesis of Marsac et al. (2000) that association with FAD.s causes bigeye tuna to be retained within areas or transported to new areas, thus creating an ecological trap. An alternative hypothesis, suggested by Hunter and Mitchell ( 1966), is that FADs function by simply providing a visual stimulus in an optically void environment. An extension of their hypothesis should include the fact that FADs provide a general sensory stimulus, including sound produced by the FAD and associated fauna, which may be the mecha- nism by which tunas locate FADs. In addition to a general sensory stimulus. FADs may also function as reference points (Freon and Dagorn, 2000). Archival tags have provided data on the type of habitat selected (light levels, depths, and tempera- tures) that should be useful for standardizing the catch per unit of effort (CPUE) of bigeye tuna by surface and longline fisheries in the EPO. The data in Table 5 indicate that bigeye tuna greater than 110 cm spend Sl^i or more of their time above 50 m, in the mixed layer at night and 53''v or more of their time between 200 and 300 m during the day. Habi- tat-based stock assessment models (Hinton and Nikano. 1996; Hinton and Deriso, 1998; Hampton et al. ') have been developed for the integration of data, such as those provided in Table 5. to adjust effort based on estimated fishing depth of longline gear (Mizuno et al.. 1999) in relation to the vertical distribution of target species by time of day. The fishing depth of longline gear has been shown to be an important source of variation in the CPUE for bigeye tuna (Hanamoto, 1987; Boggs, 1992); higher catch rates of bigeye tuna have been associated with greater fishing depths of the longline gear. This has been interpreted previously as a prefer- ence of bigeye tuna for 10° to 15°C water during daylight hours (Hanamoto. 1987; Holland et al., 1990; Boggs, 1992; Brill. 1994). We suggest that bigeye tuna are most likely not selecting their daytime and nighttime habitats based on temperature, depth, or light preferences, but on the distributions of their preferred prey. Cephalopods and mesopelagic fishes also show diel vertical migrations (as do other organisms) as- sociated with the DSL. We suggest that the depths and temperatures preferred by bigeye tuna during daylight hours when exhibiting unassociated type- 1 behavior are the environmental variables associ- ated with their preferred prey. The greater depths and lower temperatures at which bigeye tuna are caught during the daytime in the CWPO (Hampton et al.''; Miyabe") may be a function of the greater daytime depths of the DSL ' Miyabe, N. 199.5. Follow-up study on the stock status of bigeye tuna in the Pacific ocean. Western Pacific Yellowfin Research Group 5. working paper 12; 21-23 August 1995, 15 p. Oceanic Fisheries Programme, SPC, B.P D5, 98848 Noumea Cedex, New Caledonia. II N -a c fei a c ';Ot^CO'X)^i>J'-Jr-i^ CO c^ t^ Tf CA CM rH ^ T-4 CO CM C-J o o o o o o (N ca lO CM -^ o ^ O 00 CO so ^ ^ ^ — ^ ,-H o o o o o o coo'X'C£)a:)i>-t:^oai'-'0-* COr-H.-HCNi-HlOO^i-HO'-'P cqiOOoqt^^^DLOO^cocooO C^J'M'i-iO'-^CO-rr'CC'-^-^C^aJ T-l ,— (M 1-. C^.-tCJirr-C^CO'lD'-'CTiOOCOX d d CO d (X) CM •-( CD ^ d t^ '^ CO CO --* d ■^ GO Tt- ^ 00 — * d d GO O LO t-H 00 cm' d d lO lO -^ 00 CM d CO CO ;o •— ' GO c^ d d m o '^ •— I GO CM d d GOOi-HOOCO-^OrfO^iOCDXCDiOCO CMCO— Tj-.-.r-. c^ CM --H d CM CO in t^ 1-* ic CO oi Oi '-' CDr^'^0O'-'C00000tJiO-]LiOi-H|>CM coiococMicdo6ddE>o]i.O'-Hd OI>colOCMco^-•-HaJ^ CM CO cm' d d d d d d d (J:ait^uOCMCMC^)'— 1^.— < O CM CO lO CO CM OO'-'Tj'COCOC^i-'CM'-'CM ■^Oli-HiOCMCMCMCMCM'-H X ^ '-' d> d> di »0 O lO CM lO O I I I O "Xl r- ^ CM LC I> o »o o O CM iC 1" 7 7 CO ^ Xt - . O CM lO lOOiOOLCOiOOiCO t>OCMlCt^OCMint>0 i-tCMCMOJCMCOCOCOCO^ I I I I I I I i I I tr^ocMint^ocMmc- ^CMCMCMCMCOCOCOCO 786 Fishery Bulletin 100(4) in those regions, compared to the daytime depths of the equatorial EPO (Maynard et al., 1975; Tont, 1976; Davies, 1977; Kuznetsov et al., 1982; Fiedler et al., 1998; Josse et al., 1998). Farr and Best (1998) reported that DSL distri- butions are related to mesoscale oceanographic features, defined by flow and temperature variability, and are most commonly observed at the pycnocline. Geographic variation in DSL depths throughout the Pacific are possibly related to isolumes of the associated micronekton, and can potentially be estimated from the light level data recorded by the archival tags attached to bigeye tuna. Variation in daytime DSL depths is probably a function of light penetration (which is regulated by bio- logical production) and absorption of light by chlorophyll and phaeopigments (Tont, 1976). The behavior of bigeye tuna is strongly influenced by the presence of drifting FADs within their habitat. Be- cause of this behavior associated with FADs, even though it is for relatively short periods, bigeye tuna are highly vulnerable to capture by purse-seine vessels. Estimates of bigeye tuna residence times and percentages of total time associated with drifting FADs, along with estimates of FAD densities, could be used to evaluate vulnerability to capture by the surface fishery. There is a critical need for conducting a large-scale tagging program in the EPO fo- cused on bigeye tuna — a program where conventional tags are used for estimating size-specific mortality and mixing rates and archival tags are used for evaluating fine-scale movements, behavior, and habitat selection. Acknowledgments We are grateful for invaluable advice and assistance pro- vided by B. Block, T Booth, M. Braun, R. Brill, J. Gunn, R. Hill, P. Hooge, and T Williams. We are thankful to B. Blocker and the crew of Her Grace for their performance in fishing and tagging operations. We are indebted to vessel owners, captains, fishermen, unloaders, and industry representatives for returning recovered archival tags. We thank Juan Gracia for his assistance with tagging and the lATTC field office personnel for recovering archival tags. We also wish to thank B. Bayliff, R. Brill, R. Deriso, G. Watters, and two anonymous reviewers for constructive comments on drafts of the manuscript. Literature cited Baras, E., and L. Westerloppe. 1999. Transintestinal expulson of surgically implanted tags by African catfish Heterobranchus longiftlis of variable size and age. Trans. Ani. Fish. See. 128(4>:737-746. BaylifT,W.H.( editor). 2000. Inter-Am. Trop. Tuna Comm., Ann. Rep. for 1998, 358 p. lATTC, La Jolla, CA. Blackburn, M. 1968. Micronekton of the eastern tropical Pacific Ocean: family composition, distribution, abundance and relations to tuna. Fish. Bull. 67:71-115. Block, B. A., H. Dewar, S. B. Blackwell.T Williams, E. D. Prince, C. J. Farwell, A. Boustany, S. L. H. Teo, A. Seitz, A. Walli. and D. Fudge. 2001a. Migratory movements, depth preferences and thermal biology of Atlantic bluefin tuna. Sdence 293:1310-1314. Block, B.A., H. Dewar, S. B. Blackwell, T Williams, E. Prince, A. M. Boustany, C. Farwell, D. J. Dau, and A. Seitz. 2001b. Archival and pop-up satellite tagging of Atlantic bluefin tuna. In Electronic tagging and tracking in marine fisheries (J. Sibert and J. Nielsen, eds. ), p. 65-88. Kluwer Academic Pubis.. Dordrecht. Blunt, C.E.Jn 1960. Observations on the food habits of longline caught bigeye and yellowfin tuna from the tropical eastern Pacific 19.5,5-1956. Calif Fish Game 46( 1 ):69-80. Boggs, C. H. 1992. Depth, capture time, and hooked longevity of longline- caught pelagic fish: timing bites of fish with chips. Fish. Bull. 90:642-6.58. Brill, R.W. 1994. A review of temperature and oxygen tolerance stud- ies of tunas pertinent to fisheries oceanography, movement models and stock assesements. Fisheries Oceanogr. 3(3): 204-216. Brill, R. W., and M. E. Lutcavage. 2001. Understanding environmental influences on move- ments and depth distributions of tunas and billfishes can significantly improve population assessments. Am. Fish. Soc.Symp. 25:179-198, Calkins, T. P., M. Yamaguchi, and N. Miyabe. 1993. Some observations on bigeye tuna {Thunnus obesus) caught by the surface and longline fisheries for tunas in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm., Bull. 20(8 ):443-499. Colletle, B. B., and C. E. Nauen. 1983. FAO species catalogue. Scombrids of the world. An annotated and illustrated catalogue of tunas, mackerels, bonitos and related species known to date. FAO Fish. Synop. 125, vol. 2, 137 p. FAO. Rome. CoUette, B. B., C. Reeb, and B. A. Block. 2001. Systematics of the tunas and mackerels (Scombridae). In Tunas: ecological physiology and evolution. (B. A. Block, and E. D. Stevens, eds.), p. 1-33. Academic Press, San Diego, CA. Dagorn, L., P. Bach, and E. Josse. 2000. Movement patterns of large bigeye tuna (Thunnus obesus) in the open ocean, determined using ultrasonic telemetry Mar Biol. 136(2):361-371. Davies, I. E. 1977. Acoustic volume reverberation in the eastern tropi- cal Pacific Ocean and its relationship to oceanographic features. Deep-Sea Res. 24: 1049-1053. Farr, R, and T Best. 1998. Relating high frequency volume scattering distribu- tions to mesoscale oceanographic features, http://www.nnic. noaa.gov/SOCC/TVRS_paperhtml. [Access date: 9 Au- gust 2001.1 Fiedler P C. 1992. Seasonal climatologies and variability of eastern tropical Pacific surface waters. U.S. Dep. Commer, U.S., NOAATech. Rep. NMFS 109:1-65. Fiedler, P. C, J. Bariow, and T Gerrodette. 1998. Dolphin prey abundance determined from acoustic backscatter data in eastern Pacific surveys. Fish. Bull. 96: 237-247. Schaefer and Fuller: Movements, behavior, and habitat selection of Tliunnus obesus 787 Freon. P., and L. Dagorii. 2000. Associative bcliaviour of pelagic fish: fiicts and hypo- thcsi's. Ill Ppcho thoniore et dispositifs dc concentration de poissons, colloque Caraibe-Martinique; Trois-Ilcts, 15- 19 Octobre 1999 (J.-Y. Lc Gall, \'. Cayre, and M. Taquet, eds.), p. 483-491. lust. Fran. Recherche Exploitation Mer (IFREMP:R)28. (iraham, J. B., and K. A. Dickson. 2001. Morphological and physiological specializations for endothermy. In Tunas: ecological physiology and evolu- tion (B. A. Block and E. D. Stevens, eds.), p. 121-165. Aca- demic Press, San Diego, CA. Gunn, J., and B. A. Block. 2001. Advances in acoustic, archival and pop-up satellite tagging of tunas. In Tunas: ecological physiology and evolution (B. A. Block and E. D. Stevens, eds.), p. 167-224. Academic Press, San Diego, CA. 1 lampton, J., and J. Gunn. 1998. Exploitation and movements of yellowfin tuna (Thun- nus albacares) and bigeye tuna (T! obesus) tagged in the north-western Coral Sea. Mar. Freshwater Res. 49(6): 475-489. Hanamoto, E. 1987. Effect of oceanographic environment on bigeye tuna distribution. Bull. Jap. Soc. Fish. Oceanogr., 51(3):203-216. Hill.R. D. 1994. Theory of geolocation by light levels. In Elephant seals: population ecology, behavior, and physiology (B. J. LeBouef and R. M. Laws, eds. ), p. 227-236. Univ California Press, Berkeley, CA. Hill, R. D., and M. J. Braun. 2001. Geolocation by light levels — the next step: latitude. In Electronic tagging and tracking in marine fisheries (J. Sibert and J. Nielsen, eds. ), p. 315-330. Kluwer Academic Pubis., Dordrecht. Hinton, M. G., and R. D. Denso. 1998. Distribution and stock assessment of swordfish, Xiphias gladius, in the eastern Pacific Ocean from catch and effort data standardized on biological and environ- mental parameters. In Biology and fisheries of sword- fish, Xiphias gladius (I. Barrett, O. Sosa-Nishizaki, and N. Bartoo, eds.), p. 161-179. U.S. Dep. Commer, NOAA Tech. Rep. NMFS 142. Hinton, M. G., and H. Nikano. 1996. Standardizing catch and effort statistics using physi- ological, ecological, or behavioral constraints and environ- mental data, with an application to blue marlin (Makaira nigrans) catch and effort data from Japanese longline fisheries in the Pacific. Inter-Am. Trop. Tuna Comm., Bull. 21(4):69-200. Hisada, K. 1973. Investigation on tuna hand-line fishing ground and some biological observations on yellowfin and bigeye tunas in the northwestern Coral Sea. Bull. Far Seas Fish. Res. Lab. 8:35-69. 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. Holland. K. N., R. W. Brill, J. R. Sibert, and D. A. Fournier 1992. Physiological and behavioral thermoregulation in bigeye tuna Thunnus obesus. Nature 358(63851:410-412. Holland K. N., and J. R. Sibert. 1994. Physiological thermoregulation in bigeye tuna, Thun- nus obesus. Environ. Biol. Fish. 40(3):319-327. Hooge, P. N., and B. Eichenlaub. 1997. Animal movement extension to arcview, version 2.04. Alaska Biological Science Center, U.S. Geological Survey, Anchorage, AK. http://www.absc.usgs,gov/glba/gistools/ [Access date: 28 November 2000.1 Hunter, J. R., A.W. Argue. W. H. Bayliff, A. E. Dizon, A. Fonteneau, D. Goodman, and G. R. Seckel. 1986. The dynamics of tuna movements: an evaluation of past and future research. FAO Fish. Tech. Pap., 277, 78 p. FAO, Rome. Hunter, J. R., and C. T. Mitchell. 1966. As.sociation of fishes with flotsam in the offshore waters of central America. Fish. Bull. 66:13-29. Itano, D. G., and K. N. Holland. 2000. Movement and vulnerability of bigeye (Thunnus obesus) and yellowfin tuna (Thunnus albacares) in relation to FADs and natural aggregation points. Aquat. Living Res. 13(4):213-223. Josse, E., P. Bach, and L. Dagorn. 1998. Simultaneous observations of tuna movements and their prey by sonic tracking and acoustic surveys. Hydro- biologia 371/372(1-31:61-69. Kawamura, G., W. Nishimura, S. Ueda, and T. Nishi. 1981. Vision in tunas and marlins. Mem. Kagoshima Univ. Res. Cent. South Pac. 2( 1 ):3-47. Kendall, M. G.. and W. R. Buckland. 1982. A dictionary of statistical terms, 213 p. Longman Inc., New York. NY. Kuznetsov. I. L., S. R. Stefanov, and V. I. Savagov. 1982. A migrating sound scattering layer in the equatorial Pacific Ocean. Oceanology 22(6):702-703. 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 Caraibe-Martinique; Trois-Ilets, 15-19 Octobre 1999 (J.-Y. Le Gall, P Cayre, and M. Taquet, eds.), p. 78-107. Inst. Fran. Recherche Exploitation Mer (IFREMER)28. Levitus, S., and T. Boyen 1994. World ocean atlas 1994, vol. 4.. Temperature. NOAA Atlas. NESDIS 4, 117 p. NOAA, Washington, D.C. Lowe, T E., R. W. Brill, and K. L. Cousins. 2000. Blood oxygen-binding characteristics of bigeye tuna (Thunnus obesus). a high-energy -demand teleost that is tol- erant of low ambient oxygen. Mar Biol. 136(6):1087-1098. Marchal. E., and A. Lebourges. 1996. Acoustic evidence for unusual diel behaviour of a meso- pelagic fish (Vinciguerria nimbaria) exploited by tuna. ICES J. Mar Sci.53(2):443-447. Marsac, F., A. Fonteneau. and F. Menard. 2000. Drifting FADs used in tuna fisheries: an ecological trap? In Peche thoniere et dispositifs de concentration de poissons, colloque Caraibe-Martinique; Trois-Klets, 15-19 Octobre 1999 (J.-Y. Le Gall, P. Cayre, and M. Taquet, eds.), p. 537-552. Inst. Fran. Recherche Exploitation Mer(IFREMER)28. Marty, G. D., and R. C. Summerfelt. 1986. Pathways and mechanisms for expulsion of surgi- cally implanted dummy transmitters from channel catfish. Trans. Am. Fish. Soc. 115(4):577-589. Maynard, S. D., F. V. Riggs, and J. F Walters. 1975. Mesopelagic micronekton in Hawaiian waters: faunal composition, standing stock, and diel vertical migration. Fish. Bull. 73:726-736. 788 Fishery Bulletin 100(4) Menard, F., B. Stequert. A. Rubin, M. Herrera, and E. Marchal. 2000. Food consumption of tuna in the equatorial Atlan- tic Ocean: FAD-associated versus unassociated schools. Aquat. Living Res. 13(4):233-240. Miyabe, N., and W. H. Bayliff. 1998. A review of information on the biology, fisheries, and stock assessment of bigeye tuna, Thunnus obsesus, in the Pacific Ocean. Inter-Am. Trop. Tuna Comm., Spec. Rept. 9:129-170. Mizuno, K.. M. Okazaki, H. Nakano, and H. Okamura. 1999. Estimating the underwater shape of tuna longlines with micro-bathythermographs. Inter-Am. Trop. Tuna Comm., Spec. Rep. 10:1-35. Musyl, M. K., R. W. Brill, D. S. Curran, J. S. Gunn, J. R. Hartog, R. D. Hill, D. W. Welch, J. P. Eveson, C. H. Boggs, and R. E. Brainard. 2001. Ability of electronic archival tags to provide esti- mates of geographical position based on light intensity. In Electronic tagging and tracking in marine fisheries (J. Sibert and J. Nielsen, eds. ), p. 343-367. Kluwer Academic Pubis., Dordrecht. Musyl, M. K., R. W. Brill, C. H. Boggs. D. S. Curran, T. K. Kazama, and M. P Seki. In press. Vertical movements of bigeye tuna (Thunnus obesusi associated with islands, buoys, and sea mounts of the Hawaiian Archipelago from archival tagging data. Fish. Ocean. Nakano, H., M. Okazaki. and H. Okamoto. 1997. Analysis of catch depth by species for tuna longline fishery based on catch by branch lines. Bull. Nat. Res. Inst. Far Seas Fish. 34:4.3-62. NMFS (National Marine Fisheries Service). 1994. Archival tags 1994: present and future. Archival tag working group, 45"' annual tuna conference, Lake Arrow- head, CA, May 23-26, 1994. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-357, 42 p. Parin, N., and B. Fedoryako. 1999. Pelagic fish communities around floating objects in the open ocean. Inter-Am. Trop. Tuna Comm., Spec. Rep. 11:447-458. Schaefer, K. M. 1999. Comparative study of some morphological features of yellowfin iThunnus albacares) and bigeye iThunnus obesus) tunas. Inter-Am. Trop. Tuna Comm., Bull. 21 (7): 489-526. Sibert, J. R.. and D. A. Fournier. 2001. Possible models for combining tracking data with conventional tagging data. In Electronic tagging and tracking in marine fisheries (J. Sibert and J. Nielsen, eds.), p. 443-456. Kluwer Academic Pubis., Dordrecht. Suzuki, Z., and S. Kume. 1982. Fishing efficiency of deep longline for bigeye tuna in the Atlantic as inferred from the operations in the Pacific and Indian Oceans. Int. Comm. Conser Atlan. Tunas, Coll. Vol. Sci. Papers, 17(2):471-486. Syrjala, S. E. 1996. A statistical test for a difference between the spatial distributions of two populations. Ecology 77(1 ):75-80. Tont, S. A. 1976. Deep scattering layers: patterns in the Pacific. Rep. Calif Coop. Ocean. Fish. Invest. 18:112-117. Tsuji, S., T Itoh, A. Nitta, and S. Kume. 1999. The trans-Pacific migration of a young bluefin tuna, Thunnus thynnus. recorded by an archival tag. Working Paper ISC2/99/15, Interim Scientific Committee for Tuna and Tuna-like Species in the North Pacific Ocean. January 15-23. 1999. Honolulu. U. S. Naval Observatory. Astronomical Applications Department. 2001. Fraction of the moon illuminated, 2000 at midnight. Central Standard Time, http://mach.usno.navy.mil/cgi-bin/ aa_moonill.pl. (Access date: 5 April 2001.] Watters, G., and M. Maunder. 2001. Status of bigeye tuna in the eastern Pacific Ocean. Stock assessment report of the Inter-Am. Trop. Tuna Comm. 1:109-210. Welch. D. W.. and J. P Eveson. 1999. An assessment of light-based geoposition estimates from archival tags. Can. J. Fish. Aquat. Sci. 56(7):1317- 1327. Widder. E. A., and T. M. Frank. 2001, The speed of an isolume: a shrimp's eye view. Mar. Biol. 138(4):669-677. Wildlife Computers. 2002. Mk9 archival tag. http://www.wildlifccomputers.com/ Archival7r20Tags/Mk9.htm. [Access date: 6 August 2001.] 789 Abstract— I sinuilali>d somatic fjrowth and accompanyiiit,' otolith growtli using an individvial-l)as('(l bioenorgotics model in order to examine the performance ol' several back-calculation methods. Four shapes of otolith radius-total length re- lations (OK-TL) were simulated. Ten dif- ferent back-calculation equations, two different regression models of radius- length, and two schemes of annulus selection were examined for a total of 20 different methods to estimate size at age from simulated data sets of length and annulus measurements. The ac- curacy of each of the twenty methods was evaluated by comparing the back- calculated length-at-age and the true length-at-age. The best back-calculation technique was directly related to how- well the OR-TL model fitted. Wlien the OR-TL was sigmoid shaped and all annuli were used, emploving a least- squares linear regression coupled with a log-transformed Lee back-calcula- tion equation (y-intercept corrected) resulted in the least error; when only the last annulus was used, employing a direct proportionality back-calculation equation resulted in the least error. When the OR-TL was linear, emploving a functional regi-ession coupled with the Lee back-calculation equation resulted in the least error when all annuli were used, and also when only the last an- nulus was used. If the OR-TL was ex- ponentially shaped, direct substitu- tion into the fitted quadratic equation resulted in the least error when all annuli were used, and when only the last annulus was used. Finally, an asymptotically shaped OR-TL was best modeled by the individually corrected WeibuU cumulative distribution func- tion when all annuli were used, and when only the last annulus was used. An evaluation of back-calculation methodology using simulated otolith data Michael J. Schirripa Hatfield Marine Science Center Northwest Fishenes Science Center 2030 SE Marine Science Drive Newport, Oregon 97365 5296 E mail address Michael Schirripaia'noaa gov Manuscript accepted 29 May 2002. Fish. Bull. 100:789-799(20021. The average rate of growth of an indi- vidual fish in a population is critical to age-based stock assessments. The aver- age rate at which the fish within the stock increases in weight ultimately determines the level of effort required to extract a desired yield from the stock as a whole (Ricker, 1975). Furthermore, current conservation standards (Gul- land and Boerema, 1973; Goodyear, 1993) are dependent upon the rate of individual growth. Thus, errors in the estimation of growth can lead to erro- neous advice to fishery managers con- cerning the present and possible future status of a population. By far the most common method of estimating fish growth rate is by esti- mating the age of individual fish from calcified structures (scales, otoliths, spines, etc.; but for this study, however, otoliths were considered the represen- tative hard structure) and with the subsequent assumption that these fish are an unbiased representation of size at that age. Growth is then described as the change in weight or length over some unit of time. To standardize age at which size is estimated, or to obtain length-at-age data on ages not included in the sample, back-calculation tech- niques are often employed to estimate a fish's size at a previous age (Bagenal, 1978). The process of back calculation can be broken down into three steps: verification of the periodicity of annulus formation, establishment of an otolith radius-total body length (OR-TL) rela- tion, and the estimation of size at the time of annulus formation. In this study, I used simulations to examine how the establishment of the OR-TL relation and the form of the back-calculation equation used may influence growth rate estimates made from otoliths. The back-calculation process as- sumes that somatic growth is directly related to otolith growth (Bagenal, 1978). This assumption is usually vali- dated through the demonstration of a relationship between the otolith radius and body length by a least-squares regression of body length on otolith radius. A variation of this technique uses a functional (model II) regression, based on the assertion that neither body length nor otolith radius are truly independent (i.e. measured without error) (Ricker, 1973, Laws and Archie, 1981). Uncertainties can enter this pro- cess from several sources. For example, incomplete data can make it difficult to discern if this relationship is linear. Furthermore, using regression to esti- mate beyond the range of the data is not recommended. Estimating beyond the range of the data can become a problem when back-calculating to very early ages that are not represented in the sample. Furthermore, several studies have found that otolith growth and somatic growth can be uncoupled (Mosegaard et al., 1988; Reznick et al, 1989; Secor and Dean, 1989; Wright et al., 1990; Milicich and Choat, 1992; Secor and Dean, 1992). Hales and Able (1995) found that changes in somatic growth accounted for only half of the variation in otolith growth. This un- coupling of somatic and otolith growth rates challenges the assumption that back-calculation is based on. The question of what is the proper back-calculation equation to use is a question that has received considerable attention. Bagenal (1978) discussed three separate methods and suggested that a combination of methods might be helpful in some cases. Francis ( 1990) presented an in-depth review of 790 Fishery Bulletin 100(4) six different back-calculation equations and their use. Ricker ( 1992) later commented on the conclusions of Fran- cis ( 1990) to suggest yet another variation on the method. Further variation exists on exactly which combination of annuli to use. Standard method suggests the use of all available annuli within the otolith to increase sample size. However, recent literature (Vaughan and Burton, 1994), as well as older reports (Ricker, 1973), have suggested that only the most recently formed annuli should be used. A review of the literature on age and growth shows that a variety of techniques are in use today and that there is no real agreement on a definitive method. The purpose of this study was to examine how well the various back-cal- culation techniques accurately estimate lengths at previ- ous ages and to examine the biases associated with each technique. Methods Model structure I simulated somatic and otolith growth using a bioener- getics model. A detailed description of the model is pre- sented in Schirripa and Goodyear (1997). The life history and growth parameters were calibrated to fit, as closely as possible, to reported estimates of striped bass growth (Bason'); however the model is not intended to be a striped bass model per se. Because of the commercial and recre- ational importance of striped bass, a great body of litera- ture from the field and laboratory work is available. One of the most studied populations of striped bass is that of the Chesapeake Bay system (Cohen et al., 1983; Coutant et al., 1984; Goodyear, 1984, 1985; Tuncer, 1988; Coutant and Benson, 1990; Secor, 1992; Brandt and Kirsch, 1993; Rose and Cowan, 1993; Rutherford and Houde, 1995; Secor and Houde, 1995). Biological and environmental parameters reported for the populations of this system were used whenever possible. The growth model used an individually based framework, but rather than following every fish of the cohort singly, "cells" offish with identical attributes were followed instead (Rose et al., 1993). A total of 250 cells, each with eleven attributes, were modeled. Attributes examined included age, length, biomass, daily food ration, food conversion efficiency, otolith weight, oto- lith radius, maximum length attained, maximum biomass attained, brain weight, condition factor, and number of fish that the cell represented. The term "population" is used to define those fish that remained alive for the entire simulation, unaffected by either natural or fishing mortality. The term "catch" re- fers to the entire group of fish that were susceptible and killed due to fishing mortality, and "sample" refers to a subsample of individuals from the catch, selected on the 1 Bason, W. H., S. E. Allison, L. O. Horseman, W. H. Keirsey. P. E. LaCivita, R. D. Sander, and C. A. Shirey. 1976. Ecological studies in the vicinity of the proposed Summit Power Station January through December 197.5. Vol. 1, Fishes, .392 p. Ich- thyological Associates, hhaca, NY. basis of length and frequency within the catch. Frequency in the catch was a function of the selectivity of the gear under consideration and frequency in the population. For the purposes of this study, gear was considered nonselec- tive. Annulus formation within the otolith was assumed to occur at the end of every growth year and to be measured without error. The specific somatic growth rate of an individual fish was calculated by a balanced energy equation. Equations for rates of consumption, respiration, egestion, and excretion generally followed those given by Hewett and Johnson. ^ The otolith growth model used was a modification of the equations presented by Mosegaard.'^ Fish formed an otolith when they reached 90 mm in length. Daily change in otolith weight (O,,,) was modeled as a function of daily change in either brain weight (B^^ ) or brain length (B,). In the case of brain weight, weight specific brain growth rate was mod- eled as a function of the somatic growth rate as follows Growth. Brain = Growth. Somatic x a.,, (1) where a., = less than 1, denoting that brain gi-owth rate is slower than somatic growth rate. The change in B^^ then was calculated as dBJdt = Growth. Brain x B^^,. (2) The daily change in otolith weight was then calculated as dO^yldt = a, X Brain. weight x temp"' (3) where a_, = the conversion factor from brain weight to oto- lith weight (see below); temp = the average temperature for the day in degrees centigrade; and «i = 0.77, which is used to determine the overall size of the otolith. Otolith radius, O^, was then calculated from O^^ assuming a spherical shape as O. 0„ :(3/4;r)" (4) ' SpD where SpD = 2.5 and is the specific density of the otolith. Assuming a spherical shape resulted in a unique radius for a given weight (i.e. a sphere made it unnecessary to consider otolith length). When brain length was used to model otolith radius, B^^, was calculated as in Equation 2 and B, was calculated as the cube root of B,„: - Hewett, S. W.. and B. L. Johnson. 1992. Fish bioenergetics model 2. Sea Grant Institute, Technical Report WIS-SG-92-250, 79 p. University of Wisconsin, Madison, WI. ■' Mosegaard, H. 1994. A model of otolith and larval fish growth. In ICES. Report of the working group on recruitment processes, CM. 1994/L:12, p. 34-38. Schirrlpa: An evaluation of back calculation methodology using simulated otolith data 791 1500 1bUU ,-.:.■■■■"•'.;■';■•■.■. B 1000 500 00 10 20 3.0 4.0 D 3 0 4 0 0 0 1.0 Otolith radius (mm) 20 3.0 4 0 Figure 1 Scatter plot of the four simulated otolith radius-total length relations (OR-TLs). (A) OR-TL/SIG = sigmoid shaped; t B I OR-TIVLIN = linear shaped; ( C ) OR-TL/EXP = exponentially shaped; (D) OR-TL/ ASYM = asymptotically shaped. D _ D 0 333 O^ was then calculated as O^ = Bjx0.5. Otolith radius-total length relation (5) (6) The OR-TL relation was fitted as closely as possible to that reported for striped bass by Heidinger and Clodfelter 1 1987 ). Modeling the conversion factor o^ ( Eq. 3 ) as a func- tion enabled me to generate four different OR-TL relations typically found in nature. A sigmoid shaped OR-TL (OR- TL/SIG) relation (Fig. lA) relation was achieved by setting the parameter a,, = 0.08 and modeling the parameter 04 as a function of body length: a^ =9-(-(-7xSin(0.006xLen^//!)-h5). (7) A linear shaped OR-TL (OR-TL/LIN) relation (Fig. IB), as found in striped bass (Heidinger and Clodfelter, 1987) was achieved by setting the parameter Oj (from Eq. 1) to 0.85 and modeling otolith radius as a function of brain length (the parameter 04 was not necessary for this relation). An exponential OR-TL( OR-TL/EXP) relation (Fig. IC), similar to that found for vermilion snapper, Rhomhoplites aurorubens, (Grimes, 1978) was achieved by again setting the parameter a., to 0.08 and modeling the parameter a^ as a linear function of otolith weight: = 25-(1.75xO„ (8) An asymptotic OR-TL (OR-TL/ASYM) relation (Fig. ID), similar to that found for walleye, Stizostedion vitreum, (Heidinger and Clodfelter, 1987), was achieved by keeping the parameter Cg = 0.08 and modeling the parameter a^ as a function of total length: a, = Q.1\1E- 12 xLength\ (9) Mortality Mortality could occur from three sources: direct starvation, random natural mortality based on length, and fishing mor- tality. If a fish lost more than a specified percentage of its maximum attained body weight (359f for larvae and 50% for juveniles), it died from starvation. Fishing mortality was described first as an overall value (F=0.4) and then divided by 365 to calculate a daily value. In order to ensure that there would be no sampling bias due to gear selectivity, fish- ing mortality was assumed to be nonselective (random). The four simulated OR-TL relations were described by using four different functions: 1) ordinary least squares (OLS) linear regression (model I) 792 Fishery Bulletin 100(4) Table 1 The ten back-calculation and OR-TL regression equations eva uated in this study. Method number refers to all annulus/last annulus only L„is the estimated length at formation of annulus R„:L and R^ is total length offish and otolith radius at capture. respectively. OLS = : ordinary least squares. Method Back-calculation equation OR-TL fitting method 1/11 L„ = (fl„/fl,)L none 2/12 L„ =a + ihR„) OLS linear regression 3/13 L„=a + (RJR, kL, -a) OLS linear regression 4/14 L„ =a + ihR„) functional linear regression 5/15 Z,„ =a + iR„/R^.)iL^ -a) functional linear regression 6/16 log,,(L„) = log^laJ -H 6(log,,(i?„)) OLS linear regression with log transformation 7/17 log^,(L„) = log,(/.,,) + bllog/Rj- \ogjR^)) OLS linear regression with log transformation 8/18 L„ =Ka ^exp(-(RJa)i')) WeibuU cumulative function 9/19 L„=c + {dRj + ieR„-) + [fR'^) quadratic equation 10/20 /.„ = (K(LJLp))iI- exp(-lRJa)l')) WeibuU cumulative function L =a+Rh, 110) where L = the total length; and R = the otolith radius and rep- resents the independent variable (assumed to be measured with out error); 2 ) functional regression ( model II ), which has the identical formula as Equation 10 but does not assume an independent variable (i.e. both L and R are measured with error); 3) WeibuU cumulative function (WeibuU, 1951), / \ f^l'l 1-exp \ a 1 K and 4) a third order quadratic equation L =c + {d^R^) + (e.yR;-) + (f^R/) (11) (12) These four functions were fitted with the SAS NLIN pro- cedure (SAS, 1988). Ten combinations of the back-calculation formula and OR-TL fitting procedures were used (Table 1). Methods 1 through 9 were simple derivations from a standard regres- sion equation and required only a fitting of the parameters and substitution into the equation (Bagenal, 1978). Meth- od 10 however used a derivation of the WeibuU distribution function. In this method, the parameter defining the as- ymptotic limit of the function tK) was modified by LJL as [KiLJL^)\ 1-exp (13) where L = the theoretical length of the fish according to its otolith radius as predicted by the fitted OR- TL WeibuU function; and L = the actual length at capture. If, for instance, the actual length of the fish was less than the theoretical length iLJL is less than 1 ). the parameter 2000 1600 "e .§ 1200 .c O) c OJ - 800 2 o H 400 LP . ,.;' *. . *. '. .^0^-- Re O 0 12 14 Otolith radius (mm) Figure 2 Demonstration of the individually corrected WeibuU cumu- lative distribution function. The curve is the fitted equa- tion to all data. Re = otolith radius at capture, Lc = total length at capture, and Lp = total length predicted from Re and the fitted curve. The asymptote parameter of the fitted equation (K) is multiplied by Lc/Lp to arrive at that fish's particular trajectory, denoted by the large empty circles. K was corrected downward and subsequent back-calcula- tions for that fish were inade according to its own individ- ual trajectory (Fig. 2). In this way, LJL was calculated for each individual fish in the same way that the slope of the Fraser-Lee back-calculation equation was estimated for each fish. These ten combinations were used for all avail- able annuli and then repeated by using the last annulus only, for a total of twenty different methods. As a measure of bias, the back-calculated length at age 2 was regressed on the age of the fish from which the estimate came (source age). In this way. for instance, a strong "Lee's phenomenon" (the phenomenon that back- calculated lengths for a given age group become smaller as Schirripa: An evaluation of back calculation methodology using simulated otolith data 793 Table 2 Age (yr) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 n Age and a ilculated true mean lei gth-at- age inim 1 for a t> pical simulated populati on. 1 102 — — — — — — — — — — — — — 250 2 108 208 — — — — — — — — — — — — 250 ■.i 108 208 330 — — — — — — — — — — — 250 4 108 208 329 457 — — — — — — — — — — 250 5 107 207 329 457 573 — — — — — — — — — 250 6 107 207 329 457 572 674 — — — — — — — — 250 7 107 207 329 457 572 674 763 — — — — — — — 250 8 108 207 329 457 574 676 766 845 — — — — — — 250 9 108 208 330 459 576 679 769 849 917 — — — — — 250 10 108 208 330 460 578 681 772 853 921 980 — — — — 250 1! 108 208 331 460 577 680 771 852 920 981 1032 — — — 250 \2 108 208 331 460 577 680 771 851 920 980 1031 1077 — — 250 13 109 208 331 460 578 681 772 853 920 980 1031 1076 1115 — 250 14 109 208 331 459 577 678 769 849 916 974 1025 1070 1108 1139 250 Mean 104 208 330 458 575 677 768 850 918 978 1029 1073 1110 1138 Age and ca Iculated true mean length-at- age (mm ) for a typical sin ulated catch. 1 102 — — — — — — — — — — — — — 117 2 109 207 — — — — — — — — — — — — 105 3 108 207 323 — — — — — — — — — — — 101 4 108 208 323 448 — — — — — — — — — — 86 5 108 205 321 449 572 — — — — — — — — — 93 6 107 205 319 445 567 673 — — — — — — — — 95 7 108 206 323 448 569 676 765 — — — — — — — 98 8 107 206 322 450 572 680 771 853 — — — — — — 104 9 108 207 323 449 571 679 771 853 918 — — — — — 115 10 108 205 318 443 562 665 755 834 901 962 — — — — 105 11 109 206 320 442 561 667 757 836 903 964 1016 — — — 106 12 107 204 316 438 557 661 751 831 901 964 1017 1060 — — 102 13 107 204 315 439 559 664 756 838 908 973 1027 1071 1107 — 101 14 108 203 316 442 563 670 763 844 910 970 1021 1061 1096 1127 88 Mean 106 206 321 446 567 672 763 843 908 966 1020 1064 1102 1127 the fish from which they are calculated become older) or a similar effect would result in a negative slope. If there is no bias caused by this approach, the expected value of this slope is zero when randomly sampled from an unfished population. The accuracy of each of the twenty methods of back-cal- culation was evaluated by plotting the percent error of the estimated length-at-age in relation to the true value. As an overall evaluation of the method, a sum-of-squares (SS) was calculated by squaring the percent error between the estimated length-at-age and the true length-at-age and summing across all ages. tion of the age used in the back-calculation. This lack of trend, and the high degree of similarity between the mean length-at-age of the population and catch suggested that the catch was a random and representative sample of the population. Methods 2, 4, 6, 8, and 9 resulted in the least bias and method 1 the most bias when the slopes were examined across the various shapes of the OR-TL relation (Fig. 3). The linear shaped OR-TL (OR-TL/LIN) relation resulted in the least amount of bias, and the exponential shaped OR-TL (OR-TL/EXP) relation resulted in the most when the various relations were examined across methods. Results The true underljdng mean length-at-age of both the sur- viving population and the catch (Table 2) was calculated and tabulated in standard back-calculation type tables. There was no apparent trend in the estimates as a func- Sigmoid-shaped OR-TL relation Of the four functions fitted to the OR-TL/SIG relation (Table 3), the Weibull cumulative function resulted in the highest coefficient of determination (r'-=0.914); however the coefficient of determination of the quadratic fit was very similar (r^=0.913). 794 Fishery Bulletin 100(4) Table 3 Summary of the combination of OR-TL and back-calculation models that for each of the four shapes of OR-TL examined resulted in the best lack-of-fit (lowest sum of squares [SS]) OR-TL shape Best lack-of-fit result OR-TL model Back-calculation model Sigmoid All annuli Last annulus only linear regression, model I none log,(L„) = log^(L, ) -t- 6(log,,(/f„ ) - log.,(H.,)) Linear All annuli Last annulus only linear regression, model 11 linear regression, model II L„ =a -KR„/fl ML -a) L„ =a + {RJR_)(.L^-a) Exponential All annuli Last annulus only quadratic equation quadratic equation L,, =c + idR„ ) + (efl/) -1- (/7?„3) Asymptotic All annuli Last annulus only Weibull cumulative function Weibull cumulative function L„ = (A'(L, /L^))(l -exp(-(fl„/a)'')) L„ =(A'iL_ /LpDd -exp(-(i?„/a)'')l 40 r 30 20 o 10 5 6 7 Method Figure 3 Scatterplot of the slopes of the regression of back-calcu- lated length at age 2 versus age at capture for the ten back- calculation methods that used all annuli. The numbers plotted indicate the shape of the OR-TL relation examined 1 1=0R-TL/SIG [sigmoid], 2=0R-TL/LIN [linear], 3=0R-TL/ EXP [exponential], 4=0R-TL/ASYM [asymptotic]). The sigmoid shape of the OR-TL relation was evident in the shape of the percent error plots for methods 1 through 10 (Fig. 4). Wlien the OR-TL relation was sigmoid-shaped and all annuli were used, the least error resulted from em- ploying a ordinary least-squares regression coupled with the log-transformed Fraser-Lee back-calculation equation (method 7, SS=0.4913). The greatest error appeared when using the direct proportion equation (Fig. 4, method 1, SS=1.4016). Using the y-intercept of the OR-TL relation in the back-calculation equation (methods 3/13 and 5/14 Figure 4 Error in mean length at age estimates using all annuli and the ten back-calculation methods outlined in Table 1 for the sigmoid-shaped OR-TL relation. SS = sum of squares. in Table 1 ) had little effect on the total sum of squares when comparing method 2 with method 3; in addition, correcting for different limits of the Weibull function in methods 8 versus 10 had little effect. However, the log transformation of methods 6 and 7 reduced the sum of squares considerably. Schirripa: An evaluation of back calculation methodology using simulated otolith data 795 Figure 5 Error in mean length at age estimates using the last annu- lus only and the ten back-calculation methods outlined in Table 1 for the sigmoid-shaped OR-TL relation. SS = sum of squares. 0 75 ■ 0 - -0 r> - 0 7S ■ 0- f--*-^ SS = 0 0047 SS = 0 0001 SS = 0 0063 <»iii-t. «■<■>. I SS = 0 0540 9 10 11 12 13 1 -0 75 0 75 0 -0 75 0 75 0- -0 75 0 75 0 ■ -0 75 0 75 0 -0 75 Age • > — • — » ■ Method 6 SS = 0 1296 SS = 0 0873 Method 10 12 3 4 5 6 7 9 10 11 12 13 14 Figure 6 Error in mean length at age estimates using all annuli and the ten back-calculation methods outlined in Table 1 for the linear-shaped OR-TL relation. SS = sum of squares. The sigmoid shape of the OR-TL relation was not as evident in the shape of the percent error plots for methods 11 through 20 (Fig. 5). When only the last annulus was used, the least error resulted from employing a direct pro- portionality back-calculation equation (Fig. 5, method 11, SS=0.0951), and the greatest error from using direct sub- stitution into the OLS regression equation (Fig. 5, method 12, SS=1.3938). When only the last annulus was used with comparable back-calculation equations, as in methods 12 versus 13 and 18 versus 20, both the sum of squares and bias were reduced considerably. Linear-shaped OR-TL relation Of the four functions fitted to OR-TL/LIN, the ordinary least squares and functional linear regressions resulted in the highest coefficient of determination value (r'=0.916). The curvature of the Weibull and quadratic fits showed that the relation deviated slightly from a straight line. There was a high degree of similarity between the percent error plots for all twenty methods (Figs. 6 and 7), suggest- ing that the estimation of length-at-age is not as sensitive to the method of back-calculation when the OR-TL relation is linear as when it is curved. When the OR-TL relation was linear, the least error resulted from employing a functional regression coupled with the Fraser-Lee back-calculation equation when all annuli were used (Fig. 6, method 5, SS=0.0001) and when only the last annulus was used (Fig. 7, method 15, SS=0.0013). The greatest error resulted from direct substitution into the OLS regression following a natural log transformation of all parameters, both when all -0 75 0 75 -0 75 0 75 -0 75 0 75 -0 75 0 75 Method 11 SS = 0 0031 Method 13 ■0.75 ' 0 75 -0 75 0 75 T -0 75 0 75 -0 75 0 75 SS = 0 0564 SS = 0 0398 Method 18 -« — • — •— * — ^— • ■ SS = 0 0054 Age Figure 7 Error in mean length at age estimates using the last annu- lus only and the ten back-calculation methods outlined in Table 1 for the linear-shaped OR-TL relation. SS = sum of squares. annuli were used (Fig. 6, method 6, SS=0.1296) and when only the last annulus (Fig. 7, method 16. SS=0.1245). 