Ai F53 h s-H U.S. Department of Commerce Volume 95 Number 3 July 1 997 U.S. Department of Commerce William M. Daley Secretary National Oceanic and Atmospheric Administration D. James Baker Under Secretary for Oceans and Atmosphere National Marine Fisheries Service Rolland A. Schmitten Assistant Administrator for Fisheries The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, Wash., and additional offices. POSTMASTER send address changes for subscriptions to Fishery Bulletin, Super- intendent of Documents, Attn: Chief, Mail List Branch, Mail Stop SSOM, Washington, DC 20402-9373. Although the contents have not been copyrighted and may be reprinted en- tirely, reference to source is appreci- ated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary in the transaction of the public business required by law of this Department. Use of funds for printing of this periodical has been ap- proved 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: $32.00 domestic and $40.00 foreign. Cost per single issue: $13.00 domestic and $16.25 foreign. See back page for order form. Scientific Editor Dr. John B. Pearce Editorial Assistant Kimberly T. Murray Northeast Fisheries Science Center National Marine Fisheries Service, NOAA 1 66 Water Street Woods Hole, Massachusetts 02543-1097 Editorial Committee Dr. Andrew E. Dizon National Marine Fisheries Service Dr. Harlyn O. Halvorson University of Massachusetts, Dartmouth Dr. Ronald W. Hardy University of Idaho, Hagerman Dr. Richard D. Methot National Marine Fisheries Service Dr. Theodore W. Pietsch University of Washington, Seattle Dr. Joseph E. Powers National Marine Fisheries Service Dr. Harald Rosenthal Universitat Kiel, Germany Dr. Fredric M. Serchuk National Marine Fisheries Service Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C 1 5700 Seattle, Washington 98115-0070 / The Fishery Bulletin carries original research reports and technical notes on investiga- tions in fishery science, engineering, and economics. The Bulletin of the United States Fish Commission was begun 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. Begin- ning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 95 Number 3 July 1 997 Fishery Bulletin Contents The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprietary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or proprietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the adver- tised product to be used or purchased because of this NMFS publication. Articles 403 Andrew, Neil L„, and Yong Chen Optimal sampling for estimating the size structure and mean size of abalone caught in a New South Wales fishery 414 Armstrong, Michael R Seasonal and ontogenetic changes in distribution and abundance of smooth flounder, Pleuronectes putnami, and winter flounder, Pleuronectes americanus, along estuarine depth and salinity gradients 431 Brightman, Ross I., Joseph J. Torres, Joseph Donnelly, and M. Elizabeth Clarke Energetics of larval red drum, Sciaenops ocellatus. Part II: Growth and biochemical indicators 445 Cadrin, Steven X., and Douglas S. Vaughan Retrospective analysis of virtual population estimates for Atlantic menhaden stock assessment 456 Crabtree, Roy E., Derke Snodgrass, and Christopher W. Hamden Maturation and reproductive seasonality in bonefish, Albula vulpes, from the waters of the Florida Keys 466 Estrella, Bruce T„, and Thomas D. Morrissey Seasonal movement of offshore American lobster, Homarus americanus, tagged along the eastern shore of Cape Cod, Massachusetts 477 Heftier, William F., Jr., David S. Peters, David R. Colby, and Elisabeth H. Laban Daily variability in abundance of larval fishes inside Beaufort Inlet 494 Nichoi, Daniel G. Effects of geography and bathymetry on growth and maturity of yellowfin sole, Pleuronectes asper, in the eastern Bering Sea Fishery Bulletin 95(3), 1 99 7 504 fSHorcross, Brenda L„, Franz-Josef Muter, and Brenda A. Holladay Habitat models for juvenile pleuronectids around Kodiak Island, Alaska 521 Paperno, Richard, Timothy E. Targett, and Paul A. Grecay Daily growth increments in otoliths of juvenile weakfish, Cynosc/on regalis: experimental assessment of changes in increment width with changes in feeding rate, growth rate, and condition factor 530 Peters, John S., and David J. Schmidt Daily age and growth of larval and early juvenile Spanish mackerel, Scomberomorus maculatus, from the South Atlantic Bight 540 Roelke, Lynn A., and Luis A. Cifuentes Use of stable isotopes to assess groups of king mackerel, Scomberomorus cavalla, in the Gulf of Mexico and southeastern Florida 552 Rogers, Donna R., Barton D. Rogers, Janaka A. de Silva, and Vernon L. Wright Effectiveness of four industry-developed bycatch reduction devices in Louisiana's inshore waters 566 Ward, Robert D., Nicholas G. Elliott, Bronwyn H. Innes, Adam J. Smolenski, and Peter M. Grewe Global population structure of yellowfin tuna, Thunnus albacares, inferred from allozyme and mitochondrial DNA variation 576 Zeldis, John R., R. I. Chris Francis, Malcolm R. Clark, Jonathan K. V. Ingerson, Paul J. Grimes, and Marianne Vignaux An estimate of orange roughy, Hoplostethus atlanticus, biomass using the daily fecundity reduction method 598 Zimmermann, Mark Maturity and fecundity of arrowtooth flounder, Atheresthes stomias, from the Gulf of Alaska Notes 612 Munehara, Hiroyuki The reproductive biology and early life stages of Podothecus sachi (Pisces: Agonidae) 620 Roman-Rodriguez, Martha J., and M. Gregory Hammann Age and growth of totoaba, Totoaba macdonaldi (Sciaenidae), in the upper Gulf of California 629 Thedinga, John F., Adam Moles, and Jeffrey T. Fujioka Mark retention and growth of jet-injected juvenile marine fish 635 Erratum 636 Subscription form 403 Abstract. ^The fishery for blacklip abalone, Haliotis rubra , is one of the most valuable in New South Wales, Australia. An important part of the stock assessment process for this fish- ery is to quantify temporal changes in mean size and size structure of abalone in the landed catch. Variation in aba- lone growth over small spatial scales in this fishery and differences in har- vest strategy among different divers result in large variations in sizes of abalone landed. Monte Carlo simula- tions were used to investigate the in- fluence of these sources of variation on estimates of mean size and size struc- ture. Different sampling scenarios were considered — from random sampling of all diver-days to a more realistic scheme where abalone were subsampled both within and among diver-days. For a given total number of abalone mea- sured, error in estimated mean size and size structure declined asymptotically with increasing numbers of diver-days. By measuring at least 1,500 abalone from 100 diver-days, reliable estimates of size structure and mean size of aba- lone in the catch for the whole fishery were produced. This conclusion was robust with respect to the number of diver-days in the fishery. Estimated sampling intensity and probabilities of detecting differences based on simu- lated variances for the whole fishery are provided. Manuscript accepted 3 February 1997. Fishery Bulletin 95:403-413 ( 1997). Optimal sampling for estimating the size structure and mean size of abalone caught in a New South Wales fishery Neil L. Andrew* Yong Chen NSW Fisheries Research Institute PO. Box 21 Cronulla, New South Wales 2230, Australia *E-mail address: andrewn@fisheries.nsw.gov.au Sample-size determination remains a crucial exercise in all aspects of ecology and fisheries biology, and the array of analytical tools avail- able continues to grow (e.g. Ger- rodette, 1987; Kimura, 1990; Peter- man, 1990; Thompson, 1992). The great majority of these techniques are designed to optimize sampling for data derived from independent samples from a number of hierar- chical sources of variation (e.g. Schweigert et al., 1985; Sen, 1986; Andrew and Mapstone, 1987; Kitada et al., 1992; Crone, 1995). Methods for determining sample sizes for describing size- or age-frequency distributions are less common (but see Smith and Sedransk, 1982; Schweigert and Sibert, 1983; Parkin- son et al., 1988; Erzini, 1990). Sample-size determination for the simultaneous estimation of dif- ferent size classes is possible ana- lytically only under limited circum- stances in fisheries applications. If differences among individuals are the only source of variation to be contended with, then the proportion of individuals in each size class in a population may be estimated simul- taneously by using the methods de- veloped by Fitzpatrick and Scott (1987) and Thompson (1987), and the calculation of the variances of these estimates are simple. In most situations facing fisher- ies biologists, however, there are many sources of variation confound- ing simple random sampling and sample-size determination for esti- mating mean size at harvest and the underlying size structure. Typi- cally, catches come from many boats, fishermen, and fishing grounds, and samplers are almost always faced with far more fish than they could possibly measure. Under these cir- cumstances there are many sources of variation that may bias sampling. Not least of these is the likelihood of underlying spatial and temporal heterogeneity in the fished popula- tions and changes in fishing behav- ior. Monte Carlo simulations pro- vide a relatively straightforward, al- though computation-intensive, means of determining appropriate schemes in these instances. Sample- size determination for multistage survey designs relies on apportion- ing sampling effort to various lev- els on the basis of variance or cost (or both). The fishery and the problem The fishery for abalone Haliotis rubra in New South Wales (NSW) is managed by using a combination of size limits, closures, and output controls. In 1995, each of 37 divers had an annual quota of 9 metric tons (t). Since 1974, divers have 404 Fishery Bulletin 95(3), 1997 been required to provide details of daily catch weight and diving hours in each of 28 zones (Fig. 1). Divers may catch abalone in any of the zones, which range in length of coastline between 7 and 147 km. In 1994 there was a total of 3,129 diver-days in the fishery and an average of 104 diver-days per zone (Fig. 2A). Based on estimates of average weight per abalone, a catch of between 20 and 760 abalone was landed per diver-day (Fig. 2B). The mean size of abalone caught per diver-day ranged between 116 and 129 mm, al- though the majority were between 117 and 121 mm long (Fig. 2C). A minimum size limit of 115 mm has been applied to the fishery since 1987. Fishing pres- sure in this fishery is intense and the size structure of abalone in the landed catch in each zone may be described by negative exponential distributions of vary- ing instantaneous slope (see examples in Fig. 3). Determining a sampling scheme to provide reli- able estimates of the size structure and mean size of abalone in the landed catch is complicated by differ- ences among diver-days. Worthington et al. (1995) and Worthington and Andrew (in press) have de- scribed large variations in demographic parameters, such as growth rate, maximum size, mortality, and fecundity, over a range of spatial scales. These stud- ies report as much variation in the rates of growth and in the maximum sizes of abalone within sites separated by 2 km as there was among sites sepa- rated by hundreds of kilometers. Sizes of abalone in landed catches will therefore depend on how and where the diver worked as well as on the demographic attributes of the population being fished. For example, on any day, a diver may work areas where abalone are fast-growing and tend to be larger or ar- eas where abalone are slow-growing and smaller (or both). In this study we report the results of simulations in order to determine an ap- propriate allocation of sampling effort to estimate mean sizes and size structures of abalone in the landed catch in the New South Wales fishery. Sampling is consid- ered for groups of zones and for the whole fishery. A simulation approach was adopted in preference to an analytical solution (e.g. Cochran, 1977) because we were interested in simultaneously optimizing sampling across a number of size classes — all of which were nonindependent. The simulation pro- cedure allowed an estimation of the devia- tion of samples of different sizes from a known or true population. Parameters used in the simulations were based on prelimi- nary sampling in 1993-94. Materials and methods Simulation study A Monte Carlo simulation approach was used to estimate the relative efficiency of three strategies for sampling abalone: Sample all abalone from randomly se- lected diver-days; Sample a fixed number of abalone ran- domly from the catches of all diver- days; and Figure 1 Map of the lower half of New South Wales showing zones in the abalone fish- ery. The zones are coded alphabetically from north to south. Andrew and Chen: Estimating size structure and mean size of Haliotis rubra 405 Number of days per zone Number of abalone per day 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 Length (mm) Figure 2 Summary statistics from the NSW abalone fishery for 1993 and 1994: (A) Frequency distribution of the number of days per zone, (B) Frequency distribution of the number of aba- lone per diver-day calculated from known average weights per abalone, (C) Frequency distribution of the mean length of abalone caught per diver-day. Sample sizes indicated are the number of diver-days. 3 Sample a fixed number of abalone randomly from diver-days selected randomly. Of the three strategies, the third is the most logis- tically and financially reasonable. There is consider- able unpredictability in where and when divers will work both because of weather conditions and a re- luctance by divers to specify where they will work on a given day. These facts conspire to make it difficult to sample in a truly random manner. Nor is it practi- cal to stratify appropriately across either divers or days because the population of diver-days to be sampled can be determined only in retrospect. For these reasons we have used a “diver-day” as the unit of stratification. Three sources of variation are con- founded in “diver-day.” Differences among divers, as a result of their fishing behavior (e.g. experience and ability) could not be separated from the variation in- herent in where they fished, and therefore in the aba- lone caught. The third source of variation pooled into diver-day is the day itself (e.g. weather and sea con- ditions). Although these sources of variation were in- separable within the present study, the inferences drawn about a representative sampling scheme are not confounded. Strategy 1, although desirable, would limit the number of diver-days that could be sampled given a fixed total sampling effort. Strat- egy 2 represents the “ideal” sampling scheme and is used as a standard from which the remaining, more realistic, schemes are judged. Parameters of the simulation Parameters were determined for the simulations by using information collected during the 1993-94 fish- ing years. We assumed that there is as much vari- ability in parameters among diver-days within a zone as among zones. Sampling schemes for zones or groups of zones and for the whole fishery were as- sessed by varying the total number of diver-days in the “fishery” per year. We therefore ran simulations by using up to 600 diver-days to determine sampling schemes for zones and groups of zones and simulated a 4,000-d fishery to determine a sampling scheme for the fishery as a whole. Step 1 (determination of the number of abalone caught per diver-dayj Based on previous sampling (Fig. 2B), the numbers of abalone caught in all diver- days were grouped into different catch groups rang- ing from the midpoint of 20 to 760 abalone, with the interval of the catch groups being 20. Thus, the total number of catch groups is 38 (i.e. (760-20)/20 + 1 = 38). The frequency of the number of abalone caught per diver-day was then estimated. Based on these frequencies, the total catch per diver-day was deter- mined by multinominal sampling described as fol- lows. Let Pj = probability of the number of abalone harvested in a diver-day in catch group J, where J = 1, 2, ..., 38. The catch of diver-day i was determined by generating a random number R between 0 and 1 based on the uniform distribution and by assigning this number to one of the catch groups. The catch was assigned to catch group J if the random number followed 406 Fishery Bulletin 95(3), 1997 .7-1 j *=i k=\ After determining the catch group (i.e. J) for aba- lone harvested in diver-day i, the number of abalone harvested in diver-day i was determined as C, ={J- 0.5)20 + 2017, where U is a random number between 0 and 1 gen- erated from a uniform distribution. Because J in this equation has been determined from the previous equation, we have omitted the subscript rJ from C. for the sake of simplicity. This procedure was repeated for all diver-days in each simulation to determine the number of abalone harvested in each diver-day. Step 2 (determination of the mean length of aba- lone caught per diver-dayj The mean length of the catch for each diver-day was determined from the estimates derived from sampling 102 diver-days in 1993-94 (Fig. 2C). Lengths of abalone were measured to the nearest mm from catches from a range of zones and are assumed to be measured without error. The estimates of mean size ranged from 116 to 129 mm. Let Pj = probability of the mean length in length interval I (I = 1, 2, ..., 14). The mean length for diver-day i was de- termined by generating a random number R between 0 and 1 based on a uniform distribution and by assigning this number to one of the length intervals. The length was assigned to length interval I if the random number followed k=i k=i After determining the length interval (I) for the mean length of abalone harvested in diver-day i, the mean length of diver- day i was determined as Li ■ = I + 115 (mm), where 115 mm is the size limit. This pro- cedure was repeated for all diver-days in each simulation to determine the mean length of abalone caught per diver-day. Step 3 (determination of length compo- sition of abalone caught per diver- dayj The size distributions of abalone caught in a diver-day may be described by exponential distributions with varying slope, truncated at the lower limit by the legal size limit and at the upper limit by the value T, which was determined by ran- dom draws from the range of extreme val- ues observed in preliminary sampling. The density function for such an exponential distribution can be written as -(g-a) P(x) = Ae a , where a 25%). We assume that this sampling frac- tion provides a reliable estimate of mean size within diver-days. The results of these simulations suggest that the large sample sizes possible in estimating mean size 412 Fishery Bulletin 95(3), 1997 1.0 >> 0.8 — (a) (b) *j3 O 0.6 a. 10%) would have a relatively low probability of detecting changes in mean size less than 3 mm. This conclusion appeared to be robust over a realistic range in the number of diver-days sampled. Given the broad similarities between the sampling scheme described for this fish- ery and those in many commercial fisheries, concerns may be raised about the reliability of samples taken from processing plants in the absence of an under- standing of the contributions of higher level sources of variation. The simulation framework described may be directly expanded to accommodate more com- plex situations. Acknowledgments We are grateful to Geoff Gordon and Duncan Worthington for discussions, and Penny Brett and Ron Avery for help with the manuscript. We thank the abalone divers and staff from NSW Abalone, S.O.S. (Southern Oceans Seafoods), A.S.E. (Abalone Shellfish Enterprises), and Interon for assistance. This work was funded in part by grants from the Australian Fisheries Research and Development Corporation and the New South Wales commercial abalone industry for which we are grateful. Literature cited Andrew, N. L., and B. D. Mapstone. 1987. Sampling and the description of spatial pattern in marine ecology. Oceanogr. Mar. Biol. Annu. Rev. 25:3-90. Bhattacharya, C. G. 1967. A simple method of resolution of a distribution into Gaussian components. Biometrics 23:115-135. Castro, M., and K. Erzini. 1988. Comparison of two length-frequency based pack- ages for estimating growth and mortality parameters using simulated samples with varying recruitment patterns. Fish. Bull. 86:645-654. Chen, Y. 1996. A Monte Carlo study on impacts of the size of subsample catch on estimation of fish stock para- meters. Fish. Res. (Amst.) 26:207-225. Cochran, W. G. 1977. Sampling techniques, 3rd ed. John Wiley and Sons, New York, NY, 608 p. Crone, P. R. 1995. Sampling design and statistical considerations for the commercial groundfish fishery of Oregon. Can. J. Fish. Aquat. Sci. 52:716-732. Deriso, R. B., T. J. Quinn, and P. R. Neal. 1985. Catch-age analysis with auxiliary information. Can. J. Fish. Aquat. Sci. 42:815-824. Erzini, K. 1990. Sample size and grouping of data for length-frequency analysis. Fish Res. (Amst.) 9:355-366. Fitzpatrick, S., and A. Scott. 1987. Quick simultaneous confidence intervals for multi- nomial proportions. J. Am. Stat. Assoc. 82:875-878. Fournier, D., and C. P. Archibald. 1982. A general theory for analysing catch at age data. Can. J. Fish. Aquat. Sci. 39:1195-1207. Gerrodette, T. 1987. A power analysis for detecting trends. Ecology 68:1364-1372. Grant, A., P. J. Morgan, and P. J. W. Olive. 1987. Use made in marine ecology of methods for estimat- ing demographic parameters from size-frequency data. Mar. Biol. 95:201-208. Hilborn, R., and C. J. Walters. 1987. A general model for simulation of stock and fleet dy- namics in spatially heterogeneous fisheries. Can. J. Fish. Aquat. Sci. 44:1366-1369. Horppila, J., and H. Peltonen. 1992. Optimizing sampling from trawl catches: contempo- raneous multistage sampling for age and length structures. Can. J. Fish. Aquat. Sci. 49:1555-1559. Kimura, D. K. 1990. Approaches to age-structured separable sequential population analysis. Can. J. Fish. Aquat. Sci. 47:2364- 2374. Andrew and Chen: Estimating size structure and mean size of Haliotis rubra 413 Kitada, S., Y. Taga, and H. Kishino. 1992. Effectiveness of a stock enhancement program evalu- ated by a two-stage sampling survey of commercial landings. Can. J. Fish. Aquat. Sci. 49:1573-1582. McShane, P. E., and M. G. Smith. 1992. Shell growth checks are unreliable indicators of age of the abalone Haliotis rubra Mollusca: Gastropoda. Aust. J. Mar. Freshwater Res. 43:1215-1219. Megrey, B. A. 1989. Review and comparison of age-structured stock as- sessment models from theoretical and applied points of view. Am. Fish. Soc. Symp. 6:8-48. Parkinson, E. A., J. Berkowitz, and C. J. Bull. 1988. Sample size requirements for detecting changes in some fisheries statistics from small trout lakes. N. Am. J. Fish. Manage. 8:181-190. Peterman, R. M. 1990. Statistical power analysis can improve fisheries re- search and management. Can. J. Fish. Aquat. Sci. 47:2-15. Schnute, J., and D. Fournier. 1980. A new approach to length frequency analysis: growth structure. Can. J. Fish. Aquat. Sci. 37:1337-1351. Schweigert, J. F., C. W. Haegele, and M. Stocker. 1985. Optimizing sampling design for herring spawn sur- veys in the Strait of Georgia, B.C. Can. J. Fish. Aquat. Sci. 42:1806-1814. Schweigert, J. F., and J. R. Sibert. 1983. Optimising survey design for determining age struc- ture of fish stocks: an example from British Columbia Pa- cific herring Clupea harengus pallasi. Can. J. Fish. Aquat. Sci. 40:588-597. Sen, A. R. 1986. Methodological problems in sampling commercial rockfish landings. Fish. Bull. 84:409-421. Shepherd, S. A., M. J. Tegner, and S. A. Guzman del Proo (eds.). 1992. Abalone of the world: biology, fisheries and culture. Fishing News Books, Oxford, 413 p. Smith, P. J., and J. Sedransk. 1982. Bayesian optimization of the estimation of the age composition of a fish population. J. Am. Stat. Assoc. 77:707-713. Smith, S. J., and J. J. Maguire. 1983. Estimating the variance of length composition samples. Can. Spec. Publ. Fish. Aquat. Sci. 66:165-170 Sullivan, P. J., H. Lai, and V. F. Galluci. 1990. A catch-at-length analysis that incorporates a sto- chastic model of growth. Can. J. Fish. Aquat. Sci. 47:184- 198. Terceiro, M., D. A. Fournier, and J. R. Sibert. 1992. C omparative performance of MULTIFAN and Shepherd’s length composition analysis (SRLCA) on simu- lated length-frequency distributions. Trans. Am. Fish. Soc. 121:667-677. Thompson, S. K. 1987. Sample size for estimating multinomial propor- tions. Am. Stat. 41:42-46. 1992. Sampling. John Wiley and Sons, New York, NY, 343 p. Worthington, D. G., and N. L. Andrew. In press. Small scale variation in demography and its im- plications for alternative size limits in the fishery for aba- lone in NSW, Australia. Spec. Publ. Can. J. Fish. Aquat. Sci. Worthington, D. G., N. L. Andrew, and G. Hamer. 1995. Covariation between growth and morphology sug- gests alternative size limits for the abalone, Haliotis rubra , in NSW, Australia. Fish. Bull. 93:551-561. 414 Seasonal and ontogenetic changes in distribution and abundance of smooth flounder, Pleuronectes putnami, and winter flounder, Pleuronectes americanus, along estuarine depth and salinity gradients Michael R Armstrong Department of Zoology, University of New Hampshire Durham, New Hampshire 03824 Present address: Massachusetts Division of Marine Fisheries Annisquam River Marine Fisheries Station 30 Emerson Avenue, Gloucester, Massachusetts 01930 E-mail address: michael.armstrong@state.ma.us Abstract .—The distribution and abundance of two potentially compet- ing flatfish species, smooth flounder, Pleuronectes putnami, and winter flounder, Pleuronectes americanus , were examined along salinity and depth gradients in upper Great Bay Estuary, New Hampshire. Both species were abundant in the estuary but exhibited differential use of habitats along both gradients. Smooth flounder were most abundant at the mesohaline, riverine habitat, whereas winter flounder were most abundant at the polyhaline, open- bay habitat. Both species exhibited a generalized up-river movement as sa- linity increased with the seasons. Smooth flounder showed ontogenetic changes in distribution along the depth gradient, with smallest individuals oc- cupying shallowest depths. Intertidal mudflats were an important nursery area for young-of-the-year smooth flounder. Winter flounder showed little separation by size along the depth gra- dient, and few were found in the inter- tidal mudflat habitat. The potential for competition between these two species is lessened by their partial segregation along the gradients examined. Manuscript accepted 14 January 1997. Fishery Bulletin 95:414-430 (1997). Smooth flounder, Pleuronectes putnami , and winter flounder, Pleuronectes americanus, are domi- nant members of fish communities in estuaries along the east coast of North America, co-occurring from Newfoundland, Canada, to Massa- chusetts Bay, USA. These morpho- logically similar species are sympa- tric over much of their geographic ranges. However, little is known of their spatial overlap within specific estuaries. Winter flounder use estu- aries primarily as nursery grounds, whereas adults spend most of their lives in coastal waters (Bigelow and Schroeder, 1953; Pearcy, 1962; Scott and Scott, 1988). In contrast, smooth flounder complete their entire life cycle within estuaries. Smooth flounder prefer softer bot- tom substrata than winter flounder (Bigelow and Schroeder, 1953), and Jackson (1922) noted they were most abundant in the low-salinity regions within Great Bay Estuary, New Hampshire. Little else is known of their intraestuarine habi- tat preferences. Several studies have examined movements and habitat use of juvenile winter floun- der in estuaries south of Cape Cod (e.g. Pearcy, 1962; Saucerman 1991). However, because many northern estuaries differ consider- ably from those south of Cape Cod, most obviously in their temperature regimes, it is possible that juvenile winter flounder use northern estu- aries differently from ones to the south, as has been shown to be the case for adults (Hanson and Cour- tenay, 1996). The purpose of this study was to provide a quantitative comparison of the occurrence of smooth and win- ter flounder in various habitats in upper Great Bay Estuary, New Hampshire. The habitats comprised gradients defined by depth or salin- ity. Comparative studies along habi- tat gradients can define which habi- tats are important to a species, es- pecially in relation to different life history stages; such analyses can also be used to study the relative importance of physical and biotic factors in limiting species distribu- tions (Connor and Bowers, 1987). Examination of the shape of species- abundance curves along a gradient can provide inferences into whether competition or physiological limita- tions are important in setting dis- Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 415 tributions (Terborgh, 1971) and can lead to the gen- eration of testable hypotheses. Methods Study area Great Bay Estuary (Fig. 1) is a complex embayment comprising the Piscataqua River, Little Bay, and Great Bay. It is a tidally dominated system and is at the confluence of seven major rivers and several small creeks, as well as the water from the Gulf of Maine (Short, 1992). Great Bay Estuary is a drowned river valley, with high tidal energy and deep channels with fringing mud flats. The main habitat types within the estuary are mudflat, eelgrass, salt marsh, chan- nel bottom, and rocky intertidal. This study was con- ducted in the upper estuary, referred to as Great Bay, although preliminary sampling took place in the lower estuary also. Great Bay is a large, shallow embayment having an average depth of 2.7 m, with deeper channels extending to 17.7 m (Short, 1992) and a tidal range of about 2 m. The water surface of Great Bay covers 23 km2 at mean high water and 11 km2 at mean low water (Turgeon, 1976). Greater than 50% of the sediment surface of Great Bay is exposed mud or eelgrass flat at low tide. The Squamscott and Lamprey Rivers are major sources of freshwater to Great Bay. River flow varies considerably on a sea- sonal basis but is generally highest during spring runoff. Vertical stratification of Great Bay is rare because of strong tide- and wind-induced currents, although partial stratification may occur during pe- riods of high freshwater runoff, particularly at the upper tidal reaches of rivers (Short, 1992). Smooth and winter flounder were sampled monthly, May 1989 through September 1991, at five sites in upper Great Bay Estuary (Fig. 1). Ice cover prevented sampling from December through March in all study years. A 4.8-m otter trawl of 38-mm stretch mesh, with a 25-mm stretch mesh codend and a 6-mm codend liner, was used for sampling. Preliminary studies indicated that the net retained flounder as small as 25-mm total length (TL). A sample consisted of all flounder collected in one 10-minute tow at approximately 2.5 knots. Four samples were taken at each site from April to November. Two samples were taken within two hours (±) of low slack tide, one tow with the tidal current and one tow against, and two samples were taken similarly around high slack tide. All flounder collected were measured to the nearest mm TL. Bottom temperature and salin- ity were measured after each tow with a Beckman Model 510 temperature, conductivity, and salinity meter. 70 °50‘ 70 °40' Figure 1 Study area. Survey sites were all located in Great Bay Estuary, New Hampshire, as indicated. 