796 Fishery Bulletin 100(4) 9 10 11 12 13 1 Figure 8 Error in mean length at age estimates using all annuli and the ten back-calculation methods outlined in Table 1 for the exponentially shaped OR-TL relation. SS = sum of squares. Figure 9 EiTor in mean length at age estimates using the last annu- lus only and the ten back-calculation methods outlined m Table 1 for the exponentially shaped OR-TL relation. SS = sum of squares. Exponentially shaped OR-TL relation Of the four functions fitted to OR-TL/EXP, the quadratic function resuUed in the highest coefficient of determina- tion (/•'-=0.883); however the coefficient of determination of the Weibull function fit was nearly as high (/■2=0.878). The percent errors, when using all annuli and linear regres- sion, followed a pattern similar to the residuals of the OR- TL relation (Fig. 8). This trend was also evident, although not as strong, when only the last annulus was used ( Fig. 9). Using the quadratic function rather than the linear regression to fit the OR-TL relation did the most at remov- ing this bias (Fig. 8 method 9, and Fig. 9 method 19). When the OR-TL relation was exponentially shaped and all annuli were used, the least error resulted from direct substitution into the fitted quadratic equation (Fig. 8, method 9, SS=0.0580), and the greatest error from us- ing the direct proportionality equation (Fig. 8, method 1, SS=L6711). When only the last annulus was used, the least error resulted from direct substitution into the fitted quadratic equation (Fig. 9, method 19, SS=0.0662), and the greatest error from using direct substitution into the OLS regression equation (Fig. 9, method 12, SS=1.5882). Asymptotically shaped OR-TL relation Of the four functions fitted to OR-TL/ ASYM, the quadratic equation resulted in the highest coefficient of determina- tion (r-=0.963); however the coefficient of determination of the Weibull function fit was nearly as high (r-=0.958). As with the exponentially shaped OR-TL relation, when linear regression was used to model the OR-TL relation, the percent error by age followed the trend of residuals for the residuals for the regression (Fig. 10). Using the last annulus only resulted in generally lower sums-of-squares, especially when the y-intercept was corrected for log transformation of the OR-TL relation used (Fig. 11). When the OR-TL relation was asymptotically shaped and all annuli were used, the least error resulted from using the individually corrected Weibull cumulative dis- tribution function (Fig. 10, method 10, SS=0.7388), and the greatest error from using direct substitution in to the OLS regression equation (Fig. 10, method 2, SS=1.9319). Wlien only the last annulus was used, the least error again resulted from using the individually corrected Weibull cumulative distribution function (Fig. 11, method 20, SS=0.0516), and the greatest error from using direct substitution in to the OLS regression equation (Fig. 11, method 12, SS=1.9261). Discussion The most accurate estimates of length-at-age resulted from the best model fits of the OR-TL relation. Even though sampling was random, poorly fitted OR-TL regres- sions resulted in back-calculation tables with obvious "Lee's phenomenon" effects. Ricker ( 1969) pointed out that the use of an incorrect otolith radius-total length relation- ship can result in this effect. Smale and Taylor (1987) also Schirripa: An evaluation of back calculation methodology using simulated otolith data 797 Figure 10 Error in mean length at age estimates using all annuli and the ten back-calculation methods outlined in Table 1 for the asymptotically shaped OR-TL relation. SS = sum of squares. Figure 11 Error in mean length at age estimates by using the last annulus only and the ten back-calculation methods outlined in Table 1 for the asymptotically shaped OR-TL relation. SS = sum of squares. showed that using the improper back-calculation method can result in a false "Lee's phenomenon" effect. Using only the last annulus reduced this effect with some back- calculation methods in this study, but not all of them. In general, the accuracy of the estimated length-at-age was directly related to how well the particular model fitted the OR-TLrelation, suggesting that the OR-TL model is just as, if not more, important as selecting the appropriate back-calculation model. Based on the importance of the fit of the OR-TL model, it follows that the methods used to sample the catch are of equal importance. Nonrandom samples of the catch, or length-based regulations that cause the catch to misrepre- sent the population, will affect the OR-TL regression. For instance, a minimum legal size will artificially truncate the OR-TL relation in samples of the catch and selectively sample faster-growing small fish. This could eliminate the youngest ages from the regression and could necessitate extrapolation of the regression beyond the range of the data. Furthermore, a truncation of the OR-TL regression would positively bias they-intercept and lead to an overes- timation of length-at-age, especially for the younger ages. It has been pointed out that univariate statistical models, which assume independence of observations, are generally inappropriate for analysis of otolith increment data (Chambers and Miller, 1995). These authors have suggested that because otolith data constitute multiple measures, perhaps examination of the covariance is more appropriate than the comparison of individual means. In this study, however, I did not seek to emphasize the ex- istence of (or lack of) a statistical difference between the true and estimated means sizes. Given the large sample sizes made available through simulation, conclusions of significant differences resulting from any statistical tests can be misleading. More useful, I believe, is the shape, direction, and magnitude of the biases that emerged from each back-calculation method. Consequently, I chose to emphasize the percent error between the true and esti- mated mean size-at-age. Using percent error allows more freedom of interpretation and is not subject to the prob- lems associated with excessively large degrees of freedom of simulated data sets. The individually corrected Weibull cumulative distribu- tion function presented here proved to be very flexible and capable of accounting for the individual otolith radius- total length trajectories. This function is very similar to the linear y-intercept corrected back-calculation equation of Fraser-Lee but can accommodate a wide varieties of curvatures. The Weibull equation I reported (Eq. 13) has an origin at x and y of 0: however, a y-intercept term can easily be added to accommodate an OR-TL relation with a nonzero intercept. Much of the cohort's diversity in biological attributes was lost within the first few months of the life because of mortalities. By the time the cohort had completed one year of growth, the diversity in biological attributes of the indi- viduals that would ultimately represent the cohort were established. Based on the observations of Secor and Houde (199.5), the establishment of the biological attributes of a 798 Fishery Bulletin 100(4) cohort occurring after one year is a realistic representation of what occurs in the early life history of striped bass. Al- though the number of fish that a cell represented could be less than one, this number was used as a relative weight- ing to all other cells; thus, proportionally, the value was valid. Consequently, the actual starting number of fish of the cohort was irrelevant for this study but could be used to calibrate the model to a particular population of interest. I was able to simulate a number of dissimilarly shaped OR-TL relationships by modifying the parameter used to convert brain weight to otolith weight from a constant to a function. However, trial runs showed that when this param- eter was held constant, the resulting OR-TL relation was not linear I later determined that because both soma and otolith growth were modeled as functions of weight and be- cause the exponent of the weight-length equation (0.31) did not exactly equal the exponent of the equation that calcu- lates the radius of a sphere (0.333), two rates must at some point diverge from linearity. For the purposes of this study, it was not necessary that the equations precisely depict the actual bioenergetics processes, only that true length of fish at annulus formation be known with certainty This study yields several conclusions important to studies of growth estimates from otolith back-calculations. The best back-calculation technique was directly related to how well the OR-TL model fitted. The percent error of any given meth- od was rarely consistent across ages, although estimates of older ages were more accurate than those of younger ones. Younger ages were generally best estimated by using direct proportionality on the last annulus only Thus, it may be necessary to use multiple methods to accurately estimate a growth curve. However, it would be difficult to select which combination of methods would be most accurate without prior knowledge of the tioie length-at-age. Acknowledgments I would like to acknowledge the following people for their contributions to this work: J. A. Bohsack, C. P. Goodyear, P. Johnson, S. Quackenbush, and J. C. Trexler And I would like to thank the manuscript reviewers for their editorial comments. Literature cited Bagenal, T 1978. Methods for assessment of fish production in fresh waters. IBP (International Biological Programme) Hand- book 3, 3rd ed., 365 p. Blackwell Sci. Publ., Oxford. Brandt, S. B., and J. Kirsch. 1993. Spatially explicit models of striped bass growth potential in Chesapeake Bay. Trans. Am. Fish. Soc. 122: 84.5-869. Chambers, R. C. and T. J. Miller 1995. Evaluating fish growth by means if otolith increment analysis: special properties if individual-level longitudinal data. In Recent developments in fish otolith research (D. H. Secor, J. M. Dean, S. E. Campana, eds.), p. 155- 175. Univ. South Carolina Press, Columbia, SC. Cohen, J. E., S. W. Christensen, and C. P. Goodyear 1983. A stochastic age-structured population model of striped bass [Morone saxatilis) in the Potomac River Can. J. Fish. Aquat. Sci. 40:2170-2183. Coutant, C. C, and D. L. Benson. 1990. Summer habitat suitability for striped bass in Chesa- peake Bay: reflections on a population decline. Trans. Am. Fish. Soc. 119:757-778. Coutant, C. C, K. L. Zachman, D. K. Cox, and B. L. Pearman. 1984. Temperature selection by juvenile striped bass in laboratory and field. Trans. Am. Fish. Soc. 1 13:666-671. Francis, R. I. C. C. 1990. Back-calculation of fish lengths: a critical review. J. Fish Biol. 36:883-902, Goodyear, C. P 1984. Analysis of potential yield per recruit for striped bass produced in Chesapeake Bay. North Am. J. Fish. Manage. 4:488-496. 1985. Relation between reported commercial landings and abundance of young striped bass in Chesapeake Bay, Mary- land. Trans. Am. Fish. Soc. 114:92-96. 1993. Spawning stock biomass per recruit in fisheries man- agement: foundation and current use. In Risk evaluation and biological reference points for fisheries management iS. J. Smith, J. J. Hunt, and D. Rivard, eds. ), p. 76-81. Can. Spec. Publ. Fish. Aqat. Sci. 120. Gulland, J. A., and L. K. Borerema. 1973. Scientific advice on catch levels. Fish. Bull, 71:325- 335. Grimes, C. B. 1978. Age, growth, and length-weight relationship of ver- milion snapper, /?/iom6op/!7t'sa(/ron(6e?Ks, from North Caro- lina and South Carolina. Fish. Bull. 78:137-146. Hales, L. S,, Jr. and K. W. Able. 1995. Effects of oxygen concentration on somatic and oto- lith growth rates of juvenile black sea bass, Centropristis striata. In Recent developments in fish otolith research ( D. H. Secor J. M. Dean, and S. E. Campana eds. ), p. 135-153. Univ. South Carolina Press. Columbia, SC. Heidinger R. C. and K. Clodfelter 1987. Validity of the otolith for determining age and growth of walleye, striped bass, and smallmouth bass in power plant cooling ponds. In Age and growth of fish (R. C. Summerfelt and G. E. Hall, eds.), p. 241-251. Iowa State University Press, Ames, lA. Laws, L. A., and J. W. Archie. 1981. Appropriate use of regression analysis in marine biology. Mar Biol. 65:13-16. Milicich, M. J., and J.H. Choat. 1992. Do otoliths record changes in somatic growth rates? Conflicting evidence from a laboratory and field study of a temperate reef fish, Parika scaber. Aust. J. Mar Freshwa- ter Res. 43:1203-1214, Mosegaard, H., H. Svedang, and K. Taberman. 1988. Uncoupling of somatic and otolith growth rates in Arctic char (Salvelinus alpmus) as an effect of difference in temperature response. Can. J. Fish. Aquat Sci. 45:1514— 1524. Reznick D., E. Lindbeck, and H. Bryga. 1989. Slower growth results in larger otoliths: an experi- mental test with guppies iPoecilia reticulata). Can. J. Fi.sh, Aquat. Sci. 46:108-112. Ricker, W. E. 1969. Effects of size-selective mortality and sampling bias on estimates of growth, mortality, production and yield. J. Fish. Res. Board Can. 26:479-541. Schirnpa An evaluation of back calculation methodology using simulated otolith data 799 1973. Linear recessions in fishery research. J. Fish. Kes. Board Canada .■!():;M.i-409. 1975. Computation and interpretation of biological statistics offish populations. Bull. Kish. Res. Board Can. 191, :i82 p. 1992. Back-calculation offish lengths based on proportion- ality between scale and length increments. Can. J. Fish. Aquat.Sci. 49:1018-1026. Rose K. A., S. W. Christensen, and D. L. DcAngelis. 1993. Individual-based modeling for populations with high mortality: a new method based on following a fixed number of model individuals. Ecol. Modeling 68:273-292. Rose K. A., and J. C. Cowan. 1993. Individual-based model of young-of-the-year striped bass population dynamics. I. Model description and base- line simulations. Trans. Am. Fish. Soc. 122:415-438. Rutherford, E. S., and E. D. Houde. 1995. The influence of temperature on cohort-specific growth, survival, and recruitment of striped bass, Moroiie saxatilis. larvae in Chesapeake Bay Fish. Bull. 93:315-332. SAS(SAS Institute, Inc.). 1988. SAS/STAT user's guide, release 6.03 ed., 1028 p. SAS Institute, Inc.. Carv, NC. Schirnpa, M. J., and C. P. Goodyear. 1997. Simulation of alternative assumptions offish otolith- somatic growth with a bioenergetics model. Ecol. Model- ing 102:209-223. Secor, D. H. 1992. Application of otolith microchemistry analysis to in- vestigate anadromy in Chesapeake Bay striped bass. Mo- rone saxatilis. Fish. Bull. 90:798-806. Secor D. H., and J. M. Dean. 1989. Somatic growth effects on the otolith — fish size rela- tionship in young pond-reared striped bass, Morone saxa- tilis. Can. J. Fish. A(|uat. Sci. 46:1 13-121. 1992. Comparison of otolith-based back-calculation meth- ods to determine individual growth histories of larval striped bass, Morone saxatilis. Can. J. Fish. Aquat. Sci. 49:1439-14,54. Secor D. H., and E. D. Houde. 1995. Temperature effects on the timing of striped bass egg production, larval viability, and recruitment potential in the Patuxent River (Chesapeake Bay). Estuaries 18(3): 527-544. Smale, M. A., and W. W. Taylor. 1987. Sources of back-calculation error in estimating growth of lake whitefish. In Age and gi-owth of fish (R. C. Sum- merfelt and G. E. Hall, eds.), p. 189-202. Iowa State Univ. Press, Ames, lA. Tuncer, H. 1988. Growth, survival, and energetics of larval and juve- nile striped bass iMorone saxatilis) and its white bass hybrid (M. saxatilis x M. chn,'sops). M.S. thesis, 136 p. Univ. Maryland, College Park, MD. Weibull, W. 1951. A statistical distribution function of wide applica- bility. J. Appl. Mechanics 18:293-297. Wright, P J , N. B. Metcalfe, and J.E. Thorpe. 1990. Otolith and somatic growth rates in Atlantic salmon parr, Salmo salar L: evidence against coupling. J. Fish. Biol. 36:241-249. Vaughan, D. S., and M. L. Burton. 1994. Estimation of von Bertalanffy growth parameters in the presence of size-selective mortality: a simulated exam- ple with red grouper. Trans. Am. Fish. Soc. 123:1-8. 800 Abstract-The bycatch of Australia's northern prawn fishery (NPF) com- prises 56 elasmobranch species ( 16 famihes). The impact of this fishery on the sustainability of these species has not been addressed. We obtained esti- mates of catch rates and the within-net survival of elasmobranchs. Carcha- rhinus tilstoni, C. dussumieri, Rhyn- chobatus djiddensis, and Himantura toshi represented 65% of the bycatch. For most species, >50'7f of individuals in the bycatch were immature, and some species recruited to the fishery at birth. For all species combined, 66% of individuals in the bycatch died in the trawl net. The relative sustainability of elasmo- branchs caught as bycatch was exam- ined by ranking species with respect to their susceptibility to capture and mortality due to prawn trawling and with respect to their capacity to recover once the population was depleted. The species that were least likely to be sus- tainable were four species of pristids, Dasyatis brevicaudata. and Himantura jenkinsii. These are bottom-associated batoids that feed on benthic organisms and are highly susceptible to capture in prawn trawls. The recovery capacity of these species was also low according to our criteria. Our results provide a valu- able first step towards ensuring the sustainability of elasmobranchs that are caught as bycatch by highlighting species for management and research. The effectiveness of turtle excluder devices (TEDs) in reducing elasmo- branch bycatch varied greatly among species but was generally not very effective because most of the captured species were small. Sustainability of elasmobranchs caught as bycatch in a tropical prawn (shrimp) trawl fishery llona C. Stobutzki Margaret J. Miller Don S. Heales David T. Brewer CSIRO Marine Research PO Box 120 Cleveland Queensland 4163, Australia E mail address ilona stobut2ki@,csiro-au Manuscript accepted 29 May 2002. Fish. Bull. 100:800-821 12002). Worldwide, there is increasing concern over the capture of elasmobranchs (sharks and rays) as bycatch. The global landings of elasmobranchs are currently 760,000 metric tons (t) but a similar amount is part of unreported bycatch (Stevens et al., 2000). This bycatch is unmanaged in most fisher- ies and elasmobranchs are less able to sustain their populations under fishing regimes designed to sustain the target teleosts or invertebrates (Heuter, 1998). Some species have declined significantly because they are captured as bycatch, e.g. the common iDipturus batis) and barndoor iD. laevis) skates (Brander, 1981; Casey and Myers, 1998). Despite these prevailing fishery practices, there have been few evalu- ations of the ability of elasmobranch species to sustain population levels (Walker and Hislop, 1998). In general, an evaluation of the sus- tainability of any bycatch species is hampered by a lack of information. This is particularly so for elasmobranchs. Elasmobranch bycatch is often not recorded (Bonfil, 1994), or when it is recorded, the species composition is un- known. There is also limited biological information on most bycatch species, such as age at maturity, growth rate, and fecundity. This lack of information hampers the use of conventional stock assessment methods to determine the population status of these species. Australia has a highly diverse elasmo- branch fauna; almost half of the species are endemic (Last and Stevens, 1994). In northern Australian waters, elas- mobranchs are impacted by a range of fisheries. Gillnet, longline, and dropline fisheries target shark species. Sharks, rays, and sawfish are also caught as bycatch in dropline and gillnet fisheries that target teleosts and in trawl fisher- ies that target teleosts or prawns. The current levels of elasmobranch bycatch are unknown for most of these fisheries. However, we know that the retained elasmobranch bycatch has increased because of the rising value of elasmo- branch products, such as fins. The largest fishery in northern Aus- tralia is the northern prawn fishery (NPF), which covers an area over 1,000,000 km- of ocean (Fig. 1) (Mc- Loughlin et al., 1997). In the NPF, elas- mobranchs contribute about 4"r of the total bycatch weight (Stobutzki et al., 2001b). Prior to 2001, NPF trawlers were allowed to retain shark products but were restricted with respect to the amount on board at any one time. Man- agement required fishermen to record retained bycatch in trawler logbooks, but the records were not validated. In 1999, 4177 kg of fillet, trunk, and whole shark and 1531 fins were recorded (Sharp et al.M. The compulsory use of turtle excluder devices (TEDs) in NPF trawls, beginning with the year 2000, have excluded some elasmobranchs from the bycatch (Brewer et al., 1998). However species-specific exclusion has not been examined. Sharp, A., J. Malcolm, and J Bishop. 2000. Northern prawn fishery and Kim- berley prawn fishery data summary 1999, Final report to Australian Fisheries Man- agement Authority, Canberra, Australia. AFMA, PO Box 7051, Canberra BC ACT 2610, Australia. Stobutzkl et a\ Sustainabillty of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 801 129° 1 134° 139°E Northern Prawn Fishery management area 229,974 1 5 "S Figure 1 The management area of the northern prawn fishery (NPFl and the bioregions defined through the interim marine and coastal rationalization (IMCR) process (Thackway and Cresswell. 1998). The shaded area represents the regions fished by commercial prawn trawlers. The dots mark the positions of the trawls that were sampled to estimate the removal rates and total biomass of bycatch species (Table 1). The numbers refer to the bioregions (l=Oceanic Shoals, 2=Tiwi, 3=Cobourg, 4=Arnhem Wessel, 5= Arafura, 6=Groote, 7=Pellew, 8=Wellesley, 9=Karumba-Nassau, 10=West Cape York, ll=Carpentaria). This study is one of several (Milton, 2001; Stobutzki et al. , 2001a) that broadly examine the sustainabillty of bycatch species groups in the NPF. The aim of this study was to assess the relative sustainabillty of elasmobranch species taken as bycatch in the NPF. We use a broadbrush method developed by Stobutzki et al (2001a) to encompass the high diversity of and limited amount of information. This semiquantitative technique assesses the sustainabil- lty of species according to two overriding characteristics: 1) their susceptibility to capture and mortality due to trawling; and 2) the ability of a population to recover after depletion. Traditional population assessment methods have attempted to measure or model these factors. The broadbrush method uses biological and ecological criteria to rank species with respect to these two characteristics, maximizing the use of the limited information available. The method identifies species that are least likely to be sustainable in the bycatch, so that these can be the focus of further research and management. Methods Species present in the NPF and those captured as bycatch A Hst of the elasmobranchs species recorded in the area of the NPF was compiled from Last and Stevens (1994). A list of species taken as NPF bycatch has been collated from two sources: 1) fishery research surveys undertaken within the NPF fishing grounds (Crocos and Coman, 1997; Stobutzki et al., 2001b; Blaber et al.^; Crocos et al.'^); and 2 ) records of elasmobranch bycatch by observers on com- mercial vessels (these observers were either scientific staff or trained crew-members) (Stobutzki, 2001b; Pender etal.^; Stobutzki etal. 5). ■ Blaber, S., D. Brewer, C. Burridge, M. Farmer, D. Milton. J. Salini, Y-G. Wang, T. Wassenberg, C. Buxton, I. Cartwright, S. Eayrs, N. Rawlinson, R. Buckworth, N. Gill, J. MacCartie. R. Mounsey, and D. Ramm. 1997. Effects of trawl design on bycatch and benthos m prawn and finfish fisheries. Final report to Fisher- ies Research and Development Corporation (FRDC), Project 93/179, 190 p. FRDC, PO Box 222, Deakm West ACT 2600, Australia. ' Crocos, P J., D. M. Smith, and G. Marsden. 1997. Factors affecting the reproductive performance of captive and wild broodstock prawns. Final report to the Fisheries Research and Development Corporation (FRDC I. Project 92/51, 87 p. FRDC, PO Box 222, Deakin West ACT 2600, Australia. Pender, P J., R. S. Willing, and D. C. Ramm. 1992. Northern prawn fishery bycatch study: distribution, abundance, size and use of bycatch from the mixed species fishery. Northern Terri- torv Department of Primary Industry and Fisheries ( NT DPIF). fishery report 26, 97 p. NT DPIF, GPO Box 3000, Darwin NT 0801, Australia. ' See next page for footnote 5. 802 Fishery Bulletin 100(4) Estimates of current bycatch rates and size frequency of species Current bycatch rates were obtained from research and observer surveys. The research surveys and gear are described in detail in Stobutzki et al. (2001b) and Stobutzki et al.^ Briefly, the research surveys sampled the nine major NPF fishing regions (1997 and 1998) to describe the bycatch. A scientific observer conducted three trips (of one-month duration) on commercial vessels in the NPF during 1996-97. A crew member of the commer- cial fleet was trained in elasmobranch identification and recorded the elasmobranch catch on commercial vessels during 1997. All elasmobranchs caught were identified, most to species, and their total number and weight were recorded. Where possible, each individual's sex, weight, and length were recorded. Total length (TL) was recorded for sharks, rhynchobatids, and pristids, and disc width (DW) was recorded for the remaining rays. Trawls during the research survey were of 0.5-h duration and a single trawl net was used. The observer data were collected from commercial trawls, 3-4 h in duration and where two nets were towed. The overall catch rate for each species was calculated from the three sources. Catch rates were corrected for duration of the trawl and the length of the headrope. The catch rates of species in each trawl were converted into catch per swept area of the trawl as the numbers of in- dividuals per square kilometer swept (no./km^). We used the trawl speed recorded during the trawls and assumed that the prawn trawls had a spread of 0.66 of the headrope length (Bishop and Sterling, 1999). Individuals of the spe- cies Carcharhinus tilstoni and C. limbatus are difficult to distinguish. Genetic studies in this region have indicated that C. limbatus is very rare (Lavery and Shaklee, 1991); therefore all specimens were recorded as C. tilstoni. Size at first maturity and fecundity Because there is limited biological information on dasy- atidids and gymnurids (Last and Stevens, 1994), we re- tained specimens from the scientific surveys to obtain preliminary estimates of size at maturity and fecundity. For females, gonad weight, diameter of the largest ovum in the ovary, and their fecundity status (whether they were pregnant of not, and whether there were in iitero embryos present) were recorded. For pregnant individuals, the number of embryos was recorded. For males, we recorded gonad weight, clasper length, and the calcification state of the clasper (uncalcified, partially calcified, or totally calci- fied). Size at sexual maturity for females was estimated as Stobutzki, I.. S. Blaber, D. Brewer, G. Frv, D. Heales, P. Jones, M. Miller, D. Milton, J. Salini, T Van der Velde, Y-G. Wang, T. Wassenberg, M. Dredge, A. Courtney, K. Chilcott, and S. Eayrs. 2000. Ecological sustainability of bycatch and biodi- versity in prawn trawl fisheries. Final report to the Fisher- ies Research and Development Corporation (FRDC), Project 96/257, .512 p. FRDC, PO Box 222. Deakin West ACT 2600, Australia. the length of the smallest pregnant individual; for males it was determined from clasper size and calcification (Bass etal.. 197.3). Within-net survival Currently there is no information on the survival rate of elasmobranchs caught as bycatch in prawn trawlers. The October 1998 research survey and crew-member observer (Table 1) recorded whether individuals were dead or alive when landed on the deck. This record provided an estimate of the within-net mortality, which was no doubt lower than the total mortality because some individuals recorded as alive would subsequently die as a result of capture. Logis- tic regressions (PROC LOGISTIC, SAS 1997) were used to determine whether there was a relationship between the likelihood of survival and the length or sex of an indi- vidual. The species were analyzed in two groups: sharks (species where TL was recorded) and rays (species where DW was recorded). Assessment of the sustainability of elasmobranch species The assessment was based on the method developed by Stobutzki et al. (2001a) which was designed to accommo- date a high diversity of and a limited amount of informa- tion. The sustainability of the species was assumed to be dependent on two overriding features; 1 ) the susceptibility of the species to capture and mortality caused by trawl- ing and 2) the capacity of the population to recover after depletion. Biological and ecological information was col- lated from the literature (Compagno, 1984a; 1984b; Last and Stevens, 1994; Froese and Pauly^). This information was used to rank the species along two axes describing the overriding features: Axis 1: The susceptibility of a species to capture and mor- tality due to a prawn trawl (susceptibility), Axis 2: The capacity of a species to recover once the popu- lation is depleted (recovery). Each feature (or axis) was derived from several criteria (listed below) that summarized aspects of the biology of the species (six criteria for axis 1 and five criteria for axis 2). Each species was given a ranking from 1 to 3 for each criterion (the definitions of the ranks for the criteria are provided in Table 2). A rank of 1 suggested that the spe- cies was highly susceptible to capture or had little capac- ity to recover; a rank of 3 suggested that the species had a low susceptibility to capture or a high capacity to recover. Depending on the criterion, these ranks were based on cat- egorical or continuous data (Table 2). Where continuous data were used, because no information was available to assign divisions between the ranks, the range of the data was divided into thirds to create the categories. •5 Froese, R., and D. Pauly, eds. 1999. Fishbase 99. URL htpp://www.fishbase.org. (Date accessed: November 1999.] Stobutzki et a\: Sustainability of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 803 Table 1 The surveys that co ntributed to the estimate of the removal rate l' ), total biomass, and within-net survival for elasmobranch spe- cies in the Northern Prawn Fishery, Australia . • indicates the surveys whose data contributed to the list of bycatch species. No. No of of nets Year Moiilli Type Gear trawls used Reference 1998*' Sep-Oct research survey Florida flyer 366 1 Stobutzki et al.""' 1997" Oct research survey Florida flyer 424 1 Stobutzki etal. (2001b) 1997*" Sep-Oct scientific observer Florida flyer 60 2 Stobutzki etal. (2001b) 1997*" Aug-Oct crew member observer Florida flyer 141 2 Stobutzki et al."' 1997*" May-Jun scientific observer Florida flyer 76 2 Stobutzki etal. (2001b) 1997*' Feb-Mar research survey Florida flyer. Engels 248 1 Stobutzki etal. (2001b) 1996*" Sep scientific observer F'lorida flyer 83 2 Stobutzki et al. (2001b) 1995' Jun research survey Florida flyer 38 1 Blaber et al.^ 1995" Oct-Nov research survey Florida flyer 39 1 Blaber et al.2 1995" Feb-Mar research survey Florida Flyer 39 1 Blaber et al.^ 1994" Nov research survey Florida flyer 7 2 Crocos and Coman ( 1997); Crocos et al.-' 1994" Jul research survey Florida flyer 7 2 Crocos and Coman 1 1997); Crocos et al.-^ 1994' Ma research survey Florida flyer 4 2 Crocos and Coman ( 1997 ); Crocos et al.-' 1994" Mar research survey Florida flyer 5 2 Crocos and Coman 1997; Crocos et al.-* 1993" Nov research survey Florida flyer 81 1 Crocos and Coman 1997; Crocos et al.'^ 1993" Oct research survey Florida Flyer 5 2 Blaber et al.^ 1993" Aug research survey Florida Flyer 9 2 Crocos and Coman ( 1997; Crocos et al. ' 1993 Jan-Feb research survey Engels, Frank and Bryce 71 1 Milton etal. (1995) 1991 Nov research survey Frank and Bryce 62 1 Milton etal. (1995) 1990 Nov-Dec research survey Frank and Bryce 128 1 Blaber et al. ( 1994); Milton et al. ( 1995) Where species-specific information was not available, a species was given the same rank as other species within its family for the criteria water column position, diet, and day and night catchability. For the other criteria, where it was not necessarily logical that family members would be similar, or where family information was not available, a rank of 1 was assigned as a precautionary approach. Axis 1 : Susceptibility of a species to capture and mortality induced by the prawn trawl There were six criteria (water column position, survival, range, day and night catchability, diet, and depth range) on axis 1. undertaken by Stobutzki et al. ^ and Stobutzki et al. (2001b). Commercial fishing is highly aggregated within the man- aged area of the fishery. The nine regions of highest effort were surveyed in 1997 (Table 1 ) and the presence or absence of each species was recorded in each region. We assumed that species with a restricted range could be impacted more heavily by trawling than those with a broader range. Day and night catchability The tiger prawn fishery is pre- dominantly a nighttime fishery (McLoughlin et al., 1997). Species with a higher catchability at night are more sus- ceptible to capture as bycatch. The relative catch rate of species during night and daytime trawling was compared during research surveys in October 1997 (Table 1). Water column position Because prawn trawls fish close to the sea floor, demersal species are more likely to be cap- tured than pelagic species. Survival This estimate was based on the survival-in- the-net data outlined previously. The possible sui^vival range of 0-100% was divided into thirds for the divisions between the ranks. Diet This criterion reflects whether the diet of the spe- cies may attract them to trawl grounds and whether they feed within the area of the water column swept by a prawn trawl. Species that feed on commercial prawns may be attracted to the commercial fishing grounds, increasing their susceptibility to capture. Species that feed on demer- sal organisms are assumed to be more susceptible to prawn trawls than species that feed higher in the water column. Range This criterion reflects the geographic spread of a species within the NPF and was determined from the research, scientific, and crew-member observer surveys Depth range Commercial trawls in the NPF are made mainly between 15 m and 40 m (Somers, 1994). An overlap between the depth range of trawling and the preferred 804 Fishery Bulletin 100(4) Table 2 The criteria used to assess 1) the relative susceptibility of bycatch species to capture and mortality due to prawn trawls and 2) their recovery capacity after depletion due to trawling. These combine to provide the ranks for the axes in Figure 2. For each criterion the definition of the three ranks is given, as well as the weighting score and the percentage of species for which species-specific information was used to rank them. Criteria Species-specific Weight information (%) Rank 1 2 3 Susceptibility Water column position 3 100 Demersal or benthic Not applicable Benthopelagic or pelagic Survival 3 18 Probability of survival <337f Probability of survival between 33'){ and 66%, inclusive Probability of survival >66% Range 2 71 Species range <3 fishery regions 3 fishery regions < species range <6 fishery regions Species range >6 fishery regions Day and night catchability 2 32 Higher catch rate at night No difference between night and day Higher catch rate at day Diet 2 55 Known to, or capable of feeding on commercial prawns or benthic organisms Not applicable Feed on pelagic organisms Depth range 1 100 Less than 60 m Not applicable Deeper than 60 m Recovery Probability of breeding 3 42 Probability of breeding before capture <50'7<: Probability of breeding before capture not significantly different from 50% Probability of breeding before capture >50% Maximum size 3 100 Maximum disc width >1755 mm Maximum total length >4781 mm 853 mm < maximum disc width <1755 mm 1861 mm < maximum total length <4781 mm Maximum disc width <853 mm Maximum total length <1861 mm Removal rate 3 79 Removal rate >66'7f 33% < removal rate <66% 33% < removal rate Annual fecundity 1 52 Annual fecundity <5 young per year 5 young per year < annual fecundity <19 young per year Annual fecundity >19 young per year Mortality mdex 1 64 mortality index >3.47 0.92 < mortality index <3.47 mortality index <0.92 depth range of species will influence their susceptibility to capture: a higher proportion of a species' population is likely to be taken if there is an overlap. Species with a broader depth range may have a spatial refuge from trawling, making them less susceptible. The depth range of species was determined from previous research surveys in the NPF and from the literature. Axis 2: The capacity of a species to recover once the population is depleted There were five criteria (probability of breeding, maximum size, removal rate, annual fecundity, mortality index I on this axis. Probability of breeding We assumed that a species is likely to have a greater capacity to recover from a de- crease in population due to trawling if most individuals are captured after they have bred. The probability that an individual of a species had bred before capture was determined from the mean length at capture in relation to the species' recorded size at maturity. The mean length at capture of a species was recorded in the research and observer surveys 1996-98 (Table 1). Size at maturity was determined from the available literature and from our estimates outhned previously. A t-test (Sokal and Rohlf 1996) was used to determine whether the mean length at capture was significantly dif- ferent from the size at maturity for each species. Stobutzki et al.: Sustainability of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 805 Table 3 The elasmobranch species that arc known to occur in the region of the northern prawn fishery (NPF), Australia, and of these spe- cies, those that have been •2. recorded in NPF bycatch (Table 11. The labt Is in parenthe.ses refer to the species abbreviations in Figure Family flecorded in bycatch Yes No Carcharhinidae Carcharhmus albimarginatus (Cal) Carch a rh in us a m blyrhyn ch oides Carcharhinus amboinensis (Cam) Carcharhinus amblyrhynchos Carcharhinus brevipinna (Cb) Carcharhinus cautus Carcharhinus dussumieri (Cd) Carcharhinus obscurus Carcharhinus fitztroyensis (Cfl Carcharhinus plumbeus Carcharhinus leucas (Cle) Carcharias taurus Carcharhinus limbatus (Ch) Carcharinus falcifurmis Carcharhinus macloti (Cm) Carcharinus melanopterus Carcharhinus sorrah (Cs) Loxodon macrorhinus Carcharhinus tilstoni (Ct) Rhizoprionodon oligolinx Galeocerdo cuvier (Go) Triaenodon obesus Negapnon acutidens (Na) Prionace glauca (Pg) Rhizoprionodon acutus (Rac) Rhizoprionodon taylori (Rta) Dasyatidae Amphotistius annotata Dasyatis brevicaudatus Dasyatis leylandi Dasyatis kuhlii Dasyatis sp. A Dasyatis thetidis Himantura fai (Aa) (Db) (Dl) (Dk) (Dsa) (Dt) (Hf) Dasyatis fluviorum Taeniura lymma Himantura granulata Himantura jenkinsii Himantura sp. A Himantura toshi Himantura uarnak Himantura undulata (Hg) (Hj) (Hsa) (Ht) (Hua) (Hun) continued Maximum size The maximum size of a species was used as an indicator of the species' relative recovery rate. In general, larger species tend to live longer and their popu- lations recover more slowly (Roberts and Hawkins, 1999). Size appears to be a good predictor of vulnerability for marine fishes (Jennings et al., 1999), and in particular skates (Walker and Hislop, 1998; Dulvy et al., 2000). Estimates of maximum size came from the literature. Species were grouped according to whether DW or TL was measured. The range of the maximum sizes of species was calculated and divided into thirds for the divisions between the ranks. Removal rate We assumed that species with a higher proportion of their biomass removed as bycatch would have a lower capacity to recover. The estimate of removal rate was based on the catch rates from research surveys and scientific observer collections undertaken between 1996 and 1998 (Table 1). We assumed that these catch rates were representative of the overall catch rates in the commercial fishery. The catch rates of bycatch species vary spatially within the NPF (Stobutzki et al., 2001b). Therefore, the fishery was stratified before we estimated the mean catch rate, using the bioregions identified in the Interim Marine and Coastal Regionalization for Australia (IMCRA) process (Thackway and Cresswell, 1998) (Fig. 1). A mean catch rate for each species was calculated for each bioregion where commercial tiger prawn trawling occurs. The biomass (in numbers of individuals per year) of by- catch removed by the commercial fishery was estimated by multiplying the mean catch rate calculated above by the 1997 commercial tiger prawn fishery effort in each biore- gion (Table 3). Commercial fishing effort is recorded in log books in boat days (held by the Australian Fisheries Man- agement Authority). One boat day was assumed to be the 806 Fishery Bulletin 100(4) Table 3 (continued) Family Recorded in bycatch Yes No Dasyatidae (continued) Pastinachus sephen (Ps) Taeniura meyeni (Tm) Urogymnus asperrimus (Ua) Ginglymostomatidae Nehriiis ferrugineus (Nf) Gymnuridae Gymnura australis (Ga) Hemigaleidae Hemigaleus microstoma (Hm) Hemiscylliuin ocellatum Hemipristis elongatus (He) Hemiscyllium tnspeculare Hemiscylliidae Chiloscyllium punctatiim (Cp) Mobulidae Manta birostris Mobula eregoodootenkee Myliobatidae Aetobatus narinari (Ana) Aetomylaeus vespertilio (Av) Aetomyleus nichofii (Ani) Narcinidae Narcine westraliensis (Nw) Narcine sp. A Orectolobidae Orectolobus ornatus (Oo) Eucrossorhinus dasypogon Orectolobus wardi Pristidae Anoxypristis cuspidata (Ac) Pristis clavata (Pc) Pristis microdon (Pm) Pristis pectinata (Pp) Pristis zijsron (Pz) Scyliorhinidae Atelomycterus fasciatus (Af) Atelomycterus macleayi Galeus sp. A (Gsa) Sphyrnidae Eusphyra blochii (Eb) Sphyrna lewini (SI) Sphyrna mokarran (Sm) Squatinidae Squatina sp. A (Ssa) Stegostomatidae Stegostoma fasciatum (Sfl Rhincodontidae Rhiniodon typus Rhinobatidae Rhinobatos typus (Rty) Aptychotrema sp. A Rhynchobatidae Rhynchobatits djiddensis (Rd) Rhma ancylostoma (Ran) equivalent of 14 hours of trawling with two nets and with 14-fathom (25.48 m) headropes at a speed of 3.2 knots (5.9 krn/h) (Bishop and Sterling. 1999). The estimate of the total amount removed for a species within the whole fishery was calculated by summing the removal estimates for the bioregions. This estimate was then converted to a proportion of the estimated total bio- mass of the species. An estimate of the total biomass of each species in the bioregions where tiger prawn trawling occurs was gener- ated from all research and scientific observer surveys conducted in the NPF during the 1990s (Fig. 1, Table 1). The gears used were prawn trawls (Florida flyer nets) and two types offish trawls (Frank and Bryce trawls and Engel trawls). Both night and daytime trawling were undertaken. Both prawn-trawl and fish-trawl surveys were analyzed in order to cover the management area of the fishery. The catch rates of species in each trawl were converted to the catch per swept area of the trawl as described previ- ously. The fish trawls were assumed to have a spread of 0.6 of the headrope length (Blaber et al., 1994). A mean catch rate for each gear at each time (day or night) was calcu- lated in each bioregion, resulting in up to six catch rate estimates for a species in a bioregion. The highest of these means was used for each species in that bioregion. This catch rate was then multiplied by the area of the bioregion to give an estimate of total numbers of individuals in the bioregion. Currently there are no robust estimates of the catchability coefficients for the various trawl gears and therefore a catchability coefficient of one was assumed for Stobulzki et al Sustainability of elasmobranrhs caught as bycatch in a tropical prawn trawl fishery 807 all species. Such a hiixh catchahility coefficient is unlikely to be valid for most s[)ecies and results in an underesti- mate of the total biomass. P^jr the two bioregions where commercial tiger prawn trawlers operate, there was no survey data from which to estimate catch rates (bioregions 4 and 5, Fig. 1). Therefore, the mean catch rate of the other bioregions was used to allow an estimate of catch rate. The total biomass of each species was calculated by summing the estimates for the bioregions. The removal rate would range between 0^< and 100%; this range was divided into thirds for the division between the ranks. s, = ^ i n 1". (2) where S^ = the total susceptibility or recovery ranks for species /; if = the weighting for criterion 7; R^ = the rank of species ( for criterion y; and n = the number of criteria on each axes. Annual fecundity The annual fecundity of species was estimated from data in the literature and the biological samples collected during our study. The annual fecundity of a species was estimated as the average number of pups per female multiplied by the number of times the females bred per year. Where the frequency of breeding was not known, it was assumed to be annual, unless the known gestation period was longer than 12 months. The range of fecundities was calculated and divided into thirds for the divisions between the ranks. Mortality index The recovery capacity of a population is likely to be related to its fishing mortality rate (Sparre and Venema, 1992). A measure of this rate can be derived from the length-frequency of a species and the von Bert- lanffy growth parameters (Sparre and Venema, 1992). However, for most species von Bertalanffy parameters were not available and therefore an index of mortality was calculated as follows: Mortalifv index = (L„ L ,)/(L„ (1) where L„,„j = the maximum length; L^^^ = the mean length at capture in the fishery; and ^nuii - ^-he smallest length caught. The closer the mean length of captured individuals (L^^,^,) to the maximum length (L^^^) the lower the mortality the population is subject to. As mortality due to fishing increases, the mean length of species in a population approaches the minimum length (L„,,„). For our analysis, we assumed constant catchahility and mortality across the whole length range caught. The L^^,^, and L„„„ were calculated from length data collected during our study. The range of mortality estimates was calculated and divided into thirds for the divisions between the ranks. Analysis of criteria Partial correlations (Sokal and Rohlf 1996) were used to determine whether there was any redundancy in the cri- teria. Strong correlations would suggest that two or more criteria explained the same factors, which would lead to overemphasis of their effect. One of the correlated criteria was, therefore, removed. The total susceptibility, or removal ranking, of a species was determined by the following equation: The criteria were weighted to reflect the relative impor- tance of each criterion in determining the overall charac- teristic and the robustness and quality of the data (Table 2), the latter in terms of the amount of species-specific in- formation and the scale of the information available. The criteria that were seen as major determinants of suscep- tibility or recovery and for which there were more robust data were weighted highest. This weighting was done in collaboration with the NPF Fishery Assessment Group. The total susceptibility and recovery ranks for the spe- cies were graphed to determine the relative sustainability of the species caught as bycatch by prawn trawlers. The species least likely to be sustainable would be identified as the species with the lowest ranks on both axes. Contour lines were drawn on the graph to group species that would be similar with respect to their sustainablity. Because neither susceptibility, nor recovery alone, provide a complete index to the sustainability of species, the in- dex is a combination of these two features. Recovery is likely to be conditionally important on susceptibility, and therefore, a multiplicative relationship between the two axes is appropriate. We assumed that this relationship is symmetrical and given this assumption, the contour lines followed the equation 16(.v - 0.75) (.V - 0.75) = 4, 9, 16, 25, 36, 49. (3) The impact of turtle excluder devices on elasmobranch bycatch Data on the size of species captured in nets fitted with TEDs and with nets with standard codends were avail- able from two sources. The crew-member observer recorded seven pairs of trawls in which one net was fitted with a TED and one had a standard codend. The TED was a Seymour TED with 110-mm bar spacing. Previous research surveys from one area of the NPF also recorded information on elas- mobranchs captured in nets with and without TEDs. The TEDs were AusTEDs, NordMore Grids, and SuperShooters; the design of these nets is detailed in Brewer et al. ( 1998). The length frequency of elasmobranchs caught in nets with TEDs was compared to the length frequency of elas- mobranchs caught in nets without a TED. First, species were grouped into sharks (TL measured) and rays (DW measured) for analysis. The mean length of individuals captured in nets fitted with a TED was compared with that of elasmobranchs caught in nets with a standard codend by using a one-way ANOVA. The lengths were 808 Fishery Bulletin 100(4) transformed (log (length + 1)) prior to analysis to normal- ize the data. There were sufficient data for three species of shark (Rhizoprionodon acutus, Hemigaleus microstoma, and Carcharhinus dussumieri), two stingrays (Dasyatis leylandi and Himantura toshi) and a shovel-nosed ray (Rhynchobatus djiddensis) to examine them separately with one-way ANOVAs. Results Species captured as bycatch in prawn trawls of the NPF At least 79 species of elasmobranchs from 18 families, inhabit the NPF region (Table 31. Of these, 56 species (16 families) have been recorded in the prawn-trawl fishery bycatch (Table 3). The Carcharhinidae and Dasyatidae, the most species-rich families in the region, are the also the families for which the highest number of species are recorded in bycatch (Table 3). There are 9 families in which all species found in this region have been recorded in bycatch (Table 3). Current catch rates In the research and observer surveys (1996 and 1998) 44 species of elasmobranchs were recorded. The highest over- all catch rates were for Carcharhinus tilstoni, C. dussumi- eri, R. djiddensis, and H. toshi (Table 4). These four species contributed almost 65*7^ of the observed elasmobranch bycatch. Carcharhinus dussumieri and C. tilstoni were recorded in 20% of all trawls, i?. dijiddensis in 14'*, and H. toshi in 17%. Size at first maturity and fecundity Specimens of five species of ray were examined to assess size at first maturity and to provide estimates of fecun- dity (Table 5). None of the species showed a change in gonadosomatic index (GSl) or diameter of the largest egg, both of which would clearly indicate maturity. The aver- age number of embryos was low (Table 5); most species had one or two, with the exception of Gymnurus austi-alis, which had up to five embryos present. Males of most species showed an increase in GSI with calcification of the claspers. The estimates of size at ma- turity for the males were lower than the estimates for females for four of the five species (Table 5). However, this finding might have been influenced by the low numbers of pregnant females sampled (Table 5). The size at maturity of the males appeared to be between 44% and 79% of the maximum size for the species. The mean size of rays caught in bycatch ranged from 182 mm for D. leylandi to 1117 mm for H. toshi (Table 6). The mean size of sharks ranged from 541 mm for Carcharhi- nus sorrah to 1643 mm for Rhina ancylostoma (Table 6). For 30 species, a size at birth was available from the lit- erature and, of these, eight species were caught in bycatch at this -size (Table 6). Where an estimate of the size at first maturity (L„^) was available for a species, an estimate could be made of the percentage of individuals captured that were mature. In species with sufficient samples sizes, the percentage of mature individuals caught ranged from <1% for S. lewini, to 54% for R. acutus (Table 6). Species such as D. leylandi had an average size at capture not significantly different from L,,,, indicating that, on average, half the individu- als caught had reached maturity before capture. Species such as R. acutus, with an average size less than L^, were those for which the majority were unlikely to have bred before capture. At the other extreme were species such as G. australis. for which it was likely that the majority had reached maturity before capture (Table 6). The female-to-male ratio of individuals caught was close to 1:1 for the two common species, D. leylandi and C. dussumieri (Table 6). However, other species had a range from predominantly male (e.g. R. acutus) to predominantly female (e.g. H. toshi) (Table 6). Within-net survival Whether an individual was alive or dead when landed on the deck was recorded for 847 animals. Overall 56%f were dead after capture in the trawl and 44% were alive. Both sharks and rays showed that the probability of survival was lower for males than for females (sharks ;^-=19.7,P<0.001, rays ;f-=10.5, P=0.0012) and that survival increased with length of the individual (sharks ;f^=4.8, P=0.029, rays ;f-'= 11.08, P=0.0009). Two-thirds of male sharks and rays were recorded as dead after capture in the trawl (Table 7). The mean size of rays and sharks that died (sharks 684 (±10 SE) mm, rays 424 (±41 SE) mm) was smaller than the mean size of those that survived (sharks 797 (±17 SE) mm, rays 546 (±33 SE) mm). The overall percentage of individu- als of a species that died varied from 10% (R. djiddensis) to 82% (C. dussumieri and R. acutus) (Table 7). Assessment of the sustainability of elasmobranch species The 56 species of elasmobranchs recorded as bycatch in the NPF were ranked on each of the criteria on the two axes (Appendices 1 and 2). The extent to which species- specific information was available varied among the cri- teria (Table 2). Water column position, depth range, and maximum size had species-specific information for all species. Survival and day and night catchability had little species-specific information. Most of the criteria were not correlated (Table 8). On the susceptibility axis the strongest correlation was between diet and water column position (Table 8). However, both criteria were retained because we believed there was suf- ficient difference between them; the correlation coefficient ir) was only 0.67. On the recovery axis no correlations were significant (Table 8). On the susceptibility axis (Appendix 1 ) the four species of Pristidae, Atelomycterus fasciatus, Himantura jenkinsi, and Stegostoma fasciatum had a rank of 1, the lowest possible rank, suggesting they were the most susceptible to capture Stobutzki et al : Sustainability of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 809 Table 4 The percentage of trawls in which species were caught, mean catch r; te (SE=standard error), and the percentage of catch contrib- uted by each species. No./kni- %of %of Family Species trawls Mean SE catch Carcharhinidae Carrharhinus albimarginatus 0.10 0.58 0.41 0.26 Carcharhiiiiis amhoinensis 0.20 0.04 0.04 0.02 Carcharhinus dussuinieri 20.57 .38.80 4.89 17.54 Carcharbinus fitztroyensis 0.20 0.80 0.40 0.35 Carcharhinus macloti 0.20 0.98 0.50 0.43 Carcharhinua sorrah 1.67 1.47 0.57 0.65 Carcharhinus tilstoni 19.49 44.20 5.98 20.07 Galeocerdo cuvier 0.20 0.01 0.00 <0.01 Negaprion acutides 0.10 0.00 0.00 <0.01 Rhizoprionodon acutus 9.15 10.61 1.63 4.83 Rhizoprionodon taylori 0.10 0.00 0.00 <0.01 Dasyatidae Amphotislius annotata 1.97 1..56 0.41 0.74 Dasyatis kuhlii 2.56 1.48 0.56 0.69 Dasyatis leylandi 15.35 9.44 1.18 4.48 Dasyatis sp. A 0.10 0.00 0.00 <0.01 Dasyatis thetidis 0.49 0.03 0.01 0.01 Himantura fai 0.10 0.01 0.00 <0.01 Himantura granulata 0.20 0.01 0.01 <0.01 Himantura jenkinsii 0.59 2.11 0.77 0.95 Himantura sp. A 2.17 0.11 0.04 0.05 Himantura toshi 17.72 27.85 3.10 12.84 Himantura uarnak 0.98 1.44 0.58 0.70 Himantura undulata 0.89 0.96 0.40 0.43 Pastinachus sephen 3.44 0.69 0.31 0.31 Taeniura meyeni 0.10 0.40 0.28 0.18 Urogymnus asperrimus 0.39 0.40 0.28 0.18 Ginglymostomatidae Nebnus ferrugineus 0.10 0.58 0.41 0.26 Gymnuridae Gymnura australis 5.91 8.02 1.64 3.82 Hemiscylliidae Chiloscyllium punctatum 5.41 11.83 1.96 5.42 Hemigaleus microstoma 9.84 9.64 1.55 4.56 Hemipristis elongatus 0.20 0.02 0.02 0.01 Myhobatidae Aetobatus narinari 0.30 0.60 0.41 0.27 Aetomylaeus nichofii 1.08 1.57 0.61 0.74 Orectolobidae Orectolobus ornatus 0.10 0.52 0.52 0.27 Pristidae Anoxypristis cuspidata 0.98 0.71 0.42 0.32 Pristis zijsron 0.10 0.02 0.02 0.01 Rhinobatidae Rhinobatos typus 0.39 0.02 0.01 0.01 Rhynchobatidae Rhina ancylostoma 0.89 0.10 0.04 0.05 Rhynchubatus djiddensis 14.27 30.87 3.39 14.26 Scyliorhinidae Atelomycterus fasciatus 0.49 0.18 0.08 0.08 Sphyrnidae Eusphyra blochii 0.20 0.04 0.04 0.02 Sphyrna leuini 2.95 6.91 1.52 3.07 Sphyrna mokarran 0.39 0.02 0.02 0.01 Stegostomatidae Stegostoma fasciatum 2.17 2.17 1.02 1.10 810 Fishery Bulletin 100(4) Table 5 The estimated size (disc SE=standard error). width) at first maturity and mean number o f pups for ray species (sample size is shown in parentheses; Species Size at maturity ( mm) Nu mber of pups M ale Female Mean SE Amphotistius annotata 200 (9) 233 (8) 1.5 0.7 (2) Dasyatis kuhlii 300 (10) 378 (6) 2 — (1) Dasyatis leylandi 185 (103) 180 (110) 1.1 0.3 (17) Gymnura australis 350 (29) 610 (16) 3.2 1.2 (6) Himantura toshi 400 (31) 660 (21) 1.5 0.7 (2) and mortality. The next 19 species had a rank of 1.15, also low. Carcharhinus tilstoni, C. macloti, Sphyrna lewini, Prio- nace glauca. C. brevipinna. and Aetomyli'iis nichofii had the highest ranks on this axis (>1.92l, indicating that they were the least susceptible to capture and mortality. On the recovery axis (Appendix 2) Aetomyleus vespertil- io, Dasyatis brevicaudatus, Pristis clavate, and P. pectinata had the lowest ranks, indicating that they had the lowest capacity to recover Gymnura australis. H. toshi. Hemigale- us microstoma, and R. taylori had the highest ranks on this axis and therefore the highest capacity to recover. When the ranks of the species on the two axes were plot- ted (Fig. 2), Dasyatis brevicaudatus, P. pectina. P. da rata. P. microdon, P. zijsron, and Himantura jenkinsii ranked the lowest on the combination of the two axes, indicating that they were the least likely to be able to survive cap- ture as bycatch. The species Eusphyr-na blochii, H. toshi, C. macloti, and C. tilstoni ranked the highest on the two axes, indicating that they were the most likely to be able to survive capture as bycatch. The impact of turtle exclusion devices on elasmobranch bycatch Both sharks and rays taken as bycatch were significantly smaller in nets with a codend fitted with a TED (Table 9). The length frequency of the sharks and rays caught in the nets with TEDs showed a lower proportion of the larger individuals (Fig. 3). Where individual species were exam- ined, there was a decrease in the size of C. dussumieri and R. djiddensis caught in the net with a TED (Table 9). There was no significant difference in size for H. microstoma, A. annotata, and H. toshi (Table 9). However, significantly larger individuals o{ Rhizoprionodon acutus were caught in the net with a TED (Table 9). Discussion ing the impact of trawling in this region on elasmobranchs are the catch rates and survival of species. Current catch rates Although the bycatch was highly diverse, four species dominated the catch of the present study (C tilstoni, C. dussumieri, R. djiddensis. and H. toshi. Table 4), occurring in 14—20'^^ of trawls, so that one individual was seen at least every seven trawls. However, most species (75%) con- tributed <1% of the catch and had low catch rates (Table 4). However, even low catch rates can result in a large overall take of individuals. The fishery recorded 18.314 days of fishing in 1999 (Sharp et al.M and if each day con- sisted of four trawls (Bishop and Sterling, 1999), 73,256 trawls (with two nets) would have been undertaken in the year. Hence for a species occurring in 1% of trawls. 733 individuals would have been caught in the year There are no long-term catch data available that can be examined for changes in catch rates of elasmobranch species. Although shark byproducts are recorded in NPF logbooks, the data are of limited value because they are not validated and not species-specific. Pender et al.^ surveyed the bycatch in Northern Territory waters of the NPF during the 1980s. Rhynchobatids (71% of the elasmobranch catch), carcharhinids {\2'7<) and dasyatids (IKO dominated the catch (Pender et al.^). All species recorded by Pender et al.^ were recorded in our study. Direct comparisons of the catch rates of Pender et al.'* with those of our study were not pos- sible because of differences in gear, season, and region. Most elasmobranchs caught in bycatch are small (<1000 mm). For some species, this means that most individuals have not bred before capture (Table 6) and therefore the fishery will have a greater adverse impact on the species. At least eight species were caught at sizes close to their known birth size (Table 6). This finding suggests that pup- ping may occur in the area of the fishery. Whether these species have restricted pupping grounds is unknown. Of the elasmobranch species known to inhabit this region. 71% were taken as bycatch in the NPF The highly diverse bycatch is characteristic of tropical prawn trawl fisheries (Hall, 1999). Two critical pieces of information for assess- Within-net survival Our estimates of within-net survival are the first for elas- mobranchs in prawn trawls. The results suggest that most Stobutzkl et al : Sustainabillty of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 811 Table 6 Tho moan, minimum (nun), and maxunum (max) size (TL arUW ) of elasmobranch species caught in nighttime prawn trawling. The size at maturity (L,,,) and at birth (pup size) are shown based on Last and Stevens ( 1994) or Table 5. The percentage of individuals caufiht that were ma ure C^f mature) and the sex ratio are also shown SE = s tandard error ; n = sample size; P is the prob ibility that the mean length at capture is different from L^. Size (mm) Sex ratio % Pup size Family Species Mean SE Min Max n F:M n K P mature (mm) Carcharhinidae Carcharhinus alhimarginatus 850 — — — 1 — — 1700 — 100 550 Carcharhinus 1700 — — — 1 — — 2100 — 0 600 amboinensis Carcharhinus dussumieri 6,36 6 270 850 377 1.08 139 700 <0.001 41 350 Carcharhinus fitzroyensis 1045 225 820 1270 2 — — 800 >0.5 50 500 Carcharhinus macloti 745 75 670 820 2 — — 690 >0.5 20 450 Carcharhinus sorrah 542 43 300 950 25 3.03 16 900 <0.001 8 500 Carcharhinus tilstoni 794.3 9 100 19,50 344 0.95 84 1200 <0.001 0.6 600 Galeocerdo cuvier 1175 285 890 1460 2 allM 1 3000 >0.1 0 500 Negaprion acutidens 2600 — — — 1 — — 2200 — — 500 Rhizoprionodon acutus 689 14 280 960 140 0.56 81 750 <0.001 54 — Rhizoprwnodon taylon 546 — — — 1 allF 1 400 — 100 2.50 Dasyatidae Amphotistius annotata 211.4 12 140 452 25 1.43 3 200 >0.5 24 — Dasyatis kuhlii 297.3 12 190 400 24 4.00 10 300 >0.5 8 160 Dasyatis leylandi 182.2 3 110 400 206 1.05 162 180 >0.2 46 110 Dasyatis sp. A 350 — — — 1 — — 360 — 0 — Dasyatis thetidis 1162 129 800 1420 5 allM 1 — — — 350 Himantura fai 1900 — — — 1 — — — — — 550 Himantura granulata 960 — — — 1 — — — — — 280 Himantura jenkinsii 890 150 300 1140 5 allM 1 — — — — Himantura sp. A 3.50 65 80 1800 57 allF 12 — — — — Himantura toshi 456 11 150 1330 235 4.17 52 400 <0.001 12 200 Himantura uarnak 1055 132 290 1600 12 allF 2 — — — 280 Himantura undulata 1117 131 400 1.500 7 allF 1 — — — 200 Pastinachus sephen 1076 53 450 2000 43 3.03 12 — — — 180 Taeniura meyeni 1300 — — — 1 — — — — — 350 Urogymnus asperrimus 850 106 530 1150 5 allM 2 — — — — Ginglymostomatidae Nebrius ferrugmeus 2400 — — — 1 — — 2250 — 100 400 Gymnuridae Gymnura australis 462 19 120 860 87 2.00 42 350 <0.001 24 — Hemigaleidae Hemigaleus microstoma 609 19 250 950 152 0.68 91 600 >0.5 47 300 Hemipristis elongata 1340 190 1150 1.530 2 allF 1 1100 >0.5 50 520 Hemiscylliidae Chiloscyllium punctatum 668 23 2.30 1000 63 2.50 7 700 <0.001 52 170 Myliobatidae Aetobatus narinan 625 125 500 750 2 allF 1 — — — 260 Aetomylaeus nichofii 437 42 240 720 11 allF 3 — — — 170 Pristidae Anoxypristis cuspidata 1930 193 1240 2550 8 allF 1 — — — — Rhinobatidae Rhinobatos typus 1953 188 1500 2340 4 — — — — — — Rhynchobatidae Rhina ancylostoma 1643 112 1010 2090 9 0.33 3 — — — — Rhynchobatus djiddensis 869 36 230 2650 187 4.76 35 1100 <0.001 8 — Scyliorhinidae Atetomycterus fasciatus 300 0 300 300 2 — — 320 — 0 — Sphyrnidae Eusphyra btochii 990 320 670 1310 2 — — 1080 >0.5 50 450 Sphyrna lewini 832 54 400 2400 37 allF 3 1400 <0.001 3 450 Sphyrna mokarran 1780 457 400 2400 3 1.00 2 2100 >0.5 33 650 Stegostomatidae Stegostoma fasciatum 1305 82 400 2000 26 0.80 3 1700 <0.001 23 200 812 Fishery Bulletin 100(4) Table 7 The percentage of elasmobranchs that died within the trawl net, recorded on research and crew -member observer surveys. "Com- bined sharks" refers to all species where total length was recorded; 'com bined rays" refers to all species where disc width was | recorded; n = number of specimens measured. Family Taxa % dead Female n Male 'I Overall n combined sharks 23 149 66 59 61 639 combined rays 56 360 67 279 40 208 Carcharhinidae Ca rch a rhin u s d u ssii m ieri 48 207 58 114 52 321 Carcharhinus sorrah 73 15 50 8 65 23 Carcharhmus tilstoni 78 40 85 33 82 73 Rhizuprionudon acutus 75 44 86 72 82 116 Dasyatidae Dasyatis leylandi 27 22 95 19 59 41 Himantura toshi 43 40 78 18 53 58 Gymnuridae Gymnura australis 31 26 75 8 41 34 Hemigaleidae Hemigaleus microstoma 44 29 64 39 62 68 Rhynchobatidae Rhynchobatus djiddensis 21 24 20 5 10 59 Table 8 The correlations between the criteria on 1) the susceptibility axis and 2) the recovery axis. * indicates significance at P < 0.05. Day and night Susceptibility Survival Range catchability Diet Depth range Water column position 0.07 -0.18 0.13 0.67* 0.07 Survival 0.48* 0.07 -0.11 -0.00 Range 0.25 0.06 0.09 Day and night catchability -0.15 0.04 Diet 0.05 Maximum Removal Annual Mortality Recovery size rate fecundity index Probability of breeding 0.07 0.25 0.16 0.08 Maximum size -0.22 -0.25 0.17 Removal rate -0.27 0.21 Annual fecundity 0.12 sharks and rays, particularly the smaller individuals, die within the trawl net (56'^'^). The lower sui-vival rates of male individuals is possibly because the males of most elas- mobranch species are smaller than the females. The rhyn- chobatid R. djiddensis had a higher survival rate {90'^7r) than most other species, whereas the lowest survival rate was seen in C. tilstoni and R. acutus (18%). Although the larger elasmobranchs appeared to have a higher within- net survival, in the commercial fishery these were the very individuals killed for their fins and therefore their mortal- ity was ultimately higher than that predicted by their size alone. In 2001 the NPF introduced an industry-initiated ban on all shark products, so that the only mortality these species are subject to is that caused by the capture process. Differences between species in survival rates may influ- ence changes in the relative abundance of species. Assessment of the sustainability of elasmobranch species Elasmobranchs, in general, are more susceptible to over- fishing than are bony fishes, but there is likely to be a range of sensitivities among the species (Walker. 1998; Stevens et al., 2000). The process we applied in our study allowed us to examine these different sensitivities and to highlight those species whose populations were most likely to be affected by the NPF. The process was designed to deal with the high diversity of the bycatch and the Stobutzki et aL: Suslainability of elasmobranchs caught as bycatch In a tropical prawn trawl fishery 813 Low 3 High 1 3 High Figure 2 The ranking of elasniobranch species with respect to criteria that reflect their susceptibiHty to capture and mortahty due to prawn trawhng and their capacity to recover after depletion by trawling. These factors combine to reflect the rela- tive ability of species to sustain capture as prawn trawl bycatch in the northern prawn fishery and therefore their relative priority with respect to research and management. Numbers refer to species combinations that fall together on the graph. ( l=Hj, Pm, Pz; 2=Ca, Cle, Dt, Gsa, Ssa, Tm; 3=Af. Dsa, Hf Hg Hua. Oo, Rty, Ua; 4=Cf Aa; 5=Ana, Cli; 6=Rac, He). Explanations of the abbreviations for species are given in Table 3. paucity of information available for most species. Our pro- cess was similar to that used by the International Union for the Conservation of Nature and Natural Resources iIUCN) red lists (lUCN. 1995) that categorize species with respect to the threat of extinction worldwide. The lUCN uses criteria on the extent of population decrease, area of occurrence, percentage of population that is mature, and the probability of extinction (lUCN, 1995). The lUCN cri- teria have been modified for application to marine fishes and to smaller geographic scales (Musick, 1998). With respect to elasmobranchs, several authors have exam- ined the variable resilience of species to fishing pressure. These approaches have focused on life history character- istics that influence the recovery of populations, including reproductive and growth parameters ( reviewed by Stevens et al., 2000). Our process is similar to these but focuses at the level of an individual fishery, incorporating fishery- specific information on the susceptibility of species to the fishery. Of significant importance with all methods is the ability to calculate the range of parameters required for a large number of species (Stevens et al., 2000). The semi- quantitative method used in our study maximizes what can be determined from the data available and enables consistency across the species. The criteria include char- acteristics that influence the probability of extinction of a species and its sensitivity to overfishing (McKjnney, 1997: Carlton et al, 1999; Roberts and Hawkins, 1999; Stevens et al., 2000). Our analysis provides a process for highlight- ing gaps in information and for prioritizing species for future management and research. This process does not replace traditional methods of population assessment but provides a rapid assessment of the species, so that tradi- tional methods can be focused on the high-risk species. The species that were least likely to be sustainable in the bycatch of the NPF were D. brevicaudatus. P. pecti- nata, P. clavata, P. microdon, P. zijsron, and Himantura Jenkinsii (Fig. 3). The pristids and H.jenkinsii had ranks of 1 on the susceptibility axis, the lowest possible rank, and D. brevicaudatus ranked 1.15 (Appendix 1). These species are demersal, are rare in the bycatch, and at least for the pristids (which have restricted depth distributions) are likely to be rare. Nothing is known about their sur- 814 Fishery Bulletin 100(4) 12 -1 100 400 800 1200 Disc width (mm) 1600 ou B ^— — ■ ■ \^ « 20 - CO 13 \L " JIAIPUI / 1 ■ O a? 10 - / . 0 - \ ii.ii ,. .ii, .. r ■ ^ ■ 1 ■ 100 80 60 O c 3 40 <" - 20 1000 2000 3000 4000 Total length (mm) Figure 3 The length frequency and cumulative length frequency of r^ys (A) and sharks (B) caught in nets with a standard codend (black columns and solid line) and nets with a TED (grey columns and broken line). vival. Their diets include benthic organisms and are likely to include commercial prawns; their range and day and night catchability is unknown. The combination of these factors means that these species are likely to occur in trawl grounds and that they are highly susceptible to cap- ture and mortality due to trawlers. The recovery capacity of populations of these species is also low (Appendix 2). The rarity of the species in the bycatch means that no data are available to estimate the probability of breeding before capture, removal rate, total biomass, or the mortality in- dex for most of these species, and they therefore received ranks of 1 for these criteria. In general these are large animals and are therefore likely to have slower recovery rates for their population than those of smaller species. The annual fecundity was low for all species. The pristids are the focus of increasing international concern because their populations are declining worldwide (Stevens et al. 2000). They are rarely seen today in areas where they were previously abundant (Simpfendorfer, 2000). This decrease in pristid populations has resulted in four species being listed on the lUCN 1996 red list (Bailie and Groombridge, 1996). Of the species studied in our study, P. pectinala and P. microdon are listed as endan- gered. Recent demographic analysis of pristid populations has indicated that their recovery will take several decades even if they are given effective conservation (Simpfendor- fer, 2000). In comparison, the species that were most likely to be able to sustain capture in the bycatch of the NPF were H. toslii, E. blockii. C. macloti. and C. tilstoni. These species had a lower susceptibility to capture and mortality due to trawling (Appendix 1). With the exception of H. toshi, these are pelagic species and there is little likelihood of their capture in prawn trawls. For the species for which Stobutzki et a\ Sustainability of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 815 Table 9 Tho mean size of elasmobranch species caught in nets with .'odend s fitted with TEDs and with standard codends. and the ANOVA results from the comparison of these nets. "Combined sharks " refers to all species where total length was recorded; "combined rays" refers to all species where disc width was recorded SE = standard error; n = number of trawls. Species Codend Size (mm ANOVA results mean SE n F df P (\)ml)ine(i sharks standard TED 887 596 59 36 168 269 4.25 1,569 0.0398 Combined rays standard TED 330 286 19 7 157 414 26.77 1,435 <0.0001 Carrhnrluniis dussiirnieri standard TED 844.4 489.1 96.4 13.8 60 81 26.88 1,139 <0.0001 Rliuoprionodon acutus standard TED 636.9 724.2 34.2 17.4 45 91 7.15 1,134 0.0084 Hem iga leus m icrostoma standard TED 708.9 508.0 148.5 19.9 23 58 2.77 1,79 0.0988 Amphoti.stius annotata standard TED 206.4 169.8 31.1 3.4 50 156 2.97 1,200 0.4395 Himanlura toshi standard TED 371.0 351.6 19.4 9.7 51 151 0.60 1,200 0.4395 Rhynchohatus djiddensis standard TED 1076.9 611.6 125.6 32.3 19 24 20.81 1,41 <0.0001 data were available, their survival was higher in trawls. The depth range of the species was wide and their catch rates during the day were the same as or higher than at night. This range provides partial refuge from the night- time commercial trawling. The data available suggest that their recovery capacity is higher than that of most elasmo- branch species (Appendix 2). Individuals of most of these species are likely to have bred before capture and they are smaller. These species were common in the bycatch, and estimates of their removal rate (which was low) and their mortality index (average) were therefore easy to deter- mine. However, all species had low annual fecundities. This assessment of the elasmobranch bycatch is an im- portant first step in ensuring their sustainability because it provides a focus for future research and management. The current ranking is constrained, however, by the avail- able data and by the assumptions outlined in the "Meth- ods" section. The effect of the lack of species-specific infor- mation on the ranks should be taken into account because it may reduce the rank of some species. The application of our assessment has highlighted important information gaps, which should be the focus of research, particularly for the species that are least likely to be sustainable. It is also important that the assessment of the sustain- ability of elasmobranch species is extended to include the impact of other fisheries in the region. There are, for in- stance, fisheries targeting sharks, as well as other fisheries that capture elasmobranchs as bycatch. Because elasmo- branch species may have a wide distribution range, their populations could be impacted by several fisheries, which might create an unsustainable status for the population overall. For example, the pristids are likely to be impacted by the inshore and estuarine gillnet fisheries in this region. The results of our analysis, it is to be hoped, will help in the management of elasmobranch species and in earmark- ing the least sustainable of these species. Future manage- ment may include the use of exclusion devices (TEDs and BRDs), closures, or further limits on retaining shark prod- ucts. The compulsory introduction of TEDs and BRDs into the NPF in 2000 is likely to affect catch rates of elasmo- branchs. The TEDs have the potential to exclude large in- dividuals. However, the majority of elasmobranchs caught are <1000 mm (Fig. 3) and may escape through TEDs. The effectiveness of TEDs will depend on their configuration (particularly the width between the bars) and the size and shape of the bycatch species. Rhynchohatus djiddensis, a large, broad species, appeared to be excluded well by TEDs (Table 9). In comparison, the smaller rays and small, slim sharks were not excluded well (Fig. 3, Table 9). With the introduction of TEDs to the fishery, species-specific exclu- sion rates should be monitored so that these can be taken into account in assessing the sustainability of a species. Juveniles of many elasmobranch species are still likely to be captured and their capture could potentially have a large impact on their respective populations. The TEDs may also be ineffective for species, such as the pristids, that may tangle their saw in the net or the TED. Species and the life stages of species, for which exclusion devices are not effective, may require different management strat- egies, such as marine protected areas. 816 Fishery Bulletin 100(4) This research is the first large-scale assessment of its kind on elasmobranch bycatch. The results highlight the diversity of elasmobranch bycatch in the NPF and the spe- cies that are least likely to be sustainable. We have also highlighted the limited information available for making this assessment. However, our method was designed to maximize the use of the limited information. The process we have used is applicable to other fisheries and also across fisheries, particularly where bycatch diversity is high. Acknowledgments We thank the commercial fishermen who took the time to record their catch in logbooks and support this project. The skippers and crews of the boats on which our scientific observer sampled are also thanked for their cooperation and patience. We also thank the Northern Prawn Fishery Man- agement Advisory Committee for its support; the Northern Prawn Fisheiy, Fishery Assessment Group for their partici- pation in this process and feedback; C. Rose for collection of the crew-member obsei-ver data; and T. Walker for valuable comments on the manuscript. CSIRO Marine Research col- leagues helped in many ways, M. Haywood, F Manson and J. Bishop for assistance with the bioregion data and commer- cial effort data. C. Burridge and W Venables for assistance with the statistical analysis; S. Blaber, G. Fry, D. Milton, J. Salini, J. Stevens and T. Wassenberg for constructive com- ments on the manuscript. We also thank the fishing, scien- tific, and electronics crew of the RV Southern Suri'eyor who made the research sui-veys possible. This project was funded by the Australian Fisheries Research and Development Cor- poration (Project no. 96/257) and CSIRO Marine Research. Literature cited Bailie, J., and B. Groombridge. 1996. lUCN red list of threatened animals, 448 p. lUCN (International Union for the Conservation of Nature and Natural Resources), Gland, Switzerland. Bass, A. J., J. D. D'Aubrey, and N. Kistnasamy. 1973. Sharks of the east coast of southern Africa. I. The genus Carcharhinus (Carcharhinidae). Oceanographic Re- search Institute Investigation Report 33. Bishop J., and D. Sterling. 1999. Survey of technology utilised in the northern prawn fishery, 48 p. AFMA (Australian Fisheries Management Authority), Canberra. Blaber, S. J. M., D. T. Brewer, and A. N. Hams. 1994. Distribution, biomass and community structure of demersal fishes of the Gulf of Carpentaria, Australia. J. Mar. Freshwater Res. 45:375-396. Bonfil, R. 1994. Overview of world elasmobranch fisheries. FAG Fisheries Tech. Paper .341, 119 p. FAG, Rome. Brander, K. 1981. Disappearance of common skate Rata batis from Irish Sea. Nature 290:48-49. Brewer, D. T., N. Rawlinson, S. Eayrs, and C. Burridge. 1998. An assessment of bycatch reduction devices in a tropi- cal Australian prawn trawl fishery. Fish. Res. 36:195-215. Carlton, J. T., J. B. Geller, M. L. Reaka-Kudla, and E. A. Norse. 1999. Historical extinctions in the sea. Annu. Rev. Ecol. Syst. 30:515-538. Casey, J. M., and R. A. Myers. 1998. Near extinction of a large, widely distributed fish. Science 281:690-691. Compagno, L. J. V. 1984a. FAG species catalogue. Vol. 4. Sharks of the world. An annotated and illustrated catalogue of shark species known to date. Part 1. Hexanchiformes to Lamniformes, 249 p. FAG, Rome. 1984b. FAG species catalogue. Vol. 4. Sharks of the wnrld. An annotated and illustrated catalogue of shark species known to date. Part 1. Carcharhiniformes, 405 p. FAG, Rome. Crocos, P. J., and G. J. Coman. 1997. Seasonal and age variability in the reproductive per- formance of Penaeus semisiilcatus broodstock, optimising broodstock selection. Aquaculture 155:57-69. Dulvy, 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. 14:283-293. Hall, S. J. 1999. The effects of fishing on marine ecosystems and com- munities, 274 p. Blackwell Science, Gxford. Hueter, R. E. 1998. Science and management of shark fisheries — intro- duction. Fish. Res. 39:105. lUCN (International Union for the Conservation of Nature and Natural Resources). 1995. lUCN red list categories, 21 p. lUCN, Gland, Swit- zerland. Jennings S.. J. D. Reynolds, and N. V. C. Polunin. 1999. Predicting the vulnerability of tropical reef fishes to exploitation: an approach based on phytogenies and life histories. Cons. Biol. 13:1466-1475. Last, P. R., and J. D. Stevens. 1994. Sharks and rays of Australia, 513 p. CSIRG Pub- lishing. Australia. Lavery, S., and J. B. Shaklee. 1991. Genetic evidence for separation of two sharks, Carcharhinus limbatus and C. tilstoni, from Northern Australia. Mar Biol. 108:1-4. McKinney, M. L. 1997. Extinction vulnerability and selectivity: combining ecological and paleontolgical views. Annu. Rev. Ecol. Syst. 28:495-516. McLoughlin, K., B. Wallner, and D. Staples. 1997. Fishery status reports 1996, 144 p. Bureau of Re- source Sciences, Kingston, Australia. Milton, D. A.. S. A. Short, M. F G'Neill. and S. J. M. Blaber 1995. Ageing of three species of tropical snapper (Lutjani- dae) from the Gulf of Carpentaria. Australia, using radiom- etry and otolith ring counts. Fish. Bull. 93:103-115. Milton, D. A. 2001. Assessing the susceptibility to fishing of populations of rare trawl bycatch: sea snakes caught by Australia's Northern Prawn Fishery Biol. Cons. 101:281-290 Musick, J. A. 1998. Endangered marine fishes: criteria and identification of north American stocks at risk. Fish. 23:28-31. Roberts, C. M., and J. P. Hawkins. 1999. Extinction risk in the sea. Trends Envir Ecol. 14: 241-246. Stobutzki et a\ Sustainability of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 817 SAS. 2001. Logistic regression examples using the SAS system, version 6, first ed., 176 p. SAS Institute Inc., Gary, NC. Simpfendorfer, C. A. 2000. Predicting population recovery rates for endangered western Atlantic sawfishes using demographic analysis. Environ. Biol. Fish. 58:371-377. Sokal. R. R., and R J Rohlf. 1996. Introduction to biostatistics. 2""* edition, 363 p. Free- man and Company, New York, NY. Somers, I. 1994. No resources, no fishery, prawns and their habitat. In Australia's northern prawn fishery (P. C. Pownall, ed.). p. 49-65. NPF 25, Cleveland, Australia. Sparre. P., and S. C. Venema. 1992. Introduction to tropical fish stock assessment, part 1. P'AO Fisheries Technical Paper 306/1. 376 p. FAO Rome. Stevens, J. D., R. Bonfil, N. K. Dulvy, and P. A. Walker. 2000. The effects of fishing on sharks, rays and chimae- ras (chondrichthyans), and the implications for marine ecosystems. ICES J. Mar. Sci. 57:476-494. Stobutzki, I. C, M.J . Miller, and D. T. Brewer. 2001a. Sustainability of fishery bycatch: a process for deal- ing with highly diverse and numerous bycatch. Environ. Cons. 28:167-181. Stobutzki, I. C, M. J. Miller, J. P Jones, and J. P Salini, 2001b. Bycatch diversity and variation in penaeid fisheries; the implications for monitoring. Fish. Res. 53:283-301. Thackway, R., and I. D. Cresswell. 1998. Interim maiine and coastal rationalization for Australia; an ecosystem-based classification for marine and coastal envi- ronments, version 3.3, 120 p. Environment Australia, Com- monwealth Department of the Environment, Canberra. Walker, T I. 1998. Can shark resources be hai-vested sustainably? A question revisited with a review of shark fisheries. Mar. Freshwater Res. 49:553-572. Walker, PA., and J. R. G. Hislop. 1998. Sensitive skates or resilient rays? Spatial and tem- poral shifts in ray species composition in the central and north-western North Sea between 1930 and the present day ICES J. Mar. Sci. 55:392-402. 818 Fishery Bulletin 100(4) Appendix 1 The ranking of elasmobranch species that occurred in the bycatch of the northern prawn fishery with respect to criteria that influence their susceptibility to capture and mortality due to prawn trawls. The weights of the criteria are shown in parentheses; * indicates where species-specific information was not available. The information was obtained from Compagno (19984a; 1984b), | Last and Stevens ( 19941, and Froese and Pauly'^. Criteria Water Day and column night Depth position Survival Range catchability Diet range Susceptibility Family Species (3) (3) (2) (2) (2) (1) ranking Dasyatidae Himantura jenkinsii 1 1' 1' 1* 1.00 Pristidae Pristis clavata 1 1* 1* 1* V*" 1.00 Pristidae Pristis microdon 1 1* 1* 1* 1 * 1.00 Pristidae Pristis pectinata 1 1* 1* 1* 1* 1.00 Pristidae Pristis zijsron 1 1* 1* 1 '^ 1.00 Stegostomatidae Stegostoma fasciatum 1 1* 1* 1 1.00 Scyhorhinidae Atelomycterus fasciatus 1 1* 1* 1 *- 3 1.15 Carcharhinidae Carcharhinus amboinensis 1 1* 1* 3 1.15 Carcharhinidae Carcharhinus leucas 1 1'*" 1* 1* 3 1.15 Dasyatidae Dasyatis brevicaudatus 1 1* 1* 1* 1 ^ 3 1.15 Dasyatidae Dasyatis sp. A 1 1* 1* 1* 3 1.15 Dasyatidae Dasyatis thetidis 1 1* 1* 1* 3 1.15 Scyhorhinidae Galeus sp. A 1 1* 1* 1* 1 * 3 1.15 Dasyatidae Himantura fai 1 Tt 1* 1 :*; 3 1.15 Dasyatidae Himantura gran ulata 1 1 * 1* 3 1.15 Dasyatidae Himantura uarnak 1 1 * 1* 1* 1* 3 1.15 Ginglymostomatidae Nebrius ferrugineus 1 1* 1* 3 1.15 Carcharhinidae Negaprion acutidens 1 1 * 1* 3 1.15 Orectolobidae Orectolobus ornatus 1 1* 1 3 1.15 Rhinobatidae Rhmobatos typus 1 1* 1* 1* 3 1.15 Squatinidae Squatina sp. A 1 1* 1* 1* 1* 3 1.15 Dasyatidae Taeniura meyeni 1 1* 1* ■l^- 3 1.15 Dasyatidae Urogymnus asperrimus 1 1* 1* 1* 3 1.15 Pristidae Anoxypristis cuspidata 1 1* 1* 2 1* 1 1.15 Dasyatidae Pastinachus sephen 1 1* 2 1* 1 1.15 Hemiscylhidae Chiloscyllium punctatum 1 1* 2 1 3 1.31 Dasyatidae Dasyatis kuhlii 1 1* 2 1 1 *■ 3 1.31 Dasyatidae Himantura sp. A 1 1* 2 1* 1 "*' 3 1.31 Narcinidae Narcine westraliensis 1 1* 1* 1* 3 1.31 Rhynchobatidae Rhina ancylostoma 1 1 * 2 1 3 1.31 Carcharhinidae Carcharhinus fitztroyensis 3 1* 1 1* 1 1.46 Dasyatidae Amphotistius annotata 1 1* 2 2 1 :^= 3 1.46 Dasyatidae Himantura undulata 1 1* 2 2 1 -t 3 1.46 Carcharhinidae Rhizoprionodon taylori 1 1* 1 3 3 1.46 Carcharhinidae Galeocerdo cuvier 3 1* 1 1* 3 1.62 Hemiscylliidae Henugaleus microstoma 3 1* 3 1.62 Dasyatidae Dasyatis leylandi 1 2 3 1 3 1.69 Gymnuridae Gymnura australis 1 2 3 2 1 "^^ 1 1.69 Carcharhinidae Carcharhinus sorrah 3 1 o 1* 3 1.77 Carcharhinidae Rhizoprionodon acutus 1 1 3 3 3 1.77 Rhynchobatidae Rhynchobatus djiddensis 1 3 2 1* 1* 3 1.77 C07tlmued Slobutzki et al : Sustainabillty of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 819 Appendix 1 (continued) Criteria Water Day and column night Depth position Survival Range catchability Diet range Susceptibility Family Species (3) (3) (2) <2) (2) (1) ranking Myliobatidae Actohatut: narinan 3 1* 1* 2 3 1.77 Myliobatidae ActomyUu'us vespcrtilio 3 1* 1* 1* 2* 3 1.77 Carcharfiinidae Carcharhmus albunargina tus 3 1* 1* 2 3 1.77 Hcmiscylliidae Hcnii/inslia elongntus 3 1* 1* 2 3 1.77 Sphyrnidae Sphyrna mokarran 3 1* 1* 2 3 1.77 t'archarhinidae Carcharhinus limbatus 3 1 1* 1* 3 1 1.77 Sphyrnidae Eusphyra hlochii 3 1* 2 1 3 1.77 Carcharhinidae Carcharhinus dussumieri 1 2 2 3 1 3 1.85 Dasyatidae Hiinanliira toshi 1 2 3 2 1 3 1.85 Myliobatidae Aetomyleus nichofii 3 1* 2 1 2* 3 1.92 Carcharhinidae Carcharhinus brevipinna 3 1* 1* 1* 3 3 1.92 Carcharhinidae Prionace glauca 3 1* 1* 1* 3 3 1.92 Sphyrnidae Sphyrna Icwini 3 1* 1* 1* 3 3 1.92 Carcharhinidae Carcharhinus macloti 3 1* 1 3 2 3 2.08 Carcharhinidae Carcharhinus tilstoni 3 2 2 2 1 3 2.15 820 Fishery Bulletin 100(4) Appendix 2 The ranking of elasmobranch species that occurrred in the bycatch of the northern prawn fishery with respect to criteria that reflect their capacity to recover after depletion by trawling. The weights of the criteria are shown in parentheses; * indicates where species-specific information was not available. The information was obtained from Compagno (1984a; 1984b). Last and Stevens ( 1994 1, and Froese and Pauly«. Family Species Criteria Probability Maximum of breeding size (3) (3) Removal rate (3) Annual fecundity (2) Mortality index (1) Recovery ranking Dasyatidae Dasyatis brevicaudatiis 1 1* 1 2 1.08 Pristidae Pnstis pectinata 1* 1 1* 2 1* 1.17 Pristidae Prist IS clavata 1* 2 1* 1* 1 * 1.25 Myliobatidae Aetomylaeus vespertilio 1* 2 1* 1* 1.33 Dasyatidae Taeniura meyeni 1 3 1.50 Dasyatidae Hinmntura jenkinsii 1* 2 2 1* 1.50 Carcharhinidae Carcharhinus amboinensis 2 1 1.50 Carcharhinidae Carcharlurjiis leucas 3 1* 1.50 Scyliorhinidae Galeus sp. A 1 * 3 1* 1* 1* 1.50 Narcinidae Narcine westraliensis 1* 3 1* 1* 1* 1.50 Pristidae Pristis microdon 1 * 3 1* 1* 1* 1.50 Squatinidae Squatma sp. A 1* 3 1* 1* 1* 1.50 Carcharhinidae Prionace glaiica 1* 2 1* 1* 1.58 Myliobatidae Aetnbatus narinan 1* 1 3 1.58 Carcharhinidae Carcharhinus limbatus 1* 3 1* 1.58 Carcharhinidae Carcharhinus albimarginatus 1* 2 2 1* 1.67 Carcharhinidae Carcharhinus brevipinna 1* 3 1* 1* 1.67 Dasyatidae Dasyatis thetidis 1 * 2 3 1* 1.75 Pristidae Pristis zijsron 1* 2 3 1* 1* 1.75 Dasyatidae Hinmntura fai 1* 2 3 1* 1* 1.75 Dasyatidae Himantura granulata 1* 2 3 1* 1* 1.75 Dasyatidae Hinmntura uarnak 1* 2 3 1* 1.75 Dasyatidae Himantura undulata 1* 2 3 1* 1.75 Orectolobidae Orectolobus ornatus 1* 2 3 1* 1* 1.75 Dasyatidae Urogymnus asperrimus 2 3 1.75 Scyliorhinidae Atelomycterus fasciatus 2 2 2 1.75 Dasyatidae Dasyatis sp. A 2 3 1 1* 1* 1.75 Dasyatidae Amphotistius annotata 2* 1 3 1* 2 1.83 Rhynchobatidae Rhynchohatus djiddensis 1 2 3 1* 2 1.83 Carcharhinidae Carcharhinus fttztroyensis 2 2 2 2 1.83 Sphyrnidae Sphyrna mokarran 2 1 3 2* 1.92 Carcharhinidae Negaprinn acuttdens 3 2 1 1* 1.92 Rhynchobatidae Rhma ancylostoma 1* 3 3 1* 2.00 Rhinobatidae Rhinobatos typus 1* 3 3 1* 2.00 Pristidae Anoxypristis cuspidata 1 3 3 1* 2.00 Hemiscylliidae Chiloscyllium punctatiim 1* 3 3 1* 2.00 Hemiscylliidae Hemipristis elongatus 2 2 3 2.00 Carcharhinidae Rhizoprionodon acutus 1 3 3 2.00 Stegostomatidae Stegostonia fasciatum 2* 2 3 1* 2.00 Sphyrnidae Sphyrna lewini 1* 2 3 2* 2* 2.00 Myliobatidae Aetomyteus ntchofii 1* 3 3 2 2.08 Carcharhinidae Carcharhinus macloti 2 2 3 2 2.08 continued Stobutzki et al : Sustainability of elasmobranchs caught as bycatch in a tropical prawn trawl fishery 821 Appendix 2 (continued) Criteria Probability Maximum Removal Annual Mortality of breeding size rate fecundity index Recovery Family Species (3) (3) (3) (2) (1) ranking Carcharhinidae Carcharhinus sorrah 1 3 3 1* 2 2.08 Dasyatidac Pastinachus sephen 3 1 3 1* 2 2.08 Carcharhinidae Galeocerdo cuvier 2 1 3 3 2 2.17 Dasyatidae Himantura sp. A 1* 3 3 1* 3 2.17 Carcharhinidae Carcharhinus tilstoni 1 3 3 1 3 2.17 Carcharhinidae Carcharhinus dussumieri 3 2 3 1 1 2.25 Dasyatidae Dasyatis kuhlii 2 3 3 1 1* 2.25 Dasyatidae Dasyatis Icylandi 2 3 3 1* 1* 2.25 Ginglymostomatidae Nebrius ferrugineus 3 2 3 2 1* 2.42 Sphyrnidae Eusphyra blochii 2 3 3 2 2 2.50 Gymnuridae Gymnura australis 3 3 3 1 2 2.58 Dasyatidae Himantura toshi 3 3 3 1 2 2.58 Carcharhinidae Rhizoprionodon taylori 3 3 3 1 2 2.58 Hemiscylliidae Hemigaleus microstoma 2 3 3 2 3 2.58 822 Abstract— Age and growth of the swordfish {Xiphias gladius) in Taiwan waters was studied from counts of growth bands on cross sections of the second ray of the first anal fin. Data on lower jaw fork length and weight, and samples of the anal fin of male and female swordfish were collected from three offshore and coastal tuna longline fishing ports on a monthly basis between September 1997 and March 1999. In total, 685 anal fins were collected and 627 of them (293 males and 334 females) were aged suc- cessfully. The lower jaw fork lengths of the aged individuals ranged from 83.4 to 246.6 cm for the females and from 83.3 to 206 cm for the males. The radii of the fin rays and growth bands on the cross sections were mea- sured under a dissecting microscope equipped with an image analysis system. Trends in the monthly mar- ginal increment ratio indicated that growth bands formed once a year Thus, the age of each fish was deter- mined from the number of visible growth bands. Two methods were used to estimate and compare the standard and the generalized von Bertalanffy growth parameters for both males and females. The nonlinear least square estimates of the generalized von Berta- lanffy growth parameters in method II, in which a power function was used to describe the relationship between ray radius and LJFL, were recommended as most acceptable. There were sig- nificant differences in growth param- eters between males and females. The growth parameters estimated for females were the following: asymptotic length (Lj = 300.66 cm, growth coef- ficient (K) = 0.040/yr, age at zero length «o) = -0.75 yr. and the fitted fourth parameter Imi = -0.785. The growth parameters estimated for males were the following: asymptotic length iL..,) = 213.05 cm. growth coefficient lA') = 0.086/yr, age at zero length (/„) = -0.626 yr, and the fitted fourth parameter ( m ) = -0.768. Age and growth of the swordfish (Xiphias gladius L.) in the waters around Taiwan determined from anal-fin rays Chi-Lu Sun Sheng-Ping Wang Su-Zan Yeh Institute ol Oceanography National Taiwan University No 1, Sec, 4, Roosevelt Road Taipei, Taiwan 106 Email address (for C L Sun) chilu'Sccmsntuedu tw Manuscript accepted 21 February 2002. Fish. Bull. 100:822-835 (2002). The swordfish iXiphias gladtus L.) is a cosmopohtan species found in the trop- ical, subtropical, and temperate waters of the world's oceans and adjacent seas (Sakagawa, 1989). In the Pacific Ocean, swordfish is generally distributed from Asia to the Americas between .50°N and 50°S (Bartoo and Coan, 1989). In the waters of Taiwan, the swordfish is an incidental bycatch of the offshore tuna longline and harpoon fisheries. Both fisheries contributed an esti- mated 1528 metric tons (99%) to the total swordfish landings from Taiwan waters in 1999. Information on age and growth of fishes is a central element in fishery management (Brothers, 1983). Mea- surement of the age of the fish provides the key variable of time needed to es- timate life history and biology factors, such as mortality and growth. Mortal- ity and growth-rate models provide quantitative information on the status offish stocks and at the same time may be used in more sophisticated models, such as yield-per-recruit analyses and cohort analyses (Powers, 1983), which will directly contribute to the rational exploitation offish resources, as well as to the development of proper manage- ment plans. Most age determination studies of swordfish have dealt with Atlantic pop- ulations and have used different hard parts, such as anal-fin rays (Berkeley and Houde, 1983; Wilson and Dean 1983; Prince et al., 1988; Ehrhardt, 1992; Esteves et al., 1995; Ehrhardt et al., 1996), otolith (Radtke and Hurley, 1983; Wilson and Dean, 1983; Esteves et al., 1995), and vertebrae (Esteves et al., 1995). In contrast, only a few at- tempts have been made to determine the age of swordfish in the Pacific Ocean. Yabe et al. (1959) estimated the growth of swordfish caught in the western North Pacific (140°-160°E) by longline during the period from 1948 to 1956 using the modal analysis of length frequencies. Castro-Longoria and Sosa-Nishizaki (1998) compared the age estimates of swordfish caught by drift gillnet vessels off Baja Califor- nia from 1992 to 1993 based on otolith microstructure and cross sections of the second ray from the first anal fin, and highly recommended the use of cross sections of the second ray to de- termine the ages of swordfish in the Pacific Ocean. Uchiyama et al. (1998) evaluated various hard parts (includ- ing rays of the first dorsal and first anal fins, vertebrae, and sagittae) for aging swordfish in the central North Pacific by Hawaii longline fishery from 1991 to 1993, and provided preliminary estimates of length-at-age. The objectives of our study were to estimate the age and growth of sword- fish by counting the growth rings on the cross sections of the second ray of the first anal fin and to compare the generalized growth function proposed by Richards (1959) with the standard von Bertalanffy model for represent- ing the best growth model of swordfish around Taiwan waters. The informa- tion is crucial because it will allow the age composition of the catch to be determined, which in turn will allow the status of the swordfish stock in the Sun el a\ Age and growth of Xiphias gladius 823 N 25 5- 250- 245- 24.0- 23,5- 23 0- 22 5- 22 0- 21.5- 21 0- 20 5- 20 0- Tungkang" Shinkang. ^<, Jt ^^ Taiwan Pacific Ocean CO E 120 140 160 1J0 120 100 119 — I — 120 121 122 E 123 Figure 1 Three fishing ports in Taiwan where the swordfish anal-fin ray samples were collected in this study. waters around Taiwan to be assessed by using yield-per- recruit and sequential population analyses. Material and methods Data on lower jaw fork length (LJFL) and weight, and samples of the first anal fin of male and female sword- fish were collected from three offshore tuna longline and harpoon fishing ports (Fig. 1) on a monthly basis between September 1997 and March 1999. The fishing grounds for those vessels are shown in Figure 2. In total, 769 LJFLs and 685 anal fins were collected. The anal fins were frozen for approximately one month before being thawed and boiled to remove the tissue and to separate the second rays. Three cross sections ranging in thickness from 0.5 to 0.75 to 1.0 mm were taken successively along the length of each ray with a low-speed "ISOMET" saw (model no. 11-1280) and diamond wafering blades, at a location equivalent to 1/2 of the maximum width of the condyle base mea- sured above the line of maximum condyle width (Fig. 3) +^U~ "p^'fr Sat' _1 A r hi „ L^^v ^K=^ 4-H- t' • • ' • ■■■ 1^^ -n i yi- 1 C / l- \ W 7^ ^ J ^^ ~- J> A IS- > - ( VT ?J §3 ^ ~- T 1 10- v?