416 Fishery Bulletin 95(3), 1997 Site 1 (low salinity, Squamscott River at Route 51), site 2 (medium salinity, Squamscott River at Route 108), and site 3 (high salinity, middle of Great Bay) were located along a salinity gradient formed by Great Bay Estuary and one of its major tributaries (Fig. 1). The mean salinity value at each site varied considerably on a seasonal basis, but a salinity gra- dient always persisted along these sites. Table 1 sum- marizes some physical characteristics of these loca- tions. Site 1 was located in the Squamscott River about 4 km above the mouth. Although the river is still tidal in this area, the water is often fresh or ex- tremely low in salinity. Site 2 is also located in the Squamscott River but only 0.5 km above the mouth. Salinity at this site is highly variable but intermedi- ate between the other two sites. Site 3, the site with greatest salinity, was located in the middle of Great Bay proper. The depth and bottom substratum were similar at all three stations (Table 1). Sites 3 and 4 (high salinity, shallow Great Bay) and site 5 (high salinity, Great Bay intertidal flats) were located along a depth gradient in a contiguous area in the middle of Great Bay. Site 3 was the deep- est site sampled along the depth gradient. Site 4 rep- resented the intermediate depth, and site 5 was lo- cated on intertidal mudflats and therefore sampled only on high tides. All three sites had similar bottom substratum, silty mud, and owing to their proxim- ity, experienced nearly identical salinities (Table 1). Monthly length frequencies at each site were pooled over all study years. The Kolmogorov-Smirnov test was used to test for differences in length-fre- quency distributions among sites. One-way analysis of variance (ANOVA) was used to test for significant differences in catches among the three sites that made up each of the two gradients. To reduce the number of ANOVA’s performed and to increase the power of the tests by increasing sample sizes, the monthly data were grouped into three seasons: spring, summer, and autumn. Months of April, May, and June were considered spring; July and August were considered summer; and September, October, and November were considered autumn. Because many months contained zero catches and, in some cases, the variances were proportionate to the means, the data were transformed by using a square-root transformation (square root(X+l)). The Kolmogorov- Smirnov test with the Lilliefors modification and probability plots of residuals indicated no significant deviations from normality, and Levene’s test indi- cated homogeneity of variances after the transfor- mation. Where a significant difference in catches was detected among sites, the sites were compared by us- ing Tukey’s HSD test (Zar, 1984). Results A total of 8,333 smooth flounder and 2,105 winter flounder were captured during the study period. Both juvenile and adult smooth flounder were abundant in the study area in contrast to winter flounder, which were abundant only as juveniles. However, length frequencies of the two flounders were similar because adult smooth flounder are about the same size as juvenile winter flounder. Smooth flounder were cap- tured from many different year classes, whereas win- ter flounder were primarily age 0+, 1+, and 2+, based on length frequencies. Salinity followed a typical boreal estuarine sea- sonal pattern (Figs. 2 and 3). The general trend at all stations was for salinity to be lowest in April, to increase over the late spring and summer months reaching the highest levels during August and Sep- tember, and to decline during autumn. These sea- sonal patterns were especially pronounced at site 1 and site 2. Salinities in spring of 1991 were higher at all sites than in the other two years, a result of an uncharacteristically dry spring and limited spring runoff. Another salinity anomaly occurring in 1991 Table 1 Physical characteristics of the sampling sites. Sites 1, 2, and 3 make up the salinity gradient, whereas sites 4, 5, and 3 form the depth gradient. Site number and habitat type Salinity (ppt) Temperature (°C) Depth (m) Bottom type Mean Range Mean Range Mean Range 1 (low salinity) 4.2 0.0-22.4 19.4 4.7-25.7 2.7 1. 9-4.0 silty mud 2 (medium salinity) 10.9 0.4-24.0 17.1 0.0-27.8 3.7 1.8-4. 3 silty mud 3 (high salinity, greatest depth) 20.3 6.5-29.9 15.4 1.8-23.9 6.2 4. 9-7. 9 silty mud 4 (intermediate depth) 20.9 6.5-29.5 16.4 2.3-24.9 2.1 1.5-4. 4 silty mud 5 (intertidal flats) 19.8 11.0-28.5 15.2 0.2-24.2 1.5 1. 1-2.2 silty mud Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 417 was a sudden decrease in salinity in September caused by dilution from the heavy rains with Hurri- cane Bob in late August of that year. Sites compris- ing the depth gradient had similar patterns of salin- ity in all years of the study. Salinity gradient Both species were unevenly distributed along the salinity gradient, and their distributions changed seasonally (Table 2). The timing of peak abundance of smooth flounder at site 1 varied from year to year. In 1989 and 1990 smooth flounder were abundant in mid to late summer (Fig. 4). The influx of smooth flounder was associated with seasonal changes in the salinity regime from fresh to oligohaline (Fig. 2). In 1991, smooth flounder were present at site 1 in all months sampled. In this year, salinity was higher than that during the two previous years (Fig. 2). Length frequencies of smooth flounder at site 1 (Fig. 5) were significantly different (P<0.0001 in all monthly K-S tests, May-October) from those at site 2 (Fig. 6), although the difference appears to be less in the autumn than in the spring. Larger fish (>100 mm) made up a higher pro- portion of the catch at site 1 in comparison with site 2, indicating differential migration among size classes. Winter flounder were rarely col- lected at site 1 (Fig. 7). They were found there only on a few occasions in September and Oc- tober when salinity was at a seasonal high. Smooth flounder were abundant at site 2 during all months, and their average abun- dance at this site exceeded that of all other sites. Their abundance was generally high in the spring, lower in late summer to early au- tumn, and high again in mid to late autumn (Fig. 4). This trend was opposite to that ob- served for site 1. Correlation analysis of catches of smooth flounder at sites 1 and 2 in- dicated a weak but significant negative rela- tionship (P=0.032, r=-0.48). When catches were large at site 1, they tended to be small at site 2. This finding suggests that the same population of smooth flounder was migrating between the stations, although the length fre- quencies show that a greater proportion of larger smooth flounder than smaller smooth flounder travel the 3 km between the sites. Winter flounder were abundant at site 2 only during autumn (Fig. 7), although even during these periods of abundance, the catches of win- ter flounder were always lower than those for smooth flounder. The movement of winter flounder into site 2 from Great Bay proper was associated with relatively high salinities (Fig. 2) and with low abundances of smooth floun- der (Fig. 4). The length frequencies of winter flounder collected from site 2 (Fig. 6) and site 3 (Fig. 8) were similar. Smooth flounder occurred at site 3 in rela- tive abundance only in April, May, and June (Fig. 4). Catches of smooth flounder decreased significantly after June in all years. Winter flounder were most abundant at this site than at the other two sites comprising the salinity Apr May Jun Jul Aug Sop Oct Nov Apr May Jun Jul Aug Sep Oct Nov Month Figure 2 Salinity at three sites sampled along a salinity gradient in Great Bay Estuary, New Hampshire. Mean salinity was highest at site 3 (20.3 ppt) followed by site 2 (10.9 ppt) and then site 1 (4.2 ppt). Solid line = 1989; dotted line = 1990; dashed line = 1991. 418 Fishery Bulletin 95(3), 1997 gradient (Fig. 7). They were present in relatively large numbers during all months. There were no sig- nificant differences in catches of winter flounder among months for all study years. Depth gradient The two species of flounder showed a differential use of the three sites that comprised the depth gradient. There were also differences in the sizes of flounder that used the three sites. Seasonal changes in distri- bution were less pronounced than those exhibited along the salinity gradient. A broad size range of smooth flounder used site 3 (Fig. 8). However, their abundance dropped off sharply after June of each year, as previously dis- cussed (Fig. 9). Winter flounder showed few seasonal trends in abundance at this station (Fig. 10). Abroad size range of juvenile winter flounder was found here. A distinct influx of young-of-the-year winter floun- der could be seen at site 3 from August through No- vember of each year (Fig. 8). At site 4, smooth flounder showed little sea- sonal change in abundance (Fig. 9), although there was a trend for catches to be lowest in late summer and early autumn. Length fre- quencies differed between site 3 and site 4. At site 4, few larger smooth flounder were present during any season (Fig. 11), whereas young- of-the-year, which were absent from site 3, were collected at most times. Abundance of winter flounder at site 4 was lowest in all years in early summer (Fig. 10), and catches were al- ways smaller than those at site 3. Length fre- quencies indicated that smaller winter flounder made up a greater proportion of the catch at site 4 (Fig. 11) as compared to site 3 (Fig. 8). Catches at site 5 were very variable for both species and showed no clear seasonal patterns (Figs. 9 and 10). Smooth flounder catches at site 5 were dominated by young-of-the-year. Few larger (>100 mm TL) individuals were ever caught at this site (Fig. 12), in contrast to site 3 (Fig. 8) but similar to site 4 (Fig. 11). Winter flounder occurred at site 5 sporadically and in very low numbers (Fig. 10). Catches of winter flounder were a mix of different sizes of juveniles. Discussion A variety of habitats are available to smooth and winter flounder in upper Great Bay Estu- ary. It was the purpose of this study to quan- tify the occurrence of these two species in vari- ous habitats. Although the species ranges of smooth and winter flounder overlap broadly, the evidence presented here indicates that they use habitats within the estuaries differ- ently and that their habitat use is subject to seasonal variations. Salinity gradient In general, smooth flounder were most abun- dant at site 2, the mesohaline river mouth Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 419 Table 2 Results of ANOVA’s testing for differences in catches of smooth and winter flounder among three sites along the salinity gradient (sites 1, 2, and 3) and three sites along the depth gradient (5, 4, and 3). If there was a significant difference (P<0.05) in catches among sites, the results of Tukey’s HSD test are listed from lowest to highest. See Table 1 for a description of the sites, ns = not significant. Year and season Smooth flounder Winter flounder Salinity gradient (F- value; df) Salinity gradient (F-value; df) 1989 Spring Summer Autumn site l S CTl CD CD CD CD CD 1 50 CD D A3 <1) -Q E Site 3 S S S oo S S S S <5> S c5> S S S cd S 5 O) CD 5> Sampling date Figure 4 Mean number of smooth flounder, Pleuronectes putnami, caught per ten minute tow at three sites along a salinity gradient in Great Bay Estuary, New Hampshire, May 1989-September 1991. Site 1 = oligohaline; site 2 = mesohaline; site 3 = polyhaline. Error bars are one standard error of the mean. Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 42! 50 - 40 30 - 20 - 1 o — May 50 — 40 - 30 - June 1 o - 50 - 40 - 30 - July 20 - > 1 o - c 0 ° 100 mm TL) smooth flounder occurred primarily at the deep- water station (site 3). They were abundant only during April-June, before migrating upriver as salinity increased. Small numbers remained at site 4 throughout the summer and autumn. The tidal flats (site 5) and shallow bay (site 4) were impor- tant nursery areas for smooth flounder. Young-of- the-year smooth flounder did not show a dramatic decrease in abundance during the summer, as seen in the larger individuals, and did not appear to make a pronounced seasonal up-estuary move- ment. Their inferior swimming ability, compared with that of larger individuals, or their inability to osmoregulate efficiently in lower salinity areas may underlie their relatively stationary habits. The tendency for smooth flounder to segregate by size, with the smaller individuals occurring in the intertidal and shallow subtidal areas, has been found in several other flatfish species including English sole, Parophrys vetulus (Toole, 1980), and European plaice, Pleuronectes platessa (Gibson, 1973; Kuipers, 1973). Segregation by size may re- duce intraspecific competition. The intertidal zone may also function as a refuge from predators for small flatfish or as an abundant source of appro- priate-size prey items (Toole, 1980). Ruiz et al. (1993) found that shallow water functioned as a refuge from size-selective predation on juveniles of several species of fish and crustaceans in Chesa- peake Bay. Van der Veer and Bergmann (1986) found that young-of-the-year European plaice used tidal flats as a refuge from predators rather than for feeding purposes. Potential predators on smooth flounder in Great Bay Estuary include sand shrimp ( Crangon septemspinosus), grubbies (Myoxocephalus aeneus), bluefish (Pomatomus saltatrix ), striped bass (Morone saxatilis), white perch ( Morone americanus), great blue heron ( Ardea herodias), and double-crested cormorants ( Phala - crocorax auritus ). Predation by large piscine predators is probably reduced in shallow water, and avian preda- tion is likely increased. Sand shrimp were abundant in trawl samples from both channel and flats areas. The value of tidal flats as refugia from predation cannot be assessed without knowledge of the relative rates of pre- dation by these different predatory groups. 422 Fishery Bulletin 95(3), 1997 eo - 60 - 40 - 20 - Smooth flounder April i ao - 60 - -40 - 20 - j May Jk- Winter flounder 80 60 - -40 - 20 - Jl 3 June 80 - 60 - 40 - 20 - > o „ ■ July .Jl. July k. c ° 25- LL 20 - 20- August 15- August 1 O - 5 - io- 5- Jhb «.L « - O o - 25 - 20 - 15- 1 o - 5 - September O 50 1 0O 1 50 200 250 300 25- 20 - October 15- 1 O 5 - ilAli . 25 - 20 - November 1 o - 5 - o - rrTTTTTTT 1^1 l 1W1 l^l TTTTTTTTrTTTT^TnTTTT O 50 1 OO 1 50 200 250 300 Total length (mm) Figure 8 Monthly length frequencies (5-mm size classes) of smooth flounder, Pleuronectes putnami, and winter flounder, P. americanus, at site 3, pooled over 1989-91. Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 425 ( 1966) found that winter flounder, acclimated to 21°C, had an upper lethal temperature of 27°C. Pearcy (1962) found an upper lethal temperature of 30°C for flounder collected during the summer in Mystic River Estuary. Olla et al. (1969) observed that win- ter flounder exposed to temperatures above 22.2°C. buried themselves in sediment and ceased to feed. Although comparable data do not exist for smooth flounder, Huntsman and Sparks ( 1924) reported that upper lethal temperatures for smooth flounder were 2-4°C higher than those for winter flounder. In Great Bay Estuary, temperature may be a factor in deter- mining the relative distribution of the two species in late summer when water temperatures at site 5 reached 22-24. 2°C but would not be a factor during most of the year. The low abundance of winter floun- der at site 5 persisted during times of the year when temperature would not seem to be limiting. Subtrate preference may play a role in excluding winter flounder from inter- tidal flats in Great Bay Estuary. Al- though the bottom type appeared simi- lar (silty mud) at all three sites along the depth gradient (Table 1), this simi- larity was based on gross examination of core samples. No detailed sediment size analysis was conducted for this study (nor in Fried [1973] or Targett and McCleave [1974]), and therefore differ- ences in sediment structure may have been present between sites but not noted on a gross scale. Sogard (1992) found that growth of winter flounder was negatively correlated with percent silt; faster growth occurred in sandier sediments. Bigelow and Schroeder ( 1953) found that winter flounder were more abundant on coarser sediments, in comparison with smooth flounder which were more abundant in muddier sedi- ments. Thus, if the channel areas of Great Bay Estuary have coarser sedi- ments than the intertidal flats, perhaps the coarser sediment may explain the difference in distribution along the depth gradient. Summary Smooth and winter flounder are par- tially segregated as species along salin- ity and depth gradients in upper Great Bay Estuary. It appears that this is due to differential responses to the physical and chemical regime, but the effects of seasonal changes in biotic interactions cannot be excluded. Smooth and winter flounder feed on similar prey items in Great Bay Estuary (Laszlo, 1972; Armstrong, 1995). Competition or move- ments related to prey abundance may influence their respective distributions. There are many instances where com- 0) -Q £ Sampling date Figure 9 Mean number of smooth flounder, Pleuronectes putnami , caught per ten minute tow at three sites along a depth gradient in Great Bay Estuary, New Hampshire. Mean depth was greatest at site 3 (mean=6.2 m) fol- lowed by site 4 (2.1 m) and then site 5 ( 1.5 m). Error bars are one standard error of the mean. 426 Fishery Bulletin 95(3), 1 99 7 CD CT> 05 CT> 05 05 <=> C~3 <— > O C~ > O c~ > <— > — - — 00 CO 00 CO CO CO 05 05 05 05 05 0> CT> 05 0> 05 CD CD CD CD CO co of o ^ ^ in co" r~t co" of o ^ ^ iif to r~^T co of Sampling date Figure 10 Mean number of winter flounder, Pleuronectes americanus, caught per ten minute tow at three sites along a depth gradient in Great Bay Estuary, New Hampshire. Mean depth was greatest at site 3 (mean=6.2 m) fol- lowed by site 4 (2.1 m) and then site 5 (1.5 m). Error bars are one standard error of the mean. petition appears to play a role in the distribution of ecologically similar species along environmental gra- dients (Connor and Bowers, 1987). In Great Bay Es- tuary, low salinity and intertidal flats appear to pro- vide at least a partial refugium for smooth flounder from competition with winter flounder. The relation between smooth and winter flounder changes on a seasonal basis. At times their segrega- tion on a spatial scale is nearly complete, whereas at other times, particularly April-June at site 3 and September-October at site 2, they overlap consider- ably. Competition theory predicts that niches should vary temporally as a function of resource abundance and of the population densities of potential competi- tors (Llewellyn and Jenkins, 1987). The predominant temporal pattern of niche overlap seen in studies is increased overlap during resource abundance (Schoener, 1982; Ross, 1986). The periods of great- Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 427 Smooth flounder Winter flounder Total length (mm) Figure 1 1 Monthly length frequencies (5-mm size classes) of smooth flounder, Pleuronectes putnami, and winter flounder, P. americanus, at site 4, pooled over 1989-91. est overlap in habitat use seen for smooth and win- ter flounder may be associated with an abundance of some resource, for example, a shared prey item(s). The upper Great Bay Estuary is an important area for both species. This study has shown the dynamic nature of habitat use by smooth and winter floun- 428 Fishery Bulletin 95(3), 1997 der. Further studies are needed to assess experimen- tally the relative importance of abiotic versus biotic factors in determining the patterns of smooth and winter flounder spatial distributions. Acknowledgments I wish to thank the countless legions of work-study students, graduate students, and staff at the Zool- ogy Department of the University of New Hampshire who assisted at various points in this work. S. Cadrin, D. Adams, H. Howell, J. Musick, L. Harris, P. Sale, J. Taylor, and three anonymous referees provided helpful reviews of the manuscript. The research formed part of a dissertation submitted in partial fulfillment of the Ph.D. degree, Department of Zool- ogy, University of New Hampshire. This work was supported in part by a grant from the U. S. Fish and Wildlife Service Wallop-Breaux Fund and by a Cen- tral University Research Grant from the University of New Hampshire. Literature cited Armstrong, M. P. 1995. A comparative study of the ecology of smooth floun- der, Pleuronectes putnami, and winter flounder, Pleuro- nectes americanus, from Great Bay Estuary, New Hamp- shire. Ph.D. diss., Univ. New Hampshire, Durham, NH, 147 p. Bigelow, H. B., and W. C. Schroeder. 1953. Fishes of the Gulf of Maine. U.S. Fish Wildl. Serv. Fish. Bull. 53, 577 p. Black, R., and R. J. Miller. 1991. Use of the intertidal zone by fish in Nova Scotia. Environ. Biol. Fishes 31:109-121. Casterlin, M. E., and W. W. Reynolds. 1982. Thermoregulatory behavior and diel activity of year- ling winter flounder, Pseudopleuronectes americanus (Walbaum). Environ. Biol. Fishes 7(2): 177 — 180. Coggan, R. A., and P. R. Dando. 1988. Movements of juvenile Dover sole, Solea solea (L.), in the Tamar Estuary, south-western England. J. Fish Biol. 33:177-184. Connor, E. F., and M. A. Bowers. 1987. The spatial consequences of interspecific competi- tion. Ann. Zool. Fenn. 24:213-226. Dorel, D., C. Koutsikopoulos, Y. Desaunay, and J. Marchand. 1991. Seasonal distribution of young sole (Solea solea L.) in the nursery ground of the Bay of Vilaine (northern Bay of Biscay). Neth. J. Sea Res. 27:4297-4306. Dovel, W. L., J. A. Mihursky, and A. J. McErlean. 1969. Life history aspects of the hogchoker, Trinectes maculatus, in the Patuxent River estuary, Maryland. Chesapeake Sci. 10(2):104-119. Fried, S. M. 1973. Distribution of demersal fishes in Montsweag Bay- Back River and lower Sheepscot River estuaries, Wiscaccet, Maine. M.S. thesis, Univ. Maine, Orono, ME, 47 p. Gibson, R. N. 1973. The intertidal movements of young fish on a sandy beach with special reference to the plaice (Pleuronectes platessa ). J. Exp. Mar. Biol. Ecol. 12:79—102. Gordon, D. C., and M. J. Dadswell. 1984. Update on the marine environmental consequences Armstrong: Distribution and abundance of Pleuronectes putnami and Pleuronectes americanus 429 of tidal power development in the upper reaches of the Bay of Fundy. Can. Tech. Rep. Fish. Aquat. Sci. 1256, 292 p. Hanson, J. M., and S. C. Courtenay. 1996. Seasonal use of estuaries by winter flounder in the southern Gulf of St. Lawrence. Trans. Am. Fish. Soc. 125:705-718. Hoff, J. G., and J. R. Westman. 1966. The temperature tolerance of three species of ma- rine fishes. J. Mar. Res. 24(2):131— 140. Huntsman, A. G., and M. I. Sparks. 1924. Limiting factors for marine animals. 3: Relative re- sistance to high temperatures. Contrib. Can. Biol., N. S. 2(61:97-114. Jackson, C. F. 1922. Ecological features of Great Bay, New Hampshire, and a preliminary checklist of fishes. Ecology 3( 1 1:48-54. Kerstan, M. 1991. The importance of rivers as nursery grounds for 0- and 1- group flounder ( Platichthys flesus L.) in compari- son to the Wadden Sea. Neth. J. Sea Res. 27:353-366. Klein-MacPhee, G. 1978. Synopsis of biological data for the winter flounder, Pseudopleuronectes americanus (Walbaum). U.S. Dep. Commer., NOAATech. Rep. NMFS Circ. 414, 43 p. Kuipers, B. 1973. On the tidal migrations of young plaice ( Pleuronectes platessa) in the Wadden Sea. Neth. J. Sea Res. 6:376- 378. Laszlo, P. T. 1972. Age-growth, food, and reproduction of the smooth flounder, Liopsetta putnami (Gill), in Great Bay, New Hampshire. Ph.D. diss., Univ. New Hampshire, Durham, NH, 75 p. Llewellyn, J. B., and S. H. Jenkins. 1987. Patterns of niche shift in mice: seasonal changes in microhabitat breadth and overlap. Am. Nat. 129(31:365- 381. Martin, T. J. 1990. Osmoregulation in three species of Ambassidae (Osteichthyes: Perciformes) from estuaries in Natal. S. Afr. J. Zool. 25(21:229-234. McCracken, F. D. 1963. Seasonal movements of the winter flounder, Pseudopleuronectes americanus, on the Atlantic coast. J. Fish. Res. Board Can. 20(21:551-585. Menge, B. A., and A. M. Olson. 1990. Role of scale and environmental factors in regula- tion of community structure. Trends Ecol. & Evol. 5:52- 57. Moser, M. L., and L. R. Gerry. 1989. Differential effects of salinity changes on two estua- rine fishes, Leiostomus xanthurus and Micropogonias undulatus. Estuaries 12(11:35-41. Olla, B. L., R. Wicklund, and S. Wilk. 1969. Behavior of winter flounder in a natural habi- tat. Trans. Am. Fish. Soc. 4:717-720. Oviatt, C. A., and S. W. Nixon. 1973. The demersal fish of Narragansett Bay: an analysis of community structure, distribution and abundance. Estuarine Coastal Mar. Sci. 1:361-378. Pearcy, W. G. 1962. Ecology of an estuarine population of winter floun- der Pseudopleuronectes americanus (Walbaum). Bull. Bingham Oceanogr. Collect. Yale Univ. 18(1), 78 p. Peters, D. S. 1971. Growth and energy utilization of juvenile flounder, Paralichthys dentatus and Paralichthys lethostigma, as affected by temperature, salinity, and food availa- bility. Ph.D. diss., North Carolina State Univ., Raleigh, NC, 69 p. Peters, D. S., and M. T. Boyd. 1972. The effect of temperature, salinity, and availability of food on the feeding and growth of the hogchoker, Trinectes maculatus (Block and Schneider). J. Exp. Mar. Biol. Ecol 7:201-207. Peterson, M. S., and S. T. Ross. 1991. Dynamics of littoral fishes and decapods along a coastal river-estuarine gradient. Estuarine Coastal Shelf Sci. 33:467-483. Powles, H., F. Auger, and G. J. Fitzgerald. 1984. Nearshore ichthyoplankton of a north temperate estuary. Can. J. Fish. Aquat. Sci. 41:1653-1663. Poxton, M. G., and N. A. Nasir. 1985. The distribution and population dynamics of 0-group plaice ( Pleuronectes platessa L.) on nursery grounds in the Firth of Forth. Estuarine Coastal Shelf Sci. 21:845-857. Riley, J. D., D. J. Symonds, and L. Woolner. 1981. On the factors influencing the distribution of 0-group demersal fish in coastal waters. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 178:223-228. Ross, S. T. 1986. Resource partitioning in fish assemblages: a review of field studies. Copeia 2:352-388 Ruiz, G. M., A. H. Hines, and M. H. Posey. 1993. Shallow water as a refuge habitat for fish and crus- taceans in non-vegetated estuaries: an example from Chesapeake Bay. Mar. Ecol. Prog. Ser. 99:1-16. Saucerman, S. E., and L. A. Deegan. 1991. Lateral and cross-channel movement of young-of-the- year winter flounder (Pseudopleuronectes americanus) in Waquoit Bay, Massachusetts. Estuaries 14(4):440-446. Schoener, T. W. 1982. The controversy over interspecific competition. Am. Sci. 70:586-595. Scott, W. B., and M. G. Scott. 1988. Atlantic fishes of Canada. Can. Bull. Fish. Aquat. Sci. 219, 731 p. Short, F. T. (ed.). 1992. The ecology of the Great Bay Estuary, New Hamp- shire and Maine: an estuarine profile and biblio- graphy. U.S. Dep. Commer., NOAA-Coastal Ocean Pro- gram Publ., 222 p. Available from the Great Bay National Estuarine Research Reserve, Durham, NH. Smith, S. M. 1986. Reproductive ecology, population dynamics and sea- sonal movements of the hogchoker (Trinectes maculatus ) in the Elizabeth River, Virginia. M.S. thesis, College of William and Mary, Williamsburg, VA, 83 p. Sogard, S. M. 1992. Variability in growth rates of juvenile fishes in dif- ferent estuarine habitats. Mar. Ecol. Prog. Ser. 85:35-53. Targett, T. E., and J. D. McCleave. 1974. Summer abundance of fishes in a Maine tidal cove with special reference to temperature. Trans. Am. Fish. Soc.l03(2):325-330. Terborgh, J. 1971. Distribution on environmental gradients: theory and a preliminary interpretation of distributional patterns in the avifauna of the Cordillera Vilcabamba, Peru. Ecology 52:2-40. Toole, C. L. 1980. Intertidal recruitment and feeding in relation to op- 430 Fishery Bulletin 95(3), 1 997 timal utilization of nursery areas by juvenile English sole (Parophrys vetulus: Pleuronectidae). Environ. Biol. Fishes 5(4):38-390. Turgeon, D. D. 1976. Distribution of the planktonic larvae of some benthic invertebrates within the Piscataqua-Great Bay Estuary, New Hampshire. Ph.D. diss., Univ. New Hampshire, Durham, NH, 165 p. Tyler, A. V. 1971. Surges of winter flounder Pseudopleuronectes americanus into the intertidal zone. J. Fish. Res. Board Can. 28(11): 17 17-1732. van der Veer, H. W., and M. J. N. Bergmann. 1986. Development of tidally related behaviour of a newly settled 0-group plaice ( Pleuronectes platessa ) population in the western Wadden Sea. Mar. Ecol. Prog. Ser. 31(2):121-129. Wells, B., D. H. Steele, and A. V. Tyler. 1973. Intertidal feeding of winter flounders (Pseudo- pleuronectes americanus) in the Bay of Fundy. J. Fish. Res. Board Can. 30:1374-1378. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Inc., Engle- wood Cliffs, NJ, 718 p. 431 Energetics of larval red drum, Sciaenops ocellatus. Part II: Growth and biochemical indicators Ross I. Brightman Joseph J. Torres Joseph Donnelly Department of Marine Science University of South Florida 1 40 Seventh Ave South, St. Petersburg, Florida 33701 E-mail address. BrightmanOemail. spjc.cc.fi. us M. Elizabeth Clarke Division of Marine Biology and Fisheries Rosenstiel School of Marine and Atmospheric Science University of Miami 4600 Rickenbacker Causeway, Miami, Florida 33 1 49 Abstract .—The effects of ration level and temperature on growth were deter- mined for larval red drum, Sciaenops ocellatus, during its first two weeks of life. Larvae were raised in the labora- tory at 20°C at a ration level of 5.0 prey/ mL, at 25°C at ration levels of 0, 0.1, 1.0, and 5.0 prey/mL, and in growout ponds at 25°C and 32°C and at ration levels of 4-6 prey/mL. Growth was mea- sured as standard length, wet mass, and dry mass. Proximate (water, ash, protein, and lipid) and elemental (C, N) composition was determined at larval ages of 0, 2, 4, 6, 10, and 14 d to pro- vide caloric values for the growing lar- vae and to examine the relative impor- tance of protein and lipid during tissue deposition in the very early life history of these larvae. Biochemical indicators of growth, RNA-DNA ratio, and activ- ity of the metabolic enzyme lactate de- hydrogenase (LDH) were examined in larvae reared at all temperature and ration combinations. The effectiveness of the biochemical indicators as prox- ies for growth was assessed by compar- ing the directly measured growth rates with RNA:DNA levels and LDH activ- ity. Larvae fed a ration of 1.0 prey/mL or less did not survive past the age of eight days. Growth rate increased with increasing temperature, reaching a maximum of 60% body mass/d in growout ponds at 32°C. Protein level (percent ash free dry mass: %AFDM) increased with increasing age in all treatments where individuals exhibited positive growth, whereas lipid (%AFDM) showed a concomitant decline. Nitrogen (%AFDM) and carbon (%AFDM) varied directly with protein and lipid contents, respectively. Biochemical indicators of growth showed a significant correlation with growth rate. However, the char- acter of the correlation changed with temperature. RNA-DNA ratios and en- zymic activities were lower at higher temperatures for equivalent growth rates. Introduction of a temperature term into multiple regression equations improved the relation between growth and the biochemical proxies. LDH ac- tivity scaled with the size of larvae, whereas RNA:DNA showed no signifi- cant relation with size. Manuscript accepted 4 February 1997. Fishery Bulletin 95:431-444 (1997). Red drum, Sciaenops ocellatus, is an important species in commercial and recreational fisheries in the southeastern United States, par- ticularly in the Gulf of Mexico. De- clines in red drum stocks (Swingle, 1990) have stimulated considerable interest in the early life history of this species, resulting in stock en- hancement programs and larval monitoring programs designed both to improve and continually to assess the status of the fish in the field. Studies on red drum and other spe- cies indicate clearly that growth during the pretransformation pe- riod of development is particularly critical to survival (Buckley, 1980; Holt et al., 1981, a and b; Holt and Arnold, 1983; Holt, 1990). The in- crease in size and mobility that characterizes development during the early larval period results in an increase in the size range of prey items available to the larvae as for- age and a decrease in the size range of potential predators on the larvae. Two variables with great poten- tial to influence rates of growth are temperature and ration levels. As a subtropical species, red drum de- velop at temperatures greater than 20°C, grow rapidly, and have a greater energy demand for metabolic processes than do larvae developing in colder systems. For example, red drum eggs at 25°C hatch in 24 h, and larvae begin feeding in 48-72 h, whereas cold water species, such as Atlantic cod, Gadus morhua, and winter flounder, Pleuronectes ameri- canus, developing at 4— 8°C, spend 30 d as developing eggs.1 High tempera- tures during early development stimulate rapid growth in red drum but leave them potentially more vul- nerable to rapid starvation in absence of sufficient food. The interaction be- tween temperature, ration level, and growth in size and calories, is an im- portant part of the energetics of lar- val red drum, basic information which is unavailable for red drum and limited for other subtropical te- leosts (Houde and Schekter, 1983). 