^ ^£ ^ ,5' 4 -.f^ ,-i^\ rtn A }\ 1 I, "iK 1 1 I i 1 i 1 i 1 1 5 J ^ 1 1 f 1 1 n 1 1 11 M 105 110 115 120 125 130 E 1 J-^-^ ^ *^ ^JC->K^ v^^ 7 2^ aZ-j-j J ) Jn ^^ iy^ "^ S ^ ^^ A ^ It ' ' \ J^S, 1 ^J)t H ' 51' ' 1 i c' *^ 105 110 115 120 Figure 2 The fishing grounds (oblique dot line and harpoon fishing boats b (A), Nanfangao (B), and Shinkan 125 130 E lines) of the long- ased in Tungkang g(C) fishing ports. (Ehrhardt, 1992; Ehrhardt et al., 1996). The sections were immersed in 95% ethanol for several minutes, placed in a labeled (with sampling date and a number) small plastic case to air dry, and then stored for later reading. Anal- fin ray sections were then taken from the small cases randomly and examined under a dissecting microscope (model: Leica MZ 6) with transmitted light at various 824 Fishery Bulletin 100(4) A 1 ^ SPINE SHAFT I TRANSVERSE SECTION ^B SECTION B "' ML, Mm W^\ 0.5 -1mm ^_ J, _^ CONDYLE BASE ■4 — ' Figure 3 The second ray of the first anal fin showing the location of the cross section at a distance equal to a half of the width of the condyle base above the base (A), and a section of the second ray of the first anal fin(Bl. magnifications from 8x to 16x depending on the size of the section. The clearest one of the three sections from each fin ray was read three times by one reader about one to three months apart with no knowledge of fish length. The pre- cision of readings was evaluated as the average percent error (APE, Beamish and Fournier, 1981) and coefficient of variation (CV. Campana et al., 1995, 2001). For those sections resulting in three different readings, each section was reread by two to three readers simultaneously. Speci- mens whose age estimates still disagreed were omitted from further analyses. The images of the anal-fin ray sections were captured by using an Image Analysis Software package (Media Cybernetics. 1997) in combination with a dissecting mi- croscope equipped with a charged coupled device (CCD) camera (model: Toshiba IK-630) and a Pentium II com- puter equipped with a 640 x 480 pixel frame grab card and a 800 x 600 pixel monitor. The images were measured in microns after distance calibrations were incorporated. The distances from the focus to the distal edge of the sec- tion (ray radius) and from the focus to the distal edge of each growth band (annulus) were measured and recorded (Fig. 4). The focus, the growth band, and the false growth band (multiple bands) were defined according to Berkeley and Houde (1983), Tserpes and Tsimenides (1995), and Ehrhardtetal.(1996). The marginal increment ratio (MIR), which was used to validate the reading of annuli, was estimated for each specimen by the following formula (Prince et al., 1988; Esteves et al., 1995): M//? = (S-S„)/(S„-S„_i), where S = ray radius; and S and S , = the distance from ray focus to bands n and n-1, respectively. The mean MIR and the standard deviation were computed for each month by sex for all ages combined and also for each age separately. Growth was analyzed by using the back-calculation of length-at-age for each sex. For this purpose, a relation- ship was determined between the ray radius and the LJFL. This relationship and the distance from the focus Sun et a\ . Age and growth of Xiphias gladius 825 A i i » ir ^ X FOCUS im u H (1 mm) Figure 4 The section of three typical second anal fin rays of sword- fish. Ray radius (S) measured from focus to edge; annuli for estimated age 1+ (Al. age 5+ (B) and age 11+ (C). to successive rings were used to back-calculate lengths at presumed previous ages (Ehrhardt, et al., 1996). For the relationship between ray radius and LJFL and the back- calculation of lengths-at-age, the following two methods were used. Method I The relationship between ray radius (S) and LJFL (L) was determined by using the standard linear regression procedure, L = a + bS (Berkeley and Houde, 1983). This relationship and the distance from focus to successive growth bands, which we assumed to be based on annual growth events, were used to back-calculate the lengths at presumed ages by the following formula (Fraser, 1916): -a = ^(L-a), o where L = LJFL at time of capture; L^ = LJF'L when band /; was formed; a = the intercept on the length axis from the regression line of length (L) on ray radius (S), e.g. L = Q + hS; and S„ = the distance from ray focus to band n. Method II The relationship between ray radius and LJFL was determined by using a power function procedure, L = aS* (Ehrhardt, 1992; Ehrhardt et. al,, 1996). Parameters of this function were estimated by nonlinear least square fits to the observed data. This relationship and the distance from focus to successive growth bands were used to back- calculate the lengths at presumed ages by the following formula (Tserpes and Tsimenides, 1995; Ehrhardt et al., 1996): where b L.-\^\L, the exponent of the regression of length (L) on ray radius (S) which is assumed to be a power function of the form L = a S ^. The data of the back-calculated length-at-age from method I and method II were then applied to the following stan- dard von Bertalanffy growth equation (standard VB) and to the generalized growth function (generalized VB) (Rich- ards, 1959): Standard VB: L, =L„(l-e-*"-'"'); Generalized VB: L, =L,„(l-e -A'il-mK(-f„l\i-" where L, = the mean LJFL at age t; L„ = the asymptotic length; tg = the hypothetical age at length zero; k and K = the growth coefficients; and m = the fitted fourth growth-function parameter. Parameters of the above two equations for male and fe- male were estimated, respectively, by fitting a curve to the observed back-calculated LJFL-at-age by using a nonlinear least square procedure (Gauss-Newton method. NLIN of SAS Institute, 1990). The measure of goodness-of-fit chosen was r~. A multivariate statistical procedure (Hotelling's T'^) was used to test for differences in growth between males and females (Bernard, 1981) for the two growth models and two methods. The r- values were ranked between the two different growth functions with the smaller as 1. A non- parametric test (Friedman 1937, 1940 1 was then employed 826 Fishery Bulletin 100(4) ■ Female n =379 SMale n =329 Dsex unknown n =61 I I I I I I 75 95 115 135 155 175 195 215 235 255 275 115 135 155 175 195 215 235 255 275 Lower jaw fork length (cm) Figure 5 The size-frequency distribution by 5-cm intervals (A) and the proportions (B) of female and male swordfish collected from Tungkang, Nanfangao, and Shinkang fish markets, September 1997 to March 1999. on the ranked results of r~ to test the significance of the goodness-of-fit comparisons between the two growth func- tions (Chen et al., 1992). Friedman's test statistic x~r was calculated with a correction for tied ranks (Zar, 1999). The association of rank ordering of r'^ values between the two growth functions was measured nonparametrically by us- ing the Kendall coefficient of concordance (Kendall, 1962). Results Of the 685 anal fin rays sampled, 627 (334 females and 293 males) were aged successfully. The average percent error (APE) was 5.18% (5.09% for females and 5.29% for males) and the coefficient of variation (CV) was 8.50%- (8.28% for female and 8.76% for males). The LJFL of the aged fish ranged from 83.4 to 246.6 cm for the females and from 83.3 to 206 cm for the males. The remaining 58 fin rays were considered unreadable mostly because the opaque- translution zonation was so unclear that annuli could not be defined (32 specimens); 7 specimens were unreadable because of the existence of multiple bands, which made the identification of annuli difficult or resulted in aging discrepancies between readings and readers; and 19 speci- mens were unreadable because of both the above factors. For all the 769 swordfish with LJFL measured, indi- viduals ranged from 83.4 to 290 cm for females and 78 to 206.6 cm for males (Fig. 5A). The proportion of the females (Fig. 5B) varied at sizes less than 195 cm, then increased to 100% at 210 cm and thereafter. The relationship between LJFL and round weight for 227 specimens (sexes pooled) is shown in Figure 6; AN- COVA revealed no differences between males and females (P>0.05). The LJFL-EFL (eye fork length) conversion equation is LJFL = 1.0647EFL + 7.7911, with df = 563 and r'-=0.99. Sun et a! : Age and growth of Xiphias glodius 827 180 - 160 - / RW = 13528x10 "LJFL'"" / 140 - r-' = 0.9664 / CT 120 - -it n = 227 / OT 100 ■ / 0) ■o 80 - CD/ oyo ID O tr 60 ■ y° rf^ 40 ■ ^^ ° 20 ■ ..^^ 0 50 100 150 200 250 Lower jaw fork length (cm) Figure 6 Relationship between round weight and lower jaw fork length for the swordfish collected from Shinkang fish market regardless of sex. The monthly means of marginal increment ratio, MIR, for females with all ages combined, dropped drastically from the maximum of 0.86 in June to the minimum of 0.32 in July and August ( Fig. 7 ). Similarly, the monthly means of MIR value for males declined sharply from the maximum of 0.65 in June to 0.37 and 0.35 in July and August. For both females and males, the monthly means of MIR dur- ing the period from September to March were not differ- ent (ANOVA, P =0.95, P,=0.48), but the monthly means of MIR in April, May, and June were significantly higher than that in July or August, respectively (two sample /-tests, P<0.001). Also, the mean MIR in August was significantly lower than that in September (/-tests, P <0.001, P,<0.01). The trends exhibited by monthly means of MIR for females and males for ages 2 to 5, respectively, were the same as those just described for all ages combined (Fig. 8). These patterns indicated the formation of one ring per year dur- ing the period from July to August. The MIR analysis by age was not performed for age 1 and ages greater than 5 because the formula used to determine MIR does not apply to samples less than or equal to age 1 and there was a lack of a sufficient number of samples for ages greater than 5. The mean band radii, by band group for female and male, are shown in Table 1. The observed LJFLs of female and male swordfish were plotted against their corresponding ray radii for method I and method II, respectively (Fig. 9). The relationships between LJFL and ray radius are de- scribed as follows: Method I Female: LJFL = 21. 137S -i- 65.091 |r2=0.8894, «=3341; Male: LJFL = 19.966S -i- 68.160 [r2=0.8737,n=2931. 1 4 - 1 2 - Female 1 - 36 32 08 - 9 12 33 T T /^ / \ ■''• 14 14 22 \ T T T T 06 - / A y < (' \19 21 J --, ^ 9 0"- ~ 02- QJ ■ " H / ■ ■ > ■ 1 ■ 1 1 ■ 1 > c JFMAMJJASOND 1 >an marginal S 1 16 Male 40 - 20 08 06 27 8 5 ^ 13 14 1/ \jj= , ' 14 04 ^ ^ . ' \ /l ■ 02 J [ J '■ J FMAMJ JASOND Month Figure 7 Monthly means of marginal increment ratio of female and male swordfish in the waters around Taiwan for all ages combined. Vertical bars are 1 SE, numbers on the top of vertical bars are sample sizes. Method II Female: LJFL = 73.754S05i93 [r2= 0.8753, ;!=334]; Male: LJFL = 77.075S»-*"2 |r-^=0.8742,«=2931. ANCOVA revealed significant differences in the rela- tionship between the males and females for both methods (for method I, P.,.,, ,,3.^=8.36, P<0.001; for method II, P.333 292 =15.59, P=0.004). The average back-calculated lengths- at-age obtained using method I and method II are shown in Table 2. Growth rates were higher during the first year of age (mean 95.2 cm and 96.1 cm LJFL for males and females, respectively, for method I, compared to 88.6 cm and 90.4 cm LJFL for males and females, respectively, for method II). After the first year of age, the growth rates of both sexes slowed appreciably. Growth rates of females were always higher than those of males, especially after the age of three. Also, the growth rates were always higher for method II than for method I except for the first year of age. Fitted standard VB and generalized VE growth 828 Fishery Bulletin 100(4) 1 08 06 0,4 02 0 12 1 0 0.8 2 0,6 1 0,4 i 02 ^ o 0 1 5 1 05 0 2 1 5 1 05 0 Female Age 2 + i-tf't^jH^ Male 1 08 06 0.4 _ 5 - • 4 8 4 10 • • ; 11 02 - • • JFMAMJJASOND JFMAMJJASOND Age 3 » 4 6 1 • r^'t^ t _i I I I 'III J FMAMJ J ASOND V " 1 Age 4 HJ J FMAMJ J ASOND 1 2 1 3 ^ . . . 2 Age 5 1 08 06 04 02 0 1 5 1 05 0 1 5 2 J FMAMJ J ASOND 4 6 1 J FMAMJ J ASOND 1-5 •2 05 1- 0 *■ T 4 \T 4 , 1 J FMAMJ J ASOND J FMAMJ J ASOND Month Figure 8 Monthly means of marginal increment ratio of female and male swordfish in the waters around Taiwan for ages 2 to 5 respectively. Vertical bars are 1 SE, numbers on the top of vertical bars are sample sizes. curves for males and females with method I and method II back-calculation are shown in Figure 9, and the esti- mated parameters corresponding to each curve are shown in Table 3. Hotelling's T- test results showed significant difference in growth parameters between female and male swordfish for either standard or generalized VB with either method I or method II back-calculation (Table 4). The calculated T~ is considerably higher than the tabulated value in Table 4 for each case, and all the parameters, except for m of the generalized VB, significantly affect the differences in growth between male and female swordfish (The Roy-Bose simultaneous confidence intervals around differences be- tween parameter values fail to include zeroi. The results of goodness-of-fit comparison showed that the generalized VB had larger r~ ranks for method II, but had equal ranks with the standard VB for method I. Considering the tie rank groups in each sex-by-method, Friedman's x^^ sta- tistics is 2 (n=4). This was not significant at the 5')?^ level (i.e. ;|f-=3.841, df=l). which indicated no significant differ- ence in the r- rank ordering of the values between the two growth functions. Kendall's coefficient of concordance was 0.5 (;i=4), which did not indicate a good agreement in the r- ranks for all sexes-by-method. Discussion Just as reported by Berkeley and Houde (1983), Radtke and Hurley (1983), Wilson and Dean (1983), Tsimenides and Tserpes (1989), Ehrhardt (1992), Tserpes and Tsi- menides (1995), and Stone and Porter (1997), females in our study were typically larger than males although the length-weight relationship between the sexes did not differ significantly. The overall sex ratio for the sampling period in our study did not deviate significantly from 1: 1 (P<0.01) but differed substantially from the ratios of 2.3 females to 1 male and 2.7 females to 1 male reported respectively by Stone and Porter (1997) and Caton et al. ( 1998). This discrepancy may have been caused by the dif- Sun et al Age and growth of XIphias gladius 829 Table 1 Moan rac numerals ins from focus to indicate the num distal edge of each band ber of bands. of fema e (A) and male I B) swordfi sh in the waters around Taiwan. Roman A Band class Sample size M ean radius (mm) from focus to each band I II III IV V VI VII VIII IX X XI XII 0 2 1 86 1.44 2 76 1.44 2.38 3 42 1.44 2.43 3.09 4 39 1.51 2.. 34 3.08 3.52 5 26 1.43 2.39 3.26 3.84 4.23 6 20 1.51 2..35 3.14 3.77 4.24 4.75 7 13 1.43 2.40 3.33 3.90 4.40 4.80 5.20 8 13 1.58 2.60 3.47 4.10 4.52 4.95 5.39 5.84 9 7 1.60 2.36 3.18 3.81 4.35 4.86 5.39 5.80 6.26 10 2 1.49 2.37 3.35 3.96 4.46 4.84 5.. 33 5.64 6.21 6.66 11 3 1.55 2.43 3.23 3.81 4.18 4.67 5.23 5.76 6.25 6.65 6.80 12 5 1.50 2.44 3.30 3.88 4.41 5.02 5.39 5.94 6.29 6.69 6.98 7.24 Mean 1.49 2.41 3.24 3.84 4.35 4.84 5.32 5.80 6.25 6.66 6.89 7.24 SD 0.18 0.23 0.34 0.40 0.40 0.49 0.40 0.42 0.33 0.46 0.57 0.40 Growth increase 0.91 0.84 0.60 0.50 0.49 0.48 0.47 0.46 0.41 0.22 0.35 B Band class Sample size M ?an radius (mm) from focus to each band I II III IV V VI VII VIII IX X 0 2 1 84 1.47 2 57 1.32 2.16 3 32 1.28 2.36 3.06 4 43 1.31 2.32 3.04 3.51 5 27 1.32 2.32 3.11 3.65 4.05 6 22 1.32 2.44 3.21 3.71 4.11 4.47 7 9 1.39 2.41 3.21 3.75 4.28 4.62 4.98 8 8 1.33 2.52 3.24 3.82 4.26 4.66 5.02 5.37 9 4 1.25 2.25 3.35 3.92 4.29 4.70 5.05 5.37 5.74 10 5 1.40 2.36 3.16 3.93 4.41 4.80 5.11 5.47 5.76 6.06 Mean 1.34 2.35 3.17 3.76 4.23 4.65 5.04 5.40 5.75 6.06 SD 0.17 0.22 0.26 0.30 0.31 0.33 0.32 0.35 0.39 0..38 Growth increase 1.01 0.82 0.58 0.48 0.41 0.39 0.36 0.35 0.20 ference in size ranges of LJFL sampled for studying the sex ratio (Mejuto, et al.. 199.5). Most of the LJFL in our sample ranged between 100 and 18.5 cm, close to Arocha and Lee's (1995) middle size range within which the sex ratio was also almost 1:1 (Arocha and Lee, 1995). Besides the size range difference, the differences in geographical areas and seasons can also affect the sex ratio (Hoey, 1991; Mejuto et al., 1991). The proportion of females in our study, which increased to 100% at 210 cm and thereafter, was similar to those described by Turner et al. (1996). Stone and Porter (1997), and DeMartini et al. (2000). Several genetic studies (Grijalva-Chon et al., 1994; Rosel and Block, 1996; Chow et al., 1997; Chow, 1998) have been unable to reject the hypothesis that swordfish comprise a single, homogenous population in the Pa- cific. However, from recent analyses of mtDNA, Reeb et al. (2000) concluded that swordfish are not homogenous in the Pacific. They found significantly different northern and southern populations in the western Pacific and sev- eral overlapping swordfish populations may occur in the eastern Pacific, making swordfish genetically continuous there. Gene flow between the populations occurs through 830 Fishery Bulletin 100(4) 300 - 250 - Female o 200 - 150 - oo -^^ 100 - ^^ - Method 1 ? 50- LJFL = 21 137S + 65 091, r' = 0 8894 — Method II UFL = 73 754 s"'", r^ = 0 8753 o) 0 ■ i ' o 5 250 n o 200 ■ 150 - ) 123456789 10 Male °9,8'^° '^^ — ^ ^i^^^^° 100 - - Method 1 50 - UFL = 19 966S + 68 160. r^ = 0 8737 — Method II UFL = 77,075 s""'". r^ = o 8742 1 1 1 I 1 1 1 I 1 1 0 123456789 10 Spine radius (mm) Figure 9 Relationship between LJFL and anal ray radius for female and male swordfi sh in the waters around Taiwan. a horseshoe-shaped corridor, running between the north- western Pacific, across to the eastern Pacific and back to the southwestern Pacific (Ward and Elscot, 2000). Accord- ing to Reed's studies, the swordfish in our samples can be considered a part of the northern population in the west- ern Pacific Ocean. We found that anal-fin rays are useful for aging swordfish; they are easily sampled without reducing the economic value of the fish and can be read easily (the growth rings stand out clearly). This aging tool is espe- cially important because swordfish lack scales and their yery small otoliths are not amenable to traditional aging techniques (Ovchinnikov, 1971; Beckett, 1974;Tserpes and Tsimenides, 1995). Moreover, fin rays can be easily stored for future reexamination (Compean- Jimenez and Bard, 1983). One problem associated with the fin-ray method used in our study, also indicated by Berkeley and Houde (198.3) and Tserpes and Tsimenides (1995), was the pos- sible existence of multiple bands and the missing first annulus in larger fish. However, Gonzalez-Garces and Fariiia-Perez (1983) and Tserpes and Tsimenides (1995) noted that experienced readers could overcome the prob- Table 2 Mean back- calculated lower jaw fork lengths at age for swordfish in the waters around Ta iwan. Age (yr) Back-calculated length (cm Method I Method II Male Female Male Female 1 95.19 96.07 88.60 90.35 2 114.86 115.70 114.96 116.18 3 131.88 133.41 133.80 136.47 4 143.12 146.30 145.22 150.35 5 152.14 157.99 154.40 162.91 6 159.15 169.95 161.36 175.34 7 165.60 180.48 167.84 186.40 8 174.76 190.14 176.87 195.83 9 184.00 198.40 185.23 204.62 10 190.89 207.79 191.58 214.15 11 215.06 220.58 12 222.15 226.62 Sun et a\ Age and growth of Xiphias gladius 831 Parameter estimates for thi waters around Taiwan. Num Table 3 standard von Bertalanffy and the generalizi bers in parentheses are standard errors. ^d von Bertal anfFy growth models for swordfish in the Standard von Bertalanffy growth model Generalized von Bertalanffy growth model Meth Ddl Method II Method I Method II Parameter Male Female Male Female Male Female Male Female L^ 224.170 (12.802) 281.809 (6.805) 207.520 (8.465) 267.441 (6.517) 231.772 (26.937) 301.877 (11.068) 213.052 (19.153) 300.656 (38.869) k 0.140 (0.025) 0.101 (0.006) 0.198 (0.031) 0.130 (0.009) 0.066 (0.103) 0.060 (0.055) 0.086 (0.035) 0.040 (0.116) t„ -3.089 (0.523) -3.204 (0.171) -1.955 (0.406) -2.302 (0.190) -1.556 (2.942) -2.036 (1.263) -0.626 (1.196) -0.750 (2.272) m -0.625 (2.388) -0.409 (0.943) -0.768 (1.730) -0.785 (1.324) Table 4 Results of the multivariate (Hotelling's T^) tests for difference between the estimated von Bertalanffy growth parameters of female and male swordfish in the waters around Taiwan. Standard von Bertalanffy Generalized von Bertalanffv growth model growth model Method I Method II Method 1 Method 11 7-2 20715.1 56005.2 1614630 6782.34 df 3.623 3,623 4,623 4,623 '^O.Ol.df 11.48 11.48 13.46 13.46 99% CI for' L. -Z.„^ 53.842 ~ 61.436" 56.452 - 63.390" 68.845-71.365" 85.621 - 89.587" k-.-ki -0.044 •- -0.034'" -0.071 -- -0.065" K,-Ks -0.008 - -0.004" -0.050 - -0.042" ^oy-'o^ -0.216 - -0.014" -0.406 ~ -0.288" -0.542 ~ -0.418" -0.163 - -0.085"' m -m 1 0.205 ~ 0.227" -0.062 - 0.028 Roy-Bose simultaneous confidence intervals around differences between parameter values. Indicates the parameter tested that significantly affects differences in growth between the females and the males at a significant level of 0.01. lem of multiple bands by determining whether the bands were continuous around the circumference of the entire ray section and by judging their distance from preceding and following bands. In our study, 91 out of the 627 read- able specimens had "multiple bands" that were read with- out a problem by using these criteria. Of the 26 discarded specimens that had multiple unreadable bands, most were found in swordfish larger than 200 cm in size. The missing first band in larger fish can be estimated from observa- tions of its position on young specimens where the first band is visible. Similar approaches for solving the problem of missing band have also been used for Atlantic swordfish (Berkeley and Houde, 1983) and eastern Mediterranean swordfish (Tserpes and Tsimenides, 1995). Results of marginal increment ratio analysis (Figs. 7 and 8) suggest one growth ring (annulus) is formed each year from July to August, which is toward the end of the spawning period for the swordfish in the north Pacific (Yabe et al., 1959). Although the timing of annulus deposi- tion coincides with the swordfish's spawning season in the north Pacific, it may also be related to swordfish migration, as suggested by Berkeley and Houde (1983) for Atlantic 832 Fishery Bulletin 100(4) swordfish and by Tserpe and Tsimenides (1995) for east- ern Mediterranean swordfish. The relationship between annulus formation and migration for the western Pacific swordfish should be investigated. Others (Nelson and Ma- nooch, 1982; Beckman et al., 1990; Sturm and Salter, 1990; Ferreira and Russ, 1994; Franks et al.. 1999) have com- mented on the physiological nature of annulus formation and the importance of environmental factors, suggesting that reproduction may not be the sole determining factor in annulus deposition. Our results only partially validated age. Validation of ages requires either a mark-recapture study or the identification of known-age fish in the popu- lation (Beamish and McFarlane, 1983; Prince et al., 1995; Tserpes and Tsimenides, 1995; Sun et al. 2001 ). According to Hotelling's T- analysis, the female and male swordfish in either method I or method II of both standard VB and generalized VB grew differently. Those of the gen- eralized VB had a little larger /■- ranks than those of the standard VB, although the goodness-of-fit of nonparamet- ric analysis of r- ranks did not show a significant difference between two growth equations (P=0.22). In addition, the generalized VB appeared to fit the data well over the range of ages and it provided more realistic growth patterns for juveniles less than one year. However, the standard VB, commonly used to describe fish asymptotic growth, did not fit these data well, and generated grossly overestimated val- ues for individuals less than one year (Table 3 and Fig. 10 1 (Ehrhardt, 1992; Ehrhardt et al, 1996). In Table 3, the tg values estimated for the generalized VB with method II (i.e. a power function was used to describe the relationship between ray radius and LJFL) were much closer to zero than those estimated for the generalized VB with method I (i.e. a simple linear function was used to describe the relationship between ray radius and LJFL I. Also, Ehrhardt (1992), Ehrhardt et al. (1996), and Tserpes and Tsnnenides ( 1995) favored the power function (method II) because they believed its description to be more biologi- cally realistic. Therefore, the parameter estimates for the generalized VB model with method II shown in Table 3 are recommended as the most acceptable for determining the age composition of swordfish in the waters around Taiwan. Age-length relationships of swordfish (Fig. 11) are mostly based on Atlantic specimens (Berkeley and Houde, 1983; Radtke and Hurley, 1983; Wilson and Dean, 1983; Ehrhardt, 1992; Ehrhardt et al, 1996). and a few on Pacific samples (Yabe et al.. 1959; Castro-Longoria and Sosa-Nishizaki. 1998; Uchiyama et al.. 1998; Castro- Longoria'; Uchiyama-). Differences in estimates of growth parameters for males and females arise from the use of different hard parts, e.g. anal-fin rays (our study) versus otoliths (Radtke and Hurley, 1983; Wilson and Dean, 1983) or vertebrae (Esteves et al., 1995), artifacts of preparation, or interpretation. Even though we used the same method as Ehrhardt (1992), Tserpes and Tsimenides (1995), and Ehrhardt et al. (1996), observed difference could be re- lated to geographical coverage of the studies. Our results were in the mid-range of previous estimates, well within the range of variation that might be expected due to the somewhat subjective nature of the processing, measuring, and interpreting of growth rings on fin rays. Therefore, we 250 - = 200 - Female St-I^li^t^ ° 150 - •i^Hi ' \/y% ° 100 - M M -o- standard VB - IVIethod 1 /y ^Standard VB-tVlethod II ? 50- // -a-GenerahzedVB - Method 1 // -^ Generalized VB - IVIethod II £ 0- yj J ' " ' ^ -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 o 1 250 - QJ 1 "' 200 - o Male ° 1 JL==i''f^ 150 ■ i \J^^l^ jln ' 100 ■ yc //j -D- standard VB - Method 1 50 - // ^^ Standard VB- Method II / // -ft- Generalized VB - Method 1 / // / -•- Generalized VB - Method II 0 -1 / /J J " f " '* 1 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 Age (year) Figure 10 Standard and generalized von Bertalanffy growth curves for female and male swordfish in the waters around Taiwan. believe our growth parameter estimates are appropriate for use in assessment studies of the northern swordfish population in the western Pacific Ocean. Acknowledgments The authors express sincere gratitude to Nancy Lo and David Au. Southwest Fisheries Science Center of the Castro-Longoria. R. 2000. Personal communication of the length-at-age estimates mentioned in Castro-Longoria and Sosa-Nishizaki. 1998. Departmento de Investigaciones Cienti- ficas y Tecnologicas de la Universidad de Sonora, Rosales y Ninos Heroes S/N. Hermosillo. Sonora. 83000 Mexico. Uchiyama. J. H. 2000. Personal communication of the length- at-age estimates mentioned in Uchiyama et al., 1998. Hono- lulu Laboratory. Southwest Fisheries Service Center, NOAA, 2570 Dole Street, Honolulu. Hawaii 96822-2396. Sun et al : Age and growth of Xiphias gladius 833 300 1 o — Yabeetal (1959)- sex combined -♦ — Berkeley & Houde (1983) • — Wilson & Dean (1983) ■ RadlkeS Hurley (1983) < — Tsimenides & Tserpes(1989) H — Ehrhardl(1992) Tserpes & Tsimenides(1995) Ehrhardtelal (1996) - -X - Uchiyama et al (2000) -^i — Castro-Longona (2000) Present study Age (year) Figure 11 A comparison of the growth curves of female and male swordfish estimated by different authors. National Marine Fisheries Sei-vices, NOAA, for their comments on the initial draft of this paper. Our many thanks are also given to four anonymous referees for their valuable comments. This study was in part supported financially by the "Fisheries Administration, Council of Agriculture, Taiwan," through grant 88-AST-1.4-FID-04 (04)toChi-LuSun. Literature cited Arocha, F.. and D. Lee. 1995. The spawning of swordfish from the Northwest Atlantic. ICCAT (International Commission for the Con- servation of Tunasi Coll. Vol. Sci. Pap. 44(3):179-186. Bartoo, N. W., and A. L.Coan. 1989. An assessment of the Pacific swordfish resource. In Proceedings of the second international billfish sym- posium; Hawaii, August 1-5. 1988 (R. H. Stroud, ed.), p. 137-151. National Coalition for Marine Conservation, Inc., Savannah, GE. 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. Beamish, R. J., and G. A. McFarlane. 1983. Validation of age determination estimates: the forgot- ten requirement. U.S. Dep. Commer. NOAA Tech. Rep. NMFS 8:29-33. Beckett, J. S. 1974. Biology of swordfish, Xiphias gladius L., in the North- west Atlantic Ocean. U. S. Dep. Commer, NOAA Tech. Rep. NMFS SSRF-675: 103-106. Beckman, D. W., A. L. Stanley, J. H. Render, and C. A. Wilson, 1990. Age and growth of black drum in Louisiana waters of the Gulf of Mexico. Trans. Am. Fish. Soc. 119:537-544. 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. U. S. Dep. Commer, NOAA Tech. Rep. NMFS 8: 137-143. Bernard. D. R. 1981. Multivariate analysis as a means of comparing growth in fish. Can. J. Fish. Aquat. Sci. 38:233-236. Brothers, E. B. 1983. Summary of round table discussions on age validation. IJ. S. Dep. Commer, NOAA Tech. Rep. NMFS 8:35-44. Campana. S. E. 2001. Accuracy, precision and quality control in age deter- 834 Fishery Bulletin 100(4) mination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Canipana, S. E., M. C. Annand, and J. I. McMillan. 1995. Graphical and statistical methods for determining the consistency of age determinations. Tran. Am. Fish. Soc. 124:131-138. Castro-Longoria, R., and O. Sosa-Nishizaki. 1998. Age determination of .swordfish. Xiphias gladius L., ft-om waters off Baja California, Mexico, using anal fin rays and otoliths. U. S. Dep. Commer., NOAATech. Rep. NMFS 142:231-257. Caton, A., K. Colgan, R Sahlqvist, R Ward. C. Ramirez, and M. Scott. 1998. Swordfish, Xiphias gladius, and the fisheries for tunas and billfishes in the Australian fishing zone. U. S. Dep. Commer, NOAATech. Rep. NMFS 142:11-35. Chen, Y., D. A Jackson, and H H. Harvey. 1992. A comparison of von Bertalanffy and polynomial func- tions in modelling fish growth data. Can. J. Fish. Aquat. Sci. 49:1228-1235. Chow, S. 1998. Genetic comparison of Pacific and Mediterranean swordfish, Xiphias gladius. by RFLP analysis of the mito- chondrial D-loop region. U. S. Dep. Commer, NOAA Tech. Rep. NMFS 142:239-244. Chow, S., H. Okamoto, Y. Uozumi, Y. Takeuchi, and H. Takeyama. 1997. Genetic stock structure of the swordfish {Xiphias gla- dius) inferred by PCR-RFLP analysis of the mitochondrial DNA control region. Mar. Biol. 127:359-367. Compean-Jimenez, G., and F. X. Bard. 1983. Growth increments on dorsal spines of eastern Atlan- tic bluefin tuna, Thunnus thynnus, and their possible rela- tion to migration patterns. U. S. Dep. Commer, NOAA Tech. Rep. NMFS 8:77-86. DeMartini, E. E., J. H. Uchiyama, and H. A. Williams. 2000. Se.xual maturity, sex ratio, and size composition of swordfish, Xiphias gladius. caught by the Hawaii-based pelagic longline fishery Fish. Bull. 98:489-506. Ehrhardt, N. M. 1992. Age and growth of swordfish, Xiphias gladius. in the northwestern Atlantic. Bull. Mar Sci. 50(2): 292-301. Ehrhardt, N. M., R. J. Robbins. and F. Arocha. 1996. Age validation and growth of swordfish, Xiphias gladius. in the northwest Atlantic. ICCAT (International Commission for the Conservation of Tunas) Col. Vol. Sci. Pap. 45(2):358-367. Esteves, E., P. Simoes, H. M. Da Silva, and J. P. Andrade. 1995. Ageing of swordfish, Xiphias gladius Linnaeus, 1758, from the Azores, using sagittae, anal-fin spine and vertebrae. Bull. Univ. Azores, Life and Marine Sciences 13A:39-51. Ferreira. B. P., and G. R. Russ. 1994. Age validation and estimation of growth rates of the coral trout, Plectroponius leopardus (Lacepede 1802), from Lizard Island, Northern Great Barrier Reef Fish Bull, 92:46-57. Franks, J. S., J. R. Warren, and M. V. Buchanan. 1999. Age and growth of cobia, Rachycentron canadum. from the northeastern Gulf of Mexico. Fish. Bull. 97:459-471. Eraser, C. Mel. 1916. Growth of the spring salmon. Trans. Pacific Fish. Soc. 1915, p. 29-39. Friedman, M. 1937. The use of ranks to avoid the assumption of normality implicit in the analysis of variance. J. Am. Stat. Assoc. 32: 67.5-701. 1940. A comparison of alternate tests of significance for the problems of m rankings. Ann. Math. Stat. 11: 8(>-92. Gonzalez-Garces, A., and A.C. Farina-Perez. 1983. Determining age of young albacore, Thunnus alalunga, using dorsal spines. U. S. Dep. Commer, NOAA Tech. Rep. NMFS 8:117-122. Grijalva-Chon, J.M., K. Numachi, O. Sosa-Nishizaki, and J. de la Rosa-Velez. 1994. Mitochondrial DNA analysis of North Pacific sword- fish (Xiphias gladius) population structure. Mar. Ecol. Prog. Ser. 115:15-19. Hoey, J. 1991. Sex ratio data for western North Atlantic swordfish. ICCAT Coll. Vol. Sci. Pap. 34(2):429-436. Kendall, M. G. 1962. Rank correlation methods, 3'''' ed, 199 p. Charles Griffin, Condon. Media Cybernetics. 1997. Image-Pro Plus, version 3.0 for Windows, 480 p. Media Cybernetics, Silver Spring, MD. Mejuto, J., B. Garcia, and M. Quintans. 1991. A preliminary analysis of the sex-ratio of the sword- fish tXiphias gladius) in the north Atlantic by size class using space-time strata. ICCAT Coll. Vol. Sci. Pap. 35(2): 473-481. Mejuto, J.. J. M. de la Serna, and B. Garcia. 1995. An overview of the sex-ratio at size of the swordfish (Xiphias gladius L.) around the world: similarity between different strata. ICCAT Coll. Vol. Sci. Pap. 44(3):197-205. Nelson, R. S., and C. S. Manooch. 1982. Growth and mortality of red snappers in the west-cen- tral Atlantic Ocean and northern Gulf of Mexico. Trans. Am. Fish. Soc. 111:465-475. Ovchinnikov, V. V. 197 1 , Swordfishes and billfishes in the Atlantic Ocean, ecol- ogy and functional morphology, 77 p. Israel Program for Scientific Translations. U.S. Dep. Commer., NOAA NMFS TT 71-50011. Powers, J. E. 1983. Some statistical characteristics of ageing data and their ramifications in population analysis of oceanic pe- lagic fishes. U. S. Dep. Commer, NOAA Tech. Rep. NMFS 8:19-24. Prince, E. D., D. W. Lee. and S. A. Berkeley 1988. Use of marginal increment analysis to validate the ana! spine method for ageing Atlantic swordfish and other alternatives for age determination. ICCAT Coll. Vol. Sci. Pap. 27:194-201. Prince, E. D., D. W. Lee, J. L. Cort, G. A. McFariane, and A. Wild. 1995. Age validation evidence for two tag-recaptured Atlan- tic albacore, Thunnus alalunga, based on dorsal, anal, and pectoral finrays, vertebrae, and otoliths. In Recent devel- opments in fish otolith research (D. H. Secor. J. M. Dean, and S. E. Campana (eds.), p. 375-396. Univ of South Carolina Press, Columbia. SC. Radtke, R. L.. and P C. F Hurley 1983. Age estimation and growth of broadbiU swordfish, Xiphias gladius, from the northwest Atlantic based on external features of otoliths. U. S. Dep. Commer, NOAA Tech. Rep. NMFS 8:145-1.50. Reeb, C. A., L. Arcangeli, and B A. Block. 2000. Structure and migration corridors in Pacific popu- lations of the swordfish Xiphias gladius, as inferred through analyses of mitochondrial DNA. Mar. Biol. 136: 123-1131. Sun et a\ Age and growth of Xiphios glodius 835 Richards. F. J. 1959. A flexible growth Function for empirical use. J. Exp. Bot. 10:290-300. Rosel.P. E.,an(i H. A. Block. 1996. Mitochondrial control rctjion variability and global population structure in the swordfish. Xipluas gladius. Mar Biol. 125:11-22. SAS Institute. 1990. SAS/STAT user's guide, version 6, fourth ed., p. 1 135- 1 194. SAS Institute Inc., Gary, NC. Sakagawa, G. T. 1989. Trends in fisheries for swordfish in the Pacific Ocean. In Proceedings of the second international billfish sympo- sium; Hawaii, August 15, 1988 (R. H.Stroud, ed.), p. 61-79. National Coalition for Marine Conservation, Inc., Savan- nah, GE. Stone, H. H., and J. M. Porter 1997. Development of a swordfish sex-ratio-at-size relation- ship for catches from the Canadian fishery. ICCAT Col. Vol. Sci. Pap. 46(3):31 1-314. Sturm. M. G. de L., and P. Salter. 1990. Age, growth, and reproduction of the king mackerel, Scombcromorus cai'alla (Cuvier) in Trinidad waters. Fish. Bull. 88:361-370. Sun, C. L., C. L. Huang, and S. Z. Yeh. 2001. Age and growth of the bigeye tuna Thiinnus obesus in the western Pacific Ocean. Fish. Bull. 99: 502-509. Tserpes, G., and N. Tsimenides. 1995. Determination of age and growth of swordfish, X;p/i- ias gladius L., 1758, in the eastern Mediterranean using anal-fin spines. Fish. Bull. 93:594-602. Tsimenides, N., and G. T.serpes. 1989. Age determination and growth of .swordfish, Xiphias gladius L., 1758, in the Aegean Sea. Fish. Res. 8:159- 168. Turner, S. C. P Arocha. and G. P Score. 1996. U.S. swordfish catch at age by sex. ICCAT Col. Vol. Sci. Pap. 45(2):373-378. Uchiyama, J. H., R. A. Skillman, J.D. Sampaga, and E. E DeMartini. 1998. A preliminary assessment of the use of hard parts to age central Pacific .swordfish, Xiphias gladius. U. S. Dep. Commen, NOAATech. Rep. NMFS 142:261-273. Ward, F, and S. Elscot. 2000. Broadbill swordfish: status of world fisheries, 208 p. Bureau of Rural Sciences, Canberra. Wilson, C. A., and J. M. Dean. 1983. The potential use of sagittae for estimating age of Atlantic swordfish, Xiphias gladius. U. S. Dep. Commer., NOAATech. Rep. NMFS 8:151-156. Yabe. H., S. Ueyanagi, S. Kikawa, and H. Watanabe. 1959. Study on the life-history of the swordfish, Xip/jias^/a- dius Linnaeus. Rep. Nankai Reg. Fish. Res. Lab. 10:107- 150 Zar, J. H. 1999. Biostatistical analysis. 4th ed., 929 p. Prentice-Hall Inc., Englewood Cliffs, NJ. 836 Abstract-The rockfishes of the se- bastid genus Sebastes are a very impor- tant fishery resource off the coasts of CaHfomia and southern Oregon. How- ever, many of the 54 managed stocks of west coast rockfish have recently reached historically low population levels, leading fishery managers to re- examine current management prac- tices. Management of rockfish stocks as multispecies aggregates, as opposed to independent stocks within the ground- fish fishery, can be more desirable when nontargeted bycatch. discard, and man- agement complexity are considered. Rockfish assemblage structure and species co-occurrences were determined by using data from the Alaska Fisher- ies Science Center triennial continental shelf bottom trawl survey. The weight of rockfish species in trawl catches was expres.sed as a catch-per-unit-of-effort (CPUE) statistic, from which species spatial distributions, overlaps, diver- sity, and richness were analyzed. Mul- tidimensional scaling of transformed CPUE data was employed in indirect gradient and multivariate partition- ing analyses to quantify assemblage relationships. Results indicated that rockfish distributions closely match the bathymetry of coastal waters. Indirect gradient analysis suggested that depth and latitude are the principal factors in structuring the spatial distributions of rockfish on trawlable habitat. In addi- tion, four assemblages were identified through the joint evaluation of species' distributions and multivariate parti- tioning analyses: 1) deep-water slope; 2) northern shelf 3) southern shelf; and 4) nearshore. The slope, shelf and near- shore groups are found in depth ranges of 200-500 m, 100-250 m, and 50-150 m, respectively. The division of north- em and southern shelf assemblages occurs over a broad area between Cape Mendocino and Monterey Canyon. The results of this analysis are likely to have direct application in the manage- ment of rockfish stocks off the coasts of southern Oregon and California. Distribution and co-occurrence of rockfishes (family: Sebastidae) over trawlable shelf and slope habitats of California and southern Oregon Erik H. Williams Stephen Ralston Southwest Fisheries Science Center National Manne Fisheries Service 1 10 Shaffer Road Santa Cruz, California 95060 Present address (for E H Williams) Southeast Fisheries Science Center National Manne Fisheries Service 101 Pivers Island Road Beaufort, North Carolina 28516-9722 E-mail address (for E H Williams) Erik, Wiiiiams^noaa gov Manuscript accepted 6 June 2002. Fish. Bull. 100:836-855 (2002). Rockfishes (family Sebastidae accord- ing to recent work by Eschnieyer (1998) and Kendall (2000). genera Sebastes and Sebastolobus) are a very important part of the groundfish fish- ery off the United States west coast, representing a relatively high value in the market. In 1997, rockfish spe- cies accounted for IT^i of total west coast groundfish landings, but 33'X of total exvessel value (Herrick et al.'). Rockfishes on the U. S. west coast are managed under the groundfish Fishery Management Plan (FMP) by the Pacific Fishery Management Council. Some of the 52 Sebastes and 2 Sebastolobus species that are included in the FMP are managed as single-species stocks, and other, generally less well-known species, are managed as part of larger multispecies aggregations. Moreover, the 54 "rockfishes" listed in the FMP do not include all of the Sebastes spp. found in the region. More than 70 species are known from the northeast Pacific Ocean (Eschmeyer et al., 1983; Chen, 1986). The current status of the principal west coast rockfish stocks is that many have reached historically low levels and the population sizes of many of the mi- nor species remain virtually unknown. Moreover, seven rockfishes (bocaccio (S. pauaspinis), cowcod (S. levis), canary (S. pinniger) darkblotched (S. crameri), widow (S. entomelas), and yelloweye (S. ruberrimus) rockfishes, and Pacific ocean perch (S. alutus) have recently been declared overfished by the Na- tional Manne Fisheries Service. The continuing declines in rockfish popula- tions and other groundfish stocks off the U. S. west coast have prompted changes in harvest policy and other management practices (Ralston, 1998, 2002). Management of exploited fish stocks on an individual basis often results in discarded bycatch of nontargeted spe- cies, which is wasteful. In contrast, management of species as aggregates or complexes can be more practical and desirable (Ralston and Polovina, 1982; Leaman and Nagtegaal, 1986; Fujita et al., 1998). However, the multispecies management approach is only as good as the assemblage or group definitions used, which depend on the availability and accuracy of species-specific distri- butional information. Most fisheries managers rely on fishery-independent surveys or at-sea observations of fish- ery catches recorded in vessel logbooks or noted by obsei'vers to provide spatial information on fish abundance pat- terns. Data collection for groundfish population assessments off the coast Herrick, S. F., J. Hastie, and W. Jacobson. 1998. Economic status of the Washing- ton, Oregon, and California groundfish fisheries. In Pacific Fishery Management Council. Status of the Pacific Coast ground- fish fishery through 1998 and recom- mended biological catches for 1999: stock assessment and fishery evaluation. Pa- cific Fishery Management Council, 2130 SW Fifth Ave., Suite 224, Portland, Oregon 97201. Williams and Ralston: Distribution and co occurrence of Sebastidae off California and Oregon 837 of California consists primarily of sampling commercial and recreational landinKs at ports, and for these landings there is little information on catch location (Pearson and Erwin, 1997; Sampson and Crone, 1997). The National Marine Fisheries Sci-vice, Alaska Fisheries Science Cen- ter (AFSC) westcoast triennial continental shelf bot- tom-trawl survey, which enters into California waters, has yielded useful spatial information for determining groundfish distributions and co-occurrences. This survey began in 1977 as a rockfish survey but changed focus in subsequent years, depending on the particular informa- tion needs at the time (Dark and Wilkins, 1994; Wilkins et al., 1998). For the eight surveys conducted from 1977 to 1998, the shelf trawl survey covered the area from central V'ancouver Island, British Columbia, to Point Conception, California, at depths ranging from 50 to 500 m (Wilkins etal.. 1998). Along the U.S. west coast, 60-65'7i^ of the groundfish catch (exclusive of Pacific whiting) is taken off the coasts of Washington and Oregon (PFMC-). This area has been the focus of past studies examining groundfish popula- tion distributions and assemblages (e.g. Gabriel and Tyler, 1980; Leaman and Nagtegaal, 1986; Rickey and Lai, 1990; Rogers and Pikitch, 1992; Weinberg, 1994; Jay, 1996; Gunderson, 1997). In these previous studies of rockfish distributions and groupings, rockfish could be broken into shelf and slope assemblages (e.g. Rogers and Pikitch, 1992; Weinberg, 1994). These studies have all indicated that along the Oregon-Washington coast a "slope" or deep-water rockfish assemblage e.xists, consist- ing of darkblotched rockfish. Pacific ocean perch, splitnose rockfish (S. diploproa), yellowmouth rockfish (S. reedi), and shortspine thornyhead iSebastolohus alascanus). In addition, a "shelf" or bottom rockfish assemblage consists of yellowtail rockfish iS. flavidus), canary rockfish (S. pinniger), sharpchin rockfish (S. zacentrus), greenstriped rockfish (S. elongatus), rosethorn rockfish (S. helvomaciila- tus), and redstripe rockfish (S. proriger). Weinberg (1994) also examined patterns in abundance and number of rock- fish species with respect to depth gradients. Both abun- dance and the number of species increased with depth to a maximum in the range of 151-250 m but both quantities decreased at depths greater than 250 m. Groundfish assemblages off California have not been studied, primarily because of the absence of detailed at-sea fishery data collection programs and because of smaller landings. Although California landings account for only 35'^7f of the nonwhiting groundfish total (PFMC-), nearly AV/c of west coast rockfish landings are taken in California waters (Herrick et al.'). Thus, an understand- ing of the distribution and co-occurrence of rockfishes off the California coast would help with efforts to implement effective fishery management actions leading to a sus- '■^ PFMC (Pacific Fishery Management Council). 