1 Hempel, G. 1979. Early life history of marine fishes; the egg stage. Univ. Wash- ington Press, Seattle, WA, 70 p. 432 Fishery Bulletin 95(3), 1997 Objectives of the present study were four-fold. The first was to examine growth in size and energy in laboratory-reared red drum larvae, from egg to on- set of transformation, at a single ration level (5 prey/ mL) and at two temperatures (20 and 25°C). This examination was achieved by using direct measure- ments of standard length and mass with age; the biochemical composition and caloric value of the growing larvae were described by analyzing their proximate and elemental composition (water, ash, protein, lipid, carbon, and nitrogen). The second was to examine the relation of growth in size and energy as a function of ration level (0, 0.1, 1.0, and 5.0 prey/ mL) at a single temperature (25°C). The third was to compare growth in size and energy in laboratory- reared larvae at 20 and 25°C and in the more het- erogeneous conditions encountered by pond-reared larvae at 25 and 32°C. The fourth objective was to describe the relation between growth, temperature, and biochemical indicators of growth and condition: RNArDNA ratios and activity of the key intermedi- ate metabolic enzyme lactate dehydrogenase (LDH). Methods and materials Laboratory maintenance Fertilized eggs were obtained from the Florida De- partment of Environmental Protection (FDEP) hatch- ery, Port Manatee, Florida. Broodstock were main- tained at 25°C and 30 ppt. Eggs were obtained from five females and from separate spawnings, from No- vember 1990 to November 1991, for all growth ex- periments described below. Broodstock females were similar in size, kept in highly controlled conditions, and fed well. Spawning was induced naturally by manipulation of photoperiod. As a consequence, eggs were very uniform in size, 0.9 to 1.0 mm in diameter. Eggs were transported to the USF Marine Science Laboratory in St. Petersburg and sorted into 26-L ex- perimental aquaria at a concentration of 2,500-3,000 individuals per aquarium. High mortality associated with first feeding resulted in a 30-40% reduction in initial numbers by day 3. Aquaria were placed in a photoperiod- and temperature-controlled incubator and maintained at either 20°C or 25°C and at a salinity of 30 ppt. Eggs were introduced to the 20°C temperature by slow exchange of water over a 60-minute period. A 13-h light and 11-h dark photoperiod was used through- out all experiments. Larvae were fed rotifers (Brach- ionus plicatilis) beginning at day 3 posthatch until flex- ion (approximately day 14), when experiments were terminated. Aquaria were aerated and a portion of the saltwater in each was changed daily. Rotifers were obtained from Florida Aqua Farms, Dade City, Florida, and cultured according to the procedure of Hoff and Snell (1987). Rotifers were fed Chlorella once a day to avoid any loss in nutritional value. Seawater for culturing was obtained off- shore in the Gulf of Mexico. The seawater was coarse- filtered, then treated with bleach (sodium hypo- chorite, 5.25%) to remove any additional plankton, and neutralized with sodium thiosulphate. Seawa- ter salinity was adjusted with distilled water and Tropic Marine Seasalt to achieve a final salinity of 30 ppt. Pond maintenance Pond-reared red drum larvae were obtained from the FDEP growout ponds, Port Manatee, Florida. Lar- vae from a single spawn were added to the plank- ton-rich ponds within 24 hours after hatching and allowed to grow. Two ponds, one at 25°C and another at 32°C, were sampled for the first 18 days of life of the red drum larvae. Temperature was monitored twice daily; the average temperature for the two- week sampling period was used to characterize the ponds. Prey items in the ponds were monitored by siev- ing water samples into two size categories: 35-220 pm (copepod nauplii, rotifers, and small copepods) and larger than 220 pm (copepods); prey were then counted in 200-mL aliquots of each size range. The concentration of prey between 35 and 220 pm was 3-5 prey/mL, whereas that greater than 220 pm was 0.5-1 prey/mL in both the 25°C and 32°C ponds. Growth versus prey density Eggs from a single spawn were divided into four 26-L aquaria for experiments on growth versus prey den- sity at 25°C. Prey were provided at four densities, 0, 0.1, 1.0, and 5.0 prey items per mL, from first feed- ing (day 3) through the start of flexion (day 14). Prey concentrations were monitored twice daily by remov- ing a 25-mL sample from each aquarium, counting the number of prey in 5-mL aliquots, and taking the average. Prey concentrations were adjusted as nec- essary. Larvae reared at 20°C were fed prey at a ra- tion level of 5.0 prey items per mL. Standard length measurements Growth in stan- dard length was monitored according to prey con- centration. Aquaria with 0, 0. 1, and 1.0 prey/mL were sampled daily. Aquaria with 5.0 prey/mL were sampled every other day, and ponds were sampled every third day. The samples were taken each morn- ing before the larvae began to feed. Standard length Brightman et al.: Energetics of larval Sciaenops ocellatus 433 of five individual larvae that had been anesthetized with MS-222 was measured with the aid of a dis- secting microscope. Standard length was considered to be the distance from the snout to the tip of the tail in preflexion larvae and from the snout to the tip of the notochord in post-flexion larvae. Mass measurements Growth in mass was moni- tored at the same intervals as those used for stan- dard length. At each monitoring interval, 30 indi- viduals were removed for wet, dry, and ash-free dry mass analysis. To determine mass, larvae were first separated into three groups of ten. Each group of larvae was filtered onto a preweighed 0.5-cm Whatman glass fiber filter (made with an office hole- punch) that was placed in a custom-made miniatur- ized vacuum funnel. Larvae were then rinsed very briefly by introducing distilled water into the funnel with a pasteur pipette and by removing the water immediately with the vacuum filter. To minimize evaporation, samples were immediately placed in preweighed microcentrifuge tubes which were then weighed to the nearest pg on a Mettler electrobalance to determine wet mass. Specimens were dried at 60°C to a constant mass (about 24 h) to determine dry mass. Average proximate and elemental composition of prey items Rotifers were collected in bulk from two 28-L cul- ture bags (approximately 50 mg dry mass/bag) for determination of proximate and elemental composi- tion. Proximate composition (water, ash, protein, and lipid content) was determined by using the methods of Stickney and Torres (1989) and Donnelly et al. (1990). Elemental composition was determined by using a C:H:N analyzer. Average proximate and elemental composition of fish larvae Methods used to estimate the proximate and elemen- tal composition of fish larvae were the same as those used for prey. Larvae were obtained in bulk (50 mg dry mass) for each day sampled. Each pond was sampled from the hatchery at 0, 2, 6, 10, and 14 days. Laboratory-raised larvae were sampled at prey con- centrations of 0, 0.1, 1.0, and 5.0 prey/mL at 0, 2, 6, 10, and 14 days for each of four spawns. Protein and lipid values as percent ash-free dry mass (%AFDM) were multiplied by individual ash-free dry mass to obtain concentrations as mg/individual. The instan- taneous protein growth rate (Gpi) was calculated by using the formula from Buckley (1982): Gpi = lnM„-lnM„xl()0 t2 ~ tx where M - mass in mg; and t = age in d. Caloric content of prey and larvae Caloric content was calculated from proximate com- positional data of the rotifers and larvae by using a value of 0.0048 cal/pg for protein and 0.0095 cal/pg for lipid (Brett and Groves, 1979). RNA-DNA ratio Ten to 20 individuals were removed for analysis of RNA:BNA content each time sampling occurred for measurements of mass. Larvae were filtered onto preweighed Whatman glass-fiber filters, rinsed with distilled water, weighed, placed in microcentrifuge tubes, and frozen at -80°C until analysis. RNA:DNA was analyzed by first homogenizing the freshly thawed groups of larvae in 1.2 M NaCl, then by us- ing the sequential enzymatic method of Bentle et al. (1981) to determine RNA:BNA. Activity of lactate dehydrogenase Larvae were sampled in bulk (minimum 10-20 mg wet tissue mass) every day at a prey concentration of 0 prey/mL. Samples were taken at 0, 2, 6, 10, and 14 days for larvae fed 5 prey/mL and for those col- lected in the growout ponds. Tissue was introduced frozen into the homogenizing medium, ice-cold Tris/ HCL buffer (10 mM, pH 7.5 at 10°C), and homog- enized by hand at 0 to 4°C with conical glass homog- enizers having ground-glass contact surfaces ( Kontes Glass Co., “Duall” models). Homogenates were cen- trifuged at 4,500 xg for 10 minutes and the superna- tants saved for enzyme analysis. L-Lactate dehydrogenase (LDH, EC 1.1.1.27; Lac- tate: NAB+ Oxidoreductase) activity was assayed in the pyruvate reductase direction by using methods described in Torres and Somero ( 1988) at a tempera- ture of 25°C. Enzyme activity was expressed as units/ gWM (wet mass), where a unit was 1 pmole of sub- strate converted to product per minute. Statistical analyses Simple regressions for each relationship were fitted by using the least-squares method (Statgraphics Plus, Manugistics Corporation). Data from treatment groups were compared by using one-way analysis of variance (ANOVA). Differences between the means 434 Fishery Bulletin 95(3), 1 997 were determined by using the least-significant-differ- ences multiple range test. Multiple regression analy- sis (Statgraphics Plus, Manugistics Corporation) was used to examine the relation between the ob- served protein growth rates and the following combina- tions of factors: the three experimental temperatures (20°C laboratory, 25°C labo- ratory, 25°C pond, 32°C pond), the three ration lev- els at 25°C (0, 5 prey/mL, and pond), RNA:DNA, and LDH activity. £ CO Results Growth rate versus prey density Standard length Starved red drum larvae (0 prey/ mL) kept at 25°C increased in standard length even as they were declining in dry mass (Figs. 1 and 2; Table 1 ). The average size at death on day 6 was 2.89 mm, which corresponds to a daily in- crease of 0.075 mm/day for days 2-6. Surprisingly, these values were similar to those for larvae fed at 5.0 prey/mL at 25°C, which attained an average length of 2.90 mm by day 6. Growth in larvae fed ad libitum increased with in- creasing temperature in both the laboratory and ponds. At 20°C, day- 14 larvae averaged 4.13 mm, and very few of the larvae exhibited flexion of the noto- chord (Fig 1; Table 1). Laboratory-reared individu- als at 25°C grew to an average of 4.45 mm by day 14. Mean masses at day 14 were significantly different between the two temperatures (ANOVA: df= 1 , F=60.3, P<0.001). Flexion of the notochord within the tail region had begun by day 14 at 25°C, indicating the onset of transformation. Larvae raised in ponds at 25°C averaged 6.45 mm at day 14. Notochord flex- ion in these larvae began on day 9 or 10, when the 12 15 12 15 18 Age (days) Figure 1 Growth in standard length of red drum larvae raised in the laboratory and in growout ponds: (A) laboratory, 25°C and 0 prey/mL; (B) laboratory, 20°C and 5 prey/mL; (C) labo- ratory, 25°C and 5 prey/mL; (D) pond, 25°C; and (E) pond, 32°C. larvae were at a size between 4.11 and 4.50 mm. Pond-reared larvae at 32°C had reached an average standard length of 7.48 mm at day 14. Notochord flexion occurred on day 7 or 8, when larvae were at a length of 4.53 mm (similar to the size at flexion of the laboratory -raised larvae) and at an earlier chronologi- cal age than that of the laboratory-raised individuals. Mass measurements Growth in mass of laboratory- reared larvae at ration levels less than 5.0 prey/mL (0, 0.1, and 1.0 prey/mL) at 25°C was negative, and no larvae survived more than 8 days (Fig. 2). Slopes Brightman et al.: Energetics of larval Sciaenops ocellatus 435 40 1 ' 1 1 1 1 ' 1 1 ' 1 1 1 1 ' ' 1 r: A - 150 - D 16000 J rTTTr 1 ' 1 1 ' ' ! 1 1 1 I ' 1 1 I 1 1 ' l_ V - : Y= exp (3.2 - 0.236X) ! 2 30 r = 0.60 120 - Y= exp (2.14 + 0.100X) - 2 12000 . Y= exp (2.56 + 0.325X) J r =0.86 ; r =0.89 •k : • 90 20 ' \ ‘ • - 8000 - ! \ j : 60 10 - • ; • { ; ■ - . 4000 - 30 - 0 / - : d . flexion \X 0 0 0 0 2 4 6 8 0 3 6 9 12 15 0 8 10 12 14 16 18 40 | T-l-T-f 1 1 1 | 1 1 1 | 1 1 1 T B - 150 i 1 1 i 1 ' i 1 1 i 1 1 i 1 1 r e: 16000 : g" 30 ' Y = exp (2.99 - 0.204 A) ; 2 - 0.45 120 ■ /= exp (2.02 + 0.176X) • - 2 : ■ 12000 1 Y= exp (1.06 + 0.471X) 1 flexion, ‘ / ■ ■ r = 0.96 / • E 90 / - / zL /• / w 20 -v. . - ' / ' ■ 8000 / 05 . : 60 / ' “ / E I / Q 10 1 . : ! . / 4000 / 30 i V* ! / .flexion / 0 0 0 0 2 4 6 8 0 3 6 9 12 15 0 8 10 12 14 16 18 40 c: ■; Y= exp (3.21 -0.231X) 30 -« r2 = 0.62 " 20 ; 10 : ' 1 M : • 0 0 2 4 6 8 Age (days) Figure 2 Growth in dry mass of red drum larvae raised in the laboratory and in growout ponds: (A) laboratory, 25°C and 0 prey/mL; (B) laboratory, 25°C and 0.1 prey/mL; (C) laboratory, 25°C and 1 prey/mL; (D) laboratory, 20°C and 5 prey/mL; (E) laboratory, 25°C and 5 prey/mL; (F) pond, 25°C; and (G) pond, 32°C of the three curves describing the time-dependent decline in individual dry biomass at the three ration levels were not significantly different (Student’s /-test: P> 0.05). Growth in dry mass at a ration level of 5.0 prey/mL was significantly higher than at the three lower ration levels (0, 0.1, and 1.0 prey/mL). In contrast, wet masses in 2-6 day-old larvae were slightly lower at ration levels of 0, 0.1, and 1.0 prey / mL than at 5.0 prey/mL but were not significantly different. This finding indicated that body water con- tent was increasing in larvae maintained at the three lower ration levels. Larvae held at 25°C and fed 5.0 prey/mL had higher growth rates than those raised at 20°C at the same ration level (Fig. 2; Table 1). Growth averaged 2.86 pg/day for the first two weeks of life in larvae raised at 20°C and 10.09 pg/day for larvae reared at 25°C. Expressed as a percent increase in mass, larvae reared at 20°C increased 10.5 %BM/d and those reared at 25°C increased 19.3 %BM/d. 436 Fishery Bulletin 95(3), 1997 Brightman et al.: Energetics of larval Sciaenops ocellatus 437 Growth rates for pond-raised red drum larvae (Fig. 2) were far greater than those for larvae fed rotifers in the laboratory, owing almost certainly to the greater prey diversity in the ponds. Larvae raised at 25°C in the ponds increased in size an average of 81.9 pg/day over 14 days: a 47.2 %BM/d increase. Larvae raised at 32°C in the ponds increased an av- erage of 799.22 mg/day: a 60.2 %BM/d increase; an increase of 10°C in the pond environment resulted in a two- to three-fold increase in absolute daily mass gain. Proximate and elemental composition of prey Brachionus plicatilis raised on Chlorella exhibited a protein level of 32.71% and a lipid level of 9.37% of its ash-free dry mass (AFBM). Carbohydrate level was low, averaging 2.84% (AFDM). The unrecovered mass was assumed to be due to refractory structural molecules that were not assayed for, e.g. chitin. Us- ing a figure of 0.24 pg for average individual bio- mass (Hoff and Snell, 1987) and the caloric values for protein, lipid, and carbohydrate of Brett and Groves (1979, see methods section), we suggest that each rotifer has an average energetic value of 0.000526 calories. Elemental composition of the rotifers showed that the percent carbon was 42.02 %AFBM and the per- cent nitrogen was 10.41 %AFBM. The carbon-nitro- gen ratio was 3.56:1. Proximate and elemental composition of larvae Proximate composition can be expressed in three ways: as a percent of wet mass (%WM), a percent of ash-free dry mass (%AFBM), and as the total con- tent per larva (mg/individual). Table 1 shows the changes in proximate composition (%WM and %AFBM) as a function of ration level and age of the larvae; total content (mg/individual) is reported be- low in the text. Red drum eggs exhibited large water content (94.65 %WM), a high protein content (42.17 %AFBM) and an intermediate lipid content (19.35 %AFBM)(Table 1 ). Carbohydrate, generally an extremely small frac- tion of the overall proximate composition of marine species, proved to be so in this case as well (0.47 %AFBM). Viewed as a fraction of the total body mass of each larva, the protein level (%AFBM) shows an increase through time at zero ration (43.84% to 62.47%) ac- companied by a reduction in lipid ( 19.72% to 13.29%), which indicates that, in starving larvae, lipid was used in preference to protein for energy production. On a pg/individual basis, protein actually decreased in starved larvae from 8.44 pg/individual in newly hatched larvae to 4.07 pg/individual for 6-day-old starved larvae (Table 2). Lipid values declined from 2.59 pg/individual on day 3 to 0.87 pg/individual on day 6. Percent water remained high until death at day 6, averaging 91.0% throughout the survival period. Table 2 Protein content, protein growth, RNA:DNA, and LDH in red drum larvae. Protein content was calculated from values reported in Table 1. Where values for % protein or % ash-free dry mass were missing, nearest neighbor values were used. Protein growth was calculated as described in text. RNA:DNA values are the mean for all RNA:DNA measurements within the last 2 d of the age interval shown in the table. They are reported as mean ±SD (n). Values for the 25°C pond were taken at day 7 instead of day 6 and calculated accordingly. All (6-14) day RNAiDNA values are significantly different (P<0.05, Students t ). nd = no data Day 2 6 10 14 2-6 6-10 10-14 2-14 6-14 6 10 14 6-14 6 10 14 Protein (ug/indiv) Protein growth (%/d) RNAiDNA LDH (units/gWM) Starved 8.44 4.07 nd nd -18.20 nd nd nd nd 0.68+0.35 (14) nd nd nd 7.48 nd nd 5/mL at 2°C 7.66 7.55 9.29 16.34 -0.40 5.20 14.10 63.00 9.50 1.5110.43 (6) 1.4510.24 (4) 1.4910.14 (4) 1.5010.30 (14) 9.19 14.44 19.69 5/mL at 25°C 9.36 9.77 14.80 47.33 1.10 10.40 29.10 13.50 19.70 1.1510.20 (2) 1.1910.32 (3) 1.2610.19 (7) 1.2510.21 (12) 15.06 24.33 33.60 Pond 25°C 8.16 15.00 67.70 484.30 12.10 50.20 49.20 34.00 49.60 2.8910.51 (4) 3.0610.81 (4) 3.5610.27 (2) 3.0910.62 (10) 18.94 30.86 42.78 Pond 32°C 8.51 82.20 324.60 1,116.20 56.70 34.30 30.90 46.00 32.60 1.19 2.21 1.39 1.6010.54 12.69 19.44 26.19 438 Fishery Bulletin 95(3), 1997 The counterpoint to O-ration data is provided by the data at 5.0 prey/mL at 20°C and 25°C (Table 1). The data clearly demonstrate accumulation of en- ergy as protein and little storage of lipids. Larvae raised at 20°C increased in protein level (%AFBM) from 43.74% as eggs to 51.73% as day- 14 larvae. Lipid levels decreased (%AFDM) from 24.34% in eggs to 9.17% in day-14 larvae. Protein concentrations in- creased from 7.66 pg/individual at day 2 to 16.34 pg/ individual at day 14. Lipid concentrations increased from 1.53 pg/individual at day 6 to 2.89 pg/individual at day 14. An identical pattern was observed in lar- vae reared at 25°C (Table 1). The data set for proximate composition values col- lected on pond-raised larvae was smaller than ideal owing to problems in obtaining adequate sample sizes from the ponds. However, the data on accumulated protein and lipid concentrations give an excellent indi- cation of maximum growth. Pond-raised larvae at 25°C and 32°C showed faster accumulation of total protein and lipid than larvae raised in the laboratory (Table 1). Protein levels of larvae increased in %AFDM from 42.70% and 45.62%, in eggs, to 56.22% and 64.16% in day- 14 larvae, at 25°C and 32°C, respectively. Lipid lev- els (%AFDM) decreased from 24.98% to 9.87% at 32°C; decreases in lipid percentages were also observed in the 25°C pond and in the laboratory -raised larvae. Pond-reared larvae at 25°C increased in total pro- tein content from 8.16 pg/individual at day 2 to 484.30 pg/individual at day 14, whereas those kept at 32°C increased from 82.20 pg/individual at day 7 to 1,116.20 pg at day 14 (Table 2). Thus, an increase in 10°C resulted in a three-fold increase in protein (pg/individual) in 2-week-old larvae raised in the ponds. Lipid values for larvae raised at 25°C and 32°C were also much higher than those for larvae raised in the laboratory; day- 14 pond larvae on average had lipid contents of 85.06 pg/individual and 171.44 pg/ individual, respectively. Carbon (%AFDM) remained about the same with age in all rearing conditions (Table 1), whereas ni- trogen (%AFDM) remained fairly constant or in- creased with age at all ration levels. Carbon-nitro- gen (C:N) ratios were higher in larvae kept at a ra- tion level of 0 prey/mL than in larvae raised either at 5.0 prey/mL or in the ponds, indicating that pro- tein commanded the largest fraction of the starved larva’s mass. Values for C:N remained high in starved larvae until death at day 5 (4.35 ± 0.46). Pond-raised larvae had values similar to those observed in lar- vae reared on 5.0 prey/mL in the laboratory but were slightly lower at day 14 (3.56 vs. 3.73). Caloric con- tents of the larvae were calculated from protein and lipid content by using the conversion factors in Brett and Groves (1979) and are reported in Table 1. Protein-specific growth and biochemical indicators Table 2 summarizes results for growth in protein (absolute and instantaneous) and the two biochemi- cal indicators, RNA:BNA and LDH activity, for easy comparison. Instantaneous growth shows an inter- esting trend with the age interval chosen for calcu- lation. If growth in protein was calculated from day 2 to day 14, larvae exhibit the trends discussed previ- ously for growth in dry mass (see results) where low- est growth was observed in the 20°C laboratory treat- ment and highest in the 32°C ponds. If instantaneous growth was calculated instead for the interval from day 6 to day 14, the highest growth was observed in the 25°C ponds (Table 2) and would suggest that the growth spurt during the first 4 d of feeding in the 32°C pond was important in determining growth during the larvae’s first 14 d of life. RNA-DNA ratio RNA:DNA in each treatment showed a decline from a high value typical of the yolk-sac stage (day 1: grand mean for all treatments 4.27 ± 0.83; x ± SD) to a plateau at day 4 that char- acterized the treatment and showed no significant change over the remaining 10 days (Fig 3; Table 2). In starved larvae RNA:DNA reached a plateau at a value of 0.7, indicating that protein synthetic capac- ity was severely diminished after that time. RNAiBNA in larvae raised at 5.0 prey/mL in the labo- ratory showed a gradual decline to a plateau of 1.5 at 20°C and 1.3 at 25°C (Fig 3; Table 2); values at the plateau were significantly different between the two temperatures (ANOVA: df=25, F=6.31, P=0.019) Pond-raised larvae had higher growth rates than laboratory-reared individuals (Figs. 1 and 2; Tables 1 and 2), and values for RNA:BNA were much greater in the 25°C pond than in any of the laboratory treat- ments (Fig. 3). Larvae raised in the ponds at 25°C had a value of 3.6 at 2 weeks of age, whereas those reared at 32°C averaged 1.5 at day 14. RNA:BNA values were significantly different between the two pond treatments (ANOVA: df=12, F=14.19, P=0.003). Three treatments took place at a temperature of 25° C: starved, 5 prey/mL, and pond. If protein growth rates and RNA:BNA are compared between labora- tory and ponds at 25°C (Table 2), there is an excel- lent correlation between protein growth and RNA:BNA. Instantaneous protein growth rate in the laboratory was 13.5%/d from day 2 to day 14; in the pond it was 34%/d over the same interval: an increase of 2.5 fold. RNA:BNA showed an increase of 1.3 to 3.0 in laboratory- versus pond-reared larvae over the same interval with a similar 2.3-fold increase. The negative growth observed in starved larvae at 25°C Brightman et a I.: Energetics of larval Sciaenops ocellatus 439 Age (days) Figure 3 RNA:DNA of red drum larvae raised in the laboratory and in growout ponds: (A) labora- tory, 25°C and 0 prey/mL; (B) laboratory, 20°C and 5 prey/mL; (C) laboratory, 25°C and 5 prey/mL; (D) pond, 25°C; and (E) pond, 32°C. (-18%/d) also showed a much lower value for RNA- DNA ratio : 0.7:1. Differences in RNA:DNA between the three treatments were highly significant (ANOVA: df=35, F= 107.6, P=0.QQ0). Overall, RNA:DNA was only a modest predictor of protein growth. Regression analysis of protein growth ( y , % per d) versus RNA:DNA (x), by using all the values in Table 2, showed a marginal fit (y = -3.64 + 14.76x; P=0.01; r2=0.34). However, as discussed above, within a temperature, RNA:DNA was an ex- cellent predictor of instantaneous protein growth, and this was borne out in a regression using only the data collected at 25°C: y (% per d) = -12.43 + 17.39 (RNA:DNA); P=0.01; r2=0.64. The performance of RNA:BNA as an overall predictor was improved significantly by using a multiple regression equation with a temperature term (Table 3). LDH Activity LDH activities of laboratory-raised larvae increased with age at ration levels of 0 and 5.0 prey/mL and with temperatures of 20°C and 25°C (Fig 4). Larvae that were starved continued to pro- 440 Fishery Bulletin 95(3), 1997 Age (days) Figure 4 LDH activity (units/gWM) in red drum larvae raised in the laboratory and in growout ponds: (A) laboratory, 25°C and 0 prey/mL; (B) laboratory, 20°C and 5 prey/mL; (C) labora- tory, 25°C and 5 prey/mL; (D) pond, 25°C; and (E) pond, 32°C. duce LDH, although at lower concentrations than those for fed individuals, until death at day 6. Lar- vae reared at 20°C had LDH values of 20-25 units/ gWM at day 14. These LDH activities were slightly lower than those for larvae raised at 25°C, which had LDH values of between 30 and 35 units/gWM at day 14. Larvae reared at 25°C in the ponds averaged LDH activities of 40-50 units/gWM, higher than the val- ues for larvae raised at 32°C, which averaged 25-30 units/gWM, and higher than the values for larvae reared in the laboratory. Like RNA:DNA, LDH activity taken overall was only a modest predictor of protein growth rate: y (% per d) = -8.17 + 1.39 (LDH); P= 0.02; r2=0.39. How- ever, within a temperature, its performance as a pre- dictor was much improved. A regression using only the 25°C data yielded an excellent coefficient of de- termination: y (% per d) = -28.74 + 1.94 (LDH); P=0.003; r2=0.86. A multiple regression with a tem- perature term improved its use as an overall predic- tor (Table 3). Brightman et al.: Energetics of larval Sciaenops ocellatus 441 Table 3 Multiple regressions describing instantaneous protein-specific growth (Y; %/d) in red drum larvae versus temperature (T), ration (0/ml, 5/mL, and pond), RNA:DNA, and LDH activity (units/gWM). Only significant regressions are presented (P< 0.05). Equation % Y n r2 P 1 T Y= 2.7 IX,- 49.52 20 0.21 0.023 2 Ration Y = 6.35Xi - 8.53 20 0.45 0.001 3 T Ration Y=2.07Xi + 5.67X,,- 58.34 20 0.57 0.022 4 RNA:DNA Y = 14.76X - 3.64 17 0.30 0.013 5 T RNA:DNA Y = 2.59X; + 14.61X,, - 69.41 17 0.56 0.007 6 LDH Y = 1.39^-8.17 13 0.34 0.022 7 T LDH Y=2.55X; + 1.26X„ - 70.35 13 0.57 0.034 Discussion Growth versus prey density Standard length and mass measurements The basic pattern of growth and development in red drum larvae, e.g. in size at flexion, was similar for larvae under a wide variety of rearing conditions. Within the basic blueprint, growth and development of red drum larvae fed to satiation could be accelerated or retarded according to the rearing temperature. Thus, larvae raised in the laboratory and the ponds underwent metamorphosis at roughly the same size, independent of the age of the larvae. In the case of the 32°C pond, day-7 larvae were already the size of day- 14 larvae reared at 25°C in the laboratory, and were at the same stage of development. Similarly, dry mass at transformation was approximately the same in the laboratory and ponds, despite the differ- ences in chronological age. Proximate and elemental composition of larvae Red drum larvae, whether fed to satiation or starved, depleted their lipid level from 40% to 50% by day 6. The increase in protein (%AFBM) reflected the de- cline in lipid and was most evident in the starved red drum larvae. Larvae that have been starved con- serve protein as musculature until the time of death. Conservation of muscular proteins allows the ani- mal to swim as long as possible before complete muscle atrophy, or “point of no return,” allowing the larvae to search out prey in other, possibly more pro- ductive, areas. The loss of dry mass in starving larvae, compared to fed larvae of equal age, reflected the catabolism of lipid and protein (Wallace, 1986). A similar, but less severe, drop in lipid was observed in all rearing con- ditions and has been observed in other species of fish. For example, Fraser et al. (1987) found that larval Atlantic herring had a lipid level of 23% dry mass (176 pg) one day after hatching decreasing to 11% (221 pg) by day 16. Those percentages were similar to those found for red drum larvae in the present study (20.18% to 11.74%) over the first two weeks of life. It is likely that lipid serves as a buffer fuel dur- ing the early life history of red drum. It is not accu- mulated. When high-quality food energy is available in excess, larval red drum larvae grow faster rather than accumulate an energy reserve. This is best ex- emplified by the differences in larvae growing at 25°C in the laboratory and 25°C in the ponds. Elemental composition agreed well with other pub- lished values for red drum (Lee et al., 1988) and lar- val herring of similar size (Ehrlich, 1974, a and b; 1975) as well as with our own results on proximate composition (Table 1). Larvae that are growing nor- mally, as in the 5.0 prey/mL experiments and the ponds, show greater increases in protein than in lipid. The increase in %N with age, and the declining %C, mirrored the changes (protein increase, lipid de- crease) in proximate composition. This changing el- emental composition resulted in a declining C:N in normally growing larvae. Starving individuals had slightly higher C:N than fed individuals as a result of their diminished protein synthesis. Larvae raised in the ponds have the lowest C:N as a result of the high protein levels relative to lipid. Thus, the C:N can be used as an indicator of physiological status in developing fish. It should be noted, however, that this ratio applies in the opposite fashion to adult fish. A declining C:N in older fish indicates starvation where lipid is laid down as an energy reserve and is com- busted before protein. The rapidly accumulating musculature of a healthy, growing fish larva results in a declining C:N, giving the appearance of starva- tion when, instead, this ratio indicates that protein is accumulating at a faster rate than lipid. 442 Fishery Bulletin 95(3), 1997 Protein-specific growth and biochemical indicators RMIA-DNA ratio Our values for RNA:DNA fall at the low end of the range of ratios reported in the litera- ture for larvae reared under a variety of different conditions (Ferron and Legget, 1994). Wright and Martin ( 1985) found similar RNA-DNA ratios ( 1 to 2 at 19-2 1°C) for starved striped bass, whereas fed striped bass larvae had ratios of 3-3.4 during the first two weeks after hatching. Robinson and Ware (1988) observed a similar trend in RNA-DNA ratios with starvation in the early life of larval Pacific her- rings, as we did with red drum; ratios declined up to yolk-sac absorption, where the ratios leveled off. Val- ues for RNA:DNA obtained in the laboratory in this study (1 to 2) were lower than previously reported values (2 to 4) for red drum larvae (Westerman and Holt, 1994). As has been reported previously (Buckley, 1982; Ferron and Legget, 1994), the relation of growth rate and RNA:DNA changed with temperature. The higher mass-specific and protein-specific growth rates observed in the laboratory at 25°C, in compari- son with those at 20°C and in the ponds at 32°C, as well as in comparsion with those at 25°C, were ac- companied by lower RNA:DNA values (Tables 1 and 2). The inverse relation between RNA:DNA and tem- perature holds true in field-caught larvae as well. It was observed by Setzler-Hamilton et al. ( 1987), who found that in late spring, values for RNA-DNA ra- tios in striped bass larvae were higher than values measured in hotter, early summer months (spring values were about 3 and summer values were 2 to 2.5). A high growth rate accompanied by a low RNA- DNA ratio, such as we observed in the 32°C ponds, is probably due to an increase in the efficiency of ribo- somes in initiating protein synthesis and to an in- crease in the rate of chain elongation due to a direct effect of temperature, i.e., an increase in the produc- tion of protein per unit of ribosomal RNA due to a Q10 effect (cf. Westerman and Holt, 1988). Despite the effect of temperature on the relation of RNA:DNA and growth rate, RNA:DNAis a useful tool for deter- mining nutritional status of fish larvae, particularly if it is understood that temperature contributes sub- stantially to the relationship between RNA:DNA and growth (Buckley, 1982: Buckley et al., 1984; Ferron and Legget, 1994). LDH Activity LDH, the terminal enzyme in verte- brate anaerobic glycolysis, is an important factor in the ability of some fish to produce sudden bursts of swimming and is found in large quantities in white muscle (Somero and Childress, 1980). The observed increase in LDH activity with age until death of starved larvae seems at first glance to conflict with priorities expected of an energy-deprived individual, in which metabolic processes would be expected to be declining. However, it is to be expected that LDH activity would be conserved, even in starving larvae, so that the muscle would remain functional as long as possible. A larva with no capability for movement would be doomed; thus, a metabolic investment in locomotory capability makes good adaptive sense. Unlike in RNA:DNA, LDH activity showed a di- rect correlation with both mass- and protein-specific growth rate in the two fed laboratory treatments despite the increase in temperature from 20°C to 25°C. In the ponds, LDH activities showed an inter- action with temperature similar to that seen in RNA:DNA, i.e. a lower specific activity at 32°C de- spite a higher growth rate. In the case of LDH, the declining activities observed in larvae from the higher temperature pond probably indicate that a lower con- centration of enzyme is sufficient to maintain the catalytic efficiency needed by the tissues at the higher temperature (cf. Hochachka and Somero, 1984). The fact that a similar drop was not noted in the labora- tory suggests a threshold for the drop in activity be- tween 25°C and 32°C that was not present in the transition between 20°C and 25°C. Clarke et al. (1992) found similar values for LDH in red drum larvae raised on wild zooplankton. Val- ues for LDH activity in Clarke’s study, assuming 87% water content, averaged 19-26 units/gWM for two- week-old larvae, slightly lower than the values we observed in the larvae raised in the laboratory and ponds. Biochemical parameters as predictive tools Although similar in their use as biochemical proxies for growth, LDH activity and RNA:DNA are funda- mentally different in many other respects. RNA:DNA is a ratio of measured quantities, whereas LDH ac- tivity is a determination of a rate: a kinetic measure- ment. Inherent in the measurement of RNA:DNA is the assumption that the methods for determining the quantities of RNA and DNA are accurate, but there is no direct effect of temperature on the assay itself. For LDH, activities are measured in saturating con- ditions of substrate, which means that the activities are maximal activities (Vmax from Michaelis-Menten kinetics; Lehninger, 1982) for each treatment. It is tacitly assumed that if assays are performed in satu- rating conditions at the same temperature, the dif- ferences in activity, or Vmax , are due to differences in concentration of the enzyme. This assumption is Brightman et ai: Energetics of larval Sciaenops ocellatus 443 a reasonable one. It is important, however, to be aware of other potential causes of variability in the relation of both RNA:DNA and LDH activity to growth or condition in fish larvae. It has been dem- onstrated here and elsewhere (Ferron and Legget, 1994) that rearing temperature alters the relation of growth and biochemical proxies for growth. An- other potential source of variability in the relation is the scaling of each of the proxies with individual size. RNA:DNA increases slightly with individual size (Buckley, 1982) but overall is insensitive to the changes in individual mass that would be expected in a study of larval fish growth within a single field sample. This is not the case for LDH activity which scales strongly with mass in fishes (e.g. Somero and Childress, 1980; Torres and Somero, 1988). In this study, a significant relation was observed between LDH activity (y, units/gWM) and protein mass (x, pg protein): y = 2.25x 0187 ; P=0.02; r2=0.43. RNA:DNA showed no significant change with size. Our study suggests that, for maximum accuracy, direct compari- sons of field-caught larvae for LDH activity are best confined to narrow size ranges or the relation be- tween LDH and size is described empirically. On the other hand, it could be argued that since mass-spe- cific LDH activity increases with increasing mass, it is actually incorporating a growth-specific change within its scaling behavior, making it a better proxy. Either way, it shows considerable potential. Acknowledgments The authors would like to thank Bill Falls, Anne Burke, and Dan Roberts of the Florida Marine Re- search Institute for providing red drum larvae and for considerable help in teaching us culturing tech- niques. This research was supported by DNR con- tract C-7701 and NSF OCE 92-18505 to J.J. Torres, and NSF OCE 92-17523 to M.E. Clarke. Literature cited Bentle, L. A., S. Datta, and J. Metcoff. 1981. The sequential enzymatic determination of RNA and DNA. Anal. Biochem. 166:5-16. Brett, J. R., and T. D. D. Groves. 1979. Physiological energetics. In W. S. Hoar, D. J. Randall, and J. R. Brett (eds.), Fish physiology, vol. VIII: bioenergetics and growth, p. 279-351. Academic Press, New York, NY. Buckley, L. J. 1980. Changes in ribonucleic acid, deoxyribonucleic acid, and protein content during ontogenesis in winter floun- der, Psuedopleuronectes americanus, and the effect of starvation. Fish. Bull. 77:703-709 1982. Effects of temperature on growth and biochemical composition of larval winter flounder Psuedopleuronectes americanus. Mar. Ecol. Prog. Ser. 8:181-186. Buckley, L. J., S. I. Turner, T. A. Halavik, A. S. Smigielski, S. M. Drew, and G. C. Laurence. 1984. Effects of temperature and food availability on growth, survival, and RNA-DNA ratio of larval sand lance (Ammodytes americanus). Mar. Ecol. Prog. Ser. 15:91-97. Clarke, M. E., C. Calvi, M. Domeier, M. Edmonds, and P. J. Walsh. 1992. Effects of nutrition and temperature on metabolic enzyme activities in larval and juvenile red drum, Sciaenops ocellatus, and lane snapper, Lutjanus synagris. Mar. Biol. (Berl.) 112:31-36. Donnelly, J., J. J. Torres, T. L. Hopkins, and T. M. Lancraft. 1990. Proximate composition of Antarctic mesopelagic fishes. Mar. Biol. (Berl.) 106:13-23. Ehrlich, K. F. 1974a. Chemical changes during growth and starvation of herring larvae. In J. H. S. Blaxter (ed.), The early life his- tory of fish, p. 301-323. Springer- Verlag, New York, NY. 1974b. Chemical changes during growth and starvation of larval Pleuronectes plateesa. Mar. Biol. (Berl.) 24:39-48. 1975. A preliminary study of the chemical composition of sea-caught larval herring and plaice. Comp. Biochem. Physiol. 51B:25-28. Ferron, A., and W. C. Leggett. 1994. An appraisal of condition measures for marine fish larvae. Adv. Mar. Biol. 30:217-303. Fraser, A. J., J. R. Sargent, J. C. Gamble, and P. MacLachlan. 1987. Lipid class and fatty acid composition as indicators of the nutritional condition of larval Atlantic herring. Am. Fish. Soc. Symp. 2:129-143. Hochachka, P. W., and G. N. Somero. 1984. Biochemical adaptation. Princeton Univ. Press, Princeton, NJ, 537 p. Hoff, F. H„ and T. W. Snell. 1987. Plankton culture manual. Florida Aqua Farms, Dade City, Florida, 98 p. Holt, G. J. 1990. Growth and development of red drum eggs and larvae. In G. W. Chamberlain, R. J. Miget and M. G. Haby (eds.), Red drum aquaculture, p. 46-50. Texas A&M Univ, Galveston, TX. Holt, G. J., and C. R. Arnold. 1983. Effects of ammonia and nitrite on growth and devel- opment of red drum eggs and larvae. Trans. Am. Fish. Soc. 112:314-318. Holt, G. J., R. Godbout, and C. R. Arnold. 1981a. Effects of temperature and salinity on egg hatch- ing and larval survival of red drum, Sciaenops ocellatus. Fish. Bull. 79:569-573. Holt, J., A. G. Johnson, C. R. Arnold, W. A. Fable Jr., and T. D. Williams. 1981b. Description of eggs and larvae of laboratory reared red drum, Sciaenops ocellata. Copeia 4:751-756. Houde, E. D., and R. C. Schekter. 1983. Oxygen uptake and comparative energetics among eggs and larvae of three subtropical marine fishes. Mar. Biol. (Berl.) 72:283-293. Lee, W. Y., S. A. Macho, X. H. Mao, and C. R. Arnold. 1988. Dynamics of elemental and biochemical components in the early life stages of red drum ( Sciaenops ocellatus). Contrib. Mar. Sci. Suppl. 30:194. 444 Fishery Bulletin 95(3), 1 997 Lehninger, A. L. 1982. Priciples of biochemistry. Worth Pubis., NewYork, NY, 1011 p. Robinson, S. M. C., and D. M. Ware. 1988. Ontogenetic development of growth rates in larval Pacific herring, Clupea harengus pallasi, measured with RNA-DNA ratios in the Strait of Georgia, British Columbia. Can. J. Fish. Aquat. Sci. 45:1422-1429. Setzler-Hamilton, E. M., D. A. Wright, F. D. Martin, C. V. Millsaps, and S. I. Whitlow. 1987. Analysis of nutritional condition and its use in pre- dicting striped bass recruitment: field studies. Am. Fish. Soc. Symp. 2:115-128. Somero, G. N., and J. J. Childress. 1980. A violation of the metabolism-size scaling paradigm: activities of glycolytic enzymes in muscle increase in larger size fish. Physiol. Zool. 53:322-337. Stickney, D. G., and J. J. Torres. 1989. Proximate composition and energy content of meso- pelagic fishes from the eastern Gulf of Mexico. Mar. Biol. (Berl.) 103:13-24. Swingle, W. E. 1990. Status of the commercial and recreational fishery. In G. W. Chamberlain, R. J. Miget, and M. G. Haby (eds.), Red drum aquaculture, p. 22-29. Texas A&M Univ., Galveston, TX. Torres, J. J., and G. N. Somero. 1988. Metabolism, enzymatic activities and cold adaptation in Antarctic mesopelagic fishes. Mar. Biol. (Berl.) 98:169- 180. Wallace, P. D. 1986. A note on the seasonal change in fat content of the autumn-spawning herring in the northern Irish Sea. J. Mar. Biol. Assoc. (U.K.) 66:71-74. Westerman, M. E., and G. J. Holt. 1988. The RNA-DNA ratio: measurement of nucleic acids in larval Sciaenops ocellatus. Contrib. Mar. Sci. Suppl. 30:117-124. Wright, D. A., and F. D. Martin. 1985. The effect of starvation on RNA:DNA ratios and growth of larval striped bass, Morone saxatalis. J. Fish. Biol. 27:479-485. 445 Retrospective analysis of virtual population estimates for Atlantic menhaden stock assessment Steven X. Cadrin Massachusetts Division of Marine Fisheries 50 A Portside Drive Pocasset, Massachusetts 02559 Present address: Northeast Fisheries Science Center, National Marine Fisheries Service, NOAA Woods Hole, Massachusetts 02543-1097 Douglas S. Vaughan Beaufort Laboratory, National Marine Fisheries Service Beaufort, North Carolina 285 1 6 Abstract- Historical and retrospec- tive comparisons of Atlantic menhaden virtual population analyses ( VPA) from 1955 to 1995 revealed substantial in- consistency in estimates of man- agement variables in the last year of stock assessments. Estimates of man- agement variables from several histori- cal stock assessments were generally consistent throughout most of the time series. In the last two years, however, historical estimates have deviated from revised estimates. Relative performance of alternative ad hoc methods for esti- mating fully recruited fishing mortal- ity (F) in terminal years showed that all methods were imprecise, but con- ventional catch-curve estimates were unbiased and had the least retrospec- tive inconsistency. Retrospective differ- ences in terminal estimates of age-1 F by separable VPA ranged widely for eight alternative settings but were clearly minimized by using seven years of catch data. The general magnitude of retrospective difference was ±1.2 bil- lion recruits (46% relative difference), ±9,000 metric tons of spawning stock biomass (33% relative difference), and ±4.7 percent maximum spawning po- tential (106% relative difference). Ret- rospective differences in recruitment, spawning stock biomass, and spawn- ing potential were positively skewed but not biased, indicating that the fre- quency of positive and negative incon- sistencies are equal but that the posi- tive differences are much greater in magnitude. The skewed distribution of retrospective inconsistency should be considered for managing the Atlantic menhaden fishery. Manuscript accepted 26 February 1997 Fishery Bulletin 95:445-455 (1997). The Atlantic menhaden, Brevoortia tyrannus, is a planktivorous clupeid that schools in coastal waters off the east coast of the United States. At- lantic menhaden dominated total U.S. fishery landings from 1946 to 1962 ( Ahrenholz etal., 1987), yield- ing approximately 600,000 metric tons (t) per year 1953-62 (Henry, 1971), after which landings steadily declined to 162,000 t in 1969 owing to recruitment failure and subse- quent overfishing. Since then, catch has increased to an average of 330,000 t per year 1970-95 ( AMAC, 1992 L1 The Atlantic menhaden fish- ery is currently managed according to six fishery and population thresh- olds that indicate overfishing in re- lation to historic production (AMAC, 1992). Three of the minimum popu- lation thresholds (2 billion recruits, 17,000-t spawning stock biomass, and 3% maximum spawning poten- tial) are derived from virtual popula- tion analysis (VPA) (Megrey, 1989). Atlantic menhaden landings have been reported from processing plants since 1940 and have been sampled for length, weight, and age since 1955 according to a two-stage cluster sampling design in which fish were sampled weekly from each port where menhaden were pro- cessed (Nicholson, 1975; Chester, 1984; Smith et al., 1987). The fre- quency of fishery samples and the consolidated nature of the fish- ery provide an extremely reliable 41-year series of catch at age, ages 0-6+, for estimation of abundance and mortality through VPA (Table 1). Unfortunately, no independent indices of relative abundance are available to calibrate abundance estimates for the last year of catch: commercial catch per unit of effort is a biased index because commer- cial catchability is inversely related to abundance (Schaaf, 1975; Ahren- holz et ah, 1987; Vaughan and Smith, 1988; Atran and Loesch, 1995) and fishery-independent survey indices are not correlated with abundance (Ahrenholz et ah, 1989). In the ab- sence of reliable abundance indices, and therefore of a formal statistical estimator for year-class abundance in the last year of the VPA, ad hoc estimation rules have been used to approximate abundance. The error in estimates of abun- dance is progressively less in previ- ous years than in the last year of 1 1992-95 landings from Joseph Smith, Beaufort Laboratory, National Marine Fisheries Service, Beaufort, NC. Per- sonal commun. 446 Fishery Bulletin 95(3), 1 997 Table 1 Age-based stock assessments of Atlantic menhaden. Yt in- dicates the terminal year the VPA. Yt Source 1976 AMMB, 1981; Powers, 1983 1981 Vaughan et al., 1986; Ahrenholz et al., 1987 1984 AMMB, 1986; Vaughan and Smith, 1988 1988 Vaughan, 1990; Vaughan and Merriner, 1991 1990 AMAC, 1992 (p. 40-50); Vaughan, 1993 1992 AMAC, 1992 (p. 17-30) 1993 Vaughan, 19947 1994 Vaughan, 1995; 1995 Vaughan, 1996J 1 Vaughan, D. S. Trigger variables for Atlantic menhaden. Natl. Mar. Fish. Serv., NOAA, Unpubl. AMAC reports. the VPA, provided that catch at age and natural mor- tality (M) are well estimated and fishing mortality (F) is at least moderate (Jones, 1961; Tomlinson, 1970; Pope, 1972; Ulltang, 1977; Megrey, 1989). As stated in the Atlantic menhaden fishery management plan, “Trigger estimates for recent years from VPA are subject to large uncertainty, while estimates 2 to 3 years old are more reliable” (AMAC, 1992). Con- sistency in successive stock assessments can be evaluated by using “historical analysis,” which com- pares estimates from the most recent assessment with contemporary estimates from prior stock assess- ments (Sinclair et al., 1985), 2 but historical assess- ments of the menhaden stock were not conducted with a common estimation rule. Consistency of the current estimation rule can be evaluated by using “retrospective analysis,” which recreates a historical series of VPA’s with a single estimation rule (Sinclair et al, 1990). 2 * * * The first objective of the current investigation was to report the general magnitude and potential bias of retrospective differences for guidance on interpret- ing current estimates and for providing fishery man- agement advice. The second objective was to attempt alternative estimation rules to improve consistency of estimates. 2 Examples of historical and retrospective analyses, interpreta- tion, and discussion can also be found in the following two refererences: Int. Counc. Explor. Sea. 1991. Report of the working group on methods of fish stock assessments. ICES Council Meet- ing Assess., p. 25. Northeast Fisheries Science Center. 1994. Report of the 18th Northeast Regional Stock Assessment Workshop (18th SAW). NEFSC Ref. Doc. 94-22. Methods Historical comparisons Three Atlantic menhaden management variables derived from VPA (age-1 abundance [R], spawning stock biomass [SSB], and percent maximum spawn- ing potential [%MSP]) were compared among ten reported stock assessments (Table 1). The number of historical estimates of each variable differed be- cause some reports did not document all three popu- lation estimates. Consistency of successive stock assessments was measured by comparing historical estimates with revised estimates (from the 1995 VPA), which are more reliable. Inconsistency may result from histori- cal estimation error or inaccurate estimates for prior years in the current VPA (Sinclair et al., 1990). The population thresholds used to define overfishing are subject to some uncertainty because they are also VPA estimates; but they are converged estimates, which are much more certain than current estimates. In com- paring current VPA estimates with these overfishing thresholds for an annual assessment of stock status, converged and current estimates are assumed to be consistent. Estimates in the last year (Y?), and back- calculated years (Yt_v Yt_2, etc.) were compared with the time series of estimates derived in 1995. Differ- ences between historical estimates and revised esti- mates were calculated as follows: ^t,t+k = ^t,t+k ~ ^,1995’ where R( 1995 = the most recent estimate of recruit- ment in year t; Rt ,+k = recruitment in year t as estimated when t+k was the last year in the assessment; and k = is the retrospective lag between year t and the last year of the historical VPA. For example, f?1990 1993 is the 1993 estimate of 1990 recruitment, which has a three-year retrospec- tive lag (i.e. k=3). When k = 0, Rtt is an estimate of recruitment for the last year in an assessment and is referred to as a terminal estimate. Histor- ical differences in SSB and %MSP were similarly calculated: SSBt,t+k = SSBtit+k-SSBt,19 95 A%MSPt't+k = %MSPtyt+k - %MSPty 1995- Root mean square (RMS) difference was used as a measure of dispersion of historical estimates from Cadrin and Vaughan: Virtual population analysis estimates for Atlantic menhaden stock assessment 447 converged estimates for the additive properties of mean square difference. Sample sizes for historical differences were low, because of the limited number of historical stock assessments, but the following retrospective analyses have greater sample size for estimating RMS difference (rc>30). Retrospective comparisons Retrospective analysis was performed in two stages to investigate consistency of both elements of the estimation rule: 1) estimation of fully recruited F by ad hoc methods and 2) estimation of partial recruit- ment to the fishery at ages 0 and 1 by separable VPA (SVPA; Pope and Shepherd, 1982). Both ana- lytical stages assumed that menhaden were fully recruited to the fishery at age 2 and that M was 0.45 for all ages, over the entire time period. Fully recruited F was approximated by using three alternative ad hoc methods for the first element of the analysis. Conventional catch curves (Beverton and Holt, 1957; Ricker, 1975; Gulland, 1983) and modified catch curves (Chapman and Robson, 1960; Robson and Chapman, 1961) were used to estimate mortality of the age-5 cohort over the four terminal years of the catch record (i.e. ages 2-5). These two catch-curve methods assumed that F in the current year was similar to F experienced by that cohort over the previous three years. The third ad hoc method, log catch ratios (Ricker, 1975; Gulland, 1983), derived fully recruited F from the negative log ratio of age- s'1' abundance in the terminal year to age-2+ abun- dance in the previous year and assumed that F in the last year was similar to F in the previous year. All three ad hoc methods assume that menhaden are fully recruited and equally available to the fishery at age-2+, which was confirmed through inspection of back-calculated F from the 1995 VPA. The second element of the assessment, estimation of partial recruitment, was performed by using SVPA on a fixed number of years. For example, a retro- spective series of 5-year SVPA’s was produced with the following algorithm. Step 1 SVPA was run on an initial time series of catch-at-age data (e.g. 1955-60) with the appropriate estimate of fully recruited F in the terminal year. Step 2 Catch data in the starting year (e.g. 1955) were deleted, and catch data from a new terminal year (e.g. 1961) were ap- pended. Step 3 SVPA was rerun on the revised time series with the appropriate estimate of fully re- cruited F in the new terminal year. Steps 2 and 3 were repeated until 1995 was the ter- minal year. Significance of retrospective bias was tested with a conventional (-ratio test (H0: mean difference=0). Normality was tested by using the Shapiro and Wilk ( 1965) method. Results of (-ratio tests were confirmed by using non parametric chi-square and sign tests. Dis- persion of retrospective estimates from converged es- timates was compared among alternative assess- ment rules by using RMS of retrospective differences. Retrospective estimates of fully recruited F in terminal years were compared with back-calculated estimates of fully recruited F from the 1995 VPA, as the aver- age of ages 2-5 (weighted by abundance), to derive retrospective differences: Note that there is no k subscript, as there were in the formulae for historical comparisons, because all retrospective estimates of fully recruited F were for terminal years (i.e. &=0). The relative retrospective difference (A Ft JFt 1995) was also calculated to remove the magnitude of the estimate from estimates of gen- eral inconsistency. Back-calculated estimates of fully recruited F from the 1995 VPA were used in terminal years to com- pare retrospective inconsistency of SVPA settings without including retrospective inconsistency from ad hoc estimates of terminal F. The fixed number of years in each series of retrospective SVPA’s was varied from three to ten years by using the algorithm described for the five-year example above. Retrospective consis- tency was compared among the eight series of retro- spective SVPA’s according to RMS difference of age- 1 F estimates. Full recruitment of age- 2 and oldest age (6+) menhaden was confirmed through inspec- tion of back-calculated F at age from the 1995 VPA and was not adjusted for retrospective comparison. Final SVPA runs were performed with seven years of catch-at-age and catch-curve estimates of termi- nal F to emulate more realistic inconsistency and describe the general magnitude and direction of ret- rospective differences. Retrospective estimates of R were derived directly from SVPA terminal estimates of age-1 abundance. SSB was estimated from termi- nal SVPA estimates of age-3+ abundance and esti- mated weight at age of spawners. Percent MSP was calculated according to egg production per recruit (Vaughan, 1990; AMAC, 1992). Retrospective dif- ferences and relative differences of management vari- able estimates were log transformed [e.g. loge ( R+ con- stant)] to test bias, and geometric mean square was used to estimate mean square difference because dif- ferences had skewed distributions. 448 Fishery Bulletin 95(3), 1997 Results Historical comparisons Management variable estimates from past stock as- sessments were generally consistent throughout most of the time series, except for the last two years of each assessment, when some historical estimates deviated from revised estimates from the 1995 VPA (Fig. 1). Terminal estimates of age-1 abundance were greater than revised estimates for five assess- ments and less than revised estimates for three assess- ments, but positive historical differences (i.e. histori- cal estimate > revised estimate) were greater. For ex- Ol < E o cn 93 « 4 1960 1965 1970 1975 1980 1985 1990 1995 5 2 S i Year Retrospective lag (yr) Figure 1 Comparison of historical estimates of Atlantic menhaden recruitment, spawning stock biomass, and percent maximum spawning potential. In the left charts, terminal years of historical stock assessments are labeled at the end of each series, overfishing thresholds are indicated by broken horizontal lines, and values in parentheses are not plotted. Conver- gence of estimates is illustrated by reduction in root mean square of historical differences over time in the charts on the right. Cadrin and Vaughan: Virtual population analysis estimates for Atlantic menhaden stock assessment 449 T3 a> CC 1 1 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I II I I I I ! 1955 1960 1965 1970 1975 1980 1985 1990 1995 Year Figure 2 Retrospective catch-curve estimates of Atlantic menhaden fishing mortal- ity and back-calculated estimates from the 1995 VPA (above) and retro- spective differences (below). The broken line in the upper chart indicates provisional estimates. ample, in 1992, recruitment was esti- mated to be 3.4 billion greater than the 1995 estimate of 1992 recruitment. Es- timates converged to within 160 million recruits of 1995 VPA estimates, when the retrospective lag ( k ) was greater than one year. Only five terminal estimates of SSB and %MSP were available, including 1995 estimates. Two historical esti- mates of SSB were greater than revised estimates and two were less than revised estimates. Historical estimates of SSB converged to within 1,500 t of 1995 VPA estimates when k>l year. Historical estimates of %MSP were greater than revised estimates for three assessments and less than revised esti- mates for two assessments. The 1992 es- timate of 1991 %MSP (i.e. backcalculated one year) was 3.6, but subsequent esti- mates of 1991 %MSP were below the over- fishing threshold of 3 %MSP. Therefore, an overfishing trigger fired, but it was not detected until two years later. Estimates of %MSP converged to within 0.6 of 1995 VPA estimates when k>l year. Retrospective comparisons Conventional catch curves produced the most consistent estimates of fully re- cruited F among the three ad hoc meth- ods used. Retrospective differences from catch-curve estimates in terminal years ranged from —1.12 to 0.92 (Fig. 2). The RMS difference was 0.51 and the RMS relative difference was 33% (n= 36). The mean retrospective difference in fully recruited F was not significantly differ- ent from zero. Therefore, although ter- minal estimates were imprecise, they were not biased. Retrospective differences in catch- curve estimates were negatively correlated with back-calculated F from the 1995 VPA (r=-0.62)(i.e. when F was low, catch-curve estimates were gener- ally greater than revised estimates; when F was high, catch-curve F was generally less than the revised estimate). Retrospective differences in fully recruited F produced opposite inconsistencies in SSB and %MSP For example, in 1993, catch-curve F was greater than the revised F (Fig. 2), and initial VPA estimates of SSB and %MSP were less than revised estimates (Fig. 1). Alternative methods of estimat- ing fully recruited F produced even greater incon- sistency. The RMS retrospective difference from modified catch curves was 0.60 (49% relative differ- ence), and log catch ratios produced a RMS differ- ence of 0.61 (52% relative difference) (n= 36 for both methods). Retrospective differences in estimates of age-1 F from SVPA ranged from -0.59 to 0.45 for all retro- spective SVPA’s and were not significantly different from zero (n=31 for each series). RMS difference was minimized when seven years of catch-at-age data were used and increased regularly as the number of years deviated from seven (Fig. 3). The RMS retrospective difference for estimates of age-1 F was 450 Fishery Bulletin 95(3), 1997 0.20 -r Number of years in SVPA Figure 3 Root-mean-square retrospective difference of fishing mortality esti- mates for age-1 Atlantic menhaden from SVPA’s with three to ten years of catch at age. 0.13 (33% relative difference) with back-calculated values of fully recruited F and increased to 0. 18 (45% relative difference) with catch-curve estimates (Fig. 4). Retrospective differences in age-1 abundance ranged from -2.4 billion to 11.5 billion individuals in terminal years (Fig. 5). There was no significant bias in log-transformed differences, and the RMS difference was 1.2 billion recruits (n=34). The RMS relative difference for estimates of age-1 abundance was 46%. Retrospective differences in terminal SSB esti- mates ranged from -72,000 t to 448,000 t (Fig. 6). The large positive differences in 1962 and 1963 SSB were primarily due to large negative differences in fully recruited F (Fig. 2). Log- transformed retrospec- tive differences were not significantly biased, and the RMS difference was 9,000 t SSB (n=34). The RMS relative difference for estimates of SSB was 33%. Retrospective differences in %MSP ranged from -5.5 to 19.5 in terminal years (Fig. 7). The RMS difference was 4.7 %MSP, and there was no significant bias in log-transformed differences (n=34). The RMS relative difference for estimates of %MSP was 106%, because inconsistencies were larger than the estimated level of %MSP. Retrospective dif- ferences in %MSP were negatively correlated with ret- rospective differences in fully recruited F (r=-0.79). Discussion This case study illustrates how retrospective com- parisons can provide useful data for analytical deci- sions and reveal important insights for management advice, especially in situations where statistical es- timates of uncertainty are not available. For example, SVPA with seven years of catch data clearly provided more consistent results than SVPA of longer or shorter time series (Fig. 3). A time period of seven years appears to be long enough to smooth annual variation in partial recruitment, while including only years which represent the current schedule of F at age. By including more years in the analysis, there is a likelihood that catches from substantially dif- ferent exploitation patterns will be incorporated. Performance of alternative intervals of catch data was judged according to general conditions over three decades. Although such guidance is valuable, specific SVPA runs should be examined to confirm the assumption of separability. For example, targeting spe- cific cohorts, such as the superabundant 1958 year class, may change fishing patterns. Abrupt changes should be reflected in patterns of log catch-ratio residuals (Pope and Shepherd, 1982) and may necessitate a longer or shorter time series of catch at age for SVPA. Retrospective inconsistency can result from a host of systematic problems, including errors in the cur- Cadrin and Vaughan: Virtual population analysis estimates for Atlantic menhaden stock assessment 451 rent VPA (Sinclair et al., 1990). For example, Lapointe et al. (1989) simulated Atlantic menhaden catch for VPA’s to show that 50% underestimates of M produced 12% overestimation of recruitment and 6% overestimation of biomass and that overestima- tion of M produced similar underestimates of recruit- ment and biomass. Therefore, the assumption that M is constant when M varies among years and ages may cause VPA inconsistency. It appears that the inability to estimate fully recruited F accurately in terminal years accounts for a substantial portion of the retrospective differences in management vari- ables reported here. Inaccurate estimates of terminal F may result from abrupt changes in F or M. Catch- curve estimates of terminal F are not sensitive to changes in current mortality because they reflect the average mortality over the last three years more than mortality in the terminal year. Retrospective differences in management variables were positively skewed, and log transformation was necessary to test for bias. Skewed retrospective dif- ferences with no bias imply that positive and nega- tive differences occur with equal frequency, but positive differences are generally greater in magni- tude. Theoretically, normally distributed errors in F will produce lognormally distributed errors in R, 452 Fishery Bulletin 95(3), 1997 SSB, and %MSP because they are based on estimated abundance, biomass, and egg production, respec- tively, which have a negative exponential relation with fishing mortality. Therefore, small underesti- mates in F can produce large overestimates of abun- dance. Retrospective differences may also be skewed because negative values of R, SSB, and %MSP are not possible. Despite the conclusion that log-trans- formed retrospective differences were not biased, the positive skewness of retrospective differences and relative differences has important implications for management of the fishery. A skewed distribution of inconsistency from converged estimates may be con- sidered a characteristic feature of terminal estimates from future menhaden VPA’s. Therefore, manage- ment advice should account for the equal likelihood of moderate underestimation and substantial over- estimation of R, SSB, and %MSP. Management decisions must consider the differ- ence between converged estimates, which are used to define overfishing thresholds, and terminal esti- mates, which are used to determine current status. Although inconsistency does not quantify uncertainty of estimates, the large retrospective differences re- Cadrin and Vaughan: Virtual population analysis estimates for Atlantic menhaden stock assessment 453 ported here suggest large uncertainty in terminal estimates. Fredrick and Peterman (1995) simulated uncertainty in Atlantic menhaden VPA’s to show that much more conservative overfishing thresholds would be needed to incorporate high levels of uncertainty into risk-averse management decisions. An alternative to incorporating uncertainty adjust- ments is to make management decisions based on converged estimates that indicate conditions of two years earlier (AMAC, 1992). If more timely manage- ment is desirable, methods to calibrate terminal estimates of abundance are necessary. Acknowledgments We thank the entire Menhaden Team of the Beau- fort Laboratory for the regular collection and analysis of data from the Atlantic menhaden fishery. We are grateful to the menhaden plant owners and opera- tors who facilitated catch sampling. We are indebted to the Atlantic Menhaden Advisory Committee of the Atlantic States Marine Fisheries Commission and two anonymous reviewers for critiquing the manuscript. 