2000. Status of the Pacific Coast groundfish fishery through 2000 and recommended acceptable biological catches for 2001 — stock assessment and fisherv* evaluation. Pacific Fishery Manage- ment Council. 2130 SW Fifth Ave., Suite 224. Portland. Oregon 97201. 48 46 42 40 38 36 34 Vancouver Washington Columbia Oregon Eureka Cape Blanco \ N ■"Cape Mendocino I California Monterey ^fean Franc isco Bay Monterey Bay Conception '_Pl Buchon l^Pl-Conception -132 -130 128 -126 -124 -122 -120 -118 -116 -114 Longitude Figure 1 Map of the west coast of the United States, including names of important coastal features and International North Pacific Fisheries Commission management areas. tainable California rockfish fishery. In particular, an un- derstanding of rockfish distributions and co-occurrences could lead to improved definitions of species complexes. Weinberg ( 1994) analyzed rockfish assemblages of the "Co- lumbia" and U.S. portion of the "Vancouver" fishery man- agement areas, as specified by the International North Pacific Fisheries Commission (INPFC) (Fig. 1). In part to complement Weinberg's (1994) analysis, we analyzed the AFSC continental shelf trawl survey data from the more southerly waters of the Eureka, Monterey, and Concep- tion INPFC areas, i.e. all sampling conducted south of lat. 43°N(Fig. 1). Materials and methods Rockfish likely form aggregations or complexes as a response to oceanographic and bathymetric features. For this reason the bathymetry of the region encompassed by the limits of the survey area in the Eureka, Monterey, and Conception INPFC areas was characterized. The study area is the area of marine waters located from lat. 34° to 43°N within a depth range of 50-500 m. Bathymetric data for this region were obtained from the National Ocean Service Hydrographic Data Base (NOSHDB). The 838 Fishery Bulletin 100(4) NOSHDB contains data digitized from hydrographic sur- veys completed from 1930 to 1965 and from survey data acquired digitally on NOS survey vessels since 1965. The total amount of habitat by depth and latitude, as well as the location of the shelf break, was determined by analysis of the NOSHDB. Because the NOSHDB contains a vast amount of data, we computed depth profiles only at each 0.5 intei-val of latitude. Profiles were obtained by contouring a narrow (0.05°=5.56 km) swath of depth soundings by using in- verse distance to a power for interpolations (Surfer, 1995). After correcting for latitudinal differences in the relation- ship between longitude and distance, the resulting depth profiles were used to estimate the total amount of habitat [km] in 50-m depth intervals. In addition, the distance offshore and the depth of the continental shelf break were estimated. The location of the shelf break was estimated by using a three-parameter segmented linear model that minimized the sums of squared differences between a con- toured depth profile and points along two linear sections of a segmented line. The fitted join point of the two seg- ments was then used to estimate the location of the shelf break. In the estimation procedure, the offshore end of the offshore segment was fixed at the exact 500 m depth value obtained from the computed profiles. Information on rockfish abundance was obtained from trawl survey data collected by the Resource Assessment and Conservation Engineering division of the AFSC. Trawl samples were generally collected by using a sam- pling design that was stratified by depth and latitude, and where allocation of sample sizes was based on prior fish- ery catches (Wilkins et al., 1998). From 1977 to 1998, trawl samples in the Eureka, Monterey, and Conception INPFC areas (Fig. 1) were typically taken between 50 and 500 m from June to August with a standardized Nor'eastern high-opening rockfish bottom trawl rigged with roller gear. Measurements recorded for each trawl sample were the following: trawl net width and height; time of the tow; distance traveled; and the number and weight of species in the catch (Wilkins et al., 1998). The fundamental objective of the AFSC triennial conti- nental shelf trawl survey is to estimate the distribution and abundance of fishes vulnerable to capture by bottom trawl along the U. S. west coast. This basic goal has not changed since the first year of the survey in 1977, although specific objectives have changed over time, which has resulted in alterations in the distribution of sampling effort. For example, sampling effort in 1977 was stratified by depth and latitude according to rockfish fishery information. Sampling efforts in 1980, 1983, and 1986 were shifted to improve biomass estimates of canary and yellowtail rock- fish. However, the lack of any significant improvement in the precision of the rockfish biomass estimates prompted a shift in the 1989 and 1992 surveys to include all demersal groundfish and to improve estimates of Pacific hake (Mer- hicciiis prociitctus) and juvenile sablefish (Anoplopoma fimbria) abundance. More recently in 1995 and 1998 the survey was expanded to include slope rockfish found in deeper waters (to 500 m) with an emphasis on obtaining a uniform sampling density. These changes in the goals and objectives of the AFSC shelf survey significantly altered the data on the spatial distribution of samples over time, which, in turn, confounded interannual comparisons of the spatial distributions of rockfish species. Due to changes in survey design detailed above, all years of the survey were simply pooled into a single com- posite data set, from which the starting position of each haul, depth of haul (m|, net width (m), distance towed (km), numbers of species collected, and species weights (kg) were extracted for analysis. The distribution of all hauls was analyzed by depth and latitude to reveal any patterns that might affect inferences about rockfish dis- tributions or co-occurrences. Because management is pri- marily concerned with biomass estimates of abundance, only species weights were used in our analysis. All trawl- specific species weight measurements were converted to a catch-per-unit-of-effort (CPUE) statistic by dividing species catch weight by the product of the distance towed and net width, i.e. the area swept (ha). An analysis of the frequency of occurrence of each species in trawls was con- ducted to obtain a subset of the most ubiquitous rockfishes for use in all subsequent analyses. These species were se- lected based on their occurrence in at least six of the eight survey years, with the exception of halfbanded rockfish (S. semtcinctus), which was included because it yielded mod- erately frequent catches in five of eight years (Table 1). The data representing the selected subset of species were plotted by depth and latitude to display distribu- tional patterns of CPUE. Next, interspecific distributional overlaps were computed by calculating the percentage of joint occurrences with other species based on presence or absence (Krebs, 1989). Joint occurrences were determined both on a trawl-specific basis and after catches had been aggregated into 50-m depth and 0.5°-latitude intervals. In addition, Sefeas^t'.s diversity and species richness were com- puted for each haul to summarize the overall distribution of rockfishes captured in the survey For our analysis, di- versity was computed by using the Shannon-Wiener index, and the number of species was used to scale richness (Krebs, 1989). Diversity and richness measures were then spatially contoured over depth and latitude dimensions to display spatial structure (Surfer, 1995). Distributional patterns and groupings of rockfish based on the CPUE data were analyzed by indirect gradient analysis by using multivariate ordination and partitioning methods. Multivariate analyses are often strongly infiu- enced by the choice of distance or (dis)similarity measure. Members of the set of Minkowski distance measures (e.g. Manhattan, Euclidean, maximum, etc.) tend to be strongly affected by extreme values. Moreover, species composition data from trawl surveys have a high proportion of zero catches and a distance measure that is little affected by this property is desirable. The Bray-Curtis index, also known as Czekanowski's quantitative index, is a commonly used statistic in other similar applications and is robust to the presence of zero values (see Bloom, 1981; Field et al., 1982; Krebs, 1989; Rogers and Pikitch, 1992; Weinberg, 1994; Meuter, 1999). A fourth-root transformation of the data was conducted before calculation of the Bray-Curtis index, as suggested by Field et al. (1982) and as implemented by Williams and Ralston: Distribution and co-occurrence of Sebastidae off California and Oregon 839 Table 1 Frequency of trawl samples with rockfish (family: Sebastidae) pre.sen t in the Eurek; \, Monterey, ar d (^once ption areas from 1296 | AKSC triennial shelf bottom trawl survey samples. The first 26 species {in bold) were ncluded in the detailed analyses; the remain- ing 23 were not. Common name Scientific name Survey year 1977 1980 1983 1986 1989 1992 1995 1998 Total Stripetail rockfish Sebastes saxicola l.-JS 47 53 55 60 78 114 96 658 Chilipepper Sehastes goodei 119 35 39 48 53 80 87 81 542 Splitnose rockfish Sebastes diploproa 211 31 46 16 20 34 88 95 541 Shortspine thomeyhead Sebastolohus alascancus 149 22 46 19 21 28 83 95 463 Greenstriped rockfish Sebastes elongatiis 58 23 56 57 44 57 72 75 442 Bocaccio Sebastes paucispinis 127 65 46 53 40 30 41 26 428 Shortbelly rockfish Sebastes jordani 101 38 20 56 45 59 62 42 423 Darkblotchcd rockfish Sebastes crameri 122 31 56 28 36 38 56 51 418 Aurora rockfish Sebastes aurora 83 1 2 1 56 60 203 Widow rockfish Sebastes entomelas 42 23 26 10 14 28 30 28 201 Canary rockfish Sebastes pinniger 31 21 39 22 17 17 17 34 198 Redbanded rockfish Sebastes babcocki 51 12 20 8 12 17 37 21 178 Sharpchin rockfish Sebastes zacentrus 8 13 15 14 16 32 22 23 143 Bank rockfish Sebastes rufus 57 9 8 3 3 7 32 8 127 Yellowtail rockfish Sebastes flavidus 17 10 26 14 8 18 19 10 122 Pacific ocean perch Sebastes alutus 24 13 22 5 5 14 17 20 120 Blackgill rockfish Sebastes melanostomus 49 1 3 1 30 35 119 Halfbanded rockfish Sebastes semicinctus 6 13 40 31 27 117 Greenspotted rockfish Sebastes chlorostictus 23 8 6 16 14 6 24 15 112 Cowcod Sebastes lei'is 11 2 4 0 14 2 22 12 67 Rosethorn rockfish Sebastes helvomaculatus 14 3 7 4 4 7 11 12 62 Brown rockfish Sebastes auriculatus 3 5 7 6 3 4 1 29 Copper rockfish Sebastes caunnus 2 1 5 2 6 4 5 4 29 Vermilion rockfish Sebastes miniatus 2 1 6 7 3 1 20 Redstripe rockfish Sebastes proriger 3 3 2 2 2 7 19 Yelloweye rockfish Sebastes ruberrimus 2 2 3 5 3 1 1 17 Longspine thorneyhead Sebastolobus altivelis 1 19 24 44 Rougheye rockfish Sebastes aleutianus 7 1 7 12 28 Unidentified rockfish Sebastes sp. 3 5 1 4 5 3 23 Flag rockfish Sebastes rubrivinctis 20 1 1 23 Tiger rockfish Sebastes nigrocinctus 2 10 12 Pygmy rockfish Sebastes wilsoni 2 3 2 1 3 12 Squarespot rockfish Sebastes hopkinsi 5 4 1 11 Greenblotched rockfish Sebastes rosenblatti 1 4 1 3 11 Speckled rockfish Sebastes ovalis 2 1 1 1 7 Shortraker rockfish Sebastes borealis 2 1 1 2 7 Blue rockfish Sebastes mystinus 2 1 5 Pink rockfish Sebastes eos 2 1 4 Black rockfish Sebastes melanops 1 1 1 4 Olive rockfish Sebastes serranoides 1 2 3 Silvergray rockfish Sebastes brevispinis 2 2 Yellowmouth rockfish Sebastes reedi 2 2 California scorpionfish Scorpaena guttata 1 1 2 Starry rockfish Sebastes constellatus 1 Calico rockfish Sebastes dalli 1 Freckled rockfish Sebastes lentiginosus 1 Rosy rockfish Sebastes rosaceus 1 Harlequin rockfish Sebastes variegatus 1 Chameleon rockfish Sebastes ph illipsi 1 840 Fishery Bulletin 100(4) E < Depth (m) ^ ^ Estimated amount 50-500 m. Meuter (1999). A different measure of similarity, based on converting the data to presence-absence binary form and computing the proportion of nonzero values held in common, has also been used (Krebs, 1989). However, preliminary analyses indi- cated very little difference between the Bray-Curtis and binary similar- ity measures (r=0.973). Therefore, the Bray-Curtis similarity measure was used in all subsequent multi- variate analyses. Ordination techniques used in in- direct gradient analysis were prin- cipal components, detrended corre- spondence analysis, and multidimen- sional scaling (Rogers and Pikitch, 1992; Mahon et al, 1998; Meuter, 1999). Multidimensional scaling (MDS) has the advantage of not requiring an assumption about the underlying response model and has been shown to be robust to different relationships of species abundances and environmental gi-adients (Minchin, 1987; Meuter, 1999). In our study we used MDS in conjunction with the Bray-Curtis similarity measure, similar to previous analyses (Field et al., 1982; Meuter, 1999). The first three dimensions from the MDS were extracted, plotted, and correlated with suspected gradients, which were then used to numerically define as- semblage groupings in the data. Species groupings or assemblages were also determined by partitioning cluster analysis, rather than using other commonly employed hierarchical cluster analyses (e.g. Rogers and Pikitch, 1992; Weinberg, 1994; Mahon et al, 1998). Hierarchical cluster techniques result in dendro- gram trees whose shape and structure depend largely on the division and linking methods used, frequently result- ing in little similarity among the many methods (Johnson and Wichern, 1992; Ripley, 1996). Results from this type of analysis are often reported for a single method, indicating results consistent with the analyst's expectations, while neglecting to report the range of variability produced by the other alternative hierarchical methods. A more objec- tive method of determining groupings is to use a partition- ing technique such as the classical ^-means algorithm (Hartigan and Wong, 1979). where the number of groups (/;) is specified a priori and a single solution to the group- ing structure is determined. For our partitioning analysis we used a more robust variant of the ^-means method, which is termed ^-me- dians (Kaufman and Rousseeuw, 1990). The /?-means and ^-medians methods rely on the minimization of the (dis)similarity between cluster centers and their mem- bers. Specifically, the /f-means algorithm minimizes the squared (dis)similarities, and /f-medians minimizes the untransformed (dis)similarities, resulting in a measure that is less sensitive to extreme values (Kaufman and Rousseeuw, 1990; Ripley, 1996). In our study a range of ^'s >j?> o Latitude Figure 2 of habitat (km) at half degree latitude intervals and depths of or cluster numbers was evaluated and the best grouping structure for the data was determined according to the highest average silhouette measure (Rousseeuw, 1987). To detect misclassifications, the final rockfish assem- blage structure was compared to the several dimensions obtained from the MDS ordination analysis and to our mapped CPUE distributions. Results There are a number of large-scale bathymetric features in the Eureka, Monterey, and Conception INPFC areas that may influence the distribution and abundance of shelf and slope rockfish species. For example, the Mendocino Escarp- ment is a large fracture zone that forms a huge submarine ridge near Cape Mendocino that extends nearly 2500 km westward into the Pacific Ocean and that measures 100 km across at its widest point. Well to the south of the escarpment are a number of large submarine canyons in the region of Monterey Bay and Point Sur These subma- rine features, including Monterey and Sur Canyons, result in a coastal bathymetry characterized by limited shelf area and rapidly increasing depth. Farther south, in the vicinity of Point Buchon, is a large offshore area that rises to 430 m depth, i.e. the Santa Lucia Bank. Lastly, Point Conception divides zoogeographic provinces and forms the southern boundary of the area of this study. Our restricted analysis of the bathymetry at 0.5°-lati- tude intervals indicates that variability in depth profiles reflects some of the important features described above. The amount of habitat in each 50-m depth interval seems to peak in the 100 to 150-m depth range for latitudes greater than 37.0°N (Fig. 2). The observed decrease in the total amount of habitat at 40.0°N, especially in waters shallower than 200 m, is due primarily to the Mendocino Williams and Ralston; Distribution and co-occurrence of Sebastidae off California and Oregon 841 43 42 41 40 2 39 35 34 O o '®- ^ ° °^^r? «> ° ° ^ « O ,. ... oogW" oO o O „ '^ ® o „ 0)0° °°„ =<»=^ b'^O > 0 o oO o o 3 . 100 200 300 Depth (m) 400 500 Figure 3 Plot of total rockfish CPUE estimates by depth and latitude from the AFSC triennial bottom trawl survey. Circle diameters are proportional to the square root of total rockfish CPUE. Vertical bold lines represent an esti- mate of the shelf break (see text). Escarpment. Likewise, the abrupt decrease in the amount of shallow depth habitat from 35.5° to 36.5°N is directly attributable to the occurrence of submarine canyons, and the increase in habitat from 450 to 500 m at 35.0°N is due to the Santa Lucia Bank (Fig. 2). The depth at which the shelf break occurs seems to be fairly constant, generally ranging from 100 to 175 m (Fig. 3). The relatively deep (210 ml estimated shelf break at 36.5 N is due to the pres- ence of Monterey Canyon (Fig. 3). For the combined Eureka, Monterey, and Conception IN- PFC areas, the shelf trawl survey database pooled over the 1977-98 period totaled 1296 hauls that together captured 49 rockfish species, including an unidentified category (Ta- ble 1 ). Based on the frequency of positive trawl samples for each of the rockfish species, 26 were selected for detailed analysis (Table 1). In Table 1, the effect of changing survey sampling objectives is evident in the interannual variation in frequency of occurrence for some of the species. Perhaps most noticeable is the abrupt decline in samples of aurora (S. aurora ) and blackgill (S. melanostomus) rockfish in the years 1980-92 (Table 1). Both are deep-water species and, clearly, the relatively high frequencies of occurrence in 1977, 1995, and 1998 were due to increased sampling at deeper depths that resulted from the altered objectives of the sampling design discussed earlier The distribution of trawl sampling locations and CPUE by depth and latitude indicated that the sampling pat- tern followed some of the bathymetric features mentioned above (Fig. 3). One notable feature was the paucity of samples in the 36.5°N latitude region, which was partly 842 Fishery Bulletin 100(4) Table 2 Percentage overlap of rockfish taken in trawls containing the species captured in the AFSC triennial shelf bottom trawl species listed in the columns. POP = Pacific ocean perch. survey by smoothed depth (50-500 m) and stripetail chili- pepper split- nose short- spine green- striped bocaccio short- belly dark- blotched aurora widow canary stripetail 100 52 37 4 3 30 39 9 2 7 4 chilipepper 57 100 25 3 3 29 33 7 0 7 5 splitnose 19 12 100 7 1 8 14 12 4 1 1 shortspine 17 11 61 100 3 11 7 35 47 6 3 greenstriped 94 85 39 24 100 79 68 52 2 57 60 bocaccio 78 69 41 6 6 100 41 13 1 16 8 shortbelly 50 39 36 2 3 20 100 7 0 5 4 darkblotched 36 25 89 30 6 20 21 100 7 11 6 aurora 9 2 42 53 0 1 1 10 100 1 0 widow 84 88 36 15 23 80 49 35 1 100 36 canary 51 60 18 9 25 42 43 21 1 38 100 redbanded 52 36 99 89 14 42 32 92 30 23 15 sharpchin 88 80 52 17 18 76 57 48 1 41 25 bank 31 30 76 21 3 27 26 56 8 8 5 yellowtail 20 21 5 4 17 21 10 8 0 17 41 POP 44 11 91 64 13 28 6 98 8 17 8 blackgiU 18 8 75 63 1 7 5 20 83 3 2 haltbanded 50 99 31 1 11 51 97 1 0 4 7 greenspotted 98 97 65 30 75 96 84 55 3 72 64 cowcod 100 96 97 30 20 95 93 56 7 37 21 rosethorn 81 80 95 60 40 93 61 94 12 60 36 brown 14 57 18 3 20 91 12 2 0 35 54 copper 24 19 16 5 13 33 15 2 0 15 72 vermilion 71 83 47 23 21 79 65 21 7 28 36 redstripe 21 17 18 15 21 17 21 16 1 19 24 yelloweye 97 96 80 37 52 89 72 76 2 91 57 due to the precipitous Big Sur coastline, with its marked reduction in the amount of trawlable habitat in the 50-500 m depth range. Morever, the 1980, 1983, and 1986 surveys did not sample south of Monterey Bay. Similarly, the increase in the number of samples in the 37.0-39.0°N area (Fig. 3) was attributable to an increase in the total amount of shelf habitat in that region (Fig. 2). Starting at the shallowest depths, there was a tendency for the size of the total rockfish catch to increase to a maximum in deeper waters just beyond the shelf break (-200 m), fol- lowed by a slight decrease in catch in the deepest waters (Fig. 3). Not apparent to the eye in Figure 3 is a slight, but significant trend in the distribution of sample locations towards deeper waters at more southerly sites (linear slope=-0.26, P-value=0.0041 ). This was probably due to the decreasing amount of shelf habitat and the increas- ing quantity of slope habitat as one moves south along the California coast (Fig. 2). The amount of interspecific overlap in spatial distribu- tions among the subset of 26 species differed, depending on whether co-occurrence was assessed from a categori- zation of the catches into depth-latitude intervals or by specific trawl locations (Tables 2 and 3). As expected, depth-latitude overlaps were greater than the site-specific trawl catch overlaps, due to the effect of spatial smoothing. Inspection of the depth-latitude species overlaps (Table 2) indicated that stripetail (S. saxicola), splitnose, chilipep- per (S. goodei), bocaccio, and shortbelly iS. jordani) rock- fish were most widespread with respect to co-occurrence with other rockfishes. Similarly, overlaps measured from actual trawl catches indicated that bocaccio, chilipepper, stripetail, canary, and widow rockfish have relatively high likelihoods of co-occurring with other rockfish species (Table 3). The spatial distribution of rockfish over the continental shelf and slope generally indicated a ridge of increased di- versity at approximately 250 m depth, at least for samples taken at northern latitudes (38.0-43.0°N) (Fig. 4). In contrast, for southern latitudes (34.0-38.0°N), the ridge of high diversity veered well offshore to a depth of 450 Williams and Ralston: Distribution and co occurrence of Sebastidae off California and Oregon 843 Table 2 latitude (34.0-43.0°N)l(K ations. Individual table ( elements represent the percentage of occurrence of a species (reading across the row ) rod- sharp- yellow- black- half- green- cow- rose- red- yellow- banded chin bank lail POP gill banded spotted cod thorn brown copper vermilion stripe eye 1 7 5 2 2 1 1 1 2 1 0 0 0 0 0 1 7 5 2 0 1 2 1 2 1 0 0 0 0 0 1 2 6 0 2 3 0 0 1 0 0 0 0 0 0 12 6 16 2 11 23 0 1 2 2 0 0 1 1 1 13 47 14 45 15 3 6 12 11 9 0 2 3 7 8 3 16 11 4 3 1 2 1 4 2 0 0 1 0 1 1 6 5 1 0 1 2 1 2 1 0 0 0 0 0 10 15 35 2 14 6 0 1 4 3 0 0 0 1 1 4 0 7 0 2 33 0 0 1 0 0 0 0 0 0 8 42 17 17 8 2 1 5 8 5 0 1 2 2 6 5 27 10 44 4 2 1 4 5 3 0 4 2 3 4 100 23 47 5 28 33 1 5 8 12 0 0 4 3 3 8 100 26 8 13 1 0 3 8 7 0 0 1 4 5 8 13 100 0 4 10 0 1 6 3 0 0 1 0 2 2 8 1 100 3 0 1 2 1 1 1 2 1 6 2 21 28 16 6 100 2 0 1 0 4 0 0 0 3 4 12 2 21 0 1 100 0 1 2 2 0 0 1 0 0 2 2 2 6 0 0 100 4 4 0 0 3 14 0 0 26 54 43 33 6 11 13 100 44 21 0 4 16 3 18 14 38 57 3 1 11 4 14 100 20 0 1 4 0 2 49 80 61 15 19 17 1 16 48 100 0 0 3 9 12 5 3 1 76 0 0 7 1 6 0 100 64 34 0 0 1 2 3 36 0 0 13 5 2 1 10 100 22 0 0 22 12 26 24 3 20 44 17 15 4 5 18 100 1 1 7 29 7 46 12 1 0 2 1 6 0 0 0 100 13 19 91 68 27 30 3 1 20 8 17 0 0 1 29 100 m, and there was some indication of a secondary increase in diversity at 150 m for the most southerly latitudes 34.0-36.0°N (Fig. 41. The distribution of species richness indicated that the existence of a distinct ridge at depths of 200-250 m — the highest portion of the ridge occurring between 36.0 and 39.0°N latitude (Fig. 5). In that region, in excess of eight distinct species co-occurred in individual trawl samples. A solitary peak in richness occurred at a depth of 500 m near 37.5°N latitude (Fig. 5). Inspection of the raw data revealed that this peak was heavily influ- enced by a single trawl sample, but there was no indica- tion of a data recording error in this sample. The spatial distribution of CPUE for each of the rockfish species is shown in Figures 6-10. The particular sequence of species in these figures corresponded to clustering re- sults that are presented below. Careful examination of these 26 distributions revealed that the depth distribu- tions of almost all species were not related to latitude. The sole exception to this generalization was the depth distribution for shortspine thornyhead, which showed a significant interaction between depth of capture and lati- tude, based on results from a two-way factorial ANOVA. For that species, depth distribution shifts into deep water at more southerly latitudes (Fig. 6). Latitudinal boundaries of rockfish distributions ap- peared to be influenced by two of the main bathymetric features on the U.S. west coast. In particular, the Men- docino Escarpment (ME), located at approximately 40.4°N latitude, and Monterey Canyon (MC), located near 36.8°N latitude, appear to form distributional impediments for some of the species, and other species appeared to be dis- tributed more uniformly across the entire latitudinal range of the study (Figs. 1, 6-10). Examples of species whose distributional boundaries appeared to be influenced by the ME were blackgill rockfish (S. melanostomus), Pacific ocean perch, chilipepper (S. goodei), shortbelly rockfish iS. jordani), bocaccio, and greenspotted rockfish. Species whose distributions appeared to border on MC were dark- blotched, greenstriped, canary, yellowtail, widow, sharp- chin, and rosethorn rockfish (Figs. 6-10). 844 Fishery Bulletin 100(4) Table 3 Percentage overlap of rockfish species capt ured in the AFSC triennial shelf bottom trawl survey. Indiv dual table elements represent Pacific ocean perch. chill- split- short- green- short- dark- stripetail pepper nose spine striped bocaccio belly blotched aurora widow canary stripetail 100 35 11 1 3 9 12 5 0 3 1 chilipepper 29 100 4 1 3 10 8 2 0 6 2 splitnose 10 4 100 4 0 3 3 9 1 1 0 shortspine 12 8 50 100 3 7 3 30 27 3 2 greenstriped 48 57 7 4 100 26 27 9 0 18 15 bocaccio 23 31 8 2 4 100 8 7 0 8 6 shortbelly 9 7 3 0 1 2 100 1 0 1 1 darkblotched 27 15 57 14 3 15 6 100 1 3 2 aurora 3 0 24 38 0 0 0 4 100 0 0 widow 14 29 3 1 5 14 6 3 0 100 6 canary 9 22 2 1 7 19 7 3 0 11 100 redbanded 37 21 61 53 5 33 13 51 8 7 3 sharpchin 38 47 14 4 9 46 24 10 0 20 7 bank 11 11 48 7 1 9 3 39 1 3 1 yellowtail 4 16 0 0 6 12 1 1 0 20 18 POP 25 2 44 28 6 25 4 72 1 9 2 blackgill 3 2 41 32 1 2 2 9 45 0 0 halfbanded 2 34 0 0 3 5 18 0 0 1 3 greenspot 41 72 20 6 55 52 35 5 0 32 44 cowcod 64 39 73 8 10 93 21 10 2 10 8 rosethorn 26 41 37 32 37 56 29 36 2 32 20 brown 0 2 0 0 0 47 1 0 0 56 36 copper 1 4 0 0 3 23 2 0 0 11 73 vermilion 6 14 3 1 1 31 6 2 1 11 19 redstripe 4 8 0 0 4 8 15 0 0 8 4 yelloweye 16 72 10 2 34 63 18 12 0 63 75 A general comparison of interspecific relationships, based on depth-latitude distributions, can be gathered by inspection of a plot of CPUE-weighted depth-latitude cen- troids (Fig. 11). From results presented in Figure 11, some species (e.g. copper \S. cam-inus] and brown [S. auricula- tus] rockfish) appear to have very similar depth-latitude distributions. Aurora and blackgill rockfish are clearly deep-water southern species, whereas halfbanded (S. semi- cinctus) and vermilion (S. miniatus) rockfish are southern species found principally in shallow water. Although this plot is useful for identifying related species by their aver- age distribution in space, it does not fully represent actual joint co-occurrences and assemblage relationships, as does a complete multivariate community analysis. The results of the multidimensional scaling of the fourth-root transformed CPUE data, with the Bray-Cur- tis similarity measure, revealed highly significant cor- relations of the first three dimensions with mean depth (7-0.90, P<0.0001), total CPUE"-'^ (r=-0.82, P<0.0001), and mean latitude (r=-0.74, P<0.0001), respectively. How- ever, these first three dimensions accounted for only 39% of the total variance in the data. The ^-medians partitioning analysis was used to exam- ine a range of ^'s or cluster numbers, effectively predefin- ing the number of distinct assemblages. The best fits were determined by the highest average silhouette measures of 0.143 and 0.140 for /; = 8 and ^ = 4, respectively (Table 4). According to Kaufman and Rouseeuw (1990) an average silhouette measure less than 0.25 does not indicate any substantial structure in the data. Despite the low aver- age silhouette measurements, the suggested gi-oupings of four and eight clusters were used as an initial guideline in determining distinct species assemblages. The species groups defined in Table 4 by a division into four clusters (A, B, C, and D), and further delineated in Figure 11, show that these assemblages generally follow depth and the latitude distributions of their member species. The first three dimensions from the multidimensional scaling analysis, along with the k = 8 cluster divisions, are shown in Figure 12. The eight clusters comprising the 26 Williams and Ralston; Distnbution and co occurrence of Sebastidae off California and Oregon 845 Table 3 the percentage ol occurrence of a species (reading across the row) tak >n in trawls containing the species listed in th ? columns: .POP = red- sharp- yellow- black- half- green- cow- rose- red- yellow- banded chin bank tail POP gill banded spotted cod thorn brown copper vermilion stripe eye 1 3 1 1 1 0 0 0 1 0 0 0 0 0 0 0 4 1 2 0 0 1 1 0 0 0 0 0 0 0 1 1 4 0 1 1 0 0 1 0 0 0 0 0 0 11 4 7 1 9 10 0 1 1 2 0 0 0 0 0 1 13 1 15 3 0 1 9 2 4 0 0 0 3 2 1 10 2 5 2 0 0 1 2 1 1 0 1 1 1 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 5 5 19 1 11 1 0 0 1 1 0 0 0 0 0 2 0 2 0 1 21 0 0 0 0 0 0 0 0 0 1 8 1 14 1 0 0 2 0 1 2 0 1 1 1 0 5 1 21 0 0 0 4 1 1 2 4 2 1 3 100 12 15 0 17 5 0 3 2 6 0 0 1 0 0 2 100 9 1 3 0 0 2 1 3 0 0 0 9 3 3 10 100 0 2 2 0 1 2 1 0 0 0 0 0 0 0 0 100 0 0 0 2 0 0 3 1 1 0 0 10 9 8 3 100 1 0 0 0 1 0 0 0 1 0 3 0 6 0 1 100 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 100 4 1 0 0 1 3 0 0 5 21 11 29 1 1 7 100 7 4 0 1 2 2 10 3 10 22 0 0 1 1 8 100 11 0 0 1 0 2 17 47 14 9 6 3 0 8 18 100 1 1 1 23 15 0 0 0 71 0 0 0 0 0 1 100 13 12 0 0 0 0 0 24 0 0 2 1 0 1 11 100 25 0 0 1 1 2 13 0 0 6 2 1 1 8 19 100 0 0 0 20 0 0 0 0 0 1 0 3 0 0 0 100 5 0 68 0 13 2 0 0 24 5 20 0 0 1 42 100 species and listed in Table 4 closely follow the ordination plots. In particular, dimensions one and three in combination showed clear segregation of the eight distinct clusters (lower panel in Fig. 12). Clearly, the A- 1 group in Table 4 stands out as a distinct assemblage in the MDS ordinations (Fig. 12). Because dimensions one and three were highly correlated with depth and latitude, respectively, we concluded that the species groups are defined by depth and latitude. However, the low amount of variance explained by the dimensions, the small sample sizes for some species, and the relatively low average silhouette measures in the partitioning analysis, would indicate that some caution should be exercised when using these results. In fact, there appeared to be some dis- crepancies in depth-latitude distributions and partitioning groups for the rockfish species (Table 4, Fig. 11). Group A-1 in Table 4 represents the deep-water slope species of rockfish, and the species in groups B-2 and B-3 represent the nearshore species of rockfish. The separa- tion of halfbanded rockfish from the nearshore group sug- gested by the ^ = 8 clustering was likely the result of a more southerly distribution for this species than that for the other members of the group (Fig. 7). The remaining C and D groups in Table 4 represent shelf species, group C being a southern shelf group and D representing a north- ern shelf group. A likely misclassification in the clustering results was the inclusion of greenspotted rockfish in the northern shelf assemblage, which was not warranted by the distribution of catches shown in Figure 9. Within the C group, the separation of cowcod as an isolate in the k = 8 analysis was due to this species' relatively deep distribu- tion (Fig. 8). The separation of canary and yellowtail rock- fish as a distinct cluster was the result of their relatively northern distribution and their inclusion in the southern shelf species group by the partitioning analysis was prob- ably the result of a boundary effect (Fig. 9). Lastly, the division of the northern shelf group into D-7 and D-8 sub- groups, suggested by the k = 8 partitioning, was likely the result of differences in the survey's total catch of rockfish (MDS dimension 2), which is not evident in distributional patterns (Table 4). 846 Fishery Bulletin 100(4) 43 42 41 40 5 39 38 37 36- 35 1 \ J j i o \ 1 ^^miiiiijmiiaaio' o w \ ■ 1 0 - ', v^ "^ % \ J - ^^x^ ^^ [ 1 o • 1 5 , ji ' / \ 1 '' \ a \ 1 .O / '^ 1 \ C ^ 1 100 200 300 400 Depth (m) 500 Figure 4 Contour map of Shannon-Weiner diversity indices from the AFSC triennial bottom trawl survey for all samples contain- ing rockfish species (abundance is measured by biomassl. Separation of rockfish species into four and eight groups, lab of AFSC triennial bottom trawl surveys from 1977 to 1998. Table 4 ?led by letters and numbers, respectively, based on /f-medians analysis Group Species Group Species A-1 B-2 B-.3 blackgill rockfish, aurora rockfish, shortspine thurnyhead, bank rockfish, darkblotched rockfish. Pacific ocean perch, redbanded rockfish. splitnose rockfish copper rockfish, vermilion rockfish. brown rockfish halfbandcd rockfish C-4 C-5 C-6 D-7 D-8 chilipepper. shortbelly rockfish, bocaccio, stripetail rockfish, greenstriped rockfish cowcod canary rockfish, yellowtail rockfish greenspotted rockfish, widow rockfish, sharpchin rockfish, rosethorn rockfish yelloweye rockfish, redstripe rockfish Williams and Ralston Distribution and co-occurrence of Sebastidae off California and Oregon 847 i 39 100 200 300 400 Depth (m) 500 Figure 5 Contour map of species richness (number of species) from the AFSC triennial bottom trawl survey for all samples containing rockfish species. Discussion Our data were the result of pooling samples collected during AFSC triennial continental shelf trawl surveys conducted from 1977 to 1998. Because this interval rep- resents a period of substantial fishery removals (Ralston, 1998), one might expect that rockfish assemblage struc- ture changed over time as commercially important species were serially removed (e.g. bocaccio and canary rockfish). Although we did not perform year-specific analyses that would allow us to address this possibility, results from Weinberg (1994) showed that, over a comparable period of time (1977-92), the trawlable assemblages of rockfishes ofT Oregon and Washington were reasonably stable in composition. In particular, his year-specific recurrent group analyses revealed good agreement among surveys, from which he inferred the existence of three groups rep- resenting a deep-water assemblage (shortspine thorny- head. Pacific ocean perch, darkblotched, and redbanded (S. habcocki) rockfish), a mid-shelf assemblage (canary, yellowtail, and greenstriped rockfish). and a shelf-break assemblage (sharpchin. rosethorn, and redstriped rock- fish). Notably, there was substantial overlap between the latter two groups, which is consistent with our findings. It is no surprise that rockfish distributions are re- lated to bathymetric features, particularly when viewed through bottom trawl survey samples, as we did in our study. It is noteworthy that physical barriers, particularly the Mendocino Escarpment and Monterey Canyon, seem to affect the latitudinal distribution of certain rockfishes. The former may act as a barrier to dispersal because converging currents at Cape Mendocino create conditions 848 Fishery Bulletin 100(4) 42 1 1 1 1 - Blackgill 1 1 II- 40 - 38 ••" / e." " ••' ° V ■ _ O ' • " ^ ;;8+ ■ ■ r ■■;:/fV_ B 36 * . ■ ' ^ , ^ ^ ..■''- . '.. -S''^ Aurora •■• ' w, 34 1 1 1 ' 1 1 1 1 ' ■ • 1 "■ ■" 1 U - • o _ 1 '-' '■■■ • i 1 • ■ ^ 1, J . , 1 u .. . 1, , 0,0 i^ ^ _ » ^ 42 - SpJitnose?°S ""oo i° '„ ■" .'■• '.'••• •7'"/*°°-. i-'°,''>- _. _ • ■.- .. jfr.-is^; ■ ■ '*■.*"' '-'^ ■ 40 _ „■■-■:&',.■■, ■■ . '■ " ■• — . g, ..■.■■•- '-^I'l '^Z^ if • ^ ;:• K\5^."5- 0) ■D °' . •■:°^?&* .'""o"' i 38 '• "?' °j§v5^^ *°'°. ' • •• •. 0^^ .Off . • . • 38 ~ 0 o = ■ - " ;■ ; 0 ° ..? • . O OO - o.* . , ** G\ =-i . . '" -to o o'° -O ° ■ . ■■■'• • 0. „ = 36 . ■. . * . " - Darkblotched 34 1 1 *' 1 ' 1 1 1 1 1 100 200 300 400 100 200 300 400 Depth Figure 6 Plot of rockfish CPUE estimates by depth and latitude from the AFSC triennial bottom trawl survey. Circle diameters are proportional to the square root of rockfish CPUE and the bold plus symbol represents the centroid of the CPUE estimates. that result in offshore transport (Magnell et al.. 1990). Likewise, Monterey Canyon may act as a distributional barrier because continental shelf habitats constrict se- verely, creating a potential bottleneck to dispersion. Some support for this hypothesis is evident in the rockfish dis- tributional maps (Figs. 6-10), which show that shallow water species are more likely to have Monterey Canyon as a distributional boundary when compared to the deeper dwelling species, for which there apparently exists ample habitat. Overall, depth appears to be the single most important determinant of rockfish distributions. Most of the abun- dance patterns we observed followed the distribution of habitats by depth. Species diversity in the study area also seemed to follow the pattern of depth habitat distribution. However, species richness did not follow this pattern, but instead followed the region of overlap between shelf and slope rockfish assemblages. In fact, the contoured ridge of highest richness may be useful in spatially delineating the shelf and slope assemblages. This ridge appeared to Williams and Ralston Distribution and co occurrence of Sebastidae off California and Oregon 849 £ 3 ■1 : - I, ^ » ■ I • . 1 1 !■ ■ .; .1 ■ -^ 1 42 ■ ' ■ f 40 38 • 36 Pacific Ocean Perch Kedhanded " 34 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 42 Copper c Vermilion - 40 - -" - 38 ■+, O - 36 - ' +. - 34 1 1 1 1 1 1 1 1 1 1 42 Brown Halfljanded - 40 - - ■ - 38 if ■ If • • 0 - 36 ~ - 1 /I 1 1 1 1 .•■ • .Dip .11 1 1 100 200 300 400 100 Deptti (m) 200 300 400 Figure 7 Plot of rockfish CPUE estimates by depth and latitude from the AFSC triennial bottom trawl sur\'ey. Circle diameters are proportional to the square root of rockfish CPUE and the bold plus symbol represents the centroid of the CPUE estimates. be fairly constant at -200-250 m across all latitudes in- cluded in our study. The ridge of increased species richness indicated that there is substantial overlap in rockfish distributions. The analysis of rockfish overlap in our study was computed in two ways, i.e. by smoothed depth-latitude abundance and by raw trawl catches. The smoothed species-specific abundance estimates for all depth-latitude combinations could not be presented here because of space limitations. However, such data could be useful for detailed spatial analyses of rockfish co-occurrences in future analyses. The trawl-catch overlap data represent the best estimate of bottom trawl co-occur- rence probabilities for the study area. A potential limitation of the overlap estimates presented in our study is that they came from only one survey, which employed a single sam- pling gear However, the bottom trawl gear used in the AFSC triennial survey is generally similar to the fishing gear used by commercial fishermen, and previous studies have shown that catch rates from the continental shelf bottom trawl survey closely match commercial catch rates (Fox and Starr, 850 Fishery Bulletin 100(4) 42 40 38 36 34 42 40 ■D I 38 36 34 42 40 38 36 34 Chilipepper Bocaccio ooV-, ■'to' •.(3 6 •• : • o ■ => o O Greenstriped -- ~i 1 Shortbellv Stripe tail Cowcod 100 200 300 400 100 200 300 400 Depth (m) Figure 8 Plot of rockfish CPUE estimates by depth and latitude from the AFSC triennial bottom trawl survey. Circle diameters are proportional to the square root of rockfish CPUE and the bold plus symbol represents the centroid of the CPUE estimates. 1996). Another potential criticism of bottom trawl gear is its selectivity for capturing demersal fishes. Of the rockfishes included in our analysis, shortbelly and widow rockfish are two species that are known to be distributed in the water column (Lenarz, 1980; Wilkins, 1986; Chess et al., 1988). De- spite their occurrence off the bottom, however, both species were captured with regularity in the bottom trawl survey (Table 1). We conclude that, at least for the trawl sector of the groundfish fishery, the results of our study should prove useful in defining assemblages for rockfish management. As previously discussed, the assemblage descriptions given in Table 4 contain some inconsistencies with respect to the spatial distributions of the species. These apparent misclassifications may be the result of temporal variability in total rockfish abundance, as was captured by the second dimension of the MDS analysis. Because the estimated abundance of a species in our analysis was influenced by population levels of the stock at the time of each survey, which can change over time, our pooled analysis may not provide the best indication of historical assemblage asso- Williams and Ralslon: Distribution and co-occurrence of Sebastidae off California and Oregon 851 -1-4.^. ' •. 1 1 i-T':|l' ■• I. •■■ ■ 1 1 42 '.:. oQ.-., ' Canary /I.")' " -_ Ycllowtail 40 38 ■■J' ■ -« . ^ 0 .0^ ■■,..'■ 0 ^ 0 o<0 - 36 *• - - - 34 42 1 1 1 1 1 1 Greenspotted - ■ 1 . ■ k . ■ 1 -j • Widow 40 °' ° ^ • - . ■D 1 38 - 36 - - - 34 42 1 - ■ • 1 ■ ■ 1 1 1 1 1 1 Sharpchin 1 • 1- ■ 1 ■ 1 Rosethorn - 40 0.- °-- -.■■■.•Q-j-Do'. - " ••■+■ - 38 0 ■« ■ '■ - ,. . 36 1 1 . 1 1 i 1 t 1 100 200 300 400 100 200 300 400 Depth (m) Figure 9 Plot of rockfish CPUE estimates by depth and latitude from the AFSC triennial bottom trawl survey. Circle diameters are proportional to the square root of rockfish CPUE and the bold plus symbol represents the centroid of the CPUE estimates. ciations (see Weinberg, 1994). Furthermore, overfishing of rockfish stocks may alter spatial distributions. A good example is cowcod, which has shown a dramatic decline in abundance in recent years within the southern California Bight; most of the remaining population resides in rela- tively deep water (Butler et al.^). Because the total abundance of each species of rockfish may have affected the results of the assemblage analy- sis, the species distributions were carefully re-examined and a final assemblage structure determined (Table 5). The changes from Table 4 for the k=4 groups include the placement of greenspotted rockfish into the southern shelf assemblage and canary and yellowtail rockfish into the 3 Butler, J. L., L. D. Jacobson, J. T. Barnes, H. G. Moser, and R. Collins. 1999. Stock assessment of cowcod. /n Pacific Fish- ery Management Council, Status of the Pacific Coast groundfish fishery through 1999 and recommended biological catches for 2000: stock assessment and fishery evaluation. Pacific Fishery Management Council, 2130 SW Fifth Ave., Suite 224. Portland, Oregon 97201. 852 Fishery Bulletin 100(4) 1 '■ 1 1 1 111- . 1- ■ . 1 1 42 Yelloweye - Redstripe - 40 - ^^ 0 • o Latitude CO oo - ■ • 36 - - 34 Ill] 1 1 1 1 100 200 300 400 100 200 300 400 Deptti (m) Figure 10 Plot of rockfish CPUE estimates by depth and latitude from the AFSC triennial bottom trawl survey. Circle diameters are proportional to the square root of rockfish CPUE and the bold plus symbol represents the centroid of the CPUE estimates. redstripe 42 - yelloweye POP 40 J rosethorn / darkblotctied widow sha^pchin ,, canary yellowtail greenstriped shortspine redbanded chilipepper ^ <: bocaccro bank 38 - ^PP^ \ stnpetail \ greenspotted shortbelly <^<^ wcod splitnose aurora \ 1 \ 1 blackgill 36 - halfbanded ■ vermilion i 1 1 1 1 100 200 300 400 Depth (m) Figure 11 Plot of CPUE weighted mean depth and latitude for 26 rockfish species (center of names correspond to point estimates except for labeled points) from the AFSC trien- nial bottom trawl survey. Lines indicate divisions of species into clusters based on a A--medians (with k=4) partitioning analysis (see text). northern shelf assemblage. Greenspotted rockfish has a distribution that clearly warrants its inclusion in the southern shelf assemblage. Canary and yellowtail rockfish were placed into the northern shelf assemblage because of their known abundance in northern waters outside the area of our study (see Weinberg, 1994). Williams and Ralston: Distribution and co-occurrence of Sebastidae off California and Oregon 853 0 4 - 0.2 - yelloweye , \ redstripe | ^ 0.0 - E Q -0.2 - -04 03 0.2 - 0.1 - S 0.0 E -0.1 - -0.2 - -0.3 ^ yellowtail ^?n5%reenstriped [cowcod/ widow 0.2 0.4 Dimension 1 Figure 12 First three dimensions from a multidimensional scaling analysis of 26 rockfishes (center of names correspond to point estimates except for labeled points) from the AFSC triennial bottom trawl survey. Clusters suggested by a ^■medians (with ^=8) partitioning analysis are indicated by the enclosing lines (see text). The dividing line between the assemblages based on depth and latitude was determined by visual examina- tion of the distributions in Figures 6-10. The separation between shelf and slope species is roughly at 200-250 m; very few slope species were captured at depths less than 200 m. Most of the shelf species did not occur below 250 m. 854 Fishery Bulletin 100(4) Table 5 Suggested species assemblages based on distribution, ordi- nation, and partitioning analyses completed in this study. Habitat Species Deepwater slope blackgill rockfish, aurora rockfish, shortspine thomyhead, bank rockfish, darkblotched rockfish. Pacific ocean perch, redbanded rockfish, splitnose rockfish Nearshore copper rockfish, vermilion rockfish. brown rockfish, halfbanded rockfish Southern shelf chilipepper, shortbelly rockfish, bocaccio, stripetail rockfish, greenstriped rockfish, greenspotted rockfish, cowcod Northern shelf canary rockfish, ycUowtail rockfish, widow rockfish, sharpchin rockfish, rosethorn rockfish, yelloweye rockfish, redstripe rockfish leaving the 200-250 m zone as an area of overlap for the two assemblages, which is exactly where the peak in rich- ness occurred. The species in the nearshore assemblage seem to reside in waters less than 150 m depth. Perhaps the 100-150 m zone represents an area of overlap between the shelf and nearshore species. Latitudinal divisions between species in the northern and southern shelf assemblages are not as well defined as those based on depth. It appears that most of the southern shelf group are uncommon above the Mendocino Escarp- ment ( 40.8°N latitude ), with the exception of stripetail and greenspotted rockfish. The northern shelf species tend to range as far south as Monterey Canyon, leaving the area between Monterey Canyon and the Mendocino Escarp- ment as an area of overlap for these assemblages. This overlap is confirmed in the species richness contour plot which indicates the areas where the highest number of species were found in that latitude zone (Fig. 5i. The only questionable assemblage assignment was the placement of greenstriped rockfish, which does not range south of the Monterey Canyon region but whose center of its distribu- tion was observed in the 38.0°N latitude area. Overall, the results of our study indicate that rockfish can be classified into fairly distinct assemblages based on their depth and latitude distributions. This study also pro- vides estimates of the co-occurrence in the form of overlap measures for the important rockfish species off the coast of California. Both the assemblage and co-occurrence infor- mation should prove useful to fishery managers for model- ing fishery dynamics, solving bycatch issues, establishing area closures, and determining effective marine reserves. Although our study highlights some uses of spatial infor- mation, the limited amount of data that is now available prevents a more thorough analysis of interannual vari- ability in the distribution of species. We would therefore like to stress the importance of collecting more informa- tion describing the spatial distribution and co-occurrence of catches and the usefulness of those data in developing new strategies for managing west coast rockfish fisheries. Acknowledgments We gratefully acknowledge the assistance of Mark Wil- kins, who instructed us on how to properly access and interpret RACEBASE, the Alaska Fisheries Science Cen- ter's relational data base in which the triennial continen- tal shelf trawl survey data are stored. We also wish to thank Alec MacCall, Rick Methot, Ken Weinberg, and Mark Wilkins for their reviews of an early version of this manuscript. Literature cited Bloom, S. A. 1981. Similarity indices in community studies: potential pitfalls. Mar. Ecol. Prog. Ser. 5:125-U128. Chen, L. C. 1986. Meristic variation in Sebastes (Scorpaenidae), with an analysis of character association and bilateral pattern and their significance in species separation. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 4.5, 17 p. Chess, J. R., S. E. Smith, and P C. Fischer. 1988. Trophic relationships of the shortbelly rockfish, Sebastes jordani, off central California. CalCOFI Rep. 29:129-136. Dark, T. A., and M. E. Wilkins. 1994. Distribution, abundance, and biological characteris- tics of groundfish off Washington, Oregon, and California, 1977-1986. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 117, 73 p. Eschmeyer, W. N. (ed.). 1998. Catalogoffishes, 2905 p. California Academy of Sci- ences, San Francisco, CA. Eschmeyer, W. N., E. S. Herald, and H. Hammann. 1983. A field guide to Pacific coast fishes, 336 p. Houghton Mifflin Company, Boston, MA. Field. J. G., K. R. Clarke, and R. M. Wamick. 1982. A practical strategy for analyzing multispecies distri- bution patterns. Mar Ecol. Prog. Ser. 8:37-52. Fox, D. S., and R. M. Starr. 1996. Comparison of commercial fishery and research catch data. Can. J. Fish. Aquat. Sci. 53:2681-2694. Fujita, R. M., T. Foran, and I. Zevos. 1998. Innovative approaches for fostering conservation in marine fisheries. Ecol. Appl. 8(1):S139-S150. Gabriel, W. L., and A. V. Tyler 1980. Preliminary analysis of Pacific coast demersal fish assemblages. Mar. Fish. Rev 42(3-41:83-88. Gunderson, D. R. 1997. Spatial patterns in the dynamics of slope rockfish stocks and their implications for management. Fish. Bull. 95:219-230. Hartigan, J. A., and M. A. Wong. 1979. A *-means clustering algorithm. Appl. Stat. 28: 100- 108. Jay, C. V. 1996. Distribution ofbottom-trawl fish assemblages over the Williams and Ralston: Distribution and co-occurrence of Sebastidae off California and Oregon 855 continental shelf and upper slope of the U. S. west coast, 1977-1992. Can. J. Fish. Aquat. Sci. 53:1203-1225. Johnson, R. A., and D. W. Wichern. 1992. Applied multivariate .statistical analysis. 645 p. Pren- tice Hall, Englewood ClifTs, N.I, Kaufman, L., and P. J. Rousseeuw. 1990. Finding groups in data: an introduction to cluster analysis, 319 p. John Wiley & Sons, Inc., New York, NY. Kendall, A. W, Jr. 2000. An historical review of Sebastes taxonomy and sys- tematics. Mar Fish. Rev. 62(2):l-23. Krebs, C. J. 1989. Ecological methodology. 654 p. Harper Collins Pub- lishers, Inc., New York, NY. Leaman, B. M., and D. A. Nagtegaal. 1986. Identification of species assemblages and results of management applications for shelf and slope rockfishes off British Columbia. /;; Proceedings of the international rockfish symposium, p. 309-328. Alaska Sea Grant Report No. 87-2, University of Alaska, Anchorage, AK. Lenarz, W. H. 1980. Shortbelly rockfish, Sebastes jordani: a large unfished resource in waters off California. Mar. Fish. Rev. 1980: 34--10. Magnell, B. A., N. A. Bray, C. D. Winant, C. L. Greengrove, J. Largier, J. F. Borchardt, R. L. Bernstein, and C. E. Dorman. 1990. Convergent shelf flow at Cape Mendocino. Oceanog- raphy 3a):4-ll. Mahon, R., S. K. Brown, K. C. T. Zwaneburg, D. B. Atkinson, K. R. Buja, L. Claflin, G. D. Howell, M. E. Monaco. R. N. O'Boyle, and M. Sinclair 1998. Assemblages and biogeography of demersal fishes of the east coast of North America. Can. J. Fish. Aquat. Sci. 55:1704-1738. Meuter, F-J. 1999. Spatial and temporal patterns in the Gulf of Alaska groundfish community in relation to the environment. Ph.D. diss., 199 p. Univ. Alaska Fairbanks, Fairbanks, AK. Minchin, PR. 1987. An evaluation of the robustness of techniques for eco- logical ordination. Vegetatio 69:89-107. Pearson, D. E., and B. Erwin. 1997. Documentation of California's commercial market sampling data entry and expansion programs. U.S. Dep. Commer, NOAA Tech. Memo. NMFS 240, 65 p. Ralston, S. 1998. The status of federally managed rockfish on the U.S. West Coast. In Marine harvest refugia for West Coast rockfish: a workshop (M. Yoklavich, ed.), p. 6-16. U.S. Dep. Commer, NOAA Tech. Memo. NMFS 255. 2002. West coast groundfish harvest policy. N.Am. J. Fish. Manage. 22:249-2.50. Ralston, S., and J. J. Polovina. 1982. A multispecies analysis of the commercial deep-sea handline fishery in Hawaii. Fish. Bull. 80:435-448. Rickey, M. H., and H-L. Lai. 1990. The deep water complex species mix in the Washing- ton and Oregon trawl fishery. Wash. Dep. Fish. Tech. Rep. No. Ill, 40 p. Ripley, B. D. 1996. Pattern recognition and neural networks, 403 p. Cambridge Univ. Press, Cambridge. Rogers, J. B., and E. K. Pikitch. 1992. Numerical definition of groundfish assemblages caught off the coasts of Oregon and Washington using commercial fishing strategies. Can. J. Fish. Aquat. Sci. 48: 1264-1272. Rousseeuw. P. J. 1987. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J. Comp. Appl. Math. 20; 53-65. Sampson, D. P., and P. R. Crone. 1997. Commercial fisheries data collection procedures for U.S. Pacific coast groundfish. U.S. Dep. Commer, NOAA Tech. Memo. NMFS 31, 140 p. Surfer. 1995. Surfer for Windows user's guide, version 6, 538 p. Golden Software, Inc., Golden, CO. Weinberg, K. L. 1994. Rockfish assemblages of the middle shelf and upper slope off Oregon and Washington. Fish. Bull. 92:620-632. Wilkins, M. E. 1986. Development and evaluation of methodologies for assessing and monitoring the abundance of widow rockfish, Sebastes entomelas. Fish. Bull. 84:287-310. Wilkins, M. E., M. Zimmermann, and K. L. Weinberg. 1998. The 1995 Pacific Coast bottom trawl survey of ground- fish resources: estimates of distribution, abundance, and length and age composition. U.S. Dep. Commer, NOAA Tech. Memo. NMFS 89, 138 p. 856 Properties of the residuals from two tag-recovery models* Robert J. Latour John M. Hoenig Department of Fisheries Science Virginia Institute of Manne Science College of William and Mary Gloucester Point, Virginia 23062 E-mail address (for R, J, Latour): latour@vims edu Kenneth H. Pollock Biomatfiematics Graduate Program Department of Statistics Nortfi Carolina State University Raleigti, Nonh Carolina 27695 Researchers often use multiyear tag- recovery studies to assess fish popu- lations: yet deriving useful stock as- sessment parameter estimates from the resulting data can be difficult. The reliability of those parameter esti- mates generally depends on data qual- ity and meeting the assumptions in- herent to the models used for analysis. As a result, practical application of multiyear tag-recovery models gener- ally requires that a large portion of the data analysis involve investigation and evaluation of biases due to poten- tial assumption violation. Brownie et al. (1985) developed a class of models that has become widely used for the analysis of multiyear tag- recovery data. These models constitute a generalization of the class of models developed by Seber (1970), which have recently been resurrected as an impor- tant tool for the analysis of multiyear tag-recovery data by the development of the software progi'am MARK I White and Burnham, 1999). Although these models are fairly simple and robust, in practical situations at least one of the assumptions is often not supported by the data. Approaches that are commonly used to assess the fit of multiyear tag-recov- ery models include the formal good- ness-of-fit test, Akaike's information criterion (AIC) (Akaike, 1973; Burn- ham and Anderson, 1992; Burnham et al., 1995) and other related measures such as quasilikelihood AIC (Akaike, 1985). Although these measures are informative about overall model fit, they do not provide any information about why a model fit is poor or which assumption(s) is (are) possibly in viola- tion. To remedy this problem, Latour et al. (2001a) conducted a series of simu- lations and demonstrated that distinct patterns in model residuals will be evident if particular assumptions are violated. They discussed in detail the residuals associated with the time- specific parameterizations of the Se- ber (1970) and Brownie et al. (1985) models, as well as the time-specific instantaneous rates model developed by Hoenig etal.( 1998). The genesis of the work by Latour et al. (2001a) can be traced to two par- ticular applications of multiyear tag- recovery models. Specifically, Latour et al. (2001b) analyzed tag- recovery data of red drum iSciaenops ocellatits) in South Carolina and found systematic patterns along the diagonals in the upper right corner of the residuals matrix. Frusher and Hoenig (2001) ap- plied a series of tag-recovery models to Australian rock lobster {Jasus edward- sii) data and found consistent patterns in the columns of the residuals matrix. In both instances, the researchers could only speculate as to the cause of these patterns in residuals. Although the simulations conducted by Latour et al. (2001a) have since provided reasonable explanations for the ob- served patterns, the development of those diagnostic procedures led to the discovery that the residuals associated with the time-specific Seber ( 1970) and Brownie et al. (1985) models are sub- ject to several constraints. This note contains a series of simple mathematical arguments that verify the assertions made by Latour et al. (2001a) about the residuals of the time-specific parameterizations of the Seber (1970) and Brownie et al. (1985) models. Unfortunately, the constraints inherent to the residuals of those models partially cloud a researcher's ability to assess the existence of a pat- tern. As such, knowledge of the inher- ent properties of the residuals of these models is of particular importance, especially because the time-specific parameterizations are commonly used for the analysis of tag-recovery data. Materials and methods Multiyear tag-recovery models Multiyear tagging data are generally represented by an upper triangular matrix of tag recoveries. For example, the matrix for a study with / years of tagging and J years of tag-recovery would be, when / = J. 'll'l9- (1) where r = the number of tags recov- ered in year j that were released in year ; (note, / = 1,... J\j=i J). Application of multiyear tag-recov- ery models generally involves con- structing a matrix of expected values and comparing them to the observed * Contribution 2490 of the Virginia Insti- tute of Marine Science, Gloucester Point, VA 23062. Manuscript accepted 10 July 2002. Fish. Bull. 100:856-860 (2002). NOTE Ldtour et al : Properties of the residuals lioni two tag-recovery models 857 data. The matrix of expected values corrospondinK to the time-specific parameterization of Brownie et al. (1985), wliicii is referred to as model 1, takes tlie form £, - N,f, ■■■N,{S.ySj_,)f:, - - - N,f,, (2) where A^, = the number tagged in year (; f\ = the tag recovery rate in year /; and S^ = the survival rate in year /. As stated above, the Brownie et al. (1985) models constitute a generalization of those developed by Seber (1970). The only difference lies in the definition of the tag recovery rate. Specifically, Seber (1970) modeled the tag recovery rate in year i as /j = ( 1 - S, )r,, where r, is the rate at which tags are reported from killed fish in year ; re- gardless of the source of mortality. The matrix of expected values associated with time-specific parameterization of the Seber (1970) models, which we will refer to as model 1 , takes the form (when 1 = J) E = N,{l-S,)r, N,St(l-S.,)r., ■■■ iV,(S,-S;_,l(l-S., )o N.^{l-S.,)r.^ ■■■N.,{S.ySj_j)a-Sj)r, (3) The data in each row of Equation 1 follow a multinomial distribution and maximum likelihood estimation can be used to derive parameter estimates from either model 1 or model 1 . Program MARK has emerged as the leading software package for deriving these estimates (White and Burnham, 1999). Patterns in residuals Latour et al. (2001a) manipulated a hypothetical perfect data set (i.e. the observed number of tag recoveries was equal to the expected number of tag recoveries) to simulate four specific forms of assumption violation for niultiyear tag-recovery models. For each scenario, they analyzed the modified data with model 1, model 1', and a time-specific parameterization of the instantaneous rates (IR) models (Hoenig et al., 1998) and noted any patterns in the residu- als matrix that resulted from each particular assumption violation. Specifically, they found the following: 1) the presence of nonmixing (which violates the assumption that the tagged population is representative of the target population) leads to consistent patterns on the main and super diagonals of the residuals matrix (the main diago- nal contains the ( 1,1),(2,2),...,(/J) cells and the first super diagonal contains the (1,2), (2, 3), ...(/-I,/) cells in a square matrix); 2) permanent emigration from the study area of individuals within a tagged cohort (which violates the assumption that all tagged fish within a cohort are subject to the same annual survival and tag- recovery rates) leads to a pattern of negative residuals along the diagonals of the upper right corner of the residuals matrix; 3) tag- induced mortality or immediate loss of tags due to poor tagging (which violates the assumptions that tags are not lost and survival rates are not affected by tagging) leads to row patterns in the residuals matrix (note that these patterns are detectable only in the residuals matrix of the IR model); and 4) a change in the natural mortality rate (which violates the frequently imposed assumption that natural mortality is constant over time) leads to column patterns in the residuals matrix (again, this only applies to the IR model). Constraints on residuals of model 1 and model 1' Latour et al. (2001a) asserted without proof that the residuals associated with model 1 and model 1 are sub- ject to several constraints. Specifically, they stated that the relationship E/j = r'u always holds, regardless of the number of years of tagging and tag-recovery (note that £„ is the expected number of tags recovered in year j that were released in year /). This implies that the observed data and the expected value associated with the (1,1) cell are always identical and that the residual for that cell is always equal to zero. They also stated that the residuals associated with the implicit "never seen again" category are also always equal to zero (recall that under a multino- mial formulation, one of the possible outcomes is to never recapture a tagged fish). Collectively, these constraints imply that the residuals matrix derived from using model 1 or model 1' to analyze data from a study with / years of tagging and J years of tag-recovery takes the form resid -■ 0.00 {)\2-Ey2) ■■■ ( >\., ' Eij ) 0.00 - ( 7-22 - E.,2 ) ■ • ■ ( r.,j - E.,j ) 0.00 ; ; ■. ; o.oo (r,j-E,j) 0.00 (4) where r,^ and £,^ are as defined previously and the last column of the matrix represents the residuals associated with the "never seen again" category. In addition to the aforementioned zero residuals, Latour et al. (2001a) stated that the sum of each row and each column of the residuals matrix must equal zero and that for the case when / = J (i.e. the recovery matrix is square), the constraint that Ej, = r,, is also present (i.e. the residual associated with the (/, /) cell is always equal to zero). In the context of searching for patterns in residuals, these constraints have the following implications. First, the presence of residuals that are constrained to be zero essentially reduces the total number of values that are available for inspection and ultimately forces conclusions about the existence of a pattern to be based on the signs of fewer residuals. For short-term tagging studies (e.g. 3-4 years), the loss of residuals for inspection makes it extremely difficult to evaluate model performance because each row, column, and diagonal of the residuals matrix al- ready contains only a few values. Second, because the sum of each row of the residuals matrix must total zero, and 858 Fishery Bulletin 100(4) because the residuals corresponding to the "never seen again" cells are zero, it is not possible for a row pattern to be expressed in the residuals matrices of either model 1 or model 1. This constraint renders it very difficult to detect assumption violations that are cohort-specific — the most common being tag-induced and handling mortality and short-term tag loss. Results To verify that the aforementioned constraints about the structure of the residuals matrices associated with model 1 and model 1' are true, we offer the following mathemati- cal arguments. The proofs simply involve algebraic manip- ulation of equations involving the analytical formulae for the maximum likelihood parameter estimates (MLE) of model 1. The formulae for the MLEs were originally devel- oped by Seber (1970) and can be applied to both to model 1 and model 1 (because r, can be expressed as a function of S, and/",, the invariance property of MLEs implies that an MLE of r, can be obtained by the transform). Hence, the proofs are developed for model 1, and we note that similar arguments could be constructed for the residuals of model 1 . Recall that the analytical solutions for the maximum likelihood estimates of/] and S, from Seber (1970) and Brownie et al. ( 1985) are given by NT and S, = — ^ ■ ■ -^ . where R^ and C, are the row and column totals of the observed data in year /; and T,=R,^T__,-C,_, , = 2,... J Tuj = T,^j_,-C,^j_, j=l,...,J-nfJ>I. The (1,1) cell To show that the residual associated with the (1,1) cell is always zero, we must demonstrate that the difference between the obsei-ved and estimated expected value in the first cell is always zero. Hence, we have 1R C ^ — ^ — - (substituting for/", ) 7V,T, j ^ " = '"ii ~ C, (since T, = i?, ) = /jj - Tjj = 0 (because the column total in the first recovery year is 1-^). The {/,/) cell when / = J To show that the residual associated with the (/,/) cell is always zero when / = J. again we must demonstrate that the difference between the observed and estimated expected values in that cell is always zero. Hence, we have 'n-E,i =r„-N,fi ( R C \ = r,, - N, — - — - (substituting for/",) " '\N,T,j ^ " = rjj - R, ( because 7) = C; ) = fji - ;•,/ = 0 ( because the row total in the final recovery year is r^;). Column sums when I = J To show that the column sums of the residuals matrix equal zero, we must demonstrate the column sum of the observed data equals that of the expected values. Consider the sum of the expected values associated with the /"^ column of the recovery matrix, that is N.,(S,-S,_,)f,+ -+N,_,S,J,+N,f]. Now substitute for /" and S, on the right hand side: ' Ri (Ti-C^) N., . N, T, R, ^/-i T,^, R, J J N., R, (7?, -C) N., R,^ ^T,_, Ml] C,_,)N, N.-, T, R, N, R, I R£l \ N,T, J ■ + Af,_, Rj-i iT,_j -C,_i) R, ^/-i 7)-! N, j N,T,] '[n,T, Cancel terms and factor out the term C/. Q (£l' [t, (R,(T,^C,)](T,-C,] (T,_,-C,A [ T., Tj., R^iT^-C.,)' T. (T,-C, 7^7-1 -Q_i £l ( R,_,{T,_,-C,_,) 7)-! + R, iT -C ) Systematically factor out terms of the form — '■ '— ire- call thatT, =/?i): ^' NOTE Latour et al.: Properties of the residuals from two tag recovery models 859 Q 'T.,-C., [ T, ; (T,_,-C,,] \ ■'/-2 ; {Ri-Ci + R2) + Ri + «4 + ■■• + /?, + /?, Utilize the definition of T^ = R^ + T", | - C, , to systemati- cally simpliCy and cancel starting with the innermost par- enthetic expression: Q (R,C, T, ■'2 Q + iTi-Ci) ^3^3 ^2 ^1 ^2 ^2 (T -C ' Systematically factor out terms of the form — ' '' Q [t, _V ■'/-I I \T,_,-C,_,+R,] T,=C,=r„+r.„+-- + r,^ Q = \^^ c^in- + (T, -Ci) c, The expression inside the innermost square brackets is equal to 1 (recall Tj = Cj). Hence, we have which demonstrates that the column sum of expected values equals the column sum of observed recoveries, as desired for the /"^ column. Similar arguments hold for the other columns. Column sums when l>J The proof that the column sums of the residuals matrix equal zero when the recovery matrix is nonsquare is simi- lar to the proof above except for making use of the defini- Q = + { T -c ) = "^'^^ * •^'•^^ ~ ^'^' = R + r,., + ■ which shows the sum of the expected values in row 1 is equal to the sum of the observed data. Similar arguments hold for the other rows. Row sums when l> J Row sums when I = J To show that the row sums (excluding the "never seen again" cell) of the residuals matrix equal zero, we must demonstrate the sum of the observed data equals that of the expected values. Consider the sum of the expected values associated with the first row of the recovery matrix: Q = £„ + £,., + ■■ + £;„= Af,/, + iV,S/, + ■ + A'l'S, ■••S,_i)/,. Now substitute for ^ and S, on the right hand side: Q = N, {R,cA^^{R,iT,-C,)N, N,TJ •••+7V. I Ni T, R, ) ' R, (Ti-Ci) N., v^. R, (R,_,(T,^,-C,^,)N,]](R,C,] [N,_ T,^ «/ JJ I N,T, ) Cancel and factor out the term T^ - Cj (recall that T^ R,y. As with the proof of the column sums when / > J, the defin- tion T,^^ = T^^. J - C,^ jis needed to show the row sum of a nonsquare matrix (excluding the "never seen again" cellsj are zero. "Never seen again" cells The likelihood function for the Brownie-type model is a product multinomial and the parameters for each row are constrained to sum to one. Therefore, the expected values in a row are simply an apportionment of the number tagged to the years of recovery and the "never seen again" category. Hence, the sum of the estimated expected values has to equal the row sum (including the "never seen again" cell), which implies the residuals of the "never seen again" cells are always equal to zero. Discussion The residuals of multiyear tag-recovery models can be very helpful for evaluating model performance. Unfortunately, examining the residuals matrix for patterns is not a com- monly employed procedure for assessing model fit in practi- cal situations. The work by Latour et al. (2001a) was intended 860 Fishery Bulletin 100(4) to demonstrate the insight a researcher can acquire by using residuals as a diagnostic probe to gauge the possibihty of assumption violation. Similarly, the work presented here is intended to further guide researchers by explicitly delineat- ing the properties of the residuals associated with two com- monly applied multiyear tag-recovery models. Model 1 and model 1' represent parameterizations of only two classes of tag-recovery models. Properties of the residuals associated with other classes of models (e.g. move- ment models, age-structured tag-recovery models, capture- recapture models) have not been studied. We feel strongly that similar types of insight about model performance and model fit can be acquired by examining the residuals for patterns. As such, we recommend that the residuals fi'oni other classes of models be more thoroughly investigated. Literature cited Akaike, H. 1973. Information theory as an extension of the maximum likelihood principle. In Second international symposium on information theory (B. N. Petrov and F. Csaki. eds.), p. 267-281. Akademiai Kiado. Budapest. Akaike, H. 1985. Prediction and entropy. /« A celebration of statistics (A. C.Atkinson and S. E. Fienberg, eds.), p. 1-24. Springer, New York, NY. Brownie, C, D. R. Anderson, K. P. Burhnam, and D. S. Robson. 1985. Statistical inference from band recovery data: a hand- book, 2""' ed., 305 p. U.S. Fish and Wildl. Sei-v. Resour Publ. 156. Burnham, K. P., and D. R. Anderson. 1992. Data-based selection of an appropriate biological model: the key to modern data analysis. In Wildlife 2001: populations (D. R. McCuUough and R. H. Barrett, eds.), p. 16-30. Elsevier Science Publishers, London. Burnham, K. P, G. C. White, and D. R. Anderson. 1995. Model selection strategy in the analysis of capture- recapture data. Biometrics 51:888-898. Frusher, S. D., and J. M. Hoenig. 2001. Estimating natural and fishing mortality and tag reporting rate of rock lobster from a multiyear tagging model. Can. J. Fish. Aquat. Sci. 58:2490-2501. Hoenig, J. M., N. J. Barrowman, W. S. Hearn, and K. H. Pollock. 1998. Multiyear tagging studies incorporating fishing effort data. Can. J. Fish. Aquat. Sci. 55:1466-1476. Latour, R. J., J. M. Hoenig, J. E. Olney and K. H. Pollock. 2001a. Diagnostics for multiyear tagging models with appli- cation to Atlantic striped bass iMorone saxatilis). Can. J. Fish. Aquat. Sci. 58:1717-1726. Latour, R. J., K. H. Pollock, C. A. Wenner, and J. M. Hoenig. 2001b. Estimates of fishing and natural mortality for red drum Sciaenops ocellatus in South Carolina waters. N. Am. J. Fish. Manage. 21:73.3-744. Sober, G. A. F 1970. Estimating time-specific survival and reporting rates for adult birds from band returns. Biometrika 57:313- 318. White, G. C, and K. P. Burnham. 1999. Program MARK-survival estimation from popula- tions of marked animals. Bird Study 46:120-138. 861 Preliminary study on the use of neural arches in the age determination of bluntnose sixgill sharks iHexanchus griseus) Gordon A. McFarlane Jacquelynne R. King Mark W. Saunders Pacific Biological Station Fisheries and Oceans Canada Nanaimo Bntish ColumbiaV9R 5K6, Canada Email address (for G A McFarlane) McFarlaneSiq^pacdfo-mpogcca Conventional structures used for age determinations of teleost fishes (e.g. fin rays, otoliths, scales) cannot be used for elasmobranchs in which these structures are car- tilaginous. However, vertebral cen- tra with systematic deposits of calcium phosphate, have been used for age estimation in a number of elasmobranch species such as the tiger shark iGaleocerdo cuvier), scalloped hammerhead {Sphy7-7ia lewini), and several smoothhound sharks (Mustelus spp.) (Cailliet, 1990). Cailliet et al. (1983) pro- vided preparation techniques for enhancing and examining vertebral bands in these elasmobranchs. Unfortunately, the vertebral cen- tra of a number of sharks, in- cluding bluntnose sixgill sharks iHexanchiis griseus), are too poorly calcified to provide age information (Cailliet, 1990). Previous attempts to age sixgill sharks by using ver- tebral centra have been unsuc- cessful (Ebert, 1986). Sharks with poorly calcified vertebral centra tend to be either deep-water spe- cies or from relatively primitive families (Cailliet, 1990). However, systematic deposits of calcium phosphate have also been found in chondrocranium, jaws, visceral arches, fin cartilage, claspers, neu- ral and haemal arches (Clement, 1992). Bluntnose sixgill sharks (hereafter referred to as simply "sixgill sharks") are one of the largest and most primi- tive species of elasmobranchs. They have a world-wide distribution and in the north-east Pacific Ocean range from the Aleutian Islands to Baja, California (Hart, 1973). Because they are deep-wa- ter inhabitants occupying depths up to 2500 m along the outer continental shelf and upper slope waters (Compagno, 1984; Ebert, 1994), little is known about their biology. Sixgill sharks are about 65-70 cm at birth and the maximum length recorded is 482 cm (Castro, 1983 ). They are ovoviviparous and have reported litter sizes ranging from 22 to 108 (Compagno, 1984; Ebert, 1992). The estimated size-at-maturity for females is 396 em (Ebert, 1992), although cap- ture of mature sixgill sharks is rare. It is likely that the young inhabit inshore waters (Compagno, 1984). Off the west coast of Canada, an experimental fishery for sixgill sharks was initiated in the early 1990s but was terminated because of conserva- tion concerns. Recently, there has been a renewed interest in a fishery, both commercial and sport, for the species. Incomplete knowledge regarding the biology and life history of sixgill sharks remains a concern. Given the continu- ing interest in a fishery (commercial and sport) and potential conservation concerns, we investigated the use of neural arches as an alternate body structure to use for the age determina- tion of sixgill sharks. We describe the staining technique used and the bands (annuli) observed. We suggest that our technique represents a possible avenue for developing an aging technique for various elasmobranchs, particularly those with poorly calcified vertebral centra. Methods From May through September 1994, as part of a co-operative industry-govern- ment sixgill shark tagging program, 259 sixgill sharks were captured with hook and line gear off the west coast of Vancouver Island, British Columbia, Canada. Fishing occurred within five main areas: Kyoquot Sound (50°00'N and 127°20'W), Esperanza Inlet (49° 45'N 127°00'W), Nootka Sound (49° 25'N and 126°40'W), Tofino Inlet (49° 05'N and 125°40'W), and Barkley Sound (48°50'N and 125°20'W). A sample of ten sharks was obtained for age determination research. Total length (TL, cm) and sex were recorded for each shark. A portion of the vertebral column containing 15-20 vertebrae, including the neural and haemal arches, was removed just posterior to the head and immediately frozen. In the labora- tory the vertebral column section was thawed. The connective tissue and the outer layer of cartilage were removed from each vertebra and neural arch (Fig. lA) by soaking them in bleach for 15 minutes and rinsing in distilled water. Some teasing away of tissue was required. For each set of vertebrae, five neural arches were separated from the vertebral centra (Fig. IB). A single dor- soventral cut was made in each neural arch to expose the inner portions (Fig. IB). The silver nitrate staining tech- nique for revealing calcium deposits in vertebral centra of other shark spe- cies (Cailliet et al., 1983) was modified and applied to the neural arches. Each arch was soaked in 150 mL of 1% silver nitrate while exposed to wide spectrum light (320-400 nm). Soak times in silver nitrate varied according to the size of the arch, but all soak times lasted at least one hour. The degree of staining was assessed on an initial arch for each shark. If required, subsequent arches were restained in the silver nitrate so- lution and removed at 15-30 minute in- tervals after removal of the first arch. Manuscript accepted 27 May 2002. Fish. Bull. 100:861-864 (2002). 862 Fishery Bulletin 100(4) Intercalary -^ — Neural arcti , j%4 — Neural canal — Centrum A Haemal ' — -'^ arch B Plane of section Figure 1 (A) Schematic diagram of the section of vertebral column with attached neural and haemal arches from a bluntnose sixgill shark. (B) Neural arch; an initial dorsoventral cut was made perpendicular to the arch face prior to saturation with silver nitrate solution. Subsequent sections (dotted lines) were made to reveal banding patterns. (C) Sectioned neural arch showing band patterns. After soaking in silver nitrate, each neural arch was rinsed with distilled water to remove any excess solution. A scalpel was used to cut sections approximately 2-3 mm thick, parallel to the initial cut (Fig. IB). Sections were placed under the white light (400-700 nm) of a dissect- ing microscope for a few minutes to allow bands to appear (Fig. IC). A 2-3 minute soak in 5% sodium thiosulphate (Cailliet et al., 1983) was unsuccessful in halting the de- velopment process and fixing the chemical substitution of calcium salts for silver nitrate. However, a 3-5 minute soak in Kodak Stop Bath SB-la" ( 1.0 L water; 125 niL 287r acetic acid) did halt the development process. The stained neural arch sections were mounted on a microscope slide with an acrylic toluene mounting medium. Results Of the ten sharks sampled, eight of the sharks were female ( 130-349 cm TL), and two were male ( 140 and 242 cm TL). All sharks were immature. The soak time required for the silver nitrate to penetrate the whole neural arch varied between arches of similarly sized sharks and even between arches from the same shark. Generally arches from sharks less than 200 cm TL, 200-250 cm TL, 250-300 cm TL, and greater than 300 cm TL were removed after 1, 2, 3, and 4 hours respectively. It was necessary to use several neural arches per shark in order to differentiate a clear set of bands (annuli). The staining occurred only near the outer portion of each arch, and this left the central portion unstained (Figs. 2 and 3). Distinct, thin, dark bands were discernible in nine of the 10 sets of neural arches examined (e.g. Fig. 2). Bands in the neural arches from the 349 cm female were not all clearly defined even after soaking in the silver nitrate solution for eight hours. In general, bands appeared at regular intervals ( Fig. 2 ). The widths between bands were asymmetrical along the whole edge of the neural arch. These widths were great- er along the top edge compared to widths at the side edges (Fig. 3). In some neural arches, the bands were not continu- ous (Figs. 3 and 4), which made examination of different por- tions of the arch necessary in order to count the number of bands formed. In some arches, bands were identifiable but fainter towards the inner portions (Fig. 4). These bands ap- peared granular, but patterns were still detectable (Fig. 4). The number of bands increased linearly with the total length of the shark (Fig. 5). By using 67.5 cm as the median length at birth (Castro, 1983), the mean estimated growth rate was 25 cm/band (SD=4, «=9). Because all of the speci- mens in our study were immature, a change in the growth rate was not observed, as would be expected when size at maturity was reached. NOTE McFarlane et al : Use of neural arches in age determination of Hexanchus gnseus 863 Figure 2 A stained neural arcli from a 207-cm (total length) female bluntnose sixgill shark. Note that the staining occurs along the outer third portion of the arch. Five bands are indicated by dots. Figure 3 k stained neural arch from a 242-cm (total ength) male bluntnose sixgill shark. Seven lands are indicated by dots. Note that the widths between bands are asymmetrical and the bands are discontinuous along the whole perimeter Discussion The nature of the staining observed on the neural arches suggests that these structures have a potential use in the age determination of elasmobranchs. In sixgill sharks, the bands were distinct and appeared at regular intervals. The number of bands per neural arch increased with total length of the shark, suggesting their potential for age determination. The banding occurred only on the outer portions, indicating that the calcium was deposited after the proximal portion of the arches developed, probably after birth. The methodology used in our study for the staining of calcium deposits in neural arches offers some promise for age determination of sixgill sharks. Future research is required to refine the method by determin- ing sectioning methods and thickness, optimum staining times, or perhaps alternative solution concentrations. For example, it might be more effective to cut neural arches into thin sections prior to soaking in silver nitrate rather than to soak larger portions and then section them. Calcium deposits have been observed in the neural arches of several elasmobranchs (Cailliet, 1990). For ex- ample, Peignoux-Deville et al. (1982) provided an exten- sive description of these deposits for dogfish {Scyliorhinus canicula). However, age determination research for elas- mobranchs has focused on the calcium deposits in the ver- tebral centra. Use of the neural arches as aging structures may provide an alternative age determination method for species in which the vertebral centra are poorly calcified Valid age and growth information is fundamental to stock assessment and management. However, Cailliet (1990) lists only 39 species of elasmobranchs for which there are published or ongoing age verification studies. As interest and concern in commercial and recreational fishing of elasmobranchs increases, the need for age determination techniques becomes more pressing. Researchers need to investigate alternative methods for aging elasmobranchs, and we suggest that neural arches may prove to be useful. It is important to note that our observations on calcium banding in sixgill shark neural arches are preliminary and are for immature sixgill sharks only. We stress that funda- mental to all aging determination studies is the validation of the method (Beamish and McFarlane, 1983), and future work is required to determine if the bands observed in our study are formed on annual, or regular, intervals. Acknowledgments Bill Andrews and Grady O'Neill provided technical assis- tance in producing the photographs. Maria Surry pro- duced the schematic diagrams of the vertebral columns and neural arches. 864 Fishery Bulletin 100(4) Figure 4 A stained neural arch from a 330 cm (total length) female bluntnose sixgill shark. The arch was soaked in I9c silver nitrate for 7.25 hours, but the inner portions of the arch were barely stained. Ten bands are indicated by dots, but the first four are difficult to determine. The fourth band is labelled with the number four Literature cited Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fish- eries biology. Trans. Am. Fish. Soc. 112(6):735-743. Cailliet,G. M. 1990. Elasmobranch age determination and verification an updated review. In Elasniobranchs as living resources: advances in the biology, ecology, systeniatics, and the 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. NMFS 90. 100 140 180 220 260 300 Total length (cm) 340 380 Figure 5 Number of bands observed in neural arches by total length (cm) of sixgill shark. Dots represent females; open circles represent males. As the total length of the shark increases, the number of annuli observed increases. Cailliet, G. M., L. K. Martm, D. Kusher, P. Wolf and B. A. Welden. 1983. Techniques for enhancing vertebral bands in age estimation of California elasniobranchs. In Proceedings of the international workshop on age eetermination of oceanic pelagic fishes: tunas, billfishes, and sharks (E. D. Prince and L. M. Pulos, eds.), p. 157-165. U.S. Dep. Commer. NOAA Tech. Rep. NMFS 8. Castro, J. I. 1983. The sharks of North American waters, 180 p. Texas A&M Univ. Press, College Station. Clement, J. G. 1992. Re-examination of the fine structure of endoskel- etal mineralization in chondrichthyans: implications for growth, ageing and calcium homeostasis. Aust. J. Mar Freshwater Res. 43:157-181, Compagno, L. J. V. 1984. FAO species catalogue. Volume 4. Sharks of the world. An annotated and illustrated catalogue of shark species known to date. Part 1. Hexanchiformes to Lamniformes. FAO Fish. Synop. 125 (vol. 4, part 1 ): 249 p. FAO, Rome. Ebert, D. A. 1986. Biological aspects of the sixgill shark, Hexanchus gnseus. Copeia 1986( 1): 131-135. 1992. Cowsharks. In California's living marine resources and their utilization, (W. S. Leet, C. M. Dewees, and C. W. Haugen, eds.), p. 54—55. Sea Grant Extension Publication UCSGEP-92-12. 1994. Diet of the sixgill shark Hexanchus griseus off south- ern Africa. Afr J. Mar Sci. 14:213-218. Hart, J. L. 1973. Pacific fishes of Canada. Bull. Fish. Res. Board, Can- ada 180. 740 p. Peignoux-Deville. J., F. Lallier, and B. Vidal. 1982. Evidence for the presence of osseous tissue in dogfish vertebrae. Cell and Tissue Research 222:605-614. 865 Re-identification of a lamnid shark embryo Henry F. Mollet Moss Ldndicig Marine Laboratories Moss Landing, California 95039 9647 E mail address molletiaipacbell.net Antonio D. Testi Italian Sliark Research Project Via A Solan 43/2 20144 Milan, Italy Leonard J. V. Compagno Shark Research Centre PO Box 61 8000 Cape Town, South Afnca Malcolm P. Francis National Institute of Water and Atmosphenc Research PO Box 14-901 Wellington, New Zealand In August 1903. a 400-500 kg preg- nant female lamnid shark was caught in the Strait of Messina, Mediterra- nean Sea. She was reported to contain 25-30 embryos, one of which was saved and taken to the local Marine Institute, where it was subsequently examined by Sanzo (1912). The male embryo measured 36.1 cm total length (TL), weighed 800 g, and had a greatly distended abdomen, as is typical of embryos of oophagous lamnoid sharks (Gilmore, 1993). The mother and the remaining embryos were not saved. Because Sanzo was not able to exam- ine the adult female from which the embryo was taken, the embryo was identified by a process of elimina- tion, based mostly on morphometries of postnatal specimens. Sanzo (1912) concluded that the embryo was a white shark, Carcharodon carcharias (Linnaeus, 1758). According to Sanzo (1912), the embryo was requested in 1909 by E. Giglioli and was then (in 1912) conserved at the Vertebrate Museum of the Superior Institute of Studies in Florence. Sanzo's (1912) identification was questioned by many (Tortonese, 1950, 1956; Bass et al., 1975; Pratt' ) but was assumed to be correct by Gilmore (1993). Tortonese (1950) suggested that the morphometric arguments used by Sanzo (1912) did not rule out the shortfin mako ilsiirus oxyrinchus Rafinesque, 1810) but that the high fecundity of 25-30 was more consis- tent with C. carcharias than with /. oxyrinchus or a Lamna species. A lack of information on lamnid reproduction and the misidentification of a likely Galeorhinus galeiis (Stevens^) with a litter of 30 as Lamna by Neill (1811), may have led Sanzo (1912) to consider the porbeagle Lamna nasus (Bon- naterre, 1788) instead of the shortfin mako as the most likely alternative to the white shark. Shann (1911) had questioned Neill's identification but this was not available to Sanzo (1912). Bass et al. (1975) incorrectly quoted Tortonese (1956) as saying that the embryo could have been a porbeagle. Tortonese (1950) pointed out that Sanzo (1912) mistook the large yolk- filled stomach (due to oophagy) for a yolk sac. Gilmore (1993) reviewed the reproductive biology of lamnoid sharks and included a redrawn sketch of the Sanzo (1912) embryo, still identified as a white shark, and also incorrectly stated that Sanzo (1912) had docu- mented oophagy for the white shark. Francis ( 1996) reviewed lamnid fecun- dity data and showed that the shortfin mako has the highest known fecundity (18 embryos; Branstetter, 1981) in the order Lamniformes, which suggested to us that the embryo was more likely a shortfin mako. Sanzo's ( 1912) embryo, well preserved in 75% ethanol, was photographed by Storai'* in the Species Museum "La Specola" (MZUF 5911) in Florence in 1992 (Mojetta et al., 1997). The photo- graph— in color — suggested to us that the lost embryo had been found. This presented an opportunity for re-exam- ining the embryo and checking Sanzo's identification. The correct identification of the San- zo embryo is important to our under- standing of lamnid reproduction and possibly white shark conservation. Few pregnant female white sharks or em- bryos have been reported, and little is known about litter size, gestation period, or the timing and duration of the reproductive cycle (Uchida et al., 1987, 1996; Francis, 1996; Mollet et al., 2000). Such information is vital for understanding the population dynam- ics of the white shark, which is now regarded as a threatened species (Com- pagno et al., 1997). If the Sanzo embryo were a white shark, then it would be the smallest white shark embryo ever reported. Most have been greater than 100 cm TL ( Francis, 1996; Uchida et al., 1996), although Bigelow and Schroeder (1948) reported white shark embryos in the range 20-61.6 cm TL, without giving any details. A white shark litter with embryos of 61 cm TL (5.4 kg each, Ellis and McCosker, 1991) was caught in the Mediterranean Sea (Norman and Eraser, 1938). No description of the embryos was given; however, the mass of the embryo suggested that it had a ' Pratt, H. L. 1996. Personal commun. Narragansett Laboratory, National Ma- rine Fisheries Service, 28 Tarzwell Drive, Narragansett RI 02882. •^ Stevens, J. D. 1998. Personal commun. CSIRO Marine Research, P O. Box 1,538, Hobart, Tasmania 7001, Australia. ■^ Storai, T. 1992. Personal commun. Mu- seum of Natural Science of Valdinievole. Piazza L. da Vinci 1, Pescia PT Italy. Manuscript accepted 24 May 2002. Fish. Bull. 100;86.5-87,5 (2002). 