454 Fishery Bulletin 95(3), 1997 © D. CL CO © CC 1955 1960 1965 1970 1975 1980 1985 1990 1995 20-r 15 -- 10 -5 -- I ■ 1.1 llllill III Tip -iQ-H-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 I I I I I I I I I I I I I I I 1955 1960 1965 1970 1975 1980 1985 1990 1995 Year Figure 7 Estimates of percent maximum spawning potential of Atlantic menhaden derived from the 1995 VPA and terminal estimates from retrospective SVPA’s (above) and retrospective differences (below). The broken line in the upper chart indicates provisional estimates. Literature cited Ahrenholz, D. W., J. F. Guthrie, and C. W. Krouse. 1989. Results of abundance surveys of juvenile Atlantic and gulf menhaden, Brevoortia tyrannus and B. patronus. U.S. Dep. Commer., NOAATech. Rep. NMFS 84, 14 p. Ahrenholz, D. W., W. R. Nelson, and S. P. Epperly. 1987. Population and fishery characteristics of Atlantic menhaden, Brevoortia tyrannus. Fish. Bull. 85(3): 569-600. AMAC (Atlantic Menhaden Advisory Committee). 1992. Atlantic menhaden fishery management plan. Fishery Management Report 22, ASMFC Washington, D.C., 159 p. AMMB (Atlantic Menhaden Management Board). 1981. Fishery management plan for Atlantic menhaden Brevoortia tyrannus (Latrobe). Fishery Management Report 2, ASMFC Washington, D. C., 134 p. AMMB (Atlantic Menhaden Management Board). 1986. Supplement to Atlantic menhaden fishery manage- ment plan. ASMFC Washington, D.C. Fisheries Manage- ment Report No. 8, 134 p. Atran, S. M., and J. G. Loesch. 1995. An analysis of weekly fluctuations in catch- ability coefficients. Fish. Bull. 93(3):562-567. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish. Invest., ser. II, Mar. Fish. G.B. Minist. Agric. Fish. Food 19, 533 p. Cadrin and Vaughan: Virtual population analysis estimates for Atlantic menhaden stock assessment 455 Chapman, D. G., and D. S. Robson. 1960. The analysis of a catch curve. Biometrics 16:354- 368. Chester, A. J. 1984. Sampling statistics in the Atlantic menhaden fishery. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 9, 16 p. Fredrick, S. W., and R. M. Peterman. 1995. Choosing fisheries harvest policies: when does un- certainty matter? Can. J. Fish. Aquat. Sci. 52:291-306. Gulland, J. A. 1983. Fish stock assessment. FAOAViley series on food and agriculture; vol. 1. John Wiley and Sons, Chichester, UK, 223 p. Henry, K. A. 1971. Atlantic menhaden ( Brevoortia tyrannus ) resource and fishery — analysis of decline. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF 642, 32 p. Jones, R. 1961. The long term effects of changes in gear selectivity and fishing effort. Mar. Res. Dep. Agric. Fish. Scotl., ser. 2, 19 p. Lapointe, M. F., R. M. Peterman, and A. D. MacCall. 1989. Trends in fishing mortality rate along with errors in natural mortality rate can cause spurious time trends in fish stock abundances estimated by virtual population analysis (VPA). Can. J. Fish. Aquat. Sci. 46:2129-2139. Megrey, B. A. 1989. Review and comparison of age-structured stock as- sessment models from theoretical and applied points of view. Am. Fish. Soc. Symp. 6:8-48. Nicholson, W. R. 1975. Age and size composition of the Atlantic menhaden, Brevoortia tyrannus, purse seine catch, 1963-71, with a brief discussion of the fishery. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF 684, 28 p. Pope, J. G. 1972. An investigation of the accuracy of virtual popula- tion analysis using cohort analysis. ICNAF Res. Bull. 9:65-74. Pope, J. G., and J. G. Shepherd. 1982. A simple method for the consistent interpretation of catch- at-age data. J. Cons. Cons. Int. Explor. Mer 40:176-184. Powers, J. E. 1983. Report of the Southeast Fisheries Center Stock As- sessment Workshop. U.S. Dep. Commer., NOAA Tech. Memo. NMFS SEFC 127, 230 p. Ricker, W. E. 1975. Computation and interpretation of biological statistics of fish 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(21:181-189. Schaaf, W. E. 1975. Fish population models: potential and actual links to ecological models. In C. S. Russell (ed.l, Ecological modeling in a resource management framework, p. 211- 239. Resources for the Future, Washington, D.C. Shapiro, S. S., and M. B. Wilk. 1965. An analysis of variance test for normality (complete samples). Biometrika 52:591-611. Sinclair, A., D. Gascon, R. O’Boyle, D. Rivard, and S. Gavaris. 1990. Consistency of some northwest Atlantic groundfish stock assessments. NAFO SCR Doc. 90/96, 26 p. Sinclair, M., V. C. Anthony, T. D. lies, and R. N. O’Boyle. 1985. Stock assessment problems in Atlantic herring (Clupea harengus ) in the northwest Atlantic. Can. J. Fish. Aquat. Sci. 42:888-898. Smith, J. W., W. R. Nicholson, D. S. Vaughan, D. L. Dudley, and E. A. Hall. 1987. Atlantic menhaden, Brevoortia tyrannus, purse seine fishery, 1972-84, with a brief discussion of age and size composition of the landings. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 59, 23 p. Tomlinson, P. K. 1970. A generalization of the Murphy catch equation. J. Fish. Res. Board Can. 27:821-825. Ulltang, 0. 1977. Sources of errors in and limitations of virtual popu- lation analysis (cohort analysis). J. Cons. Cons. Int. Explor. Mer 37(31:249-260. Vaughan, D. S. 1990. Assessment of the status of the Atlantic menhaden stock with reference to internal waters processing. U.S. Dep. Commer., NOAA Tech. Memo. NMFS SEFC 262, 28 p. 1993. A comparison of event tree risk analysis to Ricker spawner-recruit simulation: an example with Atlantic menhaden. In S. J. Smith, J. J. Hunt, and D. Rivard (eds.), Risk evaluation and biological reference points for fisher- ies management, p. 231-241. Can. Spec. Publ. Fish. Aquat. Sci. 120. Vaughan, D. S., and J. V. Merriner. 1991. Assessment and management of Atlantic menhaden, Brevoortia tyrannus, and gulf menhaden, Brevoortia patronus, stocks. Mar. Fish. Rev. 53(4):49-57. Vaughan, D. S., J. V. Merriner, D. W. Ahrenholz, and R. B. Chapoton. 1986. Stock assessment of menhaden and coastal her- rings. U.S. Dep. Commer., NOAA Tech. Memo. NMFS SEFC 178, 34 p. Vaughan, D. S., J. W. Smith. 1988. Stock assessment of the Atlantic menhaden, Brevoortia tyrannus, fishery. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 63, 18 p. 456 Maturation and reproductive seasonality in bonefish, Albula vulpes, from the waters of the Florida Keys Roy E. Crabtree* Derke Snodgrass Christopher W. Harnden Florida Marine Research Institute, Department of Environmental Protection 100 Eighth Avenue SE, St. Petersburg, Florida 33701-5095 *E-mail address: crabtree_r@sellers.dep.state.fl.us Abstract .—We examined 528 bone- fish to estimate length and age at sexual maturity and to describe sea- sonal patterns in gonadal development. These fish ranged from 21 to 702 mm fork length (FL) and were collected in South Florida waters from 1989 to 1995. Gonads of 437 bonefish were ex- amined histologically, and gonado- somatic indices (GSI) were calculated for 449 bonefish. Male bonefish reached 50% sexual maturity (the predicted size and age at which half the individuals are expected to be sexually mature) at 418 mm FL (95% confidence interval 393-443 mm) and an age of 3.6 years (95% confidence interval 3. 3-3. 9 years). Females reached 50% sexual maturity at 488 mm FL (95% confidence inter- val 472-504 mm) and 4.2 years (95% confidence interval 3. 9-4. 6 years). Lengths and ages at 50% maturity for males and females were significantly different. The smallest sexually mature male was 425 mm FL, and the small- est sexually mature female was 358 mm FL. The youngest sexually mature male was 3 years old, and the youngest sexually mature female was 2 years old. Gonadal activity was seasonal and peaked during November-May. Vitel- logenic oocytes were present in ovaries in every month except August and Sep- tember and were most abundant dur- ing November-May. Median GSI’s were greatest during November-May and least during July-September for both males and females. No fully hydrated ovaries or postovulatory follicles were found, therefore we could not estimate spawning periodicity or batch fecundi- ties. Total fecundity ranged from 0.4 to 1.7 million oocytes and had a signifi- cant positive relation to fish weight. The absence of fully hydrated ovaries and postovulatory follicles in the bone- fish we sampled suggests that bonefish spawn outside the traditional shallow- water (<2 m) fishing grounds in the Florida Keys. Manuscript accepted 17 January 1997. Fishery Bulletin 95:456-465 (1997). Bonefish, Albula spp., frequent coastal and inshore waters of tropi- cal seas worldwide and are the ba- sis of economically important re- creational fisheries in many areas of their range. Although 23 nomi- nal Albula species have been de- scribed (Whitehead, 1986), only two Atlantic species, A. vulpes and A. nemoptera, are recognized (Rivas and Warlen, 1967). In the western Atlantic, A. vulpes is common in the Florida Keys, the Bahama Islands, and throughout the Caribbean Sea (Hildebrand, 1963 ). Albula nemop- tera appears to have a more re- stricted distribution than A. vulpes and has been reported from the Guianas, Venezuela, Columbia, Panama, Jamaica, and Hispaniola (Rivas and Warlen, 1967; Uyeno et al., 1983). The single record of A. nemoptera from Florida waters is considered questionable (Robins and Ray, 1986), and the species has not been reported from the Bahama Islands (Bohlke and Chaplin, 1993). Bonefish are esteemed for their wariness and fighting abilities, and fishing for them provides an impor- tant source of income to Florida Keys and Bahamian fishing guides. Commercial sale of bonefish in Florida is prohibited, and regula- tions on the recreational fishery re- strict catches to one fish per angler per day and the length of captured fish to a minimum total length of 457 mm (390 mm fork length). Most bonefish caught in Florida waters are released. The life history of bonefish has not been adequately described. Crabtree et al. (1996) described the age and growth of bonefish from South Florida and found that bone- fish can attain ages of 19 years. Female and male growth models were significantly different; females were slightly longer than males of the same age. Although the age and growth of Florida Keys bonefish have been studied, important ques- tions remain regarding bonefish reproduction. Bruger (1974) re- ported sexually mature females as small as 210 mm standard length (221 mm fork length) and as young as 1 year from waters off the Florida Keys, but his sample size was inad- equate to determine the age or length at 50% maturity (the pre- dicted size and age at which half the individuals are expected to be sexu- ally mature). Bruger found ripe fe- male bonefish throughout the year in waters off the Florida Keys and concluded that reproduction was not seasonal, but his sample size was small («=148) and his conclu- sions equivocal. In other areas, bone- fish reproduction appears to be sea- sonal according to patterns of lar- val and juvenile abundance (Alex- ander, 1961; Pfeiler, 1984; Pfeiler et al., 1988; Mojica et al., 1995). There Crabtree et al.: Reproduction of Albula vulpes 457 is also no published information on bonefish fe- cundity. In this article, we estimate the age and length at which sexual maturity is attained and describe the seasonal cycle of gonadal development in bonefish from waters off the Florida Keys. We also estimate the total fecundity of 33 bonefish collected from these waters. Methods Sampling We examined 528 bonefish collected from South Florida waters from February 1989 to April 1995. Most of these bonefish were caught with hook-and- line gear either by biologists or by a single profes- sional bonefish guide and his anglers from waters off the Florida Keys and in Florida and Biscayne Bays. Five bonefish caught with hook-and-line gear were obtained from taxidermists in Fort Lau- derdale and five others from tournaments in the Keys. Supplemental collections of small bonefish (<425 mm) were made with various-size seines and gill nets in waters off the Keys. Ages, based on vali- dated sectioned otoliths and growth rates of these bonefish, were described by Crabtree et al. (1996). Fork length (FL) was measured to the nearest millimeter (mm), and fish were weighed to the nearest gram. Sex, gonad condition, and gonad weight (g) were recorded. Gonad samples for his- tology and for estimation of fecundity were re- moved from the fish and preserved in 10% buff- ered formalin; they were later soaked in water for one hour and then stored in 70% ethanol. Histo- logical sections of gonads from 437 bonefish rang- ing from 228 to 702 mm were prepared and as- sessed for reproductive state. Gonad samples were processed histologically with a modification of the periodic acid Schiff’s (PAS) stain for glycol-meth- acrylate sections, with Weigert’s iron-hematoxy- lin as a nuclear stain and metanil yellow as a coun- terstain (Quintero-Hunter et al., 1991). Oocyte staging Oocytes were staged and counted from histological preparations at lOOx with a compound microscope attached to a digital image-processing system. Three oocyte stages were recognized in bonefish ovaries: primary growth, cortical alveolar, and vitellogenic (Wallace and Selman, 1981; Fig. 1A). In addition, we counted PAS-positive melanomacrophage centers (Ravaglia and Maggese, 1995), which were present in many ovaries (Fig. IB). When stained with the Figure 1 (A) A histological section from an ovary of a 677-mm-FL bone- fish, Albula vulpes, showing oocyte stages. PG = primary growth oocytes, CA = cortical alveolar oocytes, and V = vitellogenic oocytes. Scale bar = 400 microns. (B) A histologi- cal section showing PAS-positive melanomacrophage centers (PAS), cortical alveolar oocytes, and primary growth stage oocytes in a regressed ovary from a 692-mm-FL bonefish. When stained with periodic acid Schiff’s stain, melanomacrophage centers are brilliant purple. Scale bar = 100 microns. (C) A histological section showing PAS-positive melanomacrophage centers in a regressed testis from a 586-mm-FL bonefish. Scale bar = 50 microns. PAS stain, these structures are brilliant purple. Melanomacrophage centers are thought to be active in degrading atretic oocytes, postovulatory follicles, and residual cells of the spermatogenic cycle (Chan et al., 1967; Ravaglia and Maggese, 1995). At least 300 combined oocytes and melanomacrophage cen- ters per slide were staged and counted in arbitrarily chosen fields, and frequencies were expressed as a percentage of the total count. We counted all struc- tures that had at least 50% of their area visible in a field before moving to the next field. The presence of atretic hydrated oocytes was also noted. 458 Fishery Bulletin 95(3), 1 997 Length and age at sexual maturity Females were considered sexually mature if vitel- logenic oocytes were present or if the histological sections appeared disorganized, highly vascularized, and contained widespread evidence of atresia. Docu- mentation of atresia followed the classification of Hunter and Macewicz (1985). Immature females had small (gonadosomatic index <0.35), well-organized gonads that contained little evidence of atresia. We interpreted the widespread occurrence of PAS-posi- tive melanomacrophage centers in inactive ovaries as evidence of past gonadal development, and we considered bonefish that had regressed (no vitel- logenic oocytes present) ovaries containing many of these structures to be sexually mature (Fig. IB). Males were considered sexually mature if the testes contained evidence of ongoing spermatogenesis, re- sidual sperm, or widespread PAS-positive melanoma- crophage centers associated with gonadal recrudes- cence (Fig. 1C). Sometimes distinguishing between the gonads of sexually immature bonefish and the regressed go- nads of mature fish was difficult. We reduced the probability of misclassifying regressed and immature fish by eliminating all bonefish collected during June-October from our analyses of age and length at sexual maturity. June-October was the season of minimal gonad development in bonefish, and most of the regressed bonefish in our sample were cap- tured during this period. By excluding the postre- productive months from our analysis, we eliminated 84% of the regressed females and 63% of the re- gressed males in our sample. Hunter et al. (1992) recommended that only fish collected early in the spawning season be used to estimate the length at 50% maturity, but our sample size was not large enough to allow us to restrict our analysis to this extent. To describe age and length at sexual maturity, we used nonlinear regression procedures to determine the inflection point of a logistic function fitted to the percentage of males and females that were sexually mature and to their respective lengths and ages. Parameter b in Table 1 is the inflection point and is the estimate of length or age at 50% maturity. Like- lihood-ratio tests were used to compare the overall regression models and parameter estimates for males and females (Kimura, 1980). Seasonality of gonad development Monthly median gonadosomatic indices (GSI) of sexu- ally mature males and females were plotted to show seasonal reproductive patterns. GSI’s were calculated for 449 bonefish ranging from 228 to 702 mm as GSI = ( GWKTW - GW)) x 100, where GW = total gonad weight (g); and TW = total fish weight (g). We also plotted the monthly frequency of occurrence of the various oocytes stages that we counted for fish Table 1 Percentage mature-age, percentage mature fork length, and weight-fecundity regressions for bonefish, Albula vulpes, from the waters of the Florida Keys. Wt = weight (g), FL = fork length (mm), AGE = age in years, FEC = fecundity. Values in parentheses are standard errors. a b Range of X Y X n (1 SE) (1 SE) r2 for regressions y = (l/(l + e(- °'*-6))))xl00 % Mature FL 150 0.028 487.6 0.632 228-702 (Females) (0.0064) (8.14) % Mature AGE 143 1.122 4.24 0.445 1-19 (Females) (0.2345) (0.192) % Mature FL 116 0.545 417.5 0.735 322-687 (Males) (2.8227) (12.59) % Mature AGE 109 1.618 3.60 0.464 2-19 (Males) (0.3674) (0.156) Y = a + bX log10FEC log10Wt 33 1.936 1.131 0.706 1,790-5,790 (0.4708) (0.1312) Crabtree etal.: Reproduction of Albula vulpes 459 collected in the years during which we had regular monthly collections. Fecundity The total fecundity (the standing stock of advanced yolked oocytes) of 33 bonefish was estimated gravi- metrically. Ovaries were subsampled from anterior, middle, and posterior portions of each ovary to evalu- ate spatial variations in oocyte size within the ovary and between ovaries. Subsamples of ovary contain- ing 1,000-1,500 vitellogenic oocytes were weighed to the nearest 0.01 mg, and total fecundity was calcu- lated on the basis of the mean number of oocytes per gram of ovary. Ovaries that contained widespread atresia, which suggested that partial spawning might have occurred, were not used for fecundity estimation. Results Two of the bonefish that we examined were statisti- cally significant outliers (Crabtree et al., 1996); both were exceptionally small for their estimated ages and the weights of their otoliths were exceptionally light. Crabtree et al. excluded both fish from growth mod- els, age-frequency distributions, and otolith weight- age regressions, and we also excluded them from our analyses. One was a 351-mm female that was 7 years old and the other was a 458-mm female that was 18 years old. Both fish were caught with hook-and-line gear on the ocean (Florida Straits) side of North Key Largo, and they were the smallest females examined whose ovaries contained vitellogenic oocytes. The 458-mm female had oocytes that were in the nuclear migratory stage, and these were the most advanced oocytes we found in any bonefish. Length and age at sexual maturity Male bonefish reached 50% sexual maturity at a length of 418 mm (95% confidence interval 393-443 mm) and an age of 3.6 years (95% confidence interval 3. 3-3. 9 years); females reached 50% sexual maturity at a length of 488 mm (95% confidence interval 472-504 mm) and an age of 4.2 years (95% confidence interval 3. 9-4. 6 years; Fig. 2; Table 1). The lengths at 50% maturity (X2=124.43, df=l, P<0.001) and the ages at 50% matu- rity (%2=5.59, df=l, P=0.018) for males and females were significantly different. In addition, the overall logistic equations for length at 50% maturity (%2=51.18, df=2, P<0.001) and for age at 50% maturity (%2=11.55, df=2, P=0.003) for males and females were significantly dif- ferent. The smallest sexually mature male was 425 mm long, and the smallest sexually mature female was 358 mm long. The youngest sexually mature male was 3 years old, and the youngest sexually mature female was 2 years old. All males longer than 477 mm and all females longer than 594 mm were sexually mature. All males older than 5 years and all females older than 7 years were sexually mature. Primary growth stage oocytes were present in all ovaries in which we counted oocytes (Fig. 3A). Corti- cal alveolar oocytes were present only in ovaries from fish longer than about 400 mm and older than 2 years and were common only in fish longer than about 475 mm and older than 4 years (Fig. 3B). Vitellogenic oocytes were found only in fish longer than 450 mm and were common only in fish longer than 550 mm and older than 5 years (Fig. 3C). PAS-positive melanomacrophage centers were common only in females longer than about 550 mm and older than 5 years (Fig. 3D): the same length and age as those for females that contained vitellogenic oocytes. Seasonality of gonad development Bonefish gonadal activity was seasonal. Vitellogenic oocytes were present in greatest numbers during No- vember-May, and their numbers declined during May-June (Fig. 4). There were no vitellogenic oocytes in any ovaries from females collected during August- September of any of the three summers during which we sampled. Cortical alveolar oocytes were present during all months but were least abundant during July-October. Primary growth stage oocytes were present in all females examined and made up at least 20% of the total number of oocytes present. PAS-posi- tive melanomacrophage centers were most abundant in the gonads of spent and regressed males and fe- males and were most abundant in ovaries at the end of the spawning season in June-August (Fig. 5). They were least abundant in ovaries immediately before the initiation of spawning in November, when recru- descence was complete and most ovaries were ripen- ing to spawn during winter-spring. We saw no evi- dence of recent or imminent spawning, such as postovulatory follicles or fully hydrated females. Only six ovaries contained atretic hydrated oocytes, and no single histological preparation contained more than a few hydrated oocytes. Seasonal GSI patterns suggest that bonefish spawned during a prolonged period from November to June (Fig. 6). Median GSI’s were greatest during November-May and were least during July-Septem- ber. The decrease in female GSFs during July-Sep- tember corresponded with the decrease in the num- ber of vitellogenic oocytes present in ovaries and with the increased abundance of spent and regressed fish during late summer. 460 Fishery Bulletin 95(3), 1997 0> 200 400 600 800 0 5 10 15 03 C Fork length (mm) Age (yr) Figure 2 Length-maturity and age-maturity relations for male and female bonefish, Albula vulpes. Lengths plotted are the midpoints of 10-mm size classes. The solid line represents the predicted relation from the logistic regressions presented in Table 1. M50 is the length or age predicted by the logistic regression at which 50% of the bonefish were sexually mature. Fecundity Bonefish total fecundity estimates ranged from 0.4 to 1.7 million oocytes and had a significant positive rela- tion to fish weight (Table 1; Fig. 7). Relative fecundity (the number of oocytes per gram fish weight) ranged from 159 to 385 oocytes/g (mean=259 oocytes/g, SD=47.1, n=33) for fish ranging in length from 485 to 702 mm. There was no significant relation between relative fecundity and fish length (rc=33, r2=0.014, P=0.514 ) or weight (n=33, r2=0.043, P=0.247), but there was a significant positive relation between rela- tive fecundity and age (n= 32, r2=0.248, P=0.003). Oocyte development among areas within the ovary was homogeneous. We used a two-factor analysis of variance to compare oocyte densities with side (right or left) and position (anterior, middle, or posterior section of the ovary) as the effects. The number of late vitellogenic oocytes per gram of wet ovary weight was not significantly different between left and right ovaries (ANOVA, df=l, P=0.648) or among sub- samples from anterior, middle, or posterior sections of the ovary (ANOVA, df=2, P=Q. 709). Furthermore, we found no significant interaction between side and position from which subsamples were taken (ANOVA, df=2, P=0.702). Weights of left and right ovaries from sexually mature females were not significantly dif- ferent (paired £-test, n=112, £=0.480, P=0.632), but right testes from sexually mature males were significantly larger than left testes (mean difference=9. 1 g, SD=18.62; paired £-test, n- 98, £=4.826, PcO.OOl). Discussion Length and age at sexual maturity The bonefish we examined reached sexual maturity at an older age and larger size than reported by Bruger ( 1974). He reported sexually mature females that were 1 year old and ranged from 221 to 352 mm FL (re- ported as 210 to 338 mm standard length). Bruger considered these small bonefish to be sexually ma- ture on the basis of the presence of vitellogenic oo- cytes, one of the criteria that we used. He did not report how many of these small sexually mature fe- Crabtree et al.: Reproduction of Albu la vulpes 461 males were captured or the typical length and age at sexual maturity. Some of the sexually mature females he collected were smaller than the 351-mm sexually mature female that we considered an outlier and excluded from our analysis. Furthermore, the lengths of Bruger’s fish were substantially shorter than our estimated length at 50% maturity for females (488 mm). We also did not find any 1-year-old bonefish that were sexually mature; the youngest sexually mature fish we examined was 2 years old. All of the small mature females reported by Bruger were caught in deeper water (9.1-12.2 m) than that sur- veyed for bonefish in our sample; most of our fish were caught in water less than 2 m deep. Both Bruger (1974) and Crabtree et al. (1996) considered the pos- sible existence of a cryptic bonefish species in wa- ters off the Florida Keys as a potential explanation for the presence of exceptionally small and sexually mature bonefish, but additional study is needed to resolve this question. Little is known regarding bonefish maturation in other areas. Pfeiler et al. (1988) reported 12 Albula sp., ranging in length from 205 to 264 mm (SL) from the Gulf of California, that had ripe or ripening go- nads; this finding suggests a smaller length at sexual maturity there than we found in the Keys. Th e Albula 462 Fishery Bulletin 95(3), 1997 species in the Gulf of California is distinct from the Caribbean A. uulpes (Pfeiler, 1996). Most of the bonefish in our sample that were caught with hook-and-line gear were 500-700 mm long (80%) and 3-10 years old (86%; Crabtree et al., 1996). If the length and age composition of our hook- 1992 1993 1994 Month Figure 4 Monthly mean percent frequency of occurrence and standard er- rors of oocyte stages in bonefish, Albula vulpes, ovaries (n= 250) for 1992-94. * = primary growth stage oocytes, 0 = cortical alveo- lar oocytes, and □ = vitellogenic oocytes. and-line sample reflects that of the fishery, as sug- gested by Crabtree et al. (1996), then most of the fish caught in the fishery are longer than our esti- mate of length at 50% maturity and older than our estimate of age at 50% maturity. The current 390- mm-FL (457-mm-TL) legal minimum fish length imposed upon the fishery is less than our esti- mate of the length at 50% maturity for males (418 mm FL) and females (488 mm FL). Crabtree et al. (1996) found little evidence of fishing mortality in the Florida Keys because most bonefish caught by recreational anglers are released; thus, the population would prob- ably be insensitive to changes in the legal mini- mum fish length. The presence of PAS-positive melanoma- crophage centers was related to gonadal activ- ity. They appeared to be associated with the resorption of postovulatory follicles and atretic vitellogenic oocytes, as has been suggested by Ravaglia and Maggese (1995). The presence of vitellogenic oocytes is an unambiguous indica- tion of sexual maturity, and the length and age at which vitellogenic oocytes first developed in bonefish corresponded to the length and age at which PAS-positive melanomacrophage centers first appeared (Fig. 3). Melanomacrophage cen- ters were most abundant at the end of the spawning season, when spent fish were most common, and decreased in abundance during the postreproductive period (July-No- vember), when recrudescence occurred (Fig. 5). Ravaglia and Maggese (1995) found that the abundance of melanoma- crophage centers was also seasonal in Synbranchus marmoratus and peaked during the postreproductive season. Seasonality of gonad development Bonefish gonadal activity in the Florida Keys was seasonal, and development oc- curred over about eight months, from November to June. Bonefish were repro- ductively inactive for only a few months during summer. This period of inactivity roughly corresponds with the period of maximum water temperatures in the Keys (Crabtree et al., 1996). Bruger (1974) reported no evidence of a seasonal pattern of gonadal development for bone- fish off the Florida Keys, but his small sample size during July-September (n= 7 females, n= 4 males) may have obscured seasonal patterns. Alexander (1961) Month Figure 5 Monthly mean percent frequency of occurrence and standard errors of PAS- positive melanomacrophage centers in bonefish, Albula vulpes, ovaries. The numbers above the error bars are the numbers of ovaries examined. Crabtree et al.: Reproduction of Albu la vulpes 463 1992 1993 1994 Month Figure 6 Monthly median gonadosomatic indices (GSI) and interquartile ranges for sexu- ally mature male and female bonefish, Albula vulpes. Weight (g) Figure 7 The fecundity-weight (g) relation for bonefish, Albula vulpes. The fecun- dity-weight regression equation is presented in Table 1. found that the abundance of pre- metamorphic bonefish larvae in the West Indies was seasonal and that most larvae (96% of a total collec- tion of 417 premetamorphic larvae) were caught during November- April. She concluded, from the cap- ture of 17 premetamorphic larvae during August, that bonefish spawn throughout the year in the West Indies, but she suggested that bone- fish spawn at reduced levels during the hot summer months. The 17 lar- vae that she reported were captured in August and were caught far south of the Florida Keys — at 13°11'N lati- tude, where spawning may follow a different seasonal pattern than it does in the Keys. Studies from other areas also sug- gest seasonal reproduction in bone- fish. Recruitment studies off the Bahama Islands suggest that spawn- ing there has a seasonal pattern that is similar to that of spawning in the Florida Keys. Mojica et al. (1995) backcalculated hatching dates from the analysis of otolith micro- structure of field-collected larvae and found that bonefish spawned continu- ously from mid-October through early January. Recruitment patterns suggest that spawning probably continues until spring, because Mojica et al. (1995) ob- served recruitment pulses of bonefish larvae as late as June — presumably re- sulting from April and May spawning. Bonefish spawning in the Gulf of Cali- fornia also appears to be seasonal. Pfeiler et al. (1988) examined gonads from 33 bonefish ranging from 202 to 279 mm (SL) and suggested that Albula sp. in the Gulf of California spawn during late spring and early summer. Metamorphic leptocephali are abundant in coastal re- gions and hypersaline lagoons through- out the Gulf of California during winter and spring (Pfeiler, 1984; Pfeiler et al., 1988). Pfeiler et al. (1988) suggested that the premetamorphic larval phase lasts 6 or 7 months. The location of bonefish spawning grounds remains unknown, but the absence of females with post- ovulatory follicles or many hydrated oocytes in our samples suggests that bonefish do not spawn in the shallow nearshore areas where the fishery exists. It is possible that bonefish had spawned in our sam- pling area but did not feed before or after spawning and so were not available for capture with hook-and- 464 Fishery Bulletin 95(3), 1997 line gear until after they had completed the resorp- tion of recognizable postovulatory follicles. This pos- sibility seems unlikely because most collections of premetamorphic bonefish larvae are from offshore waters (Alexander, 1961); thus spawning bonefish probably move out of the shallow waters (<2 m) where fishing usually occurs. Alexander (1961) suggested that bonefish either spawn offshore or in areas where currents are likely to carry the eggs offshore. Fecundity We did not examine any bonefish ovaries containing oocytes in the final stages of oocyte maturation or showing definitive evidence of recent spawning, such as postovulatory follicles. Consequently, we do not know if bonefish are isochronal or multiple-batch spawners, and we could not estimate batch fecun- dity. If annual fecundity in bonefish is indeterminate (Hunter et al., 1992), our estimate of total fecundity may not accurately represent total annual egg pro- duction. The bonefish spawning season is prolonged, and the potential exists for additional vitellogenic oocytes to mature from the standing stock of unyolked oocytes during the spawning season. Some ovaries contained vitellogenic oocytes, widespread atresia, and were loosely organized and highly vascularized. These females may have spawned earlier in the sea- son and were developing an additional batch of oo- cytes that would have been spawned later in the sea- son. It is unclear whether these oocytes were re- cruited from the standing stock of unyolked oocytes after previous spawning or if they were vitellogenic oocytes already present in the ovary that did not ovu- late during previous spawning. Another bias of our fecundity estimates is that we could not correct them for atretic losses of vitellogenic oocytes during the spawning season; these losses could have caused us to overestimate egg production. Acknowledgments We thank Captain John Kipp, who provided us with most of the bonefish examined in this study and with- out whose efforts this work would not have been pos- sible. We also thank Captain Mike Collins, the Florida Keys Fishing Guides Association, and the Islamorada bonefish tournaments for their support. We also thank John Swanson, Bill Gibbs, and the staff at the Keys Marine Laboratory for their assistance; Jim Colvocoresses, John Hunt, and others at the South Florida Regional Laboratory for their cooperation; and David Harshany, Victor Neugebauer, and Connie Stevens for their assistance. Jim Colvocoresses, Harry Grier, Judy Leiby, Rich McBride, Jim Quinn, and Dana Winkelman made helpful comments that improved 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 Restoration Project Number F-59. Literature cited Alexander, E. C. 1961. A contribution to the life history, biology and geo- graphical distribution of the bonefish, Albula vulpes (Linnaeus). Dana-Rep. Carlsberg Found. 53, 51 p. Bohlke, J. E., and C. C. G. Chaplin. 1993. Fishes of the Bahamas and adjacent tropical waters, 2nd ed. Univ. Texas Press, Austin, TX, 771 p. Bruger, G. E. 1974. Age, growth, food habits, and reproduction of bone- fish, Albula vulpes , in South Florida waters. Fla. Mar. Res. Publ. 3, 20 p. 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. (Lond.) 153:527-539. Crabtree, R. E., C. W. Harnden, B. Snodgrass, and C. Stevens. 1996. Age, growth, and mortality of bonefish, Albula vulpes, from the waters of the Florida Keys. Fish. Bull. 94: 442-451. Hildebrand, S. F. 1963. Family Albulidae. In H. B. Bigelow (ed.), Fishes of the western North Atlantic, p. 132-147. Mem. Sears Found. Mar. Res. Yale Univ. 1(3). Hunter, J. R., and B. J. Macewicz. 1985. Measurement of spawning frequency in multiple spawning fishes. In R. Lasker, (ed.), An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax, p. 79-94. U.S. Dep. Commer., NOAATech. Rep. NMFS 36. Hunter, J. R., B. J. Macewicz, N. C. Lo, and C. A. Kimbrell. 1992. Fecundity, spawning, and maturity of female Dover sole Microstomus pacificus, with an evaluation of assump- tions and precision. Fish. Bull. 90:101-128. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Mojica, R., J. M. Shenker, C. W. Harnden, and B. E. Wagner. 1995. Recruitment of bonefish, Albula vulpes, around Lee Stocking Island, Bahamas. Fish. Bull. 93:666-674. Pfeiler, E. 1984. Inshore migration, seasonal distribution and sizes of larval bonefish, Albula, in the Gulf of California. Environ. Biol. Fishes 10:117-122. 1996. Allozyme differences in Caribbean and Gulf of California populations of bonefishes (Albula). Copeia 1996:181-183. Pfeiler, E., M. A. Mendoza, and F. A. Manrique. 1988. Premetamorphic bonefish ( Albula sp.) leptocephali from the Gulf of California with comments on life history. Environ. Biol. Fishes 21:241-249. Quintero-Hunter, I., H. Grier, and M. Muscato. 1991. Enhancement of histological detail using metanil yellow as counterstain in periodic acid Schiff’s hematoxy- Crabtree et a I.: Reproduction of Albula vulpes 465 lin staining of glycol methacrylate tissue sections. Bio- technol. 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, Synbranchidae): histological and histochemical characteri- zation. J. Fish Dis. 18:117—125. Rivas, L. R., and S. M. Warlen. 1967. Systematics and biology of the bonefish, Albula nemoptera (Fowler). Fish Bull. 66:251-258. Robins, C. R., and G. C. Ray. 1986. Afield guide to Atlantic coast fishes of North America. The Petersen field guide series, number 32. Houghton Mifflin Co., Boston, MA, 354 p. Uyeno, T., K. Matsuura, and E. Fujii. 1983. Fishes trawled off Suriname and French Guiana. Japan Marine Fishery Resource Research Center, 519 p. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am. Zool. 21: 325-343. Whitehead, P. J. P. 1986. The synonymy of Albula vulpes (Linnaeus, 1758) (Teleostei, Albulidae). Cybium 10:211-230. 466 Abstract.— A total of 1,237 tagged American lobsters, Homarus ameri- canus, with a carapace length (CL) range of 48 to 198 mm (mean CL of 104 mm ) were liberated at three release sta- tions off the eastern shore of Cape Cod, MA, between 1969 and 1971. By 1973, 332 (26.8%) of the tags were returned. Mean time at large was 112.5 days (range 0-897 d). One hundred and thirty (39.2%) of the recaptured lobsters moved less than 10 km from their points of release. One hundred and fifty-one (45.5%) were re- captured within 10 to 40 km from their points of release; 51 (15.4%) at 40 km or more. Recapture depths and distances trav- eled were significantly greater in colder months. The distribution of these re- captures with time, depth, and location indicates seasonal movement to and from the edge of the continental shelf between fall and spring. The apparent reshoaling of these in- shore-tagged lobsters to the eastern shore of Massachusetts in successive summers and the greater movement shown by females with ripe eggs at tag- ging, versus the movement of sublegal and nonovigerous female classes, sug- gest that the migration of this group of offshore lobsters is stimulated by sea- sonal changes in environmental cues in relation to hatching or reproductive needs (or both). Their relation to the Georges Bank-Southern Offshore stock unit, reproductive potential, and exten- sive seasonal movement into the south- ern and western Gulf of Maine, repre- sent important considerations for re- source managers and emphasize the need for further research on rate of stock interchange. Manuscript accepted 11 February 1997 Fishery Bulletin 95:466-476 (1997). Seasonal movement of offshore American lobster, Homarus americanus, tagged along the eastern shore of Cape Cod, Massachusetts Bruce T. Estrella Massachusetts Division of Marine Fisheries 50 A Portside Drive Pocasset, Massachusetts 02559 E-mail address: Bruce. Estrella@state.ma. us Thomas D. Morrissey 7 Horseshoe Circle Sandwich, Massachusetts 02563 The traditional American lobster, Homarus americanus, fishery con- sists of a small-boat fleet that fishes with traps within a few miles of shore in depths up to 20 fathoms. This inshore fishery is centered in the northern Gulf of Maine and pro- duces annually approximately 47% of the U.S. pounds landed. Massa- chusetts, the next largest producer, contributes about 28% of U.S. land- ings. Exploitation of inshore stocks is intensive. In the coastal waters of Maine, over 85% of the commercial inshore catch consists of new recruits (Krouse et al.1 ). In Massachusetts coastal waters, approximately 90% of the inshore catch falls into this cat- egory (Estrella and Armstrong2). Prior to 1948, small numbers of lobsters were caught incidentally by trawls in groundfish operations. These incidental catches accounted for less than 1% of U.S. landings. About 1945, trawlers began to fish specifically for lobsters, principally in deep water in the offshore region south of the Gulf between southeast Georges Bank and Hudson Canyon. This fishery was developed moder- ately and by 1968 offshore lobsters accounted for 16.9% of all U.S. lob- ster landings (Skud and Perkins, 1969). The introduction of deep-wa- ter trap fishing in the late 1960’s rapidly accelerated development of the offshore fishery. Vessels that fish with traps have largely re- placed the original trawl fleet, and the number of vessels in the off- shore lobster fishery have increased greatly. Substantial numbers of ves- sels in the 40' to 60' class, as well as larger vessels, were built or con- verted specifically for offshore trap fishing. Offshore landings averaged 24% of U.S. landings (3,400 metric tons [t]) between 1970 and 1974 but declined to a 1978-83 average of 17% or 2,500 t per year (NEFC, 1983). Despite short-term increases (5,000 t in 1990), offshore landings have not accounted for more than 18% of total U.S. lobster landings since the mid 1970’s (NEFSC, 1994). 1 Krouse, J. S., K. H. Kelly, G. E. Nutting, D. B. Parkhurst Jr., G. A. Robinson, and B. C. Scully. 1994. Maine Dep. Marine Resources lobster stock assessment project 3-IJ-61-2. Maine Dep. Mar. Resources, Ma- rine Resources Laboratory, P.O. Box 8, West Boothbay Harbor, ME 04575. Annual Rep., 53 p. 2 Estrella, B. T., and M. P. Armstrong. 1995. Massachusetts coastal commercial lobster trap sampling program, May-November, 1994. Mass. Div. Mar. Fish., 20 p. Estrella and Morrissey: Seasonal movement of Homarus americanus 467 The continued development of the offshore fishery has subjected American lobster in all remaining seg- ments of its range to intensive exploitation. Thus, stock identification and determination of any inter- relations between stocks of lobsters is of increasing importance to management of the lobster fishery (Pezzack3). Earlier studies indicated that lobsters were relatively nonmigratory. Numerous tagging experiments conducted primarily in more northern inshore areas showed that most lobsters usually re- main within a radius of 3-5 km (Templeman, 1935, 1940; Wilder and Murray, 1958; Wilder, 1963; Coo- per, 1970; Cooper et al., 1975; Krouse, 1980; Stasko, 1980; Campbell, 1982; Lawton et al., 1984). Accord- ingly, management practices were based largely on the concept of discrete local stocks. Findings of ex- tensive lobster movement in tagging experiments conducted in offshore areas (Saila and Flowers, 1968; Cooper and Uzmann, 1971, 1980; Uzmann et al., 1977; Fogarty et al., 1980; Campbell et al,. 1984) and in more southern inshore areas (Morrissey, 1971; Briggs and Muschacke, 1984) show significant move- ment of large, sexually mature lobsters which be- come intermixed with the inshore resource. Thus, the concept of discrete inshore stocks, characterized by a particular size range or maturity status (Camp- bell and Stasko, 1986; Campbell, 1989) becomes speculative. Intermingling of offshore lobsters with inshore stocks off southern New England is shown from re- captures of offshore-tagged lobsters in inshore areas. Cooper and Uzmann (1971, 1980) and Campbell (1986) hypothesized that seasonal depth-related movements are important to the biology of the spe- cies in providing optimal temperatures for mating, molting, egg extrusion, and egg development. In an effort to obtain information that would aug- ment offshore tagging studies off the Massachusetts coast, we undertook a three-year tagging experiment beginning in 1969 in the inshore waters of Cape Cod where previous work (Morrissey, 1971) showed the existence of a seasonal population of large, highly mobile lobsters. Additional lobster tagging in this area in 1984-85 also confirmed highly migratory behavior.4 Estrella and McKiernan (1989) described the ex- tensive size range of this segment of the lobster re- source, which is only seasonally available east of Cape Cod, as characteristic of an offshore migrant 3 Pezzack, D. S. 1987. Lobster ( Homarus americanus) stock- structure in the Gulf of Maine. Int. Counc. Explor. Sea. Shell- fish Comm. Council Meeting 1987/K:17, 18 p. 4 Estrella, B. T. 1997. American lobster tagging studies conducted in Massachusetts coastal waters. MA Div. Mar. Fish. In prep. group. This area exhibits the smallest percentage of sublegal-size lobsters in commercial trap catches of any other Massachusetts coastal region (10% com- pared with 89% in waters off Boston, MA, in 19955 ). Catch per trap haul of sublegal-size lobster off outer Cape Cod was four to eight times lower than that for other Massachusetts coastal regions sampled in 1994 (Estrella and Armstrong2). Outer Cape Cod habitat is not “classic” lobster habitat conducive to support- ing a resident (burrowed) resource; it is character- ized by expansive sandy bottom and is dynamic ow- ing to its exposure to strong easterly winds. Some local lobster production apparently occurs in the Nauset Marsh area of outer Cape Cod where limited numbers of early benthic-phase lobster have been found (Able et al., 1988). However, it is the larger, more common, offshore migrant lobster which sup- ports the commercial fishermen in this area and which shapes the style of fishing deployed there. Long strings of traps are set parallel to the shoreline to intercept incoming migrations each season. In late spring, traps are initially set by day-boat lobstermen approximately thirteen miles from shore. This gear is gradually fished shoalward as migrations proceed closer to land in warmer months, until declining au- tumn temperatures reverse the trend. It is informative that the intense outer Cape Cod lobster fishery has not been successful in reducing the size structure of this resource as definitively as in other inshore Massachusetts regions. A greater number of size-age groups are represented in outer Cape Cod catches. In most other inshore areas, lob- sters exhibit minimal migration and are exposed to fishing pressure throughout the year. New recruits (lobsters which, upon molting, become legal size) may represent as much as 95% of the legal catch, com- pared with only 55% in the outer Cape Cod area. The seasonal occurrence of these offshore lobsters in the outer Cape Cod area thus limits their exposure to intense fishing pressure. The size structure of this portion of the resource is similar to that from southern Georges Bank. Accord- ingly, the Sixteenth Northeast Regional Stock Assess- ment Workshop (16th SAW) of NMFS assigned this migratory group of lobsters to the Georges Bank- Southern Offshore stock unit (NEFSC, 1993). Estrella and McKiernan (1989) discussed the ge- ography as a potential factor in concentrating migrants. The outer Cape Cod area is adjacent to steeply sloping gradients which lead to a much greater depth range than that found in most other inshore regions. 5 Estrella, B. T. 1997. Massachusetts Division of Marine Fisher- ies, 50 A Portside Drive, Pocasset, MA 02559. Unpubl. data. 468 Fishery Bulletin 95(3), 1997 Materials and methods Tagged lobsters were released at three specific loca- tions along the eastern shore of Cape Cod (Fig. 1). Lobsters used in tagging were collected in the im- mediate vicinity of each release station and released within a day of capture. At station 1 (Provincetown), tagged lobsters were liberated in the periods 21-25 July 1969; 6-9 July 1970; and 23-25 June 1971. Lob- sters used in the tagging at station 1 were collected by SCUBA teams that attempted to capture all lob- sters observed on each dive. At station 2 (Truro) and station 3 (Eastham), tagged lobsters were liberated over the period 20 July to 18 August 1970. Lobsters used at these two stations were collected in the traps of a local commercial fisherman. Sphyrion anchor tags were used. These consisted of a coded polyvinyl-chloride-tubing pennant con- nected by a monofilament thread to a stainless steel anchor. The anchor was inserted in the lobsters with a hypodermic needle through the membrane connect- ing the carapace and first abdominal segment and implanted in dorsal extensor musculature below the carapace hypodermis as described in Cooper (1970). The implanted tag can endure successive molts. A reward of $1 was paid for each tag, as well as the market value of each tagged lobster returned with information on the date and location of recapture. Dur- ing 1971, the reward was increased to $5 for each tagged lobster submitted for examination, and the fisherman was permitted to retain possession of the lobster. Distance traveled by recaptured lobsters was de- termined as the shortest distance, avoiding land- mass, from point of release to point of recapture. Direction of travel was computed to the nearest 0. 1 degree true north. Results A total of 1,237 tagged lobsters with carapace length (CL) range of 48 to 198 mm (mean CL of 104 mm), Map of northeast coast of the United States and Canada showing eastern Cape Cod, Massachusetts, release stations for tagged American lobster, Homarus americanus. Estrella and Morrissey: Seasonal movement of Homarus americanus 469 were liberated at three release stations (Table 1) between 21 July 1969 and 25 June 1971. By 22 De- cember 1972, 332 or 26.8% of the tags were returned (Fig. 2). Mean distance from point of release to point of recapture was 22.6 km (median=140.4 km), mean time at large 112.5 days (median = 448.5 d). One hundred thirty (39.2%) of the recaptured lob- sters moved less than 10 km from their points of re- lease (Table 2). One hundred fifty-one (45.5%) were recaptured within 10 to 40 km, and 51 (15.4%) were recaptured at 40 km or more from point of release. Four lobsters moved farther than 100 km, one of the four as far as 281 km. The distances traveled by lobsters grouped in classes based on size, sex, and the presence or ab- sence of ripe and immature external eggs at tagging are shown in Table 3A. Legal-size females without eggs moved the shortest distance of all groups. Ripe ovigerous lobster moved farthest (average 30.3 km), followed by females with immature eggs (average 23.8 km). Analysis of variance of log-transformed data indicated there were significant differences among groups (P<0.01). Several multiple-range test procedures, Tukey-HSD test, Student-Newman- Keuls (SNK), and Duncan, were run on log-trans- formed data. A common result among the tests was Table 1 Tagged lobsters liberated and recaptured from three release stations at Cape Cod, Massachusetts. Release station Number tagged Carapace length (mm) Tags returned Mean distance traveled (km) Movement Mean time at large (days) Velocity (km/day) Number Percent recovery Mean Range SD 1 Provincetown Male 190 84 48-198 25.2 54 28.4 22.4 220.4 0.49 Female 683 113 49-189 21.2 200 29.3 21.4 83.6 0.65 2 Truro Male 39 75 67-80 3.8 8 20.5 25.3 228.9 0.37 Female 105 106 67-149 21.0 21 20.0 21.7 104.7 1.50 3 Eastham Male 75 76 69-90 3.8 14 18.7 22.6 198.1 0.49 Female 145 96 66-146 21.6 35 24.1 29.6 55.1 1.18 Total or weighted mean 1237 102 48-198 24.9 332 26.8 22.6 112.5 0.72 Table 2 Distance traveled by lobsters liberated from three release stations at Cape Cod, Massachusetts. Distance traveled (km) Provincetown Truro Eastham Total Number of tags returned Percent of total Number of tags returned Percent of total Number of tags returned Percent of total Number of tags returned Percent of total Less than 10 100 39.4 2 6.9 28 57.1 130 39.2 10-19 49 19.3 15 51.7 1 2.0 65 19.6 20-29 26 10.2 2 6.9 2 4.1 30 9.0 30-39 46 18.1 8 27.7 2 4.1 56 16.9 40-49 20 7.8 1 3.4 5 10.3 26 7.8 50 or more 13 5.2 1 3.4 11 22.4 25 7.5 Total 254 100.0 29 100.0 49 100.0 332 100.0 470 Fishery Bulletin 95(3), 1997 NANTUCKET SOUND - km 7WM that the distance traveled by legal-size nonovigerous females was significantly shorter than that of all other groups except sublegal females. Only fourteen of the returned lobsters molted while at large. These were distributed among most of the lobster classes. Sample size was insufficient to as- sess effects of molting on movement. The relatively long mean distances traveled by sublegal male and female lobster groups (22.5 km and 17.0 km, respectively) were likely due to their number of days at large being, on average, consider- Release Station 1 Release Station 2 Release Station 3 Provincetown Truro Easthain Recapture Sites troni Station 1 Recapture Sites from Station 2 Recapture Sites from Station 3 Statute Miles 34-67 40 37' 00' 40 15-70 Figure 2 Tagged American lobster, Homarus americanus , release stations and return locations off the coast of Massachusetts. Some inshore recapture sites represent multiple recaptures. ably greater than the number of days at large for other lobster groups (Table 3A). The mean time at large was less for legal-size females with and with- out external eggs than for legal-size males and sublegal males and females. Lobster “velocities” greater than 3 km/day were not exhibited by individual sublegal males or sublegal females. However, these rates were calculated for the larger lobsters, including 6.6% of legal-size males, 1.1% of legal-size nonovigerous females, 7.6% of fe- males with immature eggs, and 4.2% of females with ripe eggs. Females with imma- ture and ripe eggs exhibited greatest mean velocities (1.55 and 0.95 km/day, respectively, Table 3A). Because variability in days at large among classes of lob- ster could affect comparisons of distance traveled, standard- ization was warranted. An ad- ditional data analysis was con- ducted which was limited to lobsters at large < 200 days (Table 3B). This eliminated potential misleading recapture locations that could occur af- ter circuitous (homing) move- ment patterns, i.e. those from lobsters which, after tagging, may move offshore and return inshore in the following year, and subsequent years. Be- cause lobsters were tagged and released in the months of June through August, a 200-day limit was considered reason- able to avoid spring recaptures in our data treatments. Analysis of these “standard- ized” data reaffirmed that le- gal-size females with ripe ex- ternal eggs exhibited the great- est mean distance traveled, 25.6 km, followed by females with immature external eggs, 24.2 km. Legal-size males ranked third, with a mean of 16 km; nonovigerous females and sublegal females and males averaged 12.2 km, 13.1 km, and 14.2 km, respectively. Analysis of variance of log- transformed distance data in- dicated that there were signifi- Estrella and Morrissey: Seasonal movement of Homarus americanus 471 Table 3A Tag returns from various classes of lobsters liberated at Cape Cod, Massachusetts, for all days at large. Lobster class Number liberated Number recovered Percent tag recovery Mean carapace length (mm) Mean distance traveled (km) Mean time at large (days) Velocity (km/day) Male sublegal-size7 213 46 21.6 72 22.5 308.0 0.31 Female sublegal-size without external eggs 128 20 15.6 75 17.0 153.0 0.37 Male legal-size 91 30 33.0 106 23.1 79.0 0.74 Female legal-size without external eggs 295 91 30.8 106 13.4 77.0 0.46 Female with immature external eggs 130 26 20.0 112 23.8 34.2 1.55 Female with ripe external eggs 380 119 31.3 118 30.3 83.0 0.95 Total or weighted mean 1,237 332 26.8 104 22.6 112.5 0.72 1 Lobsters less than 81 mm carapace length. During this study, the minimum legal carapace length in Massachusetts was 3 and 3/16 inches (80.96 mm). Table 3B Tag returns from various classes of lobsters liberated at Cape Cod, Massachusetts, which were at large <200 days. Lobster class Number liberated Number recovered Percent tag recovery Mean carapace length (mm) Mean distance traveled (km) Mean time at large (days) Velocity (km/day) Male sublegal-size7 213 21 9.9 75 14.2 34.1 0.59 Female sublegal-size without external eggs 128 14 10.9 75 13.1 47.2 0.51 Male legal-size 91 26 28.6 104 16.0 38.1 0.82 Female legal-size without external eggs 295 81 27.5 106 12.2 35.0 0.51 Female with immature external eggs 130 25 19.2 112 24.2 22.8 1.61 Female with ripe external eggs 380 107 28.2 117 25.6 36.1 1.04 Total or weighted mean 1,237 274 22.2 107 19.1 35.2 0.85 1 Lobsters less than 81 mm carapace length. During this study the minimum legal carapace length in Massachusetts was 3 and 3/16 inches (80.96 mm). cant differences among groups (PcO.Ol). Tukey-HSD, SNK, and Duncan test results were similar to re- sults from tests on all data. They indicated that dis- tances traveled by sublegal and legal nonovigerous lobster groups were significantly less than those of egg-bearing female groups (P=0.05). The trend in mean velocity and proportion of lobsters traveling greater than 3 km/day was similar to that calculated for each lobster class from all data. Maximum ve- locities, calculated by lobster class, were 2.98 km/ day for a 79-mm-CL sublegal male, 2.36 km/day for sublegal females (80 mm CL), 5.19 km/day for legal size males (106 mm CL), 4.15 km/day for legal-size nonovigerous females ( 101 mm CL), 7 km/day for le- 472 Fishery Bulletin 95(3), 1997 gal-size females with immature eggs (103 mm CL), and 5.8 km/day for females with mature eggs (118 mm CL). Time at large for recaptured lobsters ranged from 0 to 897 days (Table 4). Approximately 78% of recap- tured lobsters were at large less than 60 days. The number of tags returned and the percentage recov- ery of tagged lobsters at large decreased sharply in the fourth month after release (October) coincident with the start of the fall season. Only three tags were returned from lobsters recaptured in the colder months from December through April ( all years com- bined). All three tagged lobsters were recovered in deep water. Two were recaptured on 7 and 22 March 1972, at depths of 54 m and 59 m, respectively, off Provincetown, MA, and the third was recaptured off- shore on 22 December 1972, at a depth of 95 m, NE of Veatch Canyon. Depth of recapture data were log-transformed and analyzed with equality of means test and found to be related to season (Welch: F=4.41, P=0.0411; Brown-Forsythe: F=4.41, P=0.0411). Depth of recap- ture for the combined months of June through Sep- tember (63 m) was significantly different from Octo- ber through May (93 m). Distance traveled was also significantly different by season (£=-3.50, P=0.001). Distance from release site was greatest during the October-May period of recapture (41.3 km), in comparison with June-Sep- tember period of recapture (22.5 km). Distance of travel northeastward of the Cape Cod landmass was apparently limited compared with dis- tance of travel in other directions; recapture points tend to be distributed in a northwest-southeast plane (Fig. 2). To test inshore versus offshore migrational tenden- cies, tag- re turn locations were grouped with consid- eration for the curvature of the “arm” of the Cape. Exclusive of lobster recaptured within Cape Cod Bay, recapture points east to south of all landmass be- tween 90° and 225° true north (for lobsters liberated at stations 2 and 3), and 1° and 225° (for lobsters Table 4 Tags returned by month of recapture and days at large for tagged lobsters liberated at Cape Cod, Massachusetts (recapture years combined). Recapture Mean days Number of Tags returned Percent of Cumulative Mean distance Mean depth month at large returns total percent (km) (meters) 1st season June 4.0 2 0.6 3.2 70.0 July 10.1 70 21.1 21.7 10.1 62.5 August 33.2 112 33.7 55.4 20.0 53.0 September 50.6 76 22.9 78.3 26.2 78.5 October 89.9 9 2.7 81.0 21.1 81.2 November 111.0 5 1.5 82.5 19.5 128.2 2nd season March 258.0 1 0.3 82.8 4.6 176.0 May 302.1 10 3.0 85.8 69.7 91.3 June 338.8 9 2.7 88.5 22.5 54.3 July 359.8 5 1.5 90.0 37.0 33.8 August 381.6 8 2.4 92.4 33.1 51.9 September 434.5 2 0.6 93.0 7.6 92.5 October 449.8 4 1.2 94.2 39.8 48.3 November 486.0 1 0.3 94.5 33.8 50.0 3rd season March 582.0 1 0.3 94.8 38.5 192.0 May 670.0 2 0.6 95.4 19.9 30.5 June 703.5 2 0.6 96.0 35.7 60.5 July 726.3 6 1.8 97.8 23.2 81.5 August 742.5 2 0.6 98.4 33.9 93.0 October 807.5 2 0.6 99.0 30.2 70.0 November 843.5 2 0.6 99.6 11.5 61.0 December 897.0 1 0.3 99.9 223.5 312.0 Estrella and Morrissey: Seasonal movement of Homarus americanus 473 liberated at station 1) were grouped as southeast (off- shore) in direction. All other recapture points, includ- ing those within Cape Cod Bay, were grouped as northwest (inshore) in direction. Inshore versus off- shore travel was tested with chi-square by month of recapture for lobsters at large up to one year and which traveled more than 3.7 km. Direction of travel was biased significantly toward inshore during warmer months (Table 5). Discussion Our findings of greater directed movement toward the north and west during summer months is in agreement with the summer movement along the eastern shore of Cape Cod and into Cape Cod Bay as reported by Morrissey (1971). The Cape Cod land- mass is interjacent to inshore grounds to the north and west and offshore grounds along the edge of the continental shelf to the south and east. The overall northwest-southeast distribution of recapture points suggests an interchange of inshore and offshore stocks in the Cape Cod area. Cooper and Uzmann (1971) and Uzmann et al. (1977) found that lobsters tagged in offshore canyon areas moved into shoal water in late spring and early summer, returning to deep water in late fall and early winter. Lobsters migrating from that offshore area to the inshore area of eastern Massachusetts would pass along the east- ern shore of Cape Cod. The northwest-southeast pat- tern of recapture locations is consistent with move- ment to and from the Georges Bank and southern offshore canyon area which exhibits a similar popu- lation structure to the lobster group which is only seasonally available east of Cape Cod. Lawton et al. (1984) found minimal movement in an inshore tagging study of 4,761 sublegal lobster at nearby Rocky Point, Plymouth, Massachusetts dur- ing 1970-75. Only 19 lobsters (<1% of returns) were retrieved 16 or more km from their release points and all 19 were within state territorial waters. A study by Fair6 on legal-size lobster in the same area several years later yielded similar results. Additional studies affirmed the nonmigratory nature of inshore lobsters (Templeman, 1935; Wilder, 1963; Cooper et al., 1975; Krouse, 1980; Stasko, 1980; Campbell, 1982). Ennis ( 1984) noted small-scale seasonal depth movements in relation to temperature with lobsters moving to shallow water in warmer months and deeper water in colder months. More extensive sea- sonal migrations were demonstrated by Campbell (1986) and Pezzack and Duggan (1986). We conclude that lobsters tagged in the present experiment are onshore migrants from an offshore stock that seasonably becomes “superimposed” on the endemic inshore stock. Recapture depths were sig- nificantly greater in colder months than during sum- mer. The movement of lobsters off southeastern Mas- sachusetts is cyclical, with lobsters moving to deep water in late fall. Uzmann et al. (1977) found that lobsters returned to the continental shelf margin and slope in fall and winter. In our study only three lob- sters were recaptured during December through April, consequently, we did not clearly establish if lobsters winter specifically on the edge of the shelf