866 Fishery Bulletin 100(4) substantial yolk-stomach as expected for a mid-term em- bryo (Mollet et al., 2000). Reports of pregnant females and small juveniles have indicated that white sharks breed in the Mediterranean (Fergusson, 1996). The reported litter size of the Sanzo shark (25-30) is the largest yet recorded for any lamnid shark. Therefore, correct identification of the embryo will also increase our knowledge of maximum fecundity in whichever species is involved. In this note, we report the results of our investigation into the identity of the Sanzo embryo. We first attempted to use morphomet- ries, dentition, and vertebral count and then, for unam- biguous identification, we used skeletal anatomy — namely, the chondro-neurocranium, palatoquadrate, and pectoral girdle. Materials and methods Materials Sanzo embryo (MZUF 5911) The preserved embryo in the "La Specola" Museum of the University of Florence, rediscovered and photographed by Storai-^ in 1992, was undoubtedly the embryo described by Sanzo (1912). Vanni,'* curator of fishes at La Specola, provided us with the following account: "The current collection number 5911 MZUF (= Museo Zoologico Universita di Firenze) cor- responds to precedent n.3052 of the "Collezione Centrale dei Vertebrati Italiani" (Italian Central Vertebrate Collec- tion), established by Giglioli in 1877 and now merged with the general collection. In the original catalogue ("libro di magazzino"), E. H. Giglioli himself wrote: "Carcharon Ron- deleti ?? feto. VIII. 1903 Messina. La femmina dal quale fu tolto pesava da 400 a 500 kg e oltre a questo aveva nell'utero 25-30 altri feti nelle medesime condizioni. Avuto dal dr Luigi Sanzo" ("Carcharadon Rondeletii ?? foetus VIII. 1903 Messina.") (The female from which the embryo was taken had a weight between 400 to 500 kg and m addition to this specimen, had 25-30 other fetuses in the same condition. Presented by Luigi Sanzo" ("Carcharadon Rondeletii ?? foetus VIII. 1903 Messina.")). The "Vertebrate Museum of the Superior Institute of Studies in Florence" corresponds exactly to the present "Museo Zoologico 'La Specola' deirUniversita di Firenze." The "Vertebrate Museum" is actually the "Museo Zoologico 'La Specola'"; the "Superior Institute of the Studies in Florence" in 1926 became "Universita degli Studi di Firenze" ("University of the Studies of Florence"). If in his paper Sanzo ( 1912) reported that the embryo was collected in 1903 near Mes- sina, then undoubtedly the specimen in question is that preserved at present in the "La Specola" Museum." Uchida embryo (SAM-35742) A 35.8-cm-TL female short- fin mako embryo was shipped to the South African Museum in Cape Town in November 2000 (SAM-35742). It came from a female (TL=3.37 m, 380 kg) that was caught near Okinawa, Japan, on 15 November 1984 (Uchida et al., 1987; Mollet et al., 2000). The litter comprised 16 embryos (11 females) with mean TL = 39.4 cm and mean mass = 1.456 kg and was preserved in formalin. A 38.4-cm-TL female embryo of this litter weighed 1.400 kg and the yolk-stomach content weighed 0.937 kg or 66.9% of the total mass (Uchida et al., 1987). Morphometries Morphometric measurements of the Sanzo embryo were taken by author ADT using the methods and abbrevia- tions of Compagno (1984). Total length (TOT) was mea- sured with the caudal fin in the extended position. In an embryo of this size, TOT is very close to total length (TL) measured with the caudal fin in the natural position. A flexible aluminum tape was used for measurements ex- ceeding 140 mm and rounded to the nearest millimeter Measurements less than 140 mm were made with calipers at 0.2 mm precision. The Uchida embryo was measured six years later by authors LJVC and HFM. We compared both ourandSanzo's( 1912) measurements with the morphometries of white sharks less than 4 m TL reported by Mollet et al. ( 1996) and with those of the short- fin makos summarized in Table 1. No morphometries of white sharks of less than 1.26 m TL were available; there- fore we had to use larger specimens for comparison, in- cluding two nearterm embryos reported by Francis ( 1996 ). This approach is reasonable for isometric characters, but it is inappropriate for allometric characters. The short- fin mako data included a shortfin mako litter of similar size to the Sanzo embryo (mean TL=36.2 cm, range 29.5- 39.5 cm; Putz and Gilmore litter in Table 1). Moron'' pro- vided additional eye length data from 51 shortfin makos in Moreno and Moron (1992). We focused attention on the four morphometries used by Sanzo (1912) to distinguish between white and shortfin mako sharks: snout shape; eye shape (EYL/EYH); ratio of mouth width to length (MOW/ MOL); distance between the origins of the second dorsal and anal fins (PAI^PD2)). Our preliminary analysis in- dicated that eye size (EYL) and the distance between the origin of the first dorsal fin and the pectoral fin free rear tip (PDl-PRT) might be more suitable for identification and they were also included. We tested these variables graphically for their ability to distinguish between the two species (Mollet and Cailliet, 1996). The statistical pro- gram SYSTAT-SYGRAPH (Wilkinson, 1986) was used for analysis and graph production. For specimens in which the relative positions of dorsal and anal fin origins were not measured directly, we calcu- lated them from the difference between snout to anal fin and snout to second dorsal fin measurements (DAO=PAL- PD2). For the evaluation of the relative positions of the first dorsal and pectoral fins, we calculated the distance ^ Vanni S. '2000. Personal commun. Sezione di Zoologia "La Specola," Museo di Storia Naturale deU'Universit, Via Romana, 17-50125 Firenze, Italy. Moron, J. 1994. Personal commun. Departamento doe Biologi'a Animal I, Universidad Complutense de Madrid, Madrid E- 28040, Spain. NOTE Mollet et al : Re identification of a lamnid shark embryo 867 Table 1 S aniniary ot'lsuriis oxyrinchus specimens used lor comparison of morphometries with Sanzo emhryo. TL(m) Comments Reference 0.295 Smallest of 15 embryos ' Putz, Gilmore - 0.360 Strait of Messina, male embryo This study ■* 0.361 Strait of Messina, male embryo Sanzo (1912) 0.362 Mean of 15 embryos' Putz, Gilmore - 0.395 Largest of 15 embryos ' Putz, Gilmore - 0.615 Nearterm female embryo Stevens (1983) 0.641 Nearterm male embryo Stevens (1983) 0.705 California, UMMZ 94726 Garrick(1967) 0.710 Smallest of 10 males and 8 females Basset al.( 1975) 0.847 Japan, MCA 35071 Garrick(1967) 1.130 Algoa Bay. male Smith (19.53) 1.251 Smallest of 11 males and 2 females Strasburg (1958) 1.438 New Zealand, NMNZ P.3014 Garrick(1967) 1.598 Ocean City MD, HMCZ 35899, male Bigelow and Schroeder ( 1948) 1.659 Mean of 53 males and females Moreno and Moron (1992) 1.920 South Africa, USNM 197686 Garrick(1967) 2.000 W. of Azores USNM 197706 Garrick(1967) 2.057 Mean of 1 1 males and 2 females Strasburg ( 1958) 2.169 Mean of 13 with range 2-3 m TL Gubanov(1974) 2.337 Bahamas, HMCZ 35367, male Bigelow and Schroeder (1948) 2.400 Algoa Bay, male Smith (1958) 2.579 Largest of 1 1 males and 2 females Strasburg (1958) 2.692 Provincetown MA, female Atwood ( 1869) 3.130 Largest of 10 males, 8 females Basset al.( 1975) 3.210 Carme! Bay CA, female Lea, Cailliet •* 3.366 Santa Catalina Island Ca, female Applegate(1966) 3.480 Redondo Beach CA, female Seigel « 3.507 Anacapa Island CA, female Applegate(1977) 3.800 Indian Ocean, female Gubanov(1974) ' 15 embryos from same litter. Putz and Gilmore litter - Putz. 0. 1995. Kennedy Space C Personal common, enter FL 32899. Grolmanstrasse 48, 10634 Berlin. Germany Gilmore, R, G. 1995. Personal commun. Dynamic Corp., ' Specimen identical to Sanzo 1912) embryo. ^ Lea, R.N. 1995. Landmg Road, M Personal common. California Fish & Game, Monterey CA 93940. Cailliet G. M. 1996. Personal common. MLML, 8272 Moss ass Landing CA 95039. ' Seigel, J, A, 1996. Persona common. LACM, Los Angeles CA 90007. between the first dorsal fin origin and the pectoral fin free rear tip (PD1-PRT = PD1-(PP1 + PIB + PID). IfPlBor PlI were not available for shortfin makos, we estimated (PlB + PlDas ll%ofTL(orTOT). Tissue samples for DNA sequencing Tissue samples were taken from the gill slits, the oral cavity, and the caudal peduncle. The samples were stored at room temperature in Wheaton polypropylene vials in QS^f ethanol. DNA was extracted from the samples and polymerase chain reaction (PCR) amplification and sequencing were attempted but did not yield useful results (Bernardi''). This was likely due to initial fixing of the specimen in formalin (which destroys DNA) before transferal to ethanol. X-ray analysis For vertebral counts, we used radiographs taken initially with a Siemens Triselenix 750 (in Milan) and later with a Shimadzu R20 computerized x-ray machine for high- ^ Bernard!, G. 1996. Personal commun. Dept. of Biology. Uni- versity of California, Santa Cruz, CA 95064. 868 Fishery Bulletin 100(4) resolution radiography (in Cape Town). A pin was inserted perpendicular to the upper origin of the caudal fin to count the precaudal vertebrae. Distinct shortening of the centra was used to distinguish between monospondylic and dip- lospondylic vertebrae. Dentition and mucous denticle examination We used the term "embryonic teeth" for teeth in an embryo, which do not resemble teeth in the adult (Gilmore, 1993). We used "row" for teeth at the same developmental stage in the mesial-distal direction and "file" for teeth at dif- ferent developmental stages in the labiolingual direc- tion derived from a single locus (tooth germ) (Zangerl, 1981). We followed Applegate (1965) for the terminology anterior, intermediate, lateral, and posterior teeth and for the definition of the dental formula. This terminology and definition apply to fully formed dentition in postnatal sharks but they appeared to be applicable to the embryo under investigation. If a tooth was missing in the first row, the one behind it was counted. We used the term "mucous denticles" for dermal denticles in the oral cavity (Yano et al., 1997). We examined the mucous denticles with a Leitz DMRB microscope at 25x and 45x magnification. The jaws could not be removed; therefore a nondestruc- tive examination was carried out with close attention to the positions of teeth in the functional and replacement rows in the upper jaw. The functional tooth and the first replacement tooth in the fifth file (counted from the sym- physis) of the upper left jaw were extracted and examined through a Leitz DMRB microscope at 25x and 45x magni- fication. The examination of replacement teeth required lifting the tissue that covered the developing teeth. The lower jaw was not examined as closely because only one tooth was visible to the naked eye. We measured enameloid height (E2), if possible, for all teeth in the upper jaw (Mollet et al., 1996). We estimated the total vertical height (H) of the largest tooth, from a 25x photograph showing the outline of the root, for comparison with the likely total heights reported by Sanzo ( 1912). We calculated the enameloid height of each tooth in relation to the third tooth. For comparison, we estimated tooth sizes in relation to the third tooth of postnatal shortfin makos from photogi'aphs or drawings (Bigelow and Schro- eder, 1948; Bass et al., 1975; Compagno, 1984). Dissection The Sanzo and the Uchida embryos were dissected conser- vatively in order to examine the internal structure of the head, jaws, and pectoral fins. The chondroneurocranium was exposed dorsally and on the left side by dissecting away flaps of skin, muscle, and connective tissue. We examined the structure of the ethmoid region, epiphysial area, orbital process, and otic capsule. Dissection of the left lateral surface of the embryos head exposed the pala- toquadrate and allowed observation of the proportions of the palatine process. Dorsal dissection of the pectoral fin allowed examination of the basal metapterygium's skel- etal structure. No white shark embryo of suitable size was available for dissection and direct comparison. The white shark embryo (TL=55 cm) described by Parker ( 1887) was a misidentified Carcharhinus (Francis, 1996). Results General condition and morphometries After more than 90 years of storage in a glass container in 75% ethanol, the 36.1-cm male Sanzo embryo (MZUF 5911) was curled up, and fins and other body parts were permanently bent (Fig. lA). The jaws appeared protruded, possibly because of the strong retraction and shrinkage of the snout. The large yolk stomach was hardened. It was 13.4 cm long, 8.6 cm wide, and 6.9 cm high and had an estimated volume of 416 cnr^. The embryo weighed 0.548 kg (condition factor, CF=11.7 kg/m^) compared with 0.800 kg (CF=17.1 kg/m^) reported by Sanzo (1912). The 31.5% mass loss was likely due to dehydration and dissolving and leaching of lipids from the yolk and liver into the ethanol. Despite this, the embryo did not appear to have shrunk in length because it still measured 36.0 cm TOT, 30.0 cm fork length (FOR), 27.4 cm precaudal length (PRO. After 16 years in formalin, the female Uchida embryo (SAM-35742) looked shriveled (Fig. IB). The embryo mea- sured TOT = 35.8 cm, FOR = 28.8 cm, PRC = 26.6 cm, mass = 1.227 kg on 6 June 2001. The condition factor of 26.7 kg/ m-* of this embryo was similar to that calculated from the reported mean length and mass of all the embryos of the litter by Uchida et al. (1987) (CF=23.8 kg/m*). TOT was almost the same as that of the Sanzo embryo, but this em- bryo weighed almost twice as much as the Sanzo embryo. Accordingly, the yolk stomach was considerably larger and was 18.5 cm long, 9.6 cm wide, and 11.5 cm high and had an estimated volume of 1069 cm '. Sanzo (1912) used the upturned snout, as one of four characters to distinguish his embryo from a shortfin mako, but this feature is probably an artifact of preservation (see "Skeletal anatomy" below). The three quantifiable mor- phometric characters used by Sanzo (1912) were also not suitable to distinguish between small white and shortfin mako sharks (Fig. 2). White sharks generally have a wider mouth, in relation to its length, than shortfin makos, but there is significant overlap (Fig. 2A). Furthermore, the ra- tio of mouth width to length (MOW/MOD is allometric in small shortfin makos; mouth width becomes progressively larger than mouth length in smaller embryos. White and shortfin mako sharks both have slightly oval to round eyes and an eye length-to-height ratio between 0.9 and 1.3 (Fig. 2B). The origin of the anal fin is behind the origin of the second dorsal fin (PAL-PD2 > 0) in both white and short- fin mako sharks (Fig. 2C). Two promising morphometries not considered by Sanzo (1912) also proved unsuitable for identification. The first dorsal fin origin (PDl) of the Sanzo embryo was 2.3 cm (6.4% TOT) behind the pectoral fin rear tip (PRT), which suggested it might be a shortfin mako following Compagno (1984). However, the origin of the dorsal fin in both white and shortfin mako sharks varies from slightly-in-front-of NOTE Mollet et al : Re-identification of a lamnid shark embryo 869 Figure 1 (A) Sanzo ( 1912) lamnid embryo (TOT 36.0 cm. MZUF 591 1). Smallest scale intervals = 1/32". ( B) Uchida ( 1989) shortfin mako embryo ( TOT 35.8 cm, SAM 35742 1. Smallest scale interval = 1 mm. (C) Sanzo embryo ventral view of mouth focusing on upper right teeth in first row. Smallest scale intervals = 1 mm. (D) Sanzo embryo functional 5th upper left tooth (enameloid height E2 -2 mm). (E) Sanzo embryo replacement 5th upper left tooth (E2 -2.3 mm. total vertical height H -2.9 mm). Thin layer of tissue is covering apex. to slightly-behind the pectoral fin free rear tip (Fig. 2D, PDl-PRT). The eyes of the Sanzo embryo were unexpect- edly small (EYL=1.4% and 1.8%, Sanzo's (1912) and our measurement, respectively) compared with those reported for nearterm shortfin mako embryos and neonates (e.g. EYL=2. 7-2.9% TL; Stevens, 1983). On the other hand, they were similar in size to those of the Uchida embryo (1.7% ) and the Putz and Gilmore litter (1.4-1.8%; Fig. 2E). Small shortfin mako embryos have small eyes; but rela- tive eye length increases rapidly and reaches a maximum of about 3% in near-term embryos of 60-64 cm TL, before declining in postnatal fish. No secondary caudal keel was observed in the Sanzo embryo by Sanzo (1912) or by us. Nevertheless, this does not allow the elimination of the porbeagle; secondary keels may be difficult to detect in preserved porbeagle embryos 870 Fishery Bulletin 100(4) because of wrinkling of the skin (Francis, personal observ.) Lohberger (1910) did not observe a sec- ondary keel in preserved salmon shark (Lamna ditropis ) embryos, although this keel is present in postnatal specimens (Compagno, 1984). Dentition The first observations by naked eye and magnify- ing lens suggested a tooth formula of 8-0-7 for the upper jaw. However, the functional tooth row (i.e. the outermost row containing erect, functional teeth) of the upper right jaw was not completely filled by teeth; it comprised eight visible teeth (in file positions 2-6 and 8-10) and six gaps (in file positions 1, 7 and 11-14 (Table 2). The gaps in the functional row were indicated by the presence of teeth in the replacement rows. The gap in file 1, the broken tooth in file 2 (the 2nd tooth on the left was not erect), and the teeth in files 3-6 (labeled) are in focus in Fig. IC. The largest functional tooth in file 6 had an enameloid height of 2.4 mm. The tooth formula for Sanzo's embryo indicated that an embryo of this size and developmental stage has the full adult complement of replace- ment tooth files in the upper jaw; two anteriors, one intermediate, eight laterals, and three poste- riors (Table 2). The first and second replacement rows contained teeth in all file positions, indicat- ing an eventual tooth formula of 14-0-14. The functional tooth extracted from the 5th file in the left upper jaw (E2 -2 mm) was fanglike and was without any lateral flattening and we considered it to be an embryonic tooth (Fig. ID I. The replacement tooth behind it was slightly cui-ved, had little lateral flattening, and a thin layer of tissue still covered the apex (E2 -2. .3 mm and H -2.9 mm) (Fig. IE). We suggest that this tooth is also an embryonic tooth. The relative po- sition of these two teeth is as shown in Figs. ID andE. The relative heights of the embryonic teeth in the upper jaw differed considerably from those of postnatal shortfin makos (Table 2). The first two teeth in the Sanzo embryo were much smaller than the teeth in files 3-8; the largest tooth was in file 6. In postnatal shortfin makos, the first two teeth are the largest, followed by a much smaller third tooth and smaller ones in files 4—13 (Table 2). The tooth formula for the lower jaw was less certain. Our initial obsei-vation with magnifying lens indicated 4- 0-7. Sanzo ( 1912) reported 4-0-4 and we agree with Sanzo that the third lower tooth was the most prominent. Mucous denticles Mucous denticles covered the palate and the tongue. Micro- scopic investigation revealed that the wartlike struc- tures were round and had a circular, flat base and a small upward-pointing center cusp. No ridges were noted ) 2 3 4 O :^ A ■ ° « O O - ■s. • O cP 5 • o° ^§ t ° oo ° 8 O o ,. - m :• o o© a o .!.■* o o o 9.o' 1 , ■ ,0 oo B . - • " 0 0 0%° • ■ oo O 0 ■ ■ ceo 0 1 .... 1 . ■ 8 # ■ ■ ■ ■ o o ■ ■ o - o_ ^°o oo • o oo<* o ,°(p D 1 T ■ 1 .... 1 ■ 0 o ■ ■ o° o " o o 0 o . ■ ■ o oS ° o 1 1 1 O" o o O Total length (m) Figure 2 Relationships between selected morphometries and total length of Carcharodon carcbarias lopen circle) and hums oxynnchus (filled square). Sanzo embryo data (Sanzo, 1912 and our study) are mdi- cated by dual symbols. (A) Ratio of mouth width to mouth length (MOW/MOD. (B) Ratio of eye length to eye height (EYL/EYH). (C) Relative position of the origins of anal and second dorsal fins (PAL- PD2). (D) Relative position of first dorsal fin origin and pectoral fin free rear tip (PDl-PRT). (E) Eye length (EYL as 9, TLl. between the cusp and the base. The mucous denticles were small and far apart on the tongue, slightly larger and closer together on the palate, and largest ( -0.4 mm diameter) and packed together in the region close to the cartilage of the upper jaw. There were few in the region close to the lower jaw and on the terminal part of the tongue. Skeletal anatomy The cranmm of the Sanzo embryo differed notably from that of postnatal lamnids. The chondrocranium of the em- bryo was evidently damaged and foreshortened by the gen- NOTE Mollet et al Re identification of a lamnid shark embryo 871 Table 2 Characterization ofujipor right dentition of Sanzo embryo (A and B) an< comparison of relative tooth sizes with postnatal Isurus oxyrinchi/s (C). We report approximate enameloid height (E2) of teeth in mm. Sanzo ( 1912) proh ibly re ported total height. G = gap; P = present. File number 12 3 4 5 6 7 8 9 10 11 12 13 14 Jaw position' A2 A3 I LI L2 L3 L4 L5 L6 L7 L8 PI P2 P3 A Results of this study for Sanzo embryo Functional row - G P^ 1.5 1.2 2.2 2.4 G 1.4 P P G G G G 1st replacement row 0.7 P P P 2.3^ P 1.4 P P P P P P P 2nd replacement row P P P P P P P P P P P P P P B Results ofSanzol 1912) Functional row ' «1 <1 <1 P P P 4 <1 <1 <1 1st replacement row P P C Relative Isurus oxyrinchus teeth size in functional row Sanzo {this study) -0.4« — 1.0 0.8 1.5 16 -0.9" 0.9 7 — — — — — B&S"', 1948 2.3 1.9 1.0 1.2 1.3 1.4 1.2 1.1 0.8 0.7 0.5 0.4 0.3 Bass ct al., 1975 2.1 1.9 1,0 1.3 1.5 1.5 1.2 0.9 0.7 0.6 0.5 0.4 Compagno, 1984 2.1 1.9 1.0 1.1 1.2 1.4 1.3 1.1 0.9 0.7 0.4 0.35 0.3 ' Al is missing in postnatal Isurus oxyrinchus (Applegate and Esp inosa, 1996J. - Eight teeth present (P) and six gaps (G). ' Broken tooth. 2nd tooth on left not erect. ■* Size estimate of upper left tooth from photograph of extracted tooth (F g ID). Est total height ( H i = 2.9 mm. "" Tooth sizes preceded with < and « signs are estimates based on qualitative descriptions by Sanzo (1912): 1st tooth almost invisible. 2nd and 3rd tooth a little more developed than the 1st one and about equal in size, 3rd tooth almost half the size of the fourth tooth. teeth 4-7 much better developed, 7th tooth 4 mm. '' Based on enameloid height of first replacement tooth. " Relative size could not be estimated. '' B&S, Bigelow and Schroeder. eral compression of its snout. This presumably was a result of being fixed and presetted in a nairow jar, and the weight of the massive yolk stomach providing sufficient force to compress the snout. The cranium had an extremely short ethmoid region compared with that of postnatal lamnids, which was exaggerated by snout foreshoilening. The ros- tral cartilages were only basally developed and partially crushed and had no well-developed rostral node. The pro- truding orbits were large but short and the otic capsules were more elongated than in postnatal lamnids. In other features, the cranium agreed with that of the Uchida embryo and postnatal shortfin makos (Table 3). The bases of the lateral rostral cartilages were positioned on the nasal capsules, as in white and mako sharks, rather than on the preorbital processes as in Lamna (Compagno, 1990). The ethmoid region across the nasal capsules was relatively narrow, as in shortfin makos and porbeagles; white sharks, in contrast, have notably broad nasal cap- sules (Haswell, 1885; Parker, 1887; Compagno, 1990). The cranial roof of white sharks has an epiphysial bar and epiphysial fenestrum just behind the anterior fontanelle, but this is absent in postnatal crania of shortfin makos, porbeagles, and Sanzo's embryo (Compagno, 1990). The upper jaw and the pectoral girdle of the Sanzo embryo agreed with those of the Uchida embryo and postnatal makos (Table 3). The palatine processes of the palatoquadrate were low, elongated, and ventrally bent or twisted as in shortfin makos. White sharks have higher, straight, and thicker palatine processes (Compagno, 1990). Porbeagles, postnatal shortfin makos, and Sanzo's embryo all have an unsegmented metapterygium in their pectoral fin skeletons, whereas white sharks have a transversely segmented basal metapterygium (Compagno and Gott- fried, unpubl. data). The precaudal vertebral count of Sanzo's embryo (110 centra in total, including 73 monospondylous and 37 dip- lospondylous centra) fell close to the average for shortfin makos, whereas white sharks and porbeagles have fewer precaudal vertebrae (Table 3). The caudal vertebrae count of Sanzo's embryo (77) falls in the range of both white sharks and shortfin makos and is shghtly greater than caudal counts for porbeagles. Caudal vertebral counts from radiographs are often unreliable in newborn and late fetal sharks because of poor calcification of the posterior end of the vertebral column. The caudal vertebrae of the Sanzo embryo were difficult to count without dissection because they were small and are not expected to be fully formed until late in embryonic life (Springer and Garrick, 1964). The Uchida embryo is of similar length but is considerably heavier compared to the Sanzo embryo, but its vertebral column was insufficiently calcified and we were unable to obtain a precaudal vertebral count from the x-rays taken. 872 Fishery Bulletin 100(4) Table 3 Comparison of skeletal anatomy of the Sanzo (1912) embryo with that of the Uchida et al. (1987) Isurus oxynnchus postpartum Isurus oxyrinchus, Carcharodon carcharias, and Lamna nasus. embryo, and Description Sanzo Uchida Isurus oxyrinchus Carcharodon carcharias Lamna nasus Chondrocranium Epiphysial bar and epiphysial fenestrum just behind the anterior fontanelle No No No Yes No Ethmoid region narrow across nasal capsule Yes Yes Yes No Yes Bases of the lateral rostral cartilages positioned on the nasal capsule (not preorbital processes) Yes Yes Yes Yes No Upper jaw Palatine processes low, elongated, and ventrally bent or twisted Yes Yes Yes No No Pectoral girdle Unsegmented inner basals (metapterygium) Yes Yes Yes No Yes Vertebral counts Precaudal 110 1 104-1 142 99-1082 83-912 CaudaF 77 1 79-86 68-83 68-712 Total 187 1 183-194-' 172-187- 150-1622 ^ Vertebral column not sufficiently calcified. - Combined ranges from Springer and Garrick (1964), Bass et al. (19751, and L. J. V. Compagno u ■> Caudal vertebrae may not be fully formed in early stage embryos (Springer and Garrick, 19641. npublished precai dal data). Discussion Skeletal anatomy and morphometries We had to use skeletal anatomy, including the chondro- neurocranium, palatoquadrate, and pectoral girdle for unambiguous identification of the Sanzo (1912) embryo after capture data and the vertebral count suggested that the embryo might be a shortfin mako rather than a white shark. Our attempts to use morphometries, dentition, and DNA analysis were not successful. Sanzo (19121 correctly placed the embryo in the family Lamnidae using only morphometric criteria. Only three species of the family Lamnidae normally occur in the Strait of Messina of the Mediterranean Sea: porbeagle, white, and shortfin mako sharks (Compagno, 1984; Fergus- son, 1996). However, the morphometric arguments used by Sanzo (1912) for identification to species were not charac- teristic, leading him to incorrectly eliminate the genus Isu- rus. The upturned snout was probably caused by distortion during preservation, and we have shown that the mouth width-to-length ratio, eye shape, and relative positions of the origins of second dorsal and anal fins are not suitable criteria for distinguishing between white and shortfin mako sharks. Other promising morphometries also failed to distinguish between the two species. These conclusions are tentative — confirmation will depend on obtaining mea- surements of these characters from small embryonic white sharks, which were missing from our database. Dentition Tooth shape is species-specific in postnatal Lamna, Isurus, and Carcharodott (Compagno, 1984) but not in embryos. Lamnid embryos have specialized "embryonic" teeth that are adapted for grasping and tearing the membrane of the eggcases on which they feed (Gilmore, 1993; Francis and Stevens, 2000). We observed fanglike embryonic teeth in porbeagle embryos, which lacked the characteristic cusplets of adult specimens. We observed fanglike embry- onic white shark teeth lacking serration in the intestine of a nearterm embryo similar to the embryonic teeth of the Sanzo embryo (Francis. 1996; Francis and Stevens. 2000). Embryonic teeth are similar in all lamnid embryos and do not appear to be suitable for identification. Shortfin mako embryos shed their embryonic dentition at about 45-50 cm TL and nearterm embryos have emerging adultlike teeth (Gilmore, 1993; Mollet et al.. 2000). It is difficult to describe a dentition completely without the benefit of prepared jaws, particularly in a relatively small embryo. In addition, the tooth formula of a small embryo may be different from that of a postnatal speci- men. Sanzo (1912) reported an upper jaw tooth formula of 10-0-10; we observed 14-0-14. the full adult complement of replacement tooth files. That suggests that Sanzo (1912) had not observed the four replacement teeth in files 11-14. He reported two replacement teeth behind the functional teeth in files 7-8 and 8-9. whereas we observed replace- ment teeth in all position of two rows by pulling back the dental lamina (Table 2). NOTE Mollet et al.: Re-identification of a lamnid shark embryo 873 We could not resolve other discrepancies. Sanzo (1912) reported four relatively large teeth in files 4-7, the 4th one was more than twice the size of the 3rd one, and the 7th tooth was 4 mm (probably including the root). We con- cluded that the four largest teeth were in files 3-6 and we estimated the total height of the largest tooth in file 6 to be about 3 mm (based on E2=2.4 mm). It is possible that Sanzo's ( 1912) minute first tooth was a recessive parasym- physial tooth, which we overlooked or which had disap- peared before we examined the embryo. Our description of the dentition of the Sanzo embryo agrees in general with that in similar-size salmon shark and porbeagle embryos (Lohberger, 1910; Mollet, personal observ.). DNA sequencing and mucous denticles DNA sequencing should have allowed identification of the Sanzo embryo but provided no useful results. This was likely due to initial fixing of the Sanzo embryo in formalin (which destroys DNA) before transferal to ethanol. We did not have SEM at our disposal during the initial stage of the investigation but suggest that the mucous denticles might be suitable to identify lamnid embryos^ Early development of mucous denticles is expected to occur in the oophagous lamnids (Reif 1985; Raschi and Tabit, 1992). Reif (1985) suggested that dermal denticles are family, genus, and in some cases even species specific. Postnatal shortfin makes and white sharks have different mucous denticles (Reif, 1985; Peyer, 1968). Capture information The capture information provided by the fisherman who caught the Sanzo shark was more consistent with shortfin mako than with other lamnids, based on presently known lamnid reproductive biology (Francis and Stevens, 2000; Mollet et al., 2000). The litter size of 25-30 was estimated and may have been inaccurate, but it does indicate a large litter The large litter size could be the reason that the Sanzo embryo weighed considerably less than the Uchida embryo although they had similar length. Litter size in porbeagles is nearly always four (maximum of five, Fran- cis. 1996; Francis and Stevens, 2000). Maximum litter size in white sharks is at least ten; unconfirmed reports are as high as 14 (Francis, 1996). Shortfin makos have the larg- est litters yet reported in the Lamnidae, reaching at least 18 (Branstetter, 1981; Mollet et al., 2000). The shark was estimated to weigh 400-500 kg, although this must be considered approximate. The TL of a female of this mass would be 3.58-3.85 m (Stevens, 1983; Mol- let et al., 2000). A shortfin mako of this length would undoubtedly be mature (Mollet et al., 2000) and the maxi- mum reported length is 4 m (Bigelow and Schroeder, 1948; Mollet^). Female white sharks do not mature until about 5.0 m (Mollet et al., 2000) with a corresponding mass of -1200 kg (Mollet and Cailliet, 1996). A full-term litter of 25-30 shortfin makos would weigh ca. 75 kg, which would ' Mollet, H.F. 1999. http://homepage.mac.com/mollet/lo/Io_large. html. [Access date: 9 August 2002.1 be reasonable for a female shortfin mako weighing around 500 kg (Mollet et al., 2000). A white shark litter of 25-30 would weigh about 500 kg at birth, i.e. the total mass of the female shark caught, which is not possible. We conclude that the Sanzo embryo is Isurus oxyrin- chus. This analysis corrects a long-standing error in the literature and should provide the incentive to procure and describe a white shark embryo of similar develop- mental stage to that of Sanzo's embryo. The smallest photo-documented white shark embryos were ca. 1.0-1.1 m TL (Uchida et al., 1996) and fully documented white shark embryos were all nearterm and had a TL between 1.35-1.51 m (Uchida et al, 1987, 1996; Francis, 1996). The definite identification of the Sanzo embryo suggests that the maximum litter size of the shortfin mako is hkely larger than 18 (Branstetter, 1981; Mollet et al., 2000) and possibly as large as 25-30. Acknowledgments P. Stipa and A. Young translated Sanzo (1912) which gave this study momentum. A photograph taken by T. Storai led to our rediscovery of the Sanzo embryo. S. Vanni and M. Poggesi allowed a partially destructive examination of the embryo and we are grateful to S. Vanni for the histori- cal identification of the specimen. E. Tavani and G. Sola helped with tooth histological sections and the first body radiographs. G. Bernardi spent much time attempting DNA sequencing. O. Putz, R. G. Gilmore, and J. Moron pro- vided morphometries of embryonic and postnatal shortfin mako embryos. G. Cailliet suggested the use of x-rays. We are grateful to Y. Kamei and S. Uchida for donating a shortfin mako embryo specimen and to A. Dainty and B. Human for help with the dissection. We also thank S. Branstetter, J. Bruner, J. Casey, G. Cliff, I. Fergusson, B. Habrich, R. Lea, G. Notarbartolo di Sciara, T. Otake, R. Phillips, R. Purdy, H. Pratt, O. Putz, J. Stevens, and S. Wintner for data and helpful suggestions and comments. HFM acknowledges computer support provided by the Monterey Bay Aquarium. The comments of three review- ers improved the manuscript. Literature cited Applegate, S. P. 1965. Tooth terminology and variation in sharks with spe- cial reference to the sand shark. Carcharias taiirus Rafin- esque. Contributions in Science (Los Angeles County Mu- seum) 86:3-18. 1966. A possible record-sized bonito shark, Isurus oxyrin- chus Rafinesque, from Southern California. Calif Fish Game 52:204-207. 1977. A new record-sized bonito shark, Isurus oxyrinchus Rafinesque, from Southern California. Calif Fish Game 63:126-129. Applegate, S. P., and L. Espinosa-Arrubarrena. 1996. The fossil history of Carcharodon and its possible ancestor, Cretolamna: a study in tooth identification. In Great white shark: the biology of Carcharodon carcharias 874 Fishery Bulletin 100(4) (A. P. Klimley and D. G. Ainley, eds.), p. 19-36. Academic Press, San Diego, CA. Atwood, N. E. 1869. Description of a shark, Carcharias tigris Atwood. Proceedings of the Boston Society of Natural History 12: 268-9. Bass, A. J., J. D. D'Aubrey, and N. Kistnasamy. 1975. Sharks of the east coast of southern Africa. IV. The famihes Odontaspididae, Scapanorhynchidae, Isuridae, Ce- torhinidae, Alopiidae, Orectolobidae and Rfiiniodontidae. Oceanogr. Res. Inst. Invest. Rep. 39:1-102. Bigelow, H. B., and W. C. Schroeder. 1948. Part 1. Sharks. In Fishes of the western North Atlan- tic, p. 59-546. Sears Foundation for Marine Research, Yale University. Branstetter, S. 1981. Biological notes on the sharks of the north central Gulf of Mexico. Contrib. Mar Sci. 24:13-34. Compagno, L. J. V. 1984. FAO species catalogue. Sharks of the world. An annotated and illustrated catalogue of shark species known to date. FAO Fisheries Synopsis 125, vol. 4, part 1. Hexanchiformes to Lamniformes, p. 1-249. FAO, Rome. 1990. Relationship of the megamouth shark, Megachasma pelagios (Lamniformes: Megachasmidae), with com- ments on its feeding habits. In Elasmobranchs as living resources: advances in the biology, ecology, systematics, and the status of fisheries (H. L. Pratt, S. H. Gruber, and T. Taniuchi, eds. ), p. 357-379. U.S. Dep. Commer., NOAA Technical Report NMFS 90. Compagno, L. J. V., M. A. Marks, and I. K. Fergusson. 1997. Threatened fishes of the world: Carcharodon car- charias (Linnaeus, 1758) (Lamnidae). Environ. Biol. Fish. 50:61-62. Ellis, R., and J. E. McCosker. 1991. Great white shark, 270 p. Harper Collins Pubis., New York, NY. Fergusson, I. K. 1996. Distribution and autoecology of the white shark in the eastern North Atlantic Ocean and the Mediterranean Sea. In Great white shark: the biology of Carcharodon carcharias (A. P. Klimley and D. G. Ainley, eds. ), p. 321-345. Academic Press, San Diego CA. Francis, M. P. 1996. Observations on a pregnant white shark with a review of reproductive biology. In Great white sharks: the biology of Carcharodon carcharias (A. P. Klimley and D. G. Ainley, eds.), p. 157-172. Academic Press, San Diego, CA. Francis, M. P., and J. D. Stevens. 2000. Reproduction, embryonic development and gi'owth of the porbeagle shark, Lamna nasus, in the Southwest Pacific Ocean. Fish. Bull. 98:41-63. Garrick, J. A. F. 1967. Revision of sharks of genus Isurus with description of a new species (Galeoidea, Lamnidae). Proc. U. S. Nat. Mus. 118(35371:663-694. Gilmore, R. G. 1993. Reproductive biology of lamnoid sharks. Env. Biol. Fish. 38:95-114. Gubanov,Ye. P 1974. The capture of a giant specimen of the mako shark (Isurus glaucus) in the Indian Ocean. J. Ichthyology 14: 589-591. Haswell, W. A. 1885. Studies on the elasmobranch skeleton. Proc. Linn. Soc. N.S.Wales 9:71-119, Lohberger, J. 1910. Uber zwei riesige Embryonen von Lamna. Abh. K. Bayer Akad. Wiss. IV, Suppl. Bd. 2:1-45, Mojetta, A., T Storay, and M. Zuffa. 1997. Notes on the presence and distribution of the white shark (Carcharodon carcharias) in Italian waters. Quad. Civ. Staz. Idrobiol. 22:125-128. Mollet, H. F, and G. M. Cailliet. 1996. Using allometry to predict body mass from linear measurements of the white shark. In Great white sharks: the biology oi Carcharodon carcharias (A. P. Klimley and D. G. Ainley, eds.), p. 81-89. Academic Press, San Diego CA. Mollet, H. F, G. M. Cailliet, A. P Klimley, D. A. Ebert, A. T Testi, and L. J. V. Compagno. 1996. A review of length validation methods for large white sharks. In Great white shark: the biology of Carcharodon carcharias (A. P. Klimley and D. G. Ainley, eds.), p. 91-108. Academic Press, San Diego, CA. Mollet, H. F, G. Cliff, H. L. Pratt Jr., and J. D. Stevens. 2000. Reproductive biology of the female shortfin mako Is- urus oxyrinchus Rafinesque, 1810, with comments on the embryonic development of lamnoids. Fish. Bull. 98:299— 318. Moreno, J. A., and J. Moron. 1992. Comparative study of the genus Isurus (Rafinesque, 1810) and description of a form Cmarrajo criollo') appar- ently endemic to the Azores. In Sharks: biology and fish- eries (J. G. Pepperell, ed.), p. 109-22. Aust. J. Mar. Fresh- water Res. 43. Neill, P 1811. A list of fishes found in the Firth of Forth, and rivers and lakes near Edinburgh, with remarks. Mem. Werner- ian Nat. Hist. Soc. (Edinburgh). Vol. I (For the years 1808,- 9,-10):526-555. Norman, J. R., and F C. Fraser. 1938. Giant fishes, whales and dolphins, 361 p. W. W. Nor- ton and Company Inc., New York, NY. Parker, T J. 1887. Notes on Carcharodon rondeletii. Proc. Zool. Soc. London 1:27-40. Peyer, B. 1968. Comparative odontology, 468 p. Translated and edited by R. Zangerl. Univ. Chicago Press, Chicago IL. Raschi,W., andC.Tabit. 1992. Functional aspects of placoid scales: a review and update. In Sharks: biology and fisheries (J. P. Pepperell, ed.), p. 123-47. Aust. J. Mar Freshwater Res. 43. Reif, W. E. 1985. Squamation and ecology of sharks. Courier Forsch.- Inst. Senckenberg, Frankfurt am Main. 78:1-255. Sanzo, L. 1912. Embrione di Carcharodon Rondeletii M, Hie, (?) con particolare disposizione del sacco vitellino. Regie Comi- tate Talassografico Italiano, Memoria XI:1-12. Shann, E. W. 1911. A description of the advanced embryonic stage of Lamna cornuhica. Fish. Board Scotland 28:73-79. Smith, J. L. B. 1953. The shark, Isurus oxyrinchus. in South African waters. Nature 171:977. 1958, Sharks of the genus Pterolammops Springer, 1951 with notes on isurid sharks, Rhodes Univ. Ichthyol. Bull. 10:131-134. Springer, V. G., and J. A. F. Garrick. 1964, A survey of vertebral number in sharks, Proc, U, S, Nat. Mus. 116(3496):73-96. NOTE Mollet et al.; Re-identification of a lamnid shark embryo 875 Stevens. J. D. 1983. Observations on reproduction in the shortfin mako hums oxyrinchus. Copeia 1983:126-130. Strasburg, D. W. 1958. Distribution, abundance, and habits of pelagic sharks in the Central Pacific Ocean. Fish. Bull. 58:335-361. Tortonese, E. 1950. Studi sui plagiostomi. III. La viviparita: un fonda- mentale carattere biolico degli squali. Archivio Zoologico Itahano 35:101-155. 1956. Fauna d'ltalia. Vol. II, Leptocardia ciclostomata sel- achii, 321 p. Edizioni Calderini, Bologna, Italy, Uchida, S., M. Toda, K. Teshima, and K. Yano. 1996. Pregnant white sharks virith near-term fetuses from Japan. In Great white sharks: the biology ofCarcharodon carcharias (A. P. Klimley and D. G. Ainley, eds.), p. 139-155. Academic Press, San Diego, CA. Uchida, S., F. Yasuzumi, M, Toda, and N. Okura. 1987. On the observation of reproduction in Carcharodon car- charias and Isurus oxyrinchus. Rep. of Japanese Group for Elasmobranch Studies 24:5-6. Wilkinson, L. 1986. SYSTAT: the system for statistics. 18 chapters and 4 appendices without continuous page numbers. SYSTAT Inc., Evanston IL. Yano K., M. Goto, and Y Yabumoto. 1997. Dermal and mucous denticles of a female megamouth shark, Megachasma pelagios, from Hakata Bay, Japan. In Biology of the megamouth shark (K. Yano, J. F. Morrissey, Y. Yabumoto, and K. Nakaya, eds.), p. 77-91. Tokai Uni- versity Press. Tokyo. Zangerl, R. 1981. Chondrichthyes I Paleozoic Elasmobranchii. Volume 3A. In Handbook of paleoichthyology iH.-P Schultze, ed.), p. 115 p. +116 figs. Gustav Fischer Verlag, Stuttgart, Ger- many, and New York, NY. 876 Necropsy findings in sea turtles taken as bycatch in the North Pacific longline fishery Thierry M. Work Hawaii Field Station National Wildlife Health Center U.S. Geological Survey 300 Ala Moana Blvd., Room 5-231 Honolulu, Hawaii 96850. E-mail address, thierryworkia'usgs.gov George H. Balazs Honolulu Laboratory Southwest Fisheries Science Center National Manne Fisheries Service, NOAA 2570 Dole St. Honolulu, Hawaii 96822. Concern about interactions between fisheries and marine turtles has in- creased in recent years, particularly since East Pacific leatherback turtles (Derrnochelys coriacea) may become extinct (Spotila et al., 2000). However, relatively little published information exists on interactions between sea turtles and North Pacific longline fish- eries. The most available literature on the topic focuses on modeling data from fisheries observers for estimating the probability of animals dying and fish- ery-induced mortality (McCracken'; Kleiber-). A more recent study was undertaken with satellite telemetry and remote sensing to evaluate the probability of interaction between longline fisheries and loggerhead sea turtles iCaretta caretta) and the effects of hooking (Polovina et al., 2000; Parker, in press). Necropsies on turtles caught by longline fisheries may provide ad- ditional objective data on the causes of mortality and the health of pelagic turtles. Although ample literature ex- ists on evaluating the health of benthic coastal-residing immature sea turtles in Hawaiian waters (Aguirre et al., 1994; Work and Balazs, 1999), noth- ing is known about the health status of pelagic sea turtles because of the difficulty in locating animals (Boltcn and Balazs, 1983) and the unavailabil- ity of specimens for diagnosis. Most turtles caught in longline fisheries are released alive (McCracken'; Kleiber^). The few dead turtles that area recov- ered can be returned to shore legally only by observers (who are present in only ~59c of the Hawaii-based North Pacific fishing fleet) (Balazs et al, 1995). Nevertheless, examining freshly dead turtles caught in longline fisher- ies provides a unique opportunity to gain insight into the health status and diet of pelagic sea turtles. Our objec- tive was to systematically evaluate all available carcasses of fresh-frozen sea turtles that had been caught in the Hawaii-based longline fishery for an evaluation of their health and to docu- ment their diet. Methods Free-ranging marine turtles acciden- tally taken as bycatch by the North Pacific longline fishery were landed on the fishing vessel and evaluated for signs of life by fishery observers employed by the National Marine Fisheries Service. Sea turtles that were judged to be dead by specific cri- teria (Balazs et al., 1995) were stored frozen and returned to Honolulu, Hawaii, where we recorded weight (kg) and body morphometries (cm). Gross necropsies entailed a complete external and internal exam of all organ systems. We also recorded any identifi- able stomach contents. Body condition of turtles was subjectively classified as good, fair or poor if coelomic and mesen- teric fat reserves appeared ample, mod- erate, or sparse, respectively. Postmor- tem condition was classified as good, fair, or poor depending on the gross ap- pearance of organs during the nec- ropsy. We classified turtles as lightly hooked if the longline fish hook was lodged in the mouth or externally, or as deeply hooked if the hook was pres- ent in the gastrointestinal tract caudal to the glottis. Hooks were classified as tuna (3.6 or 3.8 mm) or swordfish and mustad (offset 8/0 or 9/0). Tissue samples (heart, lung, kidney, liver, spleen, brain, stomach, small intestines, skin, trachea, salt gland, gonad, thyroid, pancreas, and brain) were fixed in 10% buffered formalin, sectioned at 5 pm and stained with hematoxylin and eosin for microscopic examination. Representative tissues were stored frozen (-70°C) in sterile plastic bags. Where gross necropsy findings suggested infectious or in- flammatory disease, swabs or tissues were processed for microbiology. For bacteriology, swabs were plated on Mc- Conkey and blood agar and incubated at 27°C and 37°C for 48 h. For virus isolation, frozen tissues were homog- enized, the supernate filtered through a 0.22-pm filter, and plated on green sea turtle embryo fibroblasts (Moore etal.. 1997). ' McCracken, M. L. 2000. Estimation of sea turtle take and mortality in thie Ha- waiian longline fisheries. Administra- tive Report H-00-06, 29 p. Southwest Fisheries Science Center, Nat. Mar. Fish. Service, NOAA. 2570 Dole St.. Honolulu, HI 96822. - Kleiber, P. 1998. Estimating annual takes and kills of sea turtles by the Hawaiian longline fishery. 1991-97, from observer program and logbook data. Administra- tive Report H-98-08, 21 p. Southwest Fisheries Science Center, Nat. Mar. Fish. Serv.. NOAA, 2570 Dole St., Honolulu, HI 96822. Manu.script accepted 28 May 2002. Fish. Bull. 100:876-880 (2002). NOTE Work et al : Necropsy findings in sea turtles taken as bycatch in the Nortin Pacific 877 Table 1 Vital necropsy statistics for turtles caught by the Hawaii- based pelagic longline fishery in the North Pacific. Age are immature ( I ), subadult (S). and adult (A). Body and post-mortem condi ion were classified as good (G) or fair (F) SOL, BC, PMC, and Set stand | for smallest carapace 1 ength, body condition, postmortem condition, and hook set. respectively. ID Species Wt (kg) SCL Age Sex BC PMC Hook Set 1 C. mydas 25.4 55.6 I M G F tuna 3.6-mni light 2 C. mydas 50 67.9 S F G G tuna 3.6-mm light 3 D. coriacea 44.5 70.4 F G G tuna 3.6mm light 4 D. coriacea 74.1 85.3 F G G tuna 3.6-mm light 5 L. olivacea 13.2 43.7 F G F tuna 3.6-mm light 6 L. olivacea 15.3 46.6 F F G offset 8/0 deep 7 L. olivacea 21 54 F G G tuna 3.6-mm light 8 L. olivacea 24.5 57.5 S F G G tuna 3.6-mm light 9 L. olivacea 33.4 62.1 A F G G tuna 3.6-mm light 10 L. olivacea 34.1 60.4 A F G G tuna 3.6-mm deep 11 L. olivacea 37.7 62.9 A F G G tuna 3.6-mm light Results We performed necropsies on seven olive ridley (Lepido- chelys olivacea), two green (Chelonia mydas). and two leatherback sea turtles (Table 1). Turtles were caught between 5.4-18.0°N and 148.5-161.3''W in the North Pacific from February 1996 through June 2000. One leatherback sea turtle (identification |ID| 3) had severe acute inflammation of the liver associated with clumps of fibrin (Fig. 1, A and B), mild acute inflammation of heart muscle, and mild diffuse pulmonary edema. No bacteria or viruses were isolated from the liver The other leatherback sea turtle (ID 4) had severe diffuse fibrosis of the pancreas (Fig. IC) and a large subcapsular hematoma of the liver. The lesions in both turtles were severe enough to cause significant impairment of organ function and probable morbidity Incidental microscopic lesions in four of the seven olive ridley sea turtles included mild acute heart muscle inflammation (ID 6), mild focal necrosis in one lung (ID 7), parasite-induced necrosis in the stomach wall (ID 8), and foreign material (probable ingesta) in the bronchioles (ID 10). Microscopic lesions in the olive ridley sea turtles were not severe enough to cause morbidity, and no microscopic lesions were seen in the remaining turtles including the green sea turtles. Cause of mortality for all sea turtles was drowning after hooking; however, in only one case (olive ridley ID 11 ) was there gross evidence of water in the lungs. Female ( 10/11) and lightly hooked (9/11) animals predominated. Most hooks were tuna 3.6-mm hooks (Table 1). In one deeply hooked olive ridley turtle (ID 10), the hook perforated the esophageal wall in two places at approximately half of the length of the wall. In the other olive ridley sea turtle, ID 6 (classified as deeply hooked by the observer), the hook perforated just caudal to the tongue. Both leatherback sea turtles had been entangled by the leader; one had the leader tightly wound around the right front flipper, the other had the leader around the neck, where it had caused lacerations . Bait (sama; Cololabias saira) was seen in the esophagus of four olive ridley sea turtles; one turtle contained three fish, indicating ingestion from more than one hook. Other items in olive ridley stomachs included cowfish, pyroso- mas, pelagic snails, bird feathers, and small fragments of plastic. Stomachs from both leatherback turtles and from one green turtle contained pyrosomas exclusively. Discussion Only two sea turtles were classified as deeply hooked; most turtles had no visible lesion indicative of hooking either to the observer present on board when the turtles were hauled into the boat, or to the observer and us at the time of necropsy. Turtles that scored as lightly hooked but that died later would suggest that deep or light hooking may not be satisfactory criteria for the probability of short-term survival in the species we studied. Similarly, Polovina et al. (2000) and Parker et al. (in press) obsei"ved that there was no significant difference in distance or speed of travel between deeply and lightly hooked loggerhead turtles that were caught by the North Pacific longline fishery and then marked with satellite tags, and followed for several months. Hence, deep versus light hooking may not be a useful indicator of long-term survival for these species. The preponderance of females in our study was notice- able. Markedly skewed sex ratios in wild sea turtles are more commonly encountered among hatchlings (Gonzales et al., 2000), presumably due to incubation well below or above the pivotal temperature (Mrosovsky and Yntema, 1980). Studies of immature green sea turtles stranded with fibropapillomatosis in Hawaii revealed a sex ratio of close to 1:1 (Koga and Balazs, 1996). Ross (1984) in Oman reported sex ratios of adult green sea turtles to be closer to 878 Fishery Bulletin 100(4) f .: Figure 1 (A) Liver: Note fibrin (f), necrotic debris (large arrow ). capsule (c), and inflammatory cells (double an-owl, bar=100 nm; (B) Liver: note granulocytes (arrow), capsule (c). necrotic debris and fibrin (fl, bar=50 |im; (C) Pancreas: Note large fibrous trabeculae coursing through pancreatic cells (arrow), bar=100 |jm; (D) Normal pancreas from leatherback for comparison, bar=100 |im. 1:1 although slightly skewed towards females. Given the small sample size for each species, interpretation of the significance of the skewed sex ratio in our sample is prob- lematic. Ross (1984) noted that bias in sampling, segrega- tion of sexes in different areas, and small sample size could be responsible for deviations from the expected sex ratio of 1:1 in sea turtles. Whether severe lesions of the digestive system occur commonly in leatherback sea turtles remains to be deter- mined. Various factors can cause acute inflammation in the liver, including viruses, bacteria, protozoa, and poisons (Kelly, 1993); however, we saw no evidence of infectious agents in our histological examinations. Pancreatic fibrosis is a chronic lesion indicating earlier insult to the organ, secondary to infectious, toxic, or metabolic processes ( Jubb, 1993). Necropsies of stranded leatherback sea turtles have revealed bacterial (Obendorf et al., 1987) and parasitic (Threlfall, 1979) infections, degenerative joint disease (Og- den et al., 1981), and struvite fecoliths (Davenport et al., 1993). The impact of these conditions to leatherback popu- lations is unknown. Given the endangered status of East Pacific leatherback sea turtles, efforts to systematically evaluate health of these animals seem justified, including performing systematic necropsies on fresh carcasses recov- ered at sea or from nesting beaches, as well as performing health assessments of live animals on nesting beaches. Lesions in the other species of sea turtles (olive ridlcy and green) were either mild, nonspecific, or absent al- NOTE Work et al.: Necropsy findings in sea turtles taken as bycatch in [he Nortii Pacific 879 together. The absence of lesions in pelagic green turtles was in contrast to what is seen in immature and adult specimens caught in nearshore habitats. In those cases, most green turtles have systemic infections with vascular flukes or fibropapillomas (Aguirre et al, 1998). This would support the hypothesis that at least for immature turtles, these diseases, while absent pelagically, are acquired once the animals enter their near shore foraging pastures. It is possible that very mild subtle lesions could have been overlooked because many tissues had freeze-thaw arti- facts. Should this be a major concern, future studies may focus on doing necropsies on freshly dead turtles caught on fishing boats; however, the logistics of doing this will be more complicated. For example, vessels of the Hawaii- based longline fishery are at sea for weeks at a time fish- ing many hundreds of km from port. Stomach contents of olive ridley sea turtles in this study (pelagic snails, pyrosomas, foreign bodies) were similar to those seen by others (Parker et al., 2002; National Marine Fisheries Service and US Fish and Wildlife Service^). The presence of multiple baits in some olive ridley sea turtles suggests that animals may graze from longline hooks. Py- rosomas are found in stomachs of leatherback and green turtles (National Marine Fisheries Service and US Fish and Wildlife Service''-^), and leatherbacks are also attract- ed to hooks baited with sama (Grant^) and squid (Skill- man and Balazs, 1992). Given that fisheries may play a significant role in the decline of leatherback sea turtles (National Marine Fisheries Service and US Fish and Wildlife Service"*), determining whether bait attraction or entanglement pose the greater threat may be of manage- ment value. Similarly, evaluating the basis of bait attrac- tion for olive ridley sea turtles caught in longline fisheries may provide clues that will help discourage interactions between this species and fisheries. Acknowledgments Thanks are due to Doug Docherty for virus isolation as- says, and Robert Rameyer, Shawn K.K. Murakawa, Shan- dell Fames, and Denise M. Parker for technical assistance. This work was supported by a National Marine Fisheries Service Contract.40JJNF900126. 2 National Marine Fisheries Service and U. S. Fish and Wildlife Service. 1998a. Recovery plan for U. S. Pacific populations of the olive ridley turtle (Lepidochelys olivacea). 63 p. National Marine Fisheries Service, Silver Springs, MD, 20814. ■» National Marine Fisheries Service and U. S. Fish and Wildlife Service. 1998. Recovery plan for U.S. Pacific populations of the green turtle (Chelonia mydas). 84 p. National Marine Fisheries Service, Silver Springs, MD, 20814. 5 National Marine Fisheries Service and U. S. Fish and Wildlife Service. 1998. Recovery plan for U.S. Pacific populations of the leatherback turtle {Dermochelys coriacea). 65 p. National Marine Fisheries Service, Silver Springs, MD, 20814. ^ Grant, G. S. 1994. Juvenile leatherback turtle caught by long-line fishing in American Samoa. Marine Turtle Newslet- ter. 66:3-5. Literature cited Aguirre A. A., G. H. Balazs, B. Zimmerman, and T. R. Spraker. 1994. Evaluation of Hawaiian green turtles (Chelonia my- das) for potential pathogens associated with fibropapil- lomas. J. Wildl. Dis. 30:8-15. Aguirre, A. A., T. R. Spraker, G. H. Balazs, and B. Zimmerman. 1998. Spirorchidiasis and fibropapillomatosis in green turtles from the Hawaiian islands. J. Widl. Dis. 34:91-98. Balazs, G. H., S. G. Pooley, and S. K. K. Murakawa. 1995. Guidelines for handling marine turtles hooked or entangled in the Hawaii longline fishery: results of an expert workshop held in Honolulu, Hawaii, March 15-17, 1995. U.S. Dep. Commer, NOAA Tech. Memo-NMFS- SWrSC-222, 41 p. Bolten.A. B. and G. H. Balazs. 1983. Biology of the early pelagic stage-the "lost year." In Biology and conservation of sea turtles (K. A. Bjorndal (ed.), p. 579-581. Smithsonian Institution Press. Wash- ington, D.C. Davenport, J., and G. H. Balazs. 1991. "Fiery bodies"-are pyrosomas an important compo- nent of the diet of leatherback turtles. British Herp. Soc. Bull. 37:33-38. Davenport, J., G. H. Balazs, J. V. Faithfull, and D. A. Williamson. 1993. Struvite faecolith in the leatherback turtle Dermo- chelys coriacea vandelli: a means of packaging garbage? Herpetological Journal 3:81-83. Gonzalez. C. V., F. de Asis Silva Batiz, and S. H. Vazquez 2000. Sex ratio of marine turtle {Lepidochelys olivacea) hatchlings in incubation corrals at La Gloria nesting beach, Jalisco, Mexico. Bol. Cent. Invest. Biol. (Maracaibo) 34: 305-313. Jubb, K. F V. 1993. The pancreas. In Pathology of domestic animals (K. F. V. Jubb, R C. Kennedy, and N. Palmer (eds.), p. 407-424. Academic Press, Inc., San Diego, CA. Kelly W. R. 1993. The liver and biliary system. In Pathology of domes- tic animals (K. F V. Jubb, P. C. Kennedy, and N. Palmer, eds.). p. 319-406. Academic Press, Inc., San Diego, CA. Koga, S. K., and G. H. Balazs. 1996. Sex ratios of green turtles stranded in the Hawaiian Islands. In Proceedings of the fifteenth annual sympo- sium on sea turtle biology and conservation, February 20-25, 1995, Hilton Head, South Carolina (J. A. Keinath, D. E. Barnard, J. A. Musick, and B. A. Bell, comps), p. 148-152. U.S. Dep. Commer, NOAA-Tech Memo NMFS- SEFSC-387, Moore, M. K., T M. Work, G. H. Balazs, and D. E. Docherty 1997. Preparation and cryopreservation, and growth of cells prepared from the green turtle {Chelonia mydas). Meth. Cell Sci. 19:161-168. Mrosovsky, N., and C. L. Yntema. 1980. Temperature dependence of sexual differentiation in sea turtles: implications for conservation practices. Biol. Cons. 18:271-280. Obendorf D. L , J. Carson, and T. J. McManus. 1987. Vibrio damsela infection in a stranded leatherback turtle {Dermochelys coriacea). J. Wildl. Dis. 23:666-668. Ogden, J. A., A. G. Rhodin, G. J. Conlogue, and T R. Light. 1981. Pathobiology of septic arthritis and contiguous osteo- myelitis in a leatherback turtle (Dermochelys coriacea ). J. Wildl. Dis. 17:277-287. Parker, D. M., G. H. Balazs, S. K. K. Murakawa, and J. J. Polvina. In press. Post-hooking survival of sea turtles taken by 880 Fishery Bulletin 100(4) pelagic longline fishing in the North Pacific. In Proceed- ings of the twenty-first annual symposium on sea turtle biology and conser\'ation, February 24—28, 2001, Philadel- phia, PA. U.S. Dep. Commer. NOAA-Tech Memo. NMFS- SEFSC. Parker D. M., W. Cooke, and G. H. Balazs. 2002. Dietary components of pelagic loggerhead turtles in the North Pacific Ocean. In Proceedings of the twentieth annual symposium on sea turtle biology and conservation, February 29-March 4, 2000, Orlando, Florida, p. 148- 151. U.S. Dep. Commer, Memo, NMFS-SEFSC-477. Polovina, J. J., D. R. Kobayashi, D. M. Parker, M. R Seki, and G. H. Balazs. 2000. Turtles on the edge: movement of loggerhead turtles ^Caretta caretta) along oceanic fronts, spanning longline fishing grounds in the central North Pacific. 1997-1998. Fish. Oceanog. 9:71-82. Ross, J. P. 1984. Adult sex ratio in the green sea turtle. Copeia 1984: 774-776. Skillman, R. A. and G. H. Balazs. 1992. Leatherback turtle captured by ingestion of squid bait on swordfish longline. Fish. Bull. 80:807-808. Spotila J. R., R. D. Reina, A. C. Steyermark, P. T. Plotkin, and F V. Paladino. 2000. Pacific leatherback turtles face extinction. Nature 405:529-30. Threlfall, W. 1979. Three species of Digenea from the Atlantic leather- back turtle (Dermochelys coriacea). Can. J. Zool. 57:1825- 1829. Work, T. M. and G. H. Balazs. 1999. Relating tumor score to hematology in green turtles with fibropapillomatosis in Hawaii. J. Wildl. Dis. 35:804-807. 881 Acknowledgment of reviewers The editorial staff of F;s/!ery Bulletin would like to thank the following referees for their time and efforts in providing reviews of the manuscripts published in 2001-2002. Their contributions have helped ensure the publication of quality science. Dr Dean W. Ahrenholz Dr. M. James Allen Mr. Andrew J. Applegate Ms. Larisa Avens Mr. M. Scott Baker Dr. M. Barange Dr Daniel D. Benetti Dr. Dave Bengston Dr Steve Berkeley Dr. Karen A. Bjorndal Dr. Barbara Block Dr Philippe Borsa Dr. S.A. Bortone Dr. Louis W. Botsford Dr W. Alex Bradbury Dr Richard W. Brill Dr J.K.T. Brodziak Mr Michael Burton Mr Michael Canino Dr N. Caputi Mr John Carlson Dr. John Carmichael Dr. James Churchill Dr John E. Hughes Clarke Dr. J. Stanley Cobb Dr A.C. Cockroft Dr Felicia Coleman Dr. William Coles Dr Dean L. Courtney Dr Robert Cowen Ms. Tara Cox Dr Roy E. Crabtree Dr Jean Cramer Dr. Debbie Grouse Dr. Laurent Dagorn Dr Edward E. DeMartini Dr M.C. DeVries Dr. Heidi Dewar Dr Andrew Dizon Dr. P.J. Doherty Dr ML. Domeier Dr Richard L. O'Dnscoll Dr Jakov Dulcic Dr Peter Dutton Dr D.A. Ebert Dr Thomas A. Edsall Dr. Nelson M. Ehrhardt Dr. Nick G. Elliot Dr Robert W. Elner Mr Robert L. Emmett Dr Charles E. Epifanio Dr Bruce T. Estrella Dr Alan R. Everson Dr. Jeff Fargo Dr James D. Felley Dr Robert Foy Dr. Malcolm P. Francis Dr Kevin Friedland Dr. Sarah Gaichas Dr. James Gelsleichter Dr. Tony Gharrett Dr. R. Grant Gil more Mr. C. Phillip Goodyear Dr. Lewis J. Haldorson Dr. Martin Hall Dr Michael Moller Hansen Dr. Gareth Harding Dr. Jon Hare Dr. Patrick J. Harris Mr. C.J. Hai-vey Dr. Euan Harvey Dr. James T. Harvey Dr. Graeme Hays Dr Edward J. Heist Dr Thomas E. Helser Dr Frank J. Hernandez Dr Sarah Hinckley Dr John M. Hoenig Dr Aleta A. Hohn Dr Kim N. Holland Dr. G.J. Holt Mr Peter B. Hood Dr Kevin Hovel Dr Robert Hueter Dr. Tomoyuki Itoh Dr Elliot Jacobson Dr Chadwick V. Jay Dr Simon Jennings Dr. A.L. Jensen Mr Brian M. Jessop Dr Susana Junquera Dr. Michel J. Kaiser Dr. Mario Katsuragawa Dr. Gregory Todd Kellison Dr. Daniel K. Kimura Dr Christopher C. Koenig Dr. Syd Kraul Dr G. Lang Ms. Anne M. Lange Dr. Mark A. Lazzari Dr Rosangela Lessa Dr. Kwang-Ming Liu Dr William F Loftus Dr. Thomas R. Loughlin Dr. Flavia M Lucena Dr D.L. Mackas Dr William K. Macy Dr. Charles S. Manooch III Dr. Gary C. Matlock Dr James D. McCleave Dr Richard McGarvey Dr John C. McGovern Dr Robert J. Miller Dr David A. Milton Dr John Mitchell Dr N. Miyabe Dr. Beatriz Morales-Nin Dr Franz-Josef Mueter Dr Robert G. Muller Mr David Munday Dr. Bruce C. Mundy Dr. Steven A. Murawski Mr Michael D. Murphy Dr K. Nagasawa Dr Hideki Nakano Dr. John D. Neilson Dr R. John Nelson Dr K.T Nilssen Dr. CP Norman Mr. Robert O'Gorman Dr T Okutani Dr Kathryn A. Ono Dr Peter Ortner Dr Hazel A. Oxenford Dr Wayne Palsson Dr Y.C. Park Dr Richard O. Parker Jr. Dr Mark A. Pegg Dr Julian G. Pepperell Mr Peter C. Perkins Dr Michael Pennington Dr Marcelo A. A. Pinheiro Dr Margaret Platell Dr Dmitri Politov 882 Fishery Bulletin 100(4) Dr. Clay E. Porch Dr. James H. Power Dr. Eric D. Prince Dr. John A. Quinlan Dr. Paul Rago Dr. William Raschi Dr. W.J. Richards Dr. Donna R. Rogers Dr. Laura Rogers-Bennett Mr. Bill Roumillat Dr. Rodney A. Rountree Dr. Michael R. Ross Dr. Stephen T. Ross Dr. Thomas L. Rutecki Dr. Susan E. Safford Dr. Robert E. Schmidt Dr. Juan Jacobo Schmitter-Soto Dr. David H. Secor Dr. James B. Shaklee Dr. Donna J. Shaver Dr. Richard F. Shaw Mr. Gary R. Shepard Dr. Michiyo Shima Dr. John Sibert Dr. G.B. Skomal Dr. Malcolm J. Smale Dr. Barry D. Smith Dr. David R. Smith Dr. Stephen Smith Dr. Susan A. Smith Ms. Melissa Snover Dr. Rick D. Stanley Dr. Jay R. Stauffer Dr. Garry B. Stenson Dr. John D. Stevens Dr. Dominic J. Tollit Dr. Andrew W. Trites Dr. Mark Tupper Dr. W.N. Tzeng Dr. Senzo Uchida Dr. Glenn Ulrich Dr. Fred M. Utter Dr. David VanVorhees Dr. Douglas S. Vaughan Dr. Michael Vecchione Dr. Eric Volk Dr. Jon H. Volstad Dr. W. Waldo Wakefield Dr. John Watson Dr. Winsor Watson Dr. Mark E. Wilkins Dr. Graham Worthy Dr. Mary M. Yoklavich Dr. S. Zanuy Dr K.C.T. Zwanenburg 883 Fishery Bulletin Index Volume 100(1-4), 2002 List of titles 100(1) 1 The use of agreement measures and latent class models to assess the reliability of classifying ther- mally marked otoliths, by D. James Blick and Peter T. Hagen. 11 Local distribution and abundance of swimming crabs (CalUnectes spp. and Arenaeiis cribrarius) on a tropical arid beach, by Carlos A. Carmona-Suarez and Jesus E. Conde 26 Age, growth, and reproduction of permit iTr-achino- tiis falcatus) in Florida waters, by Roy E. Crabtree, Peter B. Hood, and Derke Snodgrass 35 Tag-reporting levels for red drum (Sciaenops ocella- fi/s) caught by anglers in South Carolina and Georgia estuaries, by Michael R. Denson, Wallace E. Jenkins, Arnold G. Woodward, and Theodore I. J. Smith 42 Age, growth, and mortality of the Mayan cichlid (Cichlasoma urophthalmus) from the southeastern Everglades, by Craig H. Faunce, Heather M. Patter- son, and Jerome J. Lorenz 51 Population status, seasonal variation in abundance, and long-term population trends of Steller sea lions iEu-metopias jubatus) at the South Farollon Islands, California, by Kelly K. Hastings and Wil- liam J. Sydeman 63 Larval and settlement periods of the northern searo- bin iPrionotus caroltnus) and the striped searobin (P. evolans). by Richard S. McBride, Michael P. Fahay, and Kenneth W Able 74 Assessing the precision of frequency distributions estimated from trawl-survey samples, by Michael Pennington, Liza-Mare Burmeister, and Vidar Hjellvik 81 Estimated ages of red porgy (Pagrus pagrus) from fishery-dependent and fishery-independent data and a comparison of growth parameters, by Jennifer C. Potts and Charles S. Manooch III. 90 Bycatch in the tuna purse-seine fisheries of the west- ern Indian Ocean, by Evgeny V. Romanov 106 Ontogenetic shifts in natural diet during benthic stages of American lobster (Homarus americanus) off the Magdalen Islands, by Bernard Sainte-Marie and Denis Chabot 117 Age and growth of Hawaiian green seaturtles (Chelo- nia mydas): and analysis based on skeletochronology, by George R. Zug, George H. Balazs, JeiTy A. Wether- all, Denise M Parker, and Shawn K. K. Murakawa 128 Estimates of lobster-handling mortality associated with the Northwestern Hawaiian Islands lobster- trap fishery, by Gerard T. DiNardo, Edward E. DeMartini, and Wayne R. Haight 134 An evaluation of pop-up satellite tags for estimating postrelease survival of blue marlin (Makaira nigri- cans) from a recreational fishery, by John E. Graves, Brian Luckhurst, and Eric D. Prince 143 Reproduction of the blacknose shark iCarcharhinus acronotus) in coastal waters off northeastern Brazil, by Fabio H. V. Hazin, Paulo G. Oliveira, and Matt K. Broadhurst 149 A new growth model for red drum (Sciaenops ocel- latus) that accommodates seasonal and ontogenic changes in growth rates, by Clay E. Porch, Charles A. Wilson, and David L. Nieland 100(2) 155 Horizontal and vertical movements of juvenile blue- fin tuna iThunnus thynnus), in relation to oceano- graphic conditions of the western North Atlantic, determined with ultrasonic telemetry, by Richard Brill, Molly Lutcavage, Greg Metzger, Peter Bush- nell, Michael Arendt, Jon Lucy, Cheryl Watson, and David Foley 168 Differences in diet of Atlantic bluefin tuna iThun- nus thynnus) at five seasonal feeding grounds on the New England continental shelf, by Bradford C. Chase 181 Movement of American lobster (Homarus america- nus) in the southwestern Gulf of St. Lawrence, by Michel Comeau and Fernand Savoie 193 Shortspine thornyhead iSebastolobus alascanus) abundance and habitat associations in the Gulf of Alaska, by Page Else, Lewis Haldorson, and Kenneth Krieger 200 Geographic and temporal patterns in size and matu- rity of the longfin inshore squid (Loligo pealeii) off the northeastern United States, by Emma M. C. Hatfield and Steven X. Cadrin 214 Age and size composition, growth rate, reproductive biology, and habitats of the West Australian dhufish (Glaucosoma hebraicum) and their relevance to the management of this species, by Sybrand A. Hesp, Ian C. Potter, and Norman G. Hall 884 Fishery Bulletin 100(4) 228 Longitudinal logbook survey designs for estimating recreational fishery catch, with application to ayu (Plecoglossus altivelis), by Shuichi Kitada and Kiyo- shi Tezuka 386 First record of a yellowfin tuna (Thunnus albacares) from the stomach of a longnose lancetfish (Alepisau- rus ferox), by Evgeny V. Romanov and Veniamin V. Zamorov 244 Physiological ecology of juvenile chinook salmon ( On- corhynchus tshawytscha) at the southern end of their distribution, the San Francisco Estuary and Gulf of the Farallones, California, by R. Bruce MacFarlane and Elizabeth C. Norton 258 Bottlenose dolphin [Tursiops truncatus) strandings in South Carolina, 1992-1996, by Wayne McFee and Sally R. Hopkins-Murphy 266 Validated age and growth of the porbeagle shark (Lamna nasus) in the western North Atlantic Ocean, by Lisa J. Natanson, Joseph J. Mello, and Steven E. Campana 279 Food habits and consumption rates of common dolphinfish iCoryphaena hippurus) in the eastern Pacific Ocean, by Robert J. Olson and Felipe Galvan- Magana 299 Recruitment season, size, and age of young Ameri- can eels (Anguilla rostrata) entering an estuary near Beaufort, North Carolina, by Perce M. Powles and Stanley M. Warlen 307 Changes over time in the spatial distribution of walleye pollock (Theragra chacogramma) in the Gulf of Alaska, 1984-1996, by Michiyo Shima, Anne Bab- cock Hollowed, and Glenn R. VanBlaricom 324 Movements of bocaccio iSebastes paucispinis) and greenspotted (S. chlorostictus) rockfishes in a Mon- terey submarine canyon: implications for the design of marine reserves, by Richard M. Starr, John N. Heine, Jason M. Felton, and Gregor M. Cailliet 338 Efficiency of bycatch reduction devices in small otter trawls used in the Florida shrimp fishery, by Philip Steele, Theresa M. Bert, Ki-istine H. Johnston, and Sandra Levett 100(3) 391 Temporal and spatial dynamics of spawning, settle- ment, and growth of gray snapper {Lutjanus griseus) from the West Florida shelf as determined from otolith microstructures, by Robert J. Allman and Churchill B. Grimes 404 Current velocity and catch efficiency in sampling set- tlement-stage larvae of coral-reef fishes, by Todd W Anderson, Claudine T. Bartels, Mark A. Hixon, Erich Bartels, Mark H. Carr, and Jonathan M. Shenker 414 Intermingling and seasonal migrations of Greenland halibut (Reinhardtius hippoglossoides) populations determined from tagging studies, by Jesper Boje 423 Improving pinniped diet analyses through identifica- tion of multiple skeletal structures in fecal samples, by Patience Brown, Jeffrey L. Laake, and Robert L. DeLong 434 Pinniped diet composition: a comparison of estima- tion models, by Jeffrey L. Laake, Patience Browne, Robert L. DeLong, and Harriet R. Huber 448 Vertical and horizontal movements of southern blue- fin tuna (Thunnus maccoyii) in the Great Australian Bight observed with ultrasonic telemetry, by Tim L. O. Davis and Clive A. Stanley 466 Turtle excluder devices — Are the escape openings large enough?, by Sheryan P. Epperly and Wendy G. Teas 475 Habitat use by demersal nekton on the continental shelf in the Benguela ecosystem, southern Africa, by Mark J. Gibbons, Andre J. J. Goosen, and Patti A. Wickens 351 Severe decline in abundance of the red porgy (Pagrus pagrus) population off the southeastern United States, by Douglas S. Vaughan and Michael H. Prager 376 A nearsurface, daytime occurrence of two mesope- lagic fish species iStenohrachius leucopsarus and Leuroglossus schriiidti) in a glacial fjord, by Alisa A. Abookire, John F. Piatt, and Suzann G. Speckman 381 Opportunistic feeding of longhorn sculpin (Myoxo- cephalus octodeccmspinosus): Are scallop fishery discards an important food subsidy for scavengers on Georges Bank?, by Jason S. Link and Frank P. Almeida 491 Population structure of king mackerel i Scorn b- eromorus cavalla I around peninsular Florida, as revealed by microsatellite DNA, by John R. Gold, Elena Park, and Doug A. DeVries 510 Do fluctuations in the somatic growth rate of rock lobster (Jasus lalandii ) encompass all size classes? A re-assessment of juvenile growth, by R. W. Anthony Hazell, David S. Schoeman, and Mark N. NofTke 519 Stock-rebuilding time isopleths and constant-F stock-rebuilding plans for overfished stocks, by Larry D. Jacobson and Steven X. Cadrin List of titles 885 537 Nuclear and mitochondrial DNA markers for specific identification of istiophorid and xiphiid billfishes, by Jan McDowell and John E. Graves 674 Age and growth of the smooth dogfish (Mustelus canis) in the northwest Atlantic Ocean, by Christina L. Conrath, James Gelsleichter, and John A. Musick 545 Seasonal growth of King George whiting (Sillagi- nodes punctata) estimated from length-at-age sam- ples of the legal-size harvest, by Richard McGarvey and Anthony J. Fowler 559 Spatial and temporal patterns in the demersal fish community on the shelf and upper slope regions of the Gulf of Alaska, by Franz J. Mueter and Brenda L. Norcross 582 Age, growth, mortality, and distribution of pinfish (Lagodon rhomboides) in Tampa Bay and adjacent Gulf of Mexico waters, by Gary A. Nelson 593 Recovery of the Gulf of Maine-Georges Bank Atlan- tic herring (Clupea harengus) complex: perspectives based on bottom trawl survey data, by William J. Overholtz and Kevin D. Friedland 609 Recruitment of larval Atlantic menhaden (Brevoortia tyrannus) to North Carolina and New Jersey estuar- ies: evidence for larval transport northward along the east coast of the United States, by Stanley M. Warlen, Kenneth W. Able, and Elisabeth H. Laban 624 Origin of immature loggerhead sea turtles iCaretta caretta ) at Hutchinson Island, Florida: evidence from mtDNA markers, by Wayne N. Witzell. Anna L. Bass, Michael J. Bresette, David A. Singewald, and Jona- than C. Gorham 632 Spawning of American shad (Alosa sapidissima) and striped bass (Moroiie saxatilis) in the Mattaponi and Pamunkey Rivers, Virginia, by Donna Marie Bilkovic, John E. Olney, and Carl H. Hershner 641 Spawning, growth, and overwintering size of searo- bins (Triglidae: Prionotus car'oliuus and P. evolans), bv Richard S. McBride 683 Bycatch of billfishes by the European tuna purse- seine fishery in the Atlantic Ocean, by Daniel Gaert- ner, Frederic Menard, Carol Develter, and Javier Ariz 690 Life history of South African snoek, Thyrsites atun (Pisces: Gempylidae): a pelagic predator of the Ben- guela ecosystem, by Marc H. Griffiths 711 Length and age at maturity of female petrale sole (Eopsetta jordani) determined from samples col- lected prior to spawning aggregation, by Robert W. Hannah, Steven J. Parker, and Erica L. Fruh 720 The measurement error in marine sui^vey catches: the bottom trawl case, by Vidar Hjellvik, Olav Rune Godo, and Dag Tjostheim 727 The reproductive biology of the porbeagle shark (Lamna nasus) in the western North Atlantic Ocean, by Christopher F. Jensen. Lisa J. Natanson, Harold L. Pratt Jr. Nancy Kohler, and Steven E. Campana 739 Integration of submersible transect data and high- resolution multibeam sonar imagery for a habi- tat-based groundfish assessment of Heceta Bank, Oregon, by Nicole M. Nasby-Lucas, Bob. W. Embley, Mark A. Hixon, Susan G. Merle, Brian N. Tissot, and Dawn J. Wright 752 An examination of spatial and temporal genetic vari- ation in walleye pollock (Theragra chalcogramma) using allozyme. mitochondrial DNA, and microsatel- lite data, by Jeffrey B. Olsen, Susan E. Merkouris, and James E. Seeb 765 Movements, behavior, and habitat selection of bigeye tuna {Thunnus obesus) in the eastern equatorial Pacific, ascertained through archival tags, by Kurt M. Schaefer and Daniel W. Fuller 100(4) 789 An evaluation of back-calculation methodology using simulated otolith data, by Michael J. Schirripa 651 Hybridization between two serranids, the coney (Cephalopholis fulva) and the creole-fish (Paranth- ias furcifer), at Bermuda, by Meredith A. Bostrom, Bruce B. Collette, Brian E. Luckhurst, Kimberly S. Reece, and John E. Graves 662 Effects of experimental harvest on red sea urchins iStrongylocentrotus franciscaiius) in northern Wash- ington, by Sarah K. Carter and Glenn R. VanBlaricom 800 Sustainability of elasmobranchs caught as bycatch in a tropical prawn (shrimp) trawl fishery, by Ilona C. Stobutzki, Margaret J. Miller, Don S. Heales, and David T. Brewer 822 Age and growth of the swordfish (Xiphias gladius L. ) in the waters around Taiwan determined from anal-fin rays, by Chi-Lu Sun, Sheng-Ping Wang, and Su-Zan Yeh 886 Fishery Bulletin 100(4) 836 Distribution and co-occurrence of rockfishes (family: Sebastidae) over trawlable shelf and slope habitats of California and southern Oregon, by Erik H. Wil- liams and Stephen Ralston 856 Properties of the residuals from two tag-recovery models, by Robert J. Latour, John M. Hoenig, and Kenneth H. Pollock 865 Re-identification of a lamnid shark embryo, by Henry F. Mollet, Antonio D. Testi, Leonard J. V. Compagno, and Malcolm P. Francis 876 Necropsy findings in sea turtles taken as bycatch in the North Pacific longline fishery, by Thierry M. Work, and George H. Balazs 861 Preliminary study on the use of neural arches in the age determination of bluntnose sixgill sharks (Hexanchus griseus), by Gordon A. McFarlane, Jac- quelynne R. King, and Mark. W. Saunders 887 Fishery Bulletin Index Volume T00(l-4), 2002 List of authors Able, Kenneth W. 63,609 Abookire, Alisa A. 376 All man, Robert J. 391 Almeida, Frank P. 381 Anderson, Todd W. 404 Antonio D. Testi Arendt, Michael 155 Ariz, Javier 683 Babcock Hollowed, Anne 307 Balazs, George H. 117,876 Bartels, Claudine T. 404 Bartels, Erich 404 Bass, Anna L. 624 Bert, Theresa M. 338 Bilkovic, Donna Marie 632 Blick, D. James 1 Boje, Jesper 414 Bostrom, Meredith A. 651 Bresette. Michael J. 624 Brewer, David T. 800 Brill, Richard 155 Broadhurst, Matt K. 143 Browne, Patience 423, 434 Burmeister, Liza-Mare 74 Bushnell, Peter 155 Cadrin, Steven X. 200,519 Caillet, Gregor M. 324 Campana, Steven E. 266, 727 Carmona-Suarez, Carlos A. 11 Carr. MarkH. 404 Carter, Sarah K. 662 Chabot, Denis 106 Chase, Bradford C. 168 Collette, Bruce B. 651 Comeau, Michel 181 Compagno, L. J. V. 865 Conde, Jesiis E. 11 Conrath, Christina L. 674 Crabtree, Roy E. 26 Davis, Tim L. O. 448 DeLong, Robert L. 423, 434 DeMartini, Edward E. 128 Denson, Michael R. 35 Develter, Carol 683 DeVries, Doug A. 491 DiNardo, Gerard T 128 Else, Page 193 Embley, BobW. 739 Epperly, Sheryan P. 466 Fahay, Michael P. 63 Faunce, Craig H. 42 Felton, Jason M. 324 Foley, David 155 Fowler, Anthony J. 545 Francis, Malcolm P. 865 Friedland, Kevin D. 593 Fruh, Erica L. 711 Fuller, Daniel W. 765 Gaertner, Daniel 683 Galvan-Magaiia. Felipe 279 Gelsleichter, James 674 Gibbons, Mark J. 475 Godo, Olav Rune 720 Gold, John R. 491 Goosen, Andre J. J. 475 Gorham, Jonathan C. 624 Graves, John E. 134,537,651 Griffiths, Marc H. 690 Grimes, Churchill B. 391 Hagen, Peter T. 1 Haight, Wayne R. 128 Haldorson, Lewis 193 Hall, Norman G. 214 Hannah, Robert W. 711 Hastings, Kelly K. 51 Hatfield, Emma M. C. 200 Hazell, R.W.Anthony 510 Hazin, Fabio H. V. 143 Heales, Don S. 800 Heine, John N. 324 Hershner, Carl H. 632 Hesp, Sybrand A. 214 Hixon, Mark A. 404, 739 Hjellvik, Vidar 74,720 Hoenig, John M. 856 Hood, Peter B. 26 Hopkms-Murphy Sally R. 258 Huber, Harriet R. 434 Jacobson, Larry D. 519 Jenkins, Wallace E. 35 Jensen. Christopher F. 727 Johnston, Kristine H. 338 King, Jacquelynne R. 861 Kitada, Shuichi 228 Kohler, Nancy 727 Krieger, Kenneth 193 Laake, Jeffrey 423,434 Laban, Elisabeth H. 609 Latour, Robert J. 856 Levett, Sandra 338 Link, Jason S. 381 Lorenz, Jerome J. 42 Luckhurst, Brian E. 134 Luckhurst, Brian E. 651 Lucy, Jon 155 Lutcavage, Molly 155 MacFarlane, R. Bruce 244 Manooch III, Charles S. 81 McBride, Richard S. 63, 641 McDowell, Jan R. 537 McFarlane, Gordon A. 861 McFee, Wayne E. 258 McGarvey, Richard 545 Mello, Joseph J. 266 Menard, Frederic 683 Merkouris, Susan E. 752 Merle, Susan G. 739 Metzger, Greg 155 Miller, Margaret J. 800 Mollet. Henry F 865 Mueter, Franz J. 559 Murakawa, Shawn K. K. 117 Musick, John A. 674 Nasby-Lucas, Nicole M. 739 Natanson, Lisa J. 266, 727 Nelson, Gary A. 582 Nieland, David L. 149 Noffke, Mark N. 510 Norcross, Brenda L. 559 Norton, Elizabeth C. 244 Oliveira, Paulo G. 143 OIney, John E. 632 Olsen, Jeffrey B. 752 Olson, Robert J. 279 Overholtz, William J. 593 Pak, Elena 491 Parker, Denise M. 117 Parker, Steven J. 711 Patterson, Heather M. 42 Pennington, Michael 74 Piatt, John F 376 Pollock, Kenneth H. 856 Porch, Clay E. 149 Potter, Ian C. 214 Potts, Jennifer C. 81 Powles, Perce M. 299 Prager, Michael H. 351 Pratt Jr., Harold L. 727 Prince, Eric D. 134 Ralston, Stephen 836 Reece, Kimberly S. 651 Romanov, EvgenyV. 90,386 888 Fishery Bulletin 100(4) Sainte-Marie, Bernard 106 Saunders, Mark W. 861 Savoie, Fernand 181 Schaefer, Kurt M. 765 Schirripa, Michael J. 789 Schoeman, David S. 510 Seeb, James E. 752 Shenker, Jonathan M. 404 Shima, Michiyo 307 Singewald, David A. 624 Smith, Theodore I. J. 35 Snodgrass, Derke 26 Speckman, Suzann G. 376 Stanley, Chve A. 448 Starr, Richard M. 324 Steele, Philip. 338 Stobutzki, Ilona C. 800 Sun, Chi-Lu 822 Sydeman, William J. 51 Teas, Wendy G. 466 Testi, Antonio D, 865 Tezuka, Kiyoshi 228 Tissot, Brian N. 739 Tjostheim, Dag 720 VanBlaricom, Glenn R. 307, 662 Vaughan, Douglas S. 351 Wang, Sheng-Ping 822 Warlen, Stanley 299,609 Watson, Cheryl 155 Wetherall, Jerry A. 117 Wickens, Patti A. 475 Williams, Erik H. 836 Wilson, Charles A. 149 Witzell, Wayne N. 624 Woodward. Arnold G. 35 Work, Thierry M. 876 Wright, Dawn J. 739 Yeh, Su-Zan 822 Zamorov, Veniamin V. 386 Zug, George R. 117 889 Fishery Bulletin Index Volume 100(1-4), 2002 List of subjects Abundance crabs 11 estimates 739 porgy. red 351 seasonal 51 sea lions, Steller 51 squid, longfin 200 thornyhead, shortspine 193 Acoustic seafloor mapping 739 Age and growth cichlid, Mayan 42 dhufish 214 dogfish, smooth 674 eel, American 299 permit 26 pinfish 582 porgy, red 81 sea turtles, green 117 shark 266,861 swordfish 822 and length (sole, petrale) 711 validation (shark, porbeagle) 266 Alepisauriis ferox — see lancetfish, longnose Allozymes 752 Alosa sapidissima — see shad, American Anguilla rostrata — see eel, American Archival tags 765 Arenaeus cribrarius — see crabs, speckled Atlantic, Northwest 381 Ayu 228 BRDibycatch reduction device) 338 Back-calculation 789 Bass, striped 632 Bathylagidae 376 Behavior (southern bluefin tuna) 448 Benguela ecosystem 475, 690 Bermuda 651 Billfish 537 bycatch 683 Bluntnose sixgill shark 861 Bocaccio 324 Bottom trawl 720 Brazil 143 Brevoortia tyi'annus — see menhaden, Atlantic Bycatch Atlantic Ocean 683 in shrimp fishery 338, 800 in tuna purse-seine fishery 90, 683 of elasmobranchs 800 of sea turtles 876 reduction device 338 western Indian Ocean 90 California Farallon Islands 51, 836 Callinectes spp. 1 1 sapidus — see crabs, swimming Carangidae 26 Carcharhinus acronotus — see shark, blacknose Carcharodon carcharias — see shark, white Caretta caretta — see sea turtles, loggerhead Catch efficiency (larval traps) 404 Cephalopholis fulva — see coney Channel nets 404 Chelonia rijydas — see sea turtles, green Cheloniidae 117 Cichlid, Mayan 42 Cichlasoma urophthalmus — see cichlid, Mayan Clupea hareiigus — see herring, Atlantic Cluster sampling 74, 307 Cod, Arctic 720 Coney 651 Coryphaena hippurus — see dolphinfish Cowcod 519 Crabs, rock 106 speckled swimming 11 swimming 11 Creole-fish 651 Current velocity 404 Data source impacts 81 Demersal fish assemblages 475, 559 Dermal denticles 864 Dhufish 214 Diet dolphinfish 279 lobster, American 106 pinniped 423, 434 seal, harbor 423,434 tuna, bluefin 168 Distribution and abundance bass 632 crab 11 pollock, walleye 307 rockfish 836 shad 632 squid, longfin 200 Dogfish, smooth 674 Dolphin bottlenose 258 strandings 258 Dolphinfish 279 Drum, red 35, 149 Ecology, salmon 244 Eel, American 299 Effective sample size 74 Elasmobranchs 266, 674, 727, 800, 861 Endangered species 466 Eopsetta jordani — see sole, petrale Eumetopias jubatus — see sea lion, Steller Everglades 42 Exotics (Mayan cichlid) 42 Experimental harvest 662 FAD (fish aggregating device) 683, 765 Fecal analyses 423 Feeding grounds 166 opportunistic 381 Fisheries Florida shrimp 338 recreational 134, 228 tuna purse-seine 90 Fishery management 662, 683 Flatfish 711 Florida permit 26 shrimp 338 snapper 391 Flounder, yellowtail 519 Forensic identification 537 Frequency distribution 74 Frequency of occurrence 434 Gempylidae 690 Genetic variation 752 Geolocation 765 Georges Bank 381, 419 Glacier Bay 376 Glaucosoma hebraicum — see dhufish Groundfish 739 890 Fishery Bulletin 100(4) Growth 789 dhufish 214 drum, red 149 lobster, rock 510 modeling drum, red 149 sea turtles 117 pinfish 582 porgy, red 81 salmon 244 searobins 641 snapper, gray 391 whiting, King George 545 Gulf of Alaska 193,307,559 Gulf of Farallones, salmon 244 Gulf of Maine 593 Gulf of Mexico drum, red 149 pinfish 582 Gulf of St. Lawrence lobster, American 181 Habitat 632 bass, striped 632 dhufish 214 nekton 475 shad, American 632 thornyhead. shortspine 193 tuna 765 Halibut, Greenland 414 Hawaiian Islands 117, 128 HecetaBank 739 Herring, Atlantic as prey 168 recovery 593 Hexanchus griseus — see bluntnose sixgill shark Histology 711 Homarus arnericanus — see lobster, American Hutchinson Island, Florida 624 Hybridization 651 Indian Ocean 90, 386 Indirect gradient analysis 836 Intergeneric hybridization 651 Istiophorids 537 Isurus oxyrinchus — see mako, shortfin Jasus lalandii — see lobster, rock Lagodon rhoniboides — see pinfish Lamna nasus — see shark, porbeagle Lampfish, northern 376 Lancetfish, longnose 386 Larvae coral-reef fish 404 Larval settlement 63 supply 404 transport 609 Lee Stocking Island 404 Length at age. King George whiting 545 frequency, shark 266 Leuroglossus schmidti — see smooth- tongue, northern Life history searobins 63 shark, blacknose 143 snoek 690 Limanda ferruginea — see flounder, yellowtail Lipids, salmon 244 Light trap 404 Lobster American 106, 181 rock 510 slipper 128 spiny (Hawaiian) 128 Logbook surveys 228 Loligo pealeii — see squid, longfin inshore Longline fishery 875 Lutjanus griseus — see snapper, gray Mackerel, king 491 Magdalen Islands 106 Makaira nigricans — see marlin, blue Mako, shortfin 865 Management (dhufish) 214 Manned submersible transects 739 Marking, thermal (of otoliths) 1 Marine mammals 423 reserves 324 Marlin blue 134 other species 537 Maturity shark, blacknose 143 sole, petrale 711 squid, longfin 200 Measurement error 720 Mediterranean Menhaden, Atlantic 609 Mesopelagic fish 376 Microsatellite DNA 491, 752 Migration diel 376 halibut 414 vertical 376 Models diet composition 434 drum, red 149 growth 149, 545 latent class 1 surplus production 351,419 tag recovery 856 Monterey submarine canyon 324 Morone saxatilis — see bass, striped Morphometries 739 Mortality cichlid, Mayan 42 dhufish 214 from handling 128 lobster 128 longline fishery 876 pinfish 582 Movement lobster, American 181 rockfish 324 tuna 155,448,765 MtDNA mackerel 491 marlin 537 ND4 region 537 pollock 752 sea turtles, loggerhead 624 Multibeam sonar imagery 739 Mustelus canis — see dogfish, smooth Myctophidae 376 Myoxocephalus octodecemspinosus — see sculpin, longhorn Necropsies 876 Neural arches 860 New Jersey 641 Nonmetric multidimensional scaling 559 North Carolina eel 299 lampfish, northern 376 smoothtongue, northern 376 Nuclear DNA (billfishes) 537 Oncorhynchus tshawytscha — see salmon, chinook Ontogeny 106, 149 Oophagy 727 Oregon 836 Otoliths 26, 789 in fecal samples 423 marking of 1 permit 26 simulated data 789 snapper, gray 391 Overfished stocks 519 Overwintering (searobins) 641 Pagrus pagrus — see porgy, red Paiiiilirtis marginatiis — see lobster, spiny Paranthias furcifer — see creole-fish Pathology 876 891 PCR (polymerase chain reaction) 537 Permit 26 Phoca vitidina — see seal, harbor Physiology, salmon 244 Pinfish 582 Pinnipeds 423,434 Placopecten magellanicus — see sea scallop Plecoglossus altivelis — see ayu Pollock, walleye 307, 752 Population demersal fish community 559 genetics 752 status 51 structure, mackerel 491 trends, sea lions 51 Pop-up satellite tags 134 Porgy, red 81,351 Postrelease survival 134 Prawn trawl fishery 800 Predation, bydolphinfish 279 by lancetfish 386 of tuna 386 Prionotus carolinus — see searobins evolans — see searobins Rays, anal-fin 821 Recolonization 662 Recreational fishing 35 ayu 228 Recruitment eel, American 299 menhaden, Atlantic 609 porgy, red 351 snapper, gray 391 squid 200 Reinhardtius hippoglossoides — see halibut, Greenland Reproduction dhufish 214 permit 26 searobins 63, 641 shark blacknose 143 porbeagle 727 snapper, gray 391 Residuals 856 Reliability index 1 Reward study 35 Rockfish bocaccio 324 cowcod 519 greenspotted 324 spp. 836 Salmon Chinook 244 salmonids 434 Sampling design 711 Sandy beach 11 San Francisco Estuary 244 Sanzo 865 Scavenging 381 Sciaenops ocellatus — see drum, red Scomberomortis cavalla — see mackerel, king Sculpin, longhorn 381 Scyllarides squammosus — see lobster, slipper Sea hon, Steller 51,307 Sea scallop 381 Sea urchin, red 662 Seasonal variation, sea lions 51 Searobins 63, 641 Sea turtles 466 demographic composition 624 green 117,466 loggerhead 466, 624 mortality 876 MtDNA analysis 624 olive ridley 466 strandings 466 Sea urchin, red 662 Seal, harbor 423 Sebastes chlorostictus — see rockfish, greenspotted levis — see cowcod paucispinis — see bocaccio Sebastidae 836 Sebastolobiis alascanus — see thornyhead, shortspine Settlement reef-fish larvae 404 searobins 63 snapper, gray 391 Shad, American 632 Shark blacknose 143 bluntnose sixgill 861 embryo re-identification 865 mako, shortfin 865 porbeagle 266, 727, 865 sustainability of 800 white 865 Shrimp (fishery) 338 Sillagirjodes punctata — see whiting. King George Size composition, dhufish 214 eel 299 searobins 641 squid, longfin inshore 200 Skeletal morphology 865 Skeletochronology 117 Snapper, gray 391 Snoek( South African) 690 Sole, petrale 711 Southeastern U.S. 181 South Farallon Islands 51 Species assemblages 836 diversity 559 endangered 466 Spawning bass 632 searobins 641 shad 632 snapper, gray 391 Squid, longfin inshore 200 Stenobrachius leiicopsarus — see lampfish, northern Stochastic models 519 Stock assessment porgy, red 351 rebuilding plans 519 sole, petrale 711 Stock decline (red porgy) 351 Strandings, dolphin 258 Strongylocentrotus franciscanus — see sea urchins, red Submersible 193, 475, 865 Survey design ( longitudinal logbook) 228 Swordfish 822 Tagging 134 and recapture 266 drum, red 35 halibut 414 pop-up satellite 134 recovery models 856 rockfish 324 shark, porbeagle 266 sonic 324 tuna 765 Taiwan 822 Tampa Bay 582 TED (turtle excluder device) 466, 800 Telemetry, ultrasonic 155, 448 Theragra chalcogramma — see pollock, walleye Thornyhead, shortspine 193 Th linn us albacares — see tuna, yellowfin maccoyii — see tuna, southern bluefin obesus — see tuna, bigeye thynnus — see tuna, bluefin Thyrsites atun — see snoek (South African ) Time isopleths (stock rebuilding) 519 Trachinotus falcatus — see permit Trawl bottom 593, 720 herring 593 892 Fishery Bulletin 100(4) otter 338 shrimp 338 Trawl surveys 74, 593, 720, 836 Triglidae 63 Tuna behavior 155, 765 bigeye 765 bluefin 155, 168 purse-seine fishery 90, 683 southern bluefin 448 yellowfin 386 Tursiops truncatus — see dolphin, bottlenose Ultrasonic telemetry 148, 155, 448 Virginia 632 Virtual population analysis 351 Whiting, King George 545 Xiphias gladius — see swordfish Xiphiids 537,822 Fishery Bulletin 100(4) 893 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) 1 Please type or print) (Additional address/attention line I Charge your order. ITS EASY! (Street address! (City, State, ZIP Code 1 (Daytime phone including area code) (Purchase Order No. I Please Choose Method of Payment: I I Check Payable to the Superintendent of Documents □ GPO Deposit Account | | | | | | | |— H □ VISA or MasterCard Account 1 1 1 1 1 1 1 1 1 M 1 1 1 1 11 1 1 1 1 To fax your orders (202) 512-2250 (Credit card expiration date) (Authorizing Signature! Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh. PA 15250-7954 Thank you for your order! This statement is required by the Act of August 12, 1970. Section 3685, Title 39, U.S. Code, show- ing ownership, management, and circulation of the Fishery' Bulletin, publication number 366-370, and was filed on 14 September, 2002. The Bulletin is published quarterly (four issues annually) with an annual subscription price of $50.00 ( sold by the Superintendent of Documents, U.S. Gov- ernment Printing Office, Washington, DC 20402!. The complete mailing address of the office of publica- tion is NMFS Scientific Publications Office. NOAA. 7600 Sand Point Way NE. BIN C15700, Seattle, WA 98115. The complete mailing address of the head- quarters of the publishing agency is National Marine Fisheries Service, NOAA, Department of Commerce, 1335 East-West Highway, Silver Spring, MD 20910. The name of the publisher is Willis Hobart and the managing editor is Sharyn Matriotti; their mailing address is: NMFS Scientific Publications Office, 7600 Sand Point Way NE, BIN 15700, Seattle, WA 981 15. The owner is the U.S. Department of Commerce, 14th St. N.W.. Washington. DC 20230; there are no bondholders, mortgages, or other security hold- ers. The purpose, function, and nonprofit status of the organization (agency) and the exempt status for Federal income tax purposes has not changed dunng the preceding 12 months. The extent and nature of circulation is as follows: total number of copies (A) (average number of copies of each issue during the preceding 12 months) was 1918 and the actual number of copies of the single issue published nearest to the filing dates was 1906. Paid circulation (B) is handled by the U.S. (Jovem- ment Printing Office, Washington, DC 20402, and the total number printed for sales (mail subscrip- tions and individual sales ) was 488 for both the aver- age number of copies each issue during the preceding 12 months and 475 the actual number of copies of the single issue published nearest to the filing date (C). Free distribution iD) by mail; samples, compli- mentary and other free copies (average number of copies each issue during the preceding 12 months) was 1405 and the actual number of copies of the single issue published nearest to the filing date was 1418. Free distribution outside the mail (E) by carriers or other means was 0 for both average number of copies and actual number of copies. Total free distribution ( F ) was 0 for both average number of copies and actual number of copies of the single issue published nearest the filing date. The total distribution (G: sum of D and B) (average number of copies each issue during the preceding 12 months) was 1893 and the actual number of copies of the single issue published nearest to the filing date was 1881. There were 25 copies not distributed (H).The total (I: sum of G and H) is equal to the net press run figures shown in Item A: 1893 and 25 copies, respectively. I certify that the statements made by me above are correct and complete: ( Signed ) Willis Hobart, Publisher. 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 English 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 intelhgible 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 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 conunents 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 hterature (unpubhshed 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 rfhe 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 anatomical 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 abbrevia- tions 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 Fisher^' 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 publication, 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 Norman Bartoo National Marine Fisheries Service, NOAA 8604 La JoUa Shores Drive La Jolla, 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. ^viCSiN^ BCSas^^Sl.^t. .^^ m^!^y^f^^T^, !S53[^^ liiii