or in deep-water areas in general. However, four of our lobsters were recaptured in clearly offshore ar- eas: at 40°07', 70° 38’, 119 m depth, on 26 September 1970; at 40°34', 67°37’, 110 m depth, on 9 May 1970; at 41°35', 68°25', 55 m depth, on 23 May 1971; and at 40°15', 70°00’, 95 m depth, on 22 December 1972. The occurrences of these recaptures in time, depth, and location suggest seasonal move- ment to and from the edge of the continen- tal shelf between fall and spring. With an average recovery rate of only 7.0% of lob- sters tagged offshore by Cooper and Uzmann (1971), it is probable that sub- stantial numbers of our inshore tagged lob- sters wintered on the edge of the continen- tal shelf. Cooper and Uzmann (1971) concluded that offshore lobsters actively orient to optimum temperature according to season. Uzmann et al. ( 1977) provided further sup- 6 Fair, J. J., Jr. 1977. Lobster investigations in man- agement area 1: southern Gulf of Maine. NOAA, NMFS State-Federal Relationships Div., Mass. Lobster Rep. No. 8, 21 April 1975-20 Apr. 1977, 8 p. Appendix, 5 p. Table 5 Seasonal distribution of recapture points of lobsters at large up to one year and that had traveled more than 3.7 km. Direction of travel Recapture period Number inshore Number offshore Total X2 1st season of release July-August 106 43 149 PcO.0001 September-October 59 15 74 PcO.0001 1st winter of release November-May 11 5 16 P=0.134 2nd season of release June-August 8 4 12 P=0.248 Total 184 67 251 474 Fishery Bulletin 95(3), 1997 port with their finding that through random or di- rected movements (or both), the offshore lobster popu- lation maintains itself within temperatures of 8°- 14°C. Cooper and Uzmann (1971) hypothesized that seasonal shoalward migration to warmer water com- pensates for a lack of sufficiently high temperature during summer in the continental slope habitat to permit extrusion and hatching of eggs and subse- quent molting and mating. They found that offshore lobsters that demonstrated the most extensive on- shore migrations were predominately females and that the migration of offshore lobsters to inshore grounds is generally confined to areas south and west of Cape Cod (no recoveries were made north of Cape Cod in the Gulf of Maine proper). While diving to collect lobsters for tagging at sta- tion 1, we observed lobsters always to be concentrated in a narrow stratum where a thermocline intercepted the steeply sloping surfaces of a sandy escarpment paralleling the beach in that area, about 1.8 km from shore. Morrissey7 conducted semiweekly SCUBA surveys throughout the summer of 1966 at Province- town, where station 1 is located, and found lobsters only in close proximity to the thermocline-sediment interface, which occurred at 24 m in late May and ranged between 9 and 19 m during June, July, and August. During a vertical transect along the bottom from the shoreline to a depth of 22 m, 14 July 1966, 15 of 16 lobsters observed were within a stratum ( 11 to 14 m) in which a bathythermograph cast showed a change of 12.8°C in water temperature. On the basis of observed activity of individual lobsters, he con- cluded that the lobsters were not concentrated by a thermal block but rather were attracted to the warmer epilimnion layer and used the reduced light intensity associated with the thermocline as cover. These observations support the conclusion of Coo- per and Uzmann (1971) and Uzmann et al. (1977) that offshore lobsters orient to optimum temperature. Our test results indicated that sublegal and legal- size females with no eggs moved significantly less than egg-bearing female groups. An explanation for why legal-size females without eggs move less than those with eggs may be that they tend to congregate in the warmer shoal water where egg extrusion oc- curs. During this study, three tagged females, which extruded eggs after tagging, moved only an average distance of 4.2 km before recovery. Although this sample size is small, the inference from it is sup- ported by fishing activity in this area. Fishermen 7 Morrissey, T. D. 1970. Observations on behavior of the Ameri- can lobster, Homarus americanus, at Provincetown, Massachu- setts during the summer of 1966. MA Div. Mar. Fish., 50 A Portside Drive, Pocasset, MA 02559. report concentrations of females with immature eggs in the shoals east of Cape Cod during August and September. The fact that tagged inshore female lobsters with ripe external eggs moved greater distances than other classes of tagged lobsters may be a verification of the findings of Cooper and Uzmann (1971), that off- shore lobsters demonstrating the most extensive in- shore migrations are predominantly female. How- ever, unlike Cooper and Uzmann (1971), our recov- eries indicate that lobster migration occurs north of Cape Cod into at least the southwestern portion of the Gulf of Maine with one recovery made as far north as latitude 42°39'. Our subsequent tagging work in this study area (1984-85) yielded the return of a female after 362 days at large from even farther north, 43°33' (off Cape Elizabeth, Maine).4 None of our lobsters were recovered in the inshore grounds south and southwest of Cape Cod where most of the inshore recoveries were made by Cooper and Uzmann. The distribution of recapture points of the 58 lobsters recovered after their first season of release (Table 4) suggests that our inshore tagged lobsters returned to the shoal waters along eastern Massachusetts in successive summers. The movement described by the findings of Coo- per and Uzmann (1971) suggests that offshore lob- sters migrate to secure more suitable hydrographic conditions. The areas involved, i.e. the edge of the continental shelf and shoaler waters extending into inshore grounds south and west of Cape Cod, are quite generalized. The apparent return of our inshore tagged lobsters to the eastern shore of Massachu- setts in successive summers, and the greater move- ment shown by females with ripe eggs at tagging, suggest that the migration of offshore lobsters may be anastrophic or gametic (Heape, 1931; Wilkinson, 1952) in character, i.e. nonrandom, stimulated by metabolic needs or reproductive cues. Campbell (1986) provided calculations that suggest ovigerous lobster need to make seasonal deep-shallow water migrations to obtain sufficient heat units for egg de- velopment within any 9-12 month period. Using a threshold temperature of 3.4°C, he determined that shallow water had more degree days than deeper water in summer months and that the reverse was true in winter months. Although significant American lobster migrations have been reported, Saila and Flowers (1968) pro- vided the first reference in the literature to long-dis- tance homing by this species. They found a pro- nounced directional tendency toward the original area of capture in the movements of berried female lobsters transplanted from Veatch Canyon on the edge of the continental shelf to Narragansestt Bay, Estrella and Morrissey: Seasonal movement of Homarus americanus 475 Rhode Island. They concluded that the lobsters tended to remain in shoal waters in suitable spawn- ing habitat until they had shed their eggs or had molted, or both. Cooper and Uzmann ( 1971) referred to seasonal movement to and from generalized ar- eas: the edge of the continental shelf and the shoaler waters of southern New England. Campbell (1986) and Pezzack and Duggan ( 1986), however, provided evidence from Canadian waters that lobsters under- take regular migrations between widely spaced and well-defined areas. In contrast, the European lobster, Homarus gammarus, although biologically similar to H. americanus, displays minimal migratory behavior (Bannister and Addison, 1995). Tagging studies have shown that both juveniles and adults exhibit “strong site loyalty.” The distribution of the H. gammarus resource and fishery is primarily coastal; “offshore” distribution occurs only 20 km from shore. The lack of substantial long-distance movement may be due to the more moderate water temperature off the Brit- ish Isles (compared with the NW Atlantic) caused by proximity to the Gulf Stream. This may mitigate the biological “need” for extensive seasonal inshore-off- shore movement by H. gammarus ,8 There is an apparent affinity between the migra- tory group of lobsters east of Cape Cod, which are examined in this study, and those from Georges Bank and southern offshore canyons. Our evidence for movement of these lobsters north of Cape Cod into the Gulf of Maine seasonally, implies that genetic interchange between stock units continues, despite high exploitation rates. In light of this, management of fishing mortality rates on the offshore resource becomes an issue of increasing importance. There is also increasing information on movements of lobster larvae, the distribution and behavior of newly settled and juvenile lobsters, concentrations of egg-bearing females, and the occurrence of long- distance homing behavior in American lobster both in the northern Gulf of Maine and southern New England waters. Some progress has been made in roughly delineating stock structure with the help of larval dispersion, hydrodynamic, and migration stud- ies (NEFSC, 1993). Interpretation of these data, how- ever, is tentative because migratory habits of larger lobsters appear extensive and may transcend the boundaries that some researchers attempt to draw solely on the basis of larval distribution. Despite many years of larval and postlarval lobster monitor- Bannister, R. C. 1997. The Centre for Environment, Fisher- ies and Aquaculture Science, Lowestoft Laboratory, Pakefield Road, Lowestoft, Suffolk, England NR33 OHT. Personal commun. ing, a definitive stock-recruitment relation has yet to be determined, although ecological knowledge has been enhanced. The relative importance of the off- shore lobster resource to recruitment in shoaler wa- ters of the Gulf of Maine or other areas must be as- sessed. We need to know the degree of interchange between the two lobster groups in order to refine stock assessments. Acknowledgments We greatly appreciate the help of M. P. Armstrong and S. X. Cadrin with initial data collation, and the data mapping assistance received from T. B. Hoopes in cooperation with MassGIS. R. A. Cooper assisted with 1969-71 field activities, including lobster tag- ging. Tagging assistance during 1984-85 was pro- vided by M. Borgatti, D. McKiernan, M. Syslo, and J. O’Gorman. Literature cited Able, K. W., K. L. Heck Jr., M. P. Fahay, and C. T. Roman. 1988. Use of salt-marsh peat reefs by small juvenile lob- sters on Cape Cod, Massachusetts. Estuaries 11:83-86. Bannister, R. C. A., and J. T. Addison. 1995. Investigating space and time variation in catches of lobster (Homarus gammarus (L.)) in a local fishery on the east coast of England. ICES Mar. Sci. Symp. 199:334—348. Briggs, P. T., and F. M. Mushacke. 1984. The American lobster in western Long Island Sound: movement, growth and mortality. N.Y. Fish Game J. 31(11:21-37. Campbell, A. 1982. Movements of tagged lobsters released off Port Maitland, Nova Scotia, 1944-80. Can. Tech. Rep. Fish. Aquat. Sci. 1136, 41 p. 1986. Migratory movements of ovigerous lobsters, Homarus 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. Campbell, A., D. E. Graham, H. I. MacNichol, and A. M. Williamson. 1984. Movements of tagged lobsters released on the conti- nental shelf from Georges Bank to Baccaro Bank, 1971- 73. Can. Tech. Rep. Fish. Aquat. Sci. 1288, 16 p. Campbell, A., and A. B. Stasko. 1986. Movements of lobsters ( Homarus americanus) tagged in the Bay of Fundy. Mar. Biol. (Berl.) 92:393-404. Cooper, R. A. 1970. Retention of marks and their effects on growth, be- havior, and migrations of the American lobster, Homarus americanus. Trans. Am. Fish. Soc. 99:409—417. Cooper, R. A., R. A. Clifford, and C. D. Newell. 1975. Seasonal abundance of the American lobster, Homarus americanus, in the Boothbay region of Maine. Trans. Am. Fish. Soc. 104(41:669-674. 476 Fishery Bulletin 95(3), 1 99 7 Cooper, R. A., and J. R. Uzmann. 1971. Migration and growth of deep-sea lobsters, Homarus americanus. Science (Wash. D.C.) 171:288-290. 1980. Ecology of juvenile and adult Homarus. In Vol. 11: J. S. Cobb and B. F. Phillips (eds.), The biology and man- agement of lobsters, p. 97-142. Acad. Press, N.Y. Ennis, G. P. 1 984. Small-scale seasonal movements of the American lob- ster Homarus americanus. Trans. Am. Fish. Soc. 113:336- 338. Estrella, B. T., and D. J. McKiernan. 1989. Catch-per-unit-effort and biological parameters from the Massachusetts coastal lobster (Homarus americanus ) resource: description and trends. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 81, 21 p. 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. Heape, W. 1931. Emigration, migration, and nomadism. W. Heffer and Sons, Ltd., Cambridge, England, 369 p. Krouse, J. S. 1980. Summary of lobster, Homarus americanus , tagging studies in American waters (1898-1978). In Proceedings of the Canada-U.S. workshop on the status of assessment science for NW Atlantic lobster (Homarus americanus) stocks, St. Andrews, NB, Oct. 24-26, 1978, p. 135- 140. Can. Tech. Rep. Fish. Aquat. Sci. 932. Lawton, R. P., G. Luders, M. Kaplan, P. Brady, C. Sheehan, W. Sides, E. Kouloheras, M. Borgatti, and V. Malkoski. 1984. The commercial lobster pot-catch fishery in the Ply- mouth vicinity, western Cape Cod Bay. In Vol. 11: Lec- ture notes on coastal and estuarine studies: observations on the ecology and biology of western Cape Cod Bay, Mas- sachusetts, p. 131-150. Springer- Verlag, New York, N.Y. Morrissey, T. D. 1971. Movements of tagged American lobster, Homarus americanus , liberated off Cape Cod, Massachusetts. Trans. Am. Fish. Soc. 100(1):117-120. Northeast Fisheries Center. 1983. American Lobster. In Status of the fishery resources off the northeastern U.S. for 1983, NOAA Tech. Memoran- dum NMFS-F/NEC-29, 132 p. Northeast Fisheries Science Center. 1993. Report of the sixteenth northeast regional stock assess- ment workshop ( 16th SAW), Stock Assessment Review Com- mittee (SARC), consensus summary of assessments. Woods Hole, MA, NOAA/NMFS/NEFSC. NEFSC Ref. Doc. 93-18. 1994. American Lobster. In Status of the fishery resources off the northeastern U.S. for 1994, NOAA Tech. Memoran- dum NMFS-F/NEC-108, 140 p. Pezzack, D. S., and D. R. Duggan. 1986. Migration and homing tendencies of offshore lobsters (Homarus americanus) on the Scotian Shelf. Can. J. Fish. Aquat. Sci. 43:2206-2211. Saila, S. B., and J. M. Flowers. 1968. Movements and behavior of berried female lobsters displaced from offshore areas to Narragansett Bay, Rhode Island. J. Cons. Perm. Int. Explor. Mer 31(3):342-351. Skud, B. E., and H. C. Perkins. 1969. Size composition, sex ratio, and size at maturity of offshore northern lobsters. U.S. D.I. Bur. Comm. Fish. Spec. Sci. Rep. — Fisheries 598, 10 p. Stasko, A. B. 1980. Tagging and lobster movements in Canada. Can. Tech. Rep. Fish. Aquat. Sci. 932:141-150. Templeman, W. 1935. Lobster tagging in the Gulf of St. Lawrence. J. Biol. Board Can. l(4):269-278. Templeman, W. 1940. Lobster tagging on the west coast of Newfoundland, 1938. Dep. Nat. Res. Fish. Bull. No. 8, 16 p. Uzmann, J. R., R. A. Cooper, and K. J. Pecci. 1977. Migration and dispersion of tagged lobsters, Homarus americanus, on the southern New England continental shelf. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-705, 92 p. Wilder, D. G. 1963. Movements, growth and survival of marked and tagged lobsters liberated in Egmont Bay, Prince Edward Island. J. Fish. Res. Board Can. 20(2):305-318. Wilder, D. G., and R. C. Murray. 1958. Do lobsters move offshore and onshore in the fall and spring? Fish. Res. Board Can. Prog. Rep. Atl. Coast 69: 12-15. Wilkinson, D. H. 1952. The random element in bird navigation. J. Exp. Biol. 29:532-560. 477 Abstract .—We measured the daily abundance of larvae of eight species of ocean-spawned, estuarine-dependent fishes to determine the effect of sam- pling frequency on the mean and vari- ance estimates during larval immigra- tion past a permanent sampling station inside Beaufort Inlet, North Carolina, mid-November 1991 to mid-April 1992. Species of interest were Brevoortia tyrannus, Lagodon rhomboides, Leio- stomus xanthurus, Micropogonias undu- latus, Mugil cephalus, Paralichthys albigutta, P. dentatus, and P. lethostigma. Our data suggest that sampling at in- tervals >7 days can lead to excessive variance in abundance estimates. For all species, abundance varied as much as an order of magnitude from night to night. Proportional residuals from poly- nomial models of the seasonal recruit- ment pattern for a given species were used to assess the potential influence of nine environmental variables on daily densities. Twenty-seven of 72 cor- relations of proportional residuals with environmental variables were signifi- cant (P<0.05). Proportional residuals were positively correlated with time after dusk for six of eight species and were negatively correlated with turbid- ity for five of eight species. However, interpretation of correlations must be done cautiously because a species’ re- cruitment pattern may coincide with normal seasonal change in one or more environmental variables. Variability in transport of larvae, from offshore to near the inlet and then through the inlet to the station, probably influences species abundance at the sampling sta- tion more than locally acting environ- mental variables. Daily collections of B. tyrannus larvae provided otoliths (n=l,341) showing that a large number of younger larvae, averaging 55 days posthatch, arrived at the station in mid- March on the date of maximum observed daily density ( 160 larvae per 100 m3). Manuscript accepted 6 March 1997. Fishery Bulletin 95:477-493 (1997). Daily variability in abundance of larval fishes inside Beaufort Inlet Willi am F. Hettler Jr. * David S. Peters David R. CoSby Elisabeth H. Laban Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 1 0 1 Pivers Island Road Beaufort, North Carolina 28516-9722 *E-mail address: whettler@hatteras.bea.nmfs.gov From 1985 to the present, weekly sampling has been conducted near Beaufort Inlet to collect fish larvae entering the estuary during fall, winter, and spring (Warlen, 1994). Such inlets provide locations for sampling larvae in order to assess potential year-class strength of ocean-spawned but estuarine-de- pendent species. Abundance, size, and age data on early larvae in the sea, on advanced larvae in the in- lets, and on juveniles in the estuar- ies can be used to understand when significant events such as mortal- ity or rapid growth occur during a species’ early life history. To obtain accurate abundance, size, and age estimates from the population of recruiting larvae, appropriate sam- pling protocol must be employed (Morse, 1989; Davis et al., 1990). Errors resulting from sampling bias can arise when larvae selectively avoid the sampling device or when there is nonrandom spatial (patchy) distribution of the larvae (Wiebe and Holland, 1968). Decreasing the time interval between sampling and decreasing the distance between stations in ichthyoplankton surveys increases the resolution of tempo- ral or spatial patterns of species with patchy pelagic egg and larval distributions at both microscale (Houde and Lovdal, 1985) and mesos- cale levels (Rowe and Epifanio, 1994). Studies within NOAA’s Southeast Atlantic Bight Recruitment Experi- ment (SABRE) are attempting to measure fluxes of larval fishes across the continental shelf and through inlets into estuaries amid myriad cyclical phenomena that bear directly on the larvae’s abun- dance (Govoni and Pietrafesa, 1994; Stegmann and Yoder, 1996). The purpose of our SABRE study was to estimate the daily variation in abundance data collected on eight species of larval fish that seasonally ingress past the permanent sam- pling station at Pivers Island inside of Beaufort Inlet in order to deter- mine an optimum sampling fre- quency for future sampling proto- cols. For one of these species, Brevoortia tyrannus (Atlantic men- haden), which has been the focus species in SABRE studies, addi- tional analysis was conducted on age and growth with specimens col- lected daily. For all species, we used daily abundance data to calculate the decrease in precision of our rela- tive abundance estimates as the interval between sampling events increased. Daily collections of lar- vae also allowed us to measure changes in size (length) of all eight 478 Fishery Bulletin 95(3), 1997 species. Finally, environmental variables were tested for their correlation with abundance. Materials and methods Sampling location and period The sampling station for larval fish abundance, lo- cated 2 km inside of Beaufort Inlet, North Carolina (34°43'N, 76°40'W), was a platform attached to a bridge over a 6-m-deep tidal channel adjacent to the Beaufort Laboratory at Pivers Island and has been the site of weekly larval fish sampling since the 1985- 86 larvae ingress season (Warlen, 1994). We sampled every night, 20 November 1991 through 15 April 1992, a period that more or less encompasses the annual periods of recruitment of ocean-spawned es- tuarine-dependent larvae that pass through North Carolina inlets from autumn to spring. Fish and environment sampling Oblique tows (bottom to surface) of a 1-m diameter, 800-p mesh net were used to sample the water col- umn for larvae. Three consecutive 4-min tows were made at 15-min intervals during the time of predicted maximum flood-tide current. Sampling was con- ducted only between dusk and dawn and about 50 minutes later each successive night because of the advancing tide stage. Oblique tows through the en- tire water column were chosen over surface, bottom, or other single-depth tows to eliminate depth bias. Species of concern, including B. tyrannus, are re- ported to be distributed by depth even in shallow, well-mixed North Carolina inlets (Lewis and Wilkens, 1971; Hettler and Barker, 1993 ). The net was deployed by paying out the winch cable as the net, pulled downstream by the tidal current, sank to the bottom. It was then retrieved obliquely through the water column. A depth sounder with a deck readout (Standard Communications DS20) was attached to the net frame to indicate that the net had reached the bottom of the channel. Tow volume was measured with a General Oceanics model 2030R flow meter. Average tow time was 4.0 minutes. The target tow volume was 100 m3, and target net re- trieval speed was 1 m/sec. Data on several environmental variables were col- lected concurrently with biological sampling. Salin- ity and temperature measurements were taken with a Hydrolab H20 water quality multiprobe. Water clar- ity was measured with a Sea Tech 25-cm transmis- someter with a 660-nm filter. Tidal current speed was measured with a Marsh-McBirney model 201 flow meter. Wind speed and direction data were obtained from the NOAA C-MAN station at Cape Lookout, 15 km SE of the larval sampling platform. Tidal ampli- tude data were obtained from a NOAA tide gauge located on Pivers Island. Processing of larvae After preservation in 70% ethanol, fish larvae were identified, counted, and up to 20 individuals of each species were indiscriminately selected for measure- ment of standard length. Ages and birth dates of all menhaden larvae retained for length measurements (1,341 individual fish) were determined by otolith daily increment counts (estimated age in days = in- crement count + 5) following the methods of Warlen (1992). Data analysis Abundance was calculated from the number of lar- vae caught per tow and water volume sampled (density=number x 100 m-3). Densities per unit vol- ume were calculated rather than densities per unit area because all published relevant abundance data on these species is per unit volume. Mean densities by species for each date sampled were determined by averaging the densities from the three tows taken on that date. Although we sampled every night, no data were available for 10 dates during the sampling period (Fig. 1, A-D). This problem is explained by the following example. On 14 December, sampling started at 2359 h and ended around 0100 h, 15 De- cember. The next night sampling began at 0033 h, 16 December. Thus, sampling never began on 15 December. Sampling on 15 December at the time of maximum flood tide current would have occurred before sampling on 14 December had ended. The same situation occurred on nine other dates. Seasonal mean densities were determined by av- eraging the daily densities during the interval when each species was caught, including dates when no individuals of the species were caught. Variations in mean daily densities and associated variance esti- mates were derived by “sampling” individual den- sity data sets for each species at intervals of 2, 3, 4, 5, 7, 14, and 30 days and by then comparing these with the actual data set (1-day intervals between sampling). From this exercise, a mean and standard deviation was generated for each sampling scheme. A 2-day cycle, for example, beginning on 20 Novem- ber and continuing every other day at day 1, 3, 5 etc., produced one daily mean and standard devia- tion, whereas a 2-day cycle beginning a day later on 21 November and continuing every other day at day Hettler et al.: Variability in abundance of larval fishes inside Beaufort Inlet 479 Date Figure 1A Nightly mean densities (three tows each night) of larval Atlantic menhaden, Breuoortia tyrannus, and spot, Leiostomus xanthurus, caught near Beaufort Inlet, North Carolina. 2, 4, 6 etc. produced a different daily mean and stan- dard deviation. Spectral analysis (ARIMA procedure) was used to examine the time series of densities for each species for evidence of periodicities. Weekly density data based on two methods were compared by using a paired-mean Wilcoxon Signed Rank test. The Laird version (Laird et al., 1965) of the log-transformed Gompertz growth equation (Zweifel and Lasker, 1976) was used to describe the average growth of B. tyrannus larvae. The model was fitted to data for size and estimated age at time of capture. Log-trans- formed standard length (mm) and estimated age (in days) were used in the model (Warlen, 1992). In order to examine the possible effects of envi- ronmental variables on observed larval densities, we first fitted polynomial regression models to the daily densities for each of the species’ densities over time. Although a second-order polynomial was sufficient to describe the recruitment patterns of Paralichthys 480 Fishery Bulletin 95(3), 1 997 320 M. undulatus 240 a> o c: CO "O c .Q CO £ Date Figure 1 B Nightly mean densities (three tows each night) of larval Atlantic croaker, Micropogonias undulatus , and pinfish, Lagodon rhomboides, caught near Beaufort Inlet, North Carolina. dentatus and Micropogonias undulatus , most species required a fourth-order polynomial, and Breuoortia tyrannus required a fifth-order polynomial. The poly- nomial models provided a means of estimating each species’ density for each day as well as the difference between the observed density and the estimated den- sity (residual ). One would expect that if an environ- mental factor influenced observed density, it would do so in a proportional sense, i.e. its effects would be exhibited in relation to the expected density of the species at that date during the season of recruitment for that species. We therefore divided each residual by the expected density for that date to obtain a measure that took the species’ recruitment pattern into account in looking at correlations with environ- mental variables. Durban-Watson statistics and first order autocorrelation coefficients were also computed to test the presence of autocorrelation and to mea- sure its magnitude. Heftier et al.: Variability in abundance of larval fishes inside Beaufort Inlet 481 to e CO NNDDDDJJJJJFFFFMMMMAAA OOEEEEAAAAAEEEEAAAAPPP VVCCCCNNNNNBBBBRRRRRRR Date Figure 1C Nightly mean densities (three tows each night) of larval southern flounder, Paralichthys lethostigma, and gulf flounder, Paralichthys albigutta, caught near Beaufort Inlet, North Carolina. Results Species abundance The eight species selected for analysis accounted for 92% of the larvae caught during the period and are listed in Table 1 in order of decreasing abundance. Although abundance was not predictable from one night to the next for any species (Fig. 1, A-D), the Durban- Watson statistics for the polynomial regres- sion models detected significant autocorrelation in the residuals for six of the eight species. However, one could not reject the null hypothesis that, for Mugil cephalus and for Paralichthys albiguitta, the test was inconclusive. The first-order autocor- relations for the other six species ranged from 0.30 for Paralichthys dentatus to 0.52 for M. undulatus. These reflect the serial dependency between densi- ties on successive nights. On many dates, densities were two or more times less or more than the night 482 Fishery Bulletin 95(3), 1997 a >. Date Figure ID Nightly mean densities (three tows each night) of larval summer flounder, Paralichthys dentatus, and striped mullet, Mugil cephalus, caught near Beau- fort Inlet, North Carolina. before; in some cases the difference between two consecutive nights was an order of magnitude (e.g, Leiostomus xanthurus on 18-19 March). Although periods of low and high abundance can be seen along the time axis for each species, oscilla- tions in abundance by each species did not occur at the same time in all cases. However, most of the spe- cies share periods of high abundance (mid January, mid February, early March, and mid March). We pre- sumed that our data set would be a good candidate for time-series analysis. Spectral analyses of the es- timated densities for each species was employed to reveal evidence of periodicities of varying length. All eight species exhibited a strong 14-day signal, most likely dominated by the lunar cycle. However, this 14-day signal turned out to be an artifact in the sam- pling method which was unavoidable. This artifact was due to sampling 50 min later each night in the sampling scheme (sampling at the same stage in the flood tide) until dawn, at which time the next sam- pling opportunity would be either 12 h 25 min later or 37 h 15 min later. We chose the later option (to Heftier et at: Variability in abundance of larval fishes inside Beaufort Inlet 483 Table 1 Average seasonal density (=AveDen) and maximum daily density (=MaxDen) of target species listed in order of decreasing aver- age density (larvae per 100 m3). Sampling was conducted between 20 November 1991 and 15 April 1992. Scientific name Common name AveDen MaxDen Capture date Leiostomus xanthurus spot 82.8 504.0 21 Dec-15 Apr Micropogonias undulatus Atlantic croaker 37.9 274.2 20 Nov-15 Apr Lagodon rhomboides pinfish 30.0 342.0 22 Nov-15 Apr Brevoortia tyrannus Atlantic menhaden 10.0 159.8 22 Nov-15 Apr Paralichthys lethostigma southern flounder 3.6 28.2 04 Dec-11 Apr P. albigutta gulf flounder 2.7 19.7 20 Nov-15 Apr P. dentatus summer flounder 2.0 12.2 31 Dec- 15 Apr Mugil cephalus striped mullet 1.0 15.2 04 Dec- 14 Apr Table 2 Standard error of the mean abundance of larval species obtained at 2- to 30-day subsampling intervals of the actual daily sam- pling data set. Scientific name If the number of days between sampling had been 2 3 4 5 7 14 30 B. tyrannus 0.78 0.89 0.93 1.32 1.98 1.93 1.54 L. rhomboides 2.60 2.36 1.62 3.05 4.14 4.14 3.89 L. xanthurus 0.73 5.22 2.52 7.99 5.68 12.86 9.50 M. undulatus 1.96 2.28 1.17 1.37 1.90 5.90 3.65 M. cephalus 0.31 0.26 0.22 0.21 0.38 0.24 0.33 P. albigutta 0.37 0.14 0.44 0.19 0.27 0.26 0.31 P. dentatus 0.07 0.26 0.14 0.09 0.19 0.32 0.38 P. lethostigma 0.15 0.24 0.43 0.22 0.42 0.48 0.46 avoid sampling twice on the same date), but either option would have resulted in an irregular pulse in- terval in the time line every 14 days and would have precluded further time-series analysis. Within-night variability Variability between tows of the three tows each night was estimated by calculating the coefficient of varia- tion (CV) for each night and for each species. The CV averaged about 50% for each species throughout the sampling period. Late in the immigration period, the CV in densities within a night increased for Paralichthys lethostigma', for all other species daily tow-to-tow variability was relatively constant throughout the time series. Sampling interval The range in density estimates derived from sub- sampling the actual data set increased as the sub- sampling interval increased (Fig. 2), as did the stan- dard error of the mean for most species (except Mugil cephalus and Paralichthys albigutta) (Table 2). For example, sampling every night yielded a seasonal mean of 10 B. tyrannus larvae per 100 m3, but if we had sampled only every 30 days our estimate for the 1991-92 immigration season, seasonal mean abun- dance could have ranged from 3 to 43 larvae per 100 m3, depending on which date we began sampling. Similarly, had we sampled for L. xanthurus every 30 days, our estimated seasonal mean could have ranged from 11 to 249 larvae per 100 m3. Daily versus weekly sampling Sampling methods were compared to see if the in- creased effort required for daily sampling provided better abundance estimates than did weekly sam- pling. Weekly estimates were calculated by using B. tyrannus abundance data from our daily 1-m-net oblique-tows and are plotted (Fig. 3) along with 484 Fishery Bulletin 95(3), 1997 CD O c 03 "D c -Q CO "CO £ _co c CO 0 E ■O 0) co E 50 25 0- 225 150 75 0- 12 1 6- O) LU B. tyrannus L. xanthurus P. dentatus P. albigutta —i 1 100 50 0 130 65 0 12 10 M. undulatus I 5 i i i i L. rhomboides P. lethostigma M. cephalus -i r~ 14 30 — l 1 1 1 1 1 r 2 3 4 5 7 14 30 Sampling interval (days) Figure 2 Estimated effect, based on daily sampling data, of sampling interval on estimated larval fish abundance. B. tyrannus data obtained with a 1 x 2 m net fished near the surface (Warlen, 1994). Sampling with a 1 x 2 m surface net occurred at night, at approximately the same hour, from the same sampling platform, and during the same immigration season (1991-92) as our basic study. The seasonal weekly mean of the weekly means calculated from daily means obtained with the 1-rm net was significantly different (P<0.Q5) from the seasonal weekly mean density obtained with the 1 x 2 m surface net. Sampling daily with the 1-m oblique net resulted in an average of 10.0 B. tyrannus larvae per 100 m3 compared with 7.2 larvae in weekly sampling during the same season with the 1 x 2 m sur- face net (Warlen, 1994). The abundance of B. tyrannus in catches of the 1-m oblique net tows made only on the nights that the 1 x 2 m surface net was fished appeared similar to the weekly average of the daily catches with the 1-m oblique net tows (10.2 vs. 10.0 larvae per 100 m3), but this similarity could not be statistically tested because the samples were not independent. Hettler et a I.: Variability in abundance of larval fishes inside Beaufort Inlet 485 Size of larvae Plots of mean length for the larval species produced a variety of seasonal patterns (Fig. 4). Although B. tyrannus increased in average size until mid-March and then decreased, the mean length of M. undulatus, L. xanthurus, and Lagodon rhomboides peaked in mid-February, then decreased. M. undulatus seemed to share peaks in abundance with peaks in increas- ing mean lengths (e.g. 30 December, 17 and 26 Janu- ary, and 28 February). Larvae of Paralichthys or Mugil did not change substantially in length over the season. Age and growth of menhaden larvae Because SABRE studies have centered around B. tyrannus, daily collections of this species provided a unique opportunity to test correlations of observed larval age structure with waves of immigrating lar- vae and environmental conditions. Otoliths of B. tyrannus (10-32 mm SL) showed a range in esti- mated age of 16 to 106 days (Fig. 5). The Gompertz growth equation predicted a size at hatching of 4.96 mm SL, which is above the reported size at hatching of 3. 2-3. 4 mm SL for laboratory-reared specimens (Powell, 1993). Average daily growth rate declined from 0.32 mm/day between days 30 and 40 to 0.03 mm/day between days 80 and 90. According to back calculations from capture date, B. tyrannus ingressing Beaufort Inlet spawned from 12 October 1991 to 16 February 1992. Expressed as a percentage of the total Atlantic menhaden caught, two age cohorts, one in mid-December and another in late January, made up about 50% of the year’s recruitment (Fig. 6). Almost 5% of the Atlantic men- haden larvae captured at the sampling station dur- ing the season were hatched on 13 December 1991. The distribution of estimated ages of larvae by col- lection date is shown with an overlay plot of mean daily density (Fig. 7). The largest daily mean den- sity of 159 larvae per 100 m3 (18 March 1992) oc- curred during a decrease in age distribution of about 25 days and would suggest a significant import of a younger cohort of B. tyrannus. Environmental variables The environmental conditions (except barometric pressure) recorded at the time of sampling are shown in Figure 8. Spearman correlation coefficients were computed for each of the eight species with each of the nine environmental variables (Table 3). Twenty- seven of the 72 coefficients were significant (P<0.Q5). 486 Fishery Bulletin 95(3), 1997 2201 120012201 1201 12001 0741851852952964185185 NNDDDDJ J J J J FFFFMMMMAAA OOEEEEAAAAAEEEEAAAAPPP VVCCCCNNNNNBBBBRRRRRRR 2201 120012201 1201 12001 0741851852952964185185 NNDDDDJ J J J J FF FFMMMMAAA OOEEEEAAAAAEEEEAAAAPPP VVCCCCNNNNNBBBBRRRRRRR Date Figure 4 Standard lengths of each species; n = number of specimens measured. The proportional residuals were positively correlated with hour of capture after dusk for six species and negatively correlated with turbidity for five species. As an example, the regressions of abundance of B. tyrannus on each environmental condition are shown in Figure 9. Significant correlations for this species were found with tidal amplitude, current speed, moon phase (=spring vs. neap tide), water clarity, and hours after dark when sampling occurred. Paralichthys dentatus, however, exhibited no correlations. As would be expected, there were significant cor- relations between the nine environmental variables. The highest was between tidal amplitude and sur- face current (0.60), and the next highest was between atmospheric pressure and wind velocity (-0.44). How- ever a principal components analysis yielded four Heftier et a I.: Variability in abundance of larval fishes inside Beaufort Inlet 487 -C U) c 0) ~o CC "O c 2 CO 35 30 25 ^ 20 15 10 B. tyrannus L = 4.96e1'745(1-e"°051') n = 1,341 0 10 20 30 40 50 60 70 80 90 100 110 Estimated age (days) Figure 5 Growth of B. tyrannus larvae as described by the Gompertz model where L = stan- dard length in mm and t = estimated age in days. Dots represent length-at-age measurement on each larva. Hatch date Figure 6 Density-weighted percentage of B. tyrannus larvae ingressing through Beaufort Inlet by back-calculated hatch date. 488 Fishery Bulletin 95(3), 1 997 eigenvalues greater than 1.0, cumulatively account- ing for only 66% of the variation, leading us to con- clude that the structure of the environmental data was not simple. Discussion The periodic nature of seasonal changes in some en- vironmental variables and in reproduction of the dif- Table 3 Spearman correlation coefficients between proportional residuals for species’ densities and various environmental variables: amplitude = tidal amplitude; pressure = barometric pressure; current = flood tide current; hours = hours after sunset; spring tide = spring tide (versus neap tide); temperature = water temperature; wind velocity = average daily wind speed. Bt=B. tyrannus, Lr =L. rhomboides, L x=L. xanthurus , Mu=M. undulatus, Mc=M. cephalus , Pa=P. albigutta, Pd=P. dentatus, P1=P lethostigma. *=P< 0.05. **=P<0.01. ***=P<0.001. Variable Bt Lr Lx Mu Me Pa Pd PI Amplitude 0 40*** 0.19* 0.12 0.08 -0.24 -0.02 0.20 0.08 Pressure 0.11 -0.09 0.04 0.03 -0.24 -0.08 -0.02 -0.20* Current 0 29*** 0.22* 0.06 0.14 -0.11 0.15 0.10 0.20* Hours 0.19* 0.30*** 0.30** 0.43*** 0.01 0.24** 0.17 0.22* Salinity 0.06 0.01 -0.20* -0.11 -0.08 -0.18* 0.14 -0.19* Spring tide -0.26** 0.04 -0.22* -0.24** -0.00 0.00 -0.19 0.09 Temperature -0.07 -0.24** -0.05 -0.06 0.18 0.02 0.08 0.00 Turbidity -0.28*** -0.42*** -0.32*** -0.29*** 0.10 -0.19* 0.03 -0.17 Wind velocity 0.14 0.28* 0.22* 0.06 0.31* 0.11 0.16 0.15 (Sample size) 133 116 102 133 61 138 91 125 Heftier et at: Variability in abundance of larval fishes inside Beaufort Inlet 489 ferent species implies that care must be taken in the interpretation of correlations between these variables and fish densities. For example, temperature may be inappropriate to associate with observed densi- ties owing to the spawning-season periodicities in- volved with life history strategies of each species. Time of spawning (e.g. early winter) and the arrival at the inlet after a 2-3 month cross-shelf transport time, could result in higher abundance correspond- ing with rising temperatures. The multiple-regres- (luo) apruiiduuv (oas/uio) luaxinQ o o o o o m uooiai pui.M aouemiusuej} % (oas/oi) paads pu|/v\ C 5 52 u 3 hi ^ P o, 0 c 43 § 'S, 03 be ^ 'C T3 . S 5 a £ a- o ' be *.S o i '■a ' T3 ; c c3 L ^3 Sh O O CD a > ►5* l“s P a bb £ S >> ’C 03 -4-J dS a> ^ B 'S § Cfl u -a g c/3 XJ T3 _Q cK o ^ 490 Fishery Bulletin 95(3), 1997 a> O c 03 ■O _Q 03 16 £ 03 03 ■D C 03 CD 160 120 80 40 10 15 Temperature 160 120 .2 80 40 20 120 80 40 D A M .. . V ^ 0 45 90 135 180 225 270 315 360 Wind direction 160 120 80 40 2 4 6 8 10 12 14 Wind speed (m/sec) i l : I i I 160 120 80 40 0.2 0.4 06 0.8 Current speed (m/sec) 160 120 80 40 1 2 3 4 5 6 7 8 9 10 11 12 1st Hours since sunset FULL Lsl NEW Moon phase Environmental factors Figure 9 Abundance of B. tyrannus collected during 138 sampling nights versus nine environmental factors. Wind direction is direction wind is coming from. Vertical axis is abundance (number larvae per 100 m3). sion analysis and the principal-component analysis suggested that unknown factors on a scale larger than the locally measured environmental conditions are probably more important in causing peaks in abundance of the immigrating larvae. Originating from wide-spread spawning sites during the fall-win- ter period, larvae of different spawning cohorts ar- rive near the inlet from many possible transport routes. The most likely factor is onshore displace- ment of warm Gulf Stream filaments containing patches of larvae (Stegmann and Yoder, 1996). Once under the influence of tidal exchange at the inlet, patches of larvae merge as they ingress the estuary. Variables including temperature, salinity, turbu- lence, turbidity, odors, and currents differentially affect survival or active transport behavior into the estuary (Boehlert and Mundy, 1988). Larval behav- ior may have contributed to the fact that some spe- cies were more abundant in catches made later in the night (e.g. L. rhomboides, L. xanthurus, M. cephalus, and P. dentatus), acting to disperse larvae into the water column from the edges and bottom, thus making them more vulnerable to the sampling gear. Hettler et a I.: Variability in abundance of larval fishes inside Beaufort Inlet 491 Daily ageing of B. tyrannus caught during the study revealed that a rapid and distinct shift in lar- val populations occurred in mid-March (Fig. 7). In early March, the recruiting larvae were primarily from a cohort that was spawned in mid-December, first reached the inlet in mid-January, and suddenly disappeared on 16 March. On 18 March a new co- hort, spawned in mid-January, appeared. Its appear- ance was coincident with the year’s highest daily mean density. This change in age structure coincided with a 3-day shift to southwesterly winds and full- moon spring tides. Advanced very high resolution radiometer sea surface temperature (AVHRR-SST) imagery revealed that sea surface temperatures 15 km off Beaufort Inlet warmed to 15°C on 9 March 1992, up from 11°C a week earlier (and down to 12.3°C a week later), possibly bringing in the younger B. tyrannus larvae from warmer offshore water to the vicinity of Beaufort Inlet (Stegmann1). This warming was also detected by our temperature mea- surements at the sampling platform, rising to 17.9°C on 9 March followed by a decrease to 12°C on 18 March (Fig. 8), when the large number of younger larvae were caught. Until that date, the average es- timated age of larvae caught in Beaufort Inlet in- creased from about 35 days in late November to about 80 days in mid-March. About 57% of the menhaden captured in the 1991-92 season were spawned in two 2-week periods (mid-December and late January). Although substantial spawning may have occurred at other times, the population from which these lar- vae were caught contained the survivors of the cross- shelf transport. It appears that size at estimated age was significantly larger, about 3 mm SL, for ages between 40 and 80 days for the 1991-92 collections than the larval size reported by Warlen ( 1992) for B. tyrannus collected mainly offshore of Beaufort Inlet in 1979-80. Because the daily ageing method was the same, this observed difference in growth rates of the 1991-92 ingressing larvae is attributed to a higher growth rate among the larvae that survive to reach the inlet. Coastal marine environments experience periods of diurnal or semidurnal tidal cycles imbedded within lunar and semilunar cycles, and these have been shown to influence a broad range of organisms and processes (Hutchinson and Sklar, 1993). Cyclical phenomena impose a particular set of requirements for their adequate measurement and for the avoid- ance of bias arising from aliasing (Kelly, 1976) be- cause the temporal sequence of observations will be 1 Stegmann, P. 1996. Graduate School of Oceanography, Univ. Rhode Island, S. Ferry Road, Narragansett, RI 02882. Per- sonal commun. autocorrelated. A circannual rhythm is a feature of the life history of most vertebrates, and one mani- festation of this is the restriction of reproduction of a species to a season of several weeks or months. If the purpose of a sampling program is to compare recruitment of a species of larval fish from year to year, then it should be designed so that it describes each year’s temporal pattern accurately. In this re- spect it differs from the normal random sampling situation designed to estimate the mean and vari- ance of some statistical population. Because its pur- pose is to enable description of a temporal pattern, a systematic design is normally chosen to ensure equal (or near equal) spacing of samples. Equal spacing of samples is required for many approaches for the analysis of a time series. However, if there are other, shorter cycles within the seasonal pattern, then care must be taken to avoid spurious results (aliasing) that can arise when the sampling interval is greater than one half the wavelength of a significant compo- nent cycle. Sampling to determine a seasonal flux of larvae should be designed to detect temporal patterns, as well as estimate a mean abundance and variance. The question of sampling frequency may have a lower priority than considerations of sampling costs and vessel availability, and thus it is important to quantify the effect that sampling frequency has on estimates of larval abundance, size, and age. For example, one may be interested in estimating the flux of larvae across a boundary over some unit of time. If one is interested in the strength of a year class, the sampling effort must include the entire season of larval recruitment of that species. If one is interested in evaluating the influence of a meteoro- logical event on larval distribution, then the appro- priate sampling interval would be measured in hours. As we shall see, both sampling designs also require due consideration of various physical and biological rhythms that have an important bearing on the num- ber of larvae collected at a given point in space and time (e.g. tidal, circadian, circannual). Prior to es- tablishing sampling protocol for future studies, we attempted to determine a sampling interval that would provide acceptable larval fish abundance es- timates. Larval fish surveys have been made in Beau- fort Inlet and other North Carolina inlets on differ- ent sampling intervals, i.e. weekly (Warlen, 1994), bi-weekly (Lewis and Mann, 1971), every new and full moon period (Hettler and Chester, 1990) and every new moon period (Hettler and Barker, 1993 ). When we com- pared the weekly sampling method that has been used for monitoring at Pi vers Island for the past 10 years (Warlen, 1994) with our daily sampling experiment, a difference in estimated abundance ofR. tyrannus was detected. Differences in the two types of nets may have 492 Fishery Bulletin 95(3), 1997 been responsible for the differences in catches. The 1-m net is an active gear, retrieved obliquely at a rate of about 1 m/sec through the water column, whereas the 1 x 2 m net is fished passively in the surface cur- rent (flowing 0.2-0. 5 m/sec, Warlen2). Also, the mesh of the 1-m net is 200 m smaller and may have reduced extrusion of larvae. Density estimates derived from sampling one night per week with the 1-m net were similar to estimates derived from sampling every night per week with the 1-m net made on the same night as the 1 x 2 m surface net sets and further support our suspicions regarding the differences in sampling with the two gears. Finally, we have assumed for 10 years (Warlen2) that the Pivers Island station serves as a proxy for Beaufort Inlet. An intensive SABRE study conducted in March 1996, dining which larval fish were sampled synchronously at seven locations in and near the inlet, including Pivers Island, is under analysis to determine how closely Pivers Island larval fish densi- ties reflect Beaufort Inlet densities. We conclude that at least one sampling event each week is required for estimating late autumn through early spring seasonal abundance of fish larvae in North Carolina inlets, although more frequent sam- pling is preferred if the standard error of estimates is to be reduced. Bias may be introduced by sam- pling only at a specific period of the lunar cycle, be- cause larval abundance appeared to decrease follow- ing spring tides. Studies where sampling is done ex- clusively at the same lunar phase may consistently over or under estimate abundance. This attribute may be acceptable for interannual comparisons, if the same methods are followed from year to year, but would not be acceptable for calculating the flux of larvae through an inlet. Sampling at intervals of 7 days or less can reduce this bias. Acknowledgments We would like to thank L. Barker, J. Burke, M. Buzzell, J. Cambalik, V. Comparetta, J. Govoni, M. Hesik, P. Murphy, Beeda Pawlik, Brian Pawlik, R. Robbins, J. Scope, L. Settle, A. Walker, H. Walsh, P. Whitefield, and L. Wood of the Beaufort Laboratory who assisted with night sampling. We especially thank V. Comparetta and M. Buzzell for sorting, iden- tifying, and measuring the larvae. J. Hare and S. Warlen provided in-house reviews of this manuscript. Support for this study was received from the SABRE 2 Warlen, S. 1996. Southeast Fish. Sci. Center, Natl. Mar. Fish. Serv., 101 Pivers Island Rd., Beaufort, NC 28516. Personal commun. program of the National Oceanic and Atmospheric Administration’s Coastal Ocean Program / Coastal Fisheries Ecosystems Studies. Literature cited Boehlert, G. W., and B. C. Mundy. 1988. Roles of behavioral and physical factors in larval and juvenile fish recruitment to estuarine nursery areas. Am. Fish. Soc. Symp. 3:51-67. Davis, T. L. O., G. P. Jenkins, and J. W. Young. 1990. Patterns of horizontal distribution of the larvae of southern bluefin ( Thunnus maccoyii) and other tuna in the Indian Ocean. J. Plankton Res. 12:1295-1314. Govoni, J. J., and L. J. Pietrafesa. 1994. Eulerian views of layered water currents, vertical distribution of some larval fishes, and inferred advective transport over the continental shelf off North Carolina, USA, in winter. Fish. Oceanogr. 3:120-132. Hettler, W. F., Jr., and D. L. Barker. 1993. Distribution and abundance of larval fishes at two North Carolina inlets. Estuarine Coastal Shelf Sci. 37:161-179. Hettler, W. F., Jr., and A. J. Chester. 1990. Temporal distribution of ichthyoplankton near Beau- fort Inlet, North Carolina. Mar. Ecol. Prog. Ser. 68:157-168. Houde, E. D., and J. D. A. Lovdal. 1985. Patterns of variability in ichthyoplankton occurrence and abundance in Biscayne Bay, Florida. Estuarine Coastal Shelf Sci. 20:79-103. Hutchinson, S. E., and F. H. Sklar. 1993. Lunar periods as grouping variables for temporally fixed sampling regimes in a tidally dominated estuary. Estuaries 16:789-798. Kelly, J. C. 1976. Sampling the sea. In D. H. Cushing and J. J. Walsh (eds.), The ecology of the seas, p. 361-387. W. B. Saunders, Philadelphia, PA. Laird, A. K., S. A. Tyler, and A. D. Barton. 1965. Dynamics of normal growth. Growth 29:233-248. 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. Lewis, R. M. and E. P. H. Wilkens. 1971. Abundance of Atlantic menhaden larvae and associ- ated species during a diel collection at Beaufort, North Carolina. Chesapeake Sci. 12:185-187. Morse, W. W. 1989. Catchability, growth, and mortality of larval fishes. Fish. Bull. 87:417-446. 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. Rowe, P. M., and C. E. Epifanio. 1994. Flux and transport of larval weakfish in Delaware Bay, USA. Mar. Ecol. Prog. Ser. 110:115-120. Stegmann, P. M., and J. A. Yoder. 1996. Variability of sea-surface temperature in the South Atlantic Bight as observed from satellite: implications for offshore-spawning fish. Cont. Shelf Res. 16:843-861. Heftier et al.: Variability in abundance of larval fishes inside Beaufort Inlet 493 Warlen, 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 Carolina estuary. Fish. Bull. 92:420-433. Wiebe, P. H., and W. R. Holland. 1968. Plankton patchiness: effects of repeated net tows. Limnol. Oceanogr. 13:315-321. Zweifel, J. R., and R. Lasker. 1976. Prehatch and posthatch growth of fishes - a general model. Fish. Bull. 74:609-621. 494 Abstract-Length at age, length at maturity, and age at maturity of yel- lowfin sole, Pleuronectes asper, in the eastern Bering Sea, are influenced by area of sampling and bottom depth. Yellowfin sole sampled during spring and summer bottom trawl surveys (1982-94) grew faster in the northwest area compared with the southeast area. Mean lengths at age were generally more than 2 cm greater than those for southeast fish at ages greater than 10 years. Length at 50% maturity in fe- males during 1993 and 1994 was re- spectively 2.3 cm and 0.94 cm larger in the northwest area than in the south- east area. In contrast, there was no apparent difference in age at 50% ma- turity between areas. Spring-summer patterns in bathy- metric habitation of yellowfin sole dif- fer for immature and mature individu- als and cause a potential bias in esti- mates of growth and maturity. There is a clear relation between length and depth for immature fish, with older, immature fish inhabiting deeper water. In contrast, mature fish distribute simi- larly by size across a wide range of bot- tom depths. As a result, estimates of length and age at 50% maturity (L50, A50) tended to increase with increasing bottom depth. Because current resource assessment surveys do not sample the shallowest areas of the summer distri- bution of yellowfin sole, estimates of L50 and A50 are inherently biased high. Manuscript accepted 27 February 1997 Fishery Bulletin 95:494-503 (1997). Effects of geography and bathymetry on growth and maturity of yellowfin sole, Pleuronectes asper, in the eastern Bering Sea Daniel G. Nichol Resource Assessment and Conservation Engineering Division Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE Seattle, Washington 981 15-0070 E-mail address: Dan.Nichol@noaa.gov Yellowfin sole, Pleuronectes asper, inhabit the nearshore shelf areas of the eastern Bering Sea, from Bristol Bay to north of Nunivak Island (60°N lat.XFig. 1), during spring and summer months. Adult indi- viduals overwinter near the shelf- slope break at approximately 200 m. Two main eastern Bering Sea over- wintering groups, composed mainly of sexually mature adults (Krivobok and Tarkovskaya, 1964) have been identified: a southern complex near Unimak Island and a central com- plex located west of the Pribilof Is- lands (Fadeev, 1970; Bakkala, 1981; Wakabayashi, 1989). During spring, as the edge of the shelf ice recedes toward the coast, yellowfin sole mi- grate across the shelf to nearshore spawning areas in Bristol Bay and off Nunivak Island (Bakkala, 1981). Yellowfin sole generally spawn at bottom depths less than 50 m (Wil- derbuer et al., 1992); most spawn- ing activity, however, occurs at depths less than 30 m, May through August (Nichol, 1995). Juvenile yel- lowfin sole probably do not undergo the long cross-shelf migration. At least some juveniles (<6 years) are known to overwinter nearshore (Fadeev, 1970; Wilderbuer et al., 1992), whereas relatively few juve- niles overwinter offshore (Krivobok and Tarkovskaya, 1964; Fadeev, 1970). Bottom trawl surveys for ground- fish resource assessment are con- ducted annually in the eastern Bering Sea to obtain abundance es- timates of commercially important fish and invertebrate species. Dif- ferences in fish distributions and oceanographic factors have prompted scientists who analyze survey results to stratify the eastern Bering Sea into discrete northwest and south- east areas, and three different depth regimes (Walters and McPhail, 1982; Walters, 1983; Wakabayashi, 1989; Bakkala, 1993). Thus, both geo- graphic and bathymetric factors af- fect fish distribution and abundance in the eastern Bering Sea. In this paper I describe regional differences in growth of yellowfin sole (P. asper ) from the eastern Bering Sea and effects of geographic area and bot- tom depth on estimates of length and age at maturity. Commercial catch records (Norris et al.1) indicate that yellowfin sole occur in substantial numbers in waters shallower than 30 m, where 1 Norris, J. G., J. D. Berger, and K. T. Black. 1991. Fisherman’s guide to catch per unit effort and bycatch data from the National Marine Fisheries Service Ob- server Program: Bering Sea/Aleutian Is- land yellowfin sole trawl fishery. AFSC Proc. Rep. NOAA-NMFS 91-07, 200 p. Alaska Fisheries Science Center, Natl. Mar. Fish. Serv., NOAA, 7600 Sand Point Way NE, Seattle, WA 98115-0070. Nichol: Effects of geography and bathymetry on growth and maturity of Pleuronectes asper 495 Figure 1 Map of the survey stations where yellowfin sole were sampled by the eastern Bering Sea crab-groundfish bottom trawl survey of the Alaska Fisheries Science Center from 1982 to 1994. Northwest (NW) and southeast (SE) areas are delineated by the strata boundary line extending from north of Kuskokwim Bay, southeast through the Pribilof Islands. research surveys are not rou- tinely carried out. Thus, cur- rent survey biomass esti- mates underestimate popula- tion abundance of the spe- cies; exclusion of juveniles in shallow water, as well as sexu- ally mature yellowfin sole that inhabit these nearshore spawning areas during the survey period (June-August), may introduce a sampling bias into estimates of growth and estimates of size at matura- tion. To determine the extent of this potential problem, I considered the effects of bot- tom depth on fish size. Materials and methods Survey area Resource assessment surveys were conducted in the eastern Bering Sea (Bakkala, 1993; Wakabayashi et al., 1985), from June to mid-August. Standard survey stations were based on a 20 by 20 nautical mile grid that cov- ered the area from inner Bristol Bay west to the con- tinental slope edge and from the Alaska Peninsula north to approximately latitude 61°N (Fig. 1). Bot- tom trawl tows of approximately 1.5 nautical miles and of 30-min duration were made at each station. Surveys began in inner Bristol Bay and generally followed north- and south-directed transects, pro- ceeding westward with each finished transect. Data Yellowfin sole otoliths were collected from 3,891 males and 5,209 females during AFSC surveys, 1982- 94 (Table 1). Fish were measured to the nearest cen- timeter total length (TL) and sagittal otoliths were removed and stored in 50% ethanol for subsequent age determination. Ages were determined by using the break-and-burn technique (Chilton and Beamish, 1982). Female maturity data were collected during 1992- 94 surveys (Table 2). Maturity codes were based on macroscopic gonadal appearance (Nichol, 1995). For the purpose of this study, codes were simplified to either mature or immature. Females were consid- ered mature if ovaries contained yolked or hydrated oocytes, or were recently spent. Immature ovaries contained no visible oocytes. Bata collected during 1993 (Table 2) were best for examining spatial dif- ferences in female length at maturity because samples were collected at nearly all survey stations within the 50-m contour line, and maturity code as- signments were verified by histological examination (Nichol, 1995). Analysis The total survey area was subdivided into northwest and southeast areas (Wakabayashi, 1989; Bakkala, 1993) extending from north of Kuskokwim Bay south- east through the Pribilof Islands (Fig. 1). Length at age Factorial analysis of variance (ANOVA) models (Reish et al., 1985) was constructed independently for males and females to examine the variance in length at each age due to geographic area (northwest or southeast) and to bottom depth. For ANOVA, bottom depths were grouped into three lev- els (<30 m, 30-49 m, and >50 m; Table 1). A model, which included interaction terms and which was pooled across years, was analyzed for both males and 496 Fishery Bulletin 95(3), 1 997 females. Preferred models were estimated by a least- squares backward stepwise procedure that sequen- tially removed the highest-order nonsignifcant fac- tor until only significant terms remained. Nonsig- nificant (P>0.05) main effects were retained if they were included in one or more significant interaction terms. Residual plots were examined to test assump- tions of homoscedasticity and normality of error terms. Least squares estimates of model coefficients were obtained by means of the Statistical Analysis System procedure GLM (SAS Institute, 1989). Length at maturity Logistic regression was used to assess the effects of area and bottom depth on the Tabie 1 Number of male and female yellowfin sole sampled for age and length by the Alaska Fisheries Science Center during resource assessment surveys conducted in northwest and southeast areas of the eastern Bering Sea from 1982 to 1994. Bottom depth (m) Northwest Southeast Year <30 30-49 >50 <30 30-49 >50 Males 1982 26 58 26 65 97 45 1983 0 52 85 56 87 27 1984 14 102 31 31 109 42 1985 28 99 36 26 65 84 1986 0 100 42 39 74 66 1987 133 30 13 38 111 31 1988 44 33 13 27 28 98 1989 29 35 39 23 105 87 1990 49 58 48 36 74 92 1991 82 64 26 0 65 97 1992 32 0 25 73 79 39 1993 31 50 18 47 0 46 1994 24 64 21 72 13 37 Total 492 745 423 533 907 791 Females 1982 15 110 49 65 97 83 1983 0 80 129 57 102 41 1984 0 143 71 33 137 83 1985 41 154 40 43 57 129 1986 0 113 97 48 57 103 1987 152 67 24 36 112 51 1988 58 38 46 47 31 100 1989 0 82 113 39 92 96 1990 47 81 86 54 38 129 1991 85 73 35 0 69 146 1992 35 0 91 86 79 67 1993 28 67 100 61 0 101 1994 29 55 49 61 20 76 Total 490 1,063 930 630 891 1,205 probability that an individual of a given length (cm) was mature or immature. The following equation was fitted independently to 1992, 1993, and 1994 female yellowfin sole length-maturity data by using the Sta- tistical Analysis System maximum likelihood proce- dure LOGISTIC (SAS Institute, 1989): _j_ g-(n+(5 L+a area+8 D+A L D) 5 (l) where MAT = mature proportion of female yellowfin sole given its length (L), area, and bottom depth (D) of capture. Area was treated as a factor indicating either northwest or southeast areas. Depth was treated as a continuous variable. Length, area, and depth coef- ficients were represented by a, and S, respectively, and /j denoted the intercept (on the logit scale). A length x depth interaction term ( L-D ; Eq. 1) with coefficient A was also tested for significance. Non- significant (P>0.05) highest order terms were re- moved from the model. Length at 50% maturity (L50) was calculated by substituting 0.5 for MAT in Equation 1 and solving for L as follows: •^50 // ■ '/ area + 5 D + A • L D (2) Due to sample-size inequalities (Table 2), length at maturity comparisons between northwest and south- east areas were limited to females sampled in 1993 and 1994. Table 2 Summary of female yellowfin sole length-maturity collec- tions during the 1992-94 Alaska Fisheries Science Center eastern Bering Sea groundfish bottom trawl surveys. A = Age, length, and maturity data collected by sex-cm inter- val; L = Length and maturity (random measurements); O = Ovaries, lengths, and maturity data collected by size category, >25 cm TL. Ages were determined for 53 of these specimens. Number of samples Year Sample type Northwest Southeast 1992 A 107 218 L 0 1,260 1993 A 98 162 O 256 512 1994 A 133 158 Nichol: Effects of geography and bathymetry on growth and maturity of Pleuronectes asper 497 Age at maturity Analysis of female age at matu- rity was treated in the same manner as with length at maturity, substituting age (A) in years for length (L). Owing to smaller sample sizes (Table 2), data for years 1992-94 were pooled. Results Length at age The general effects of area and bottom depth on length at age were highly significant for both males and females (Table 3). Mean length-at-age plots for both male and female yellowfin sole indicated greater 45 Males E o or c CD 45 40 35 30 25 20 15 10 Females ss l2 5 i ■ • NW n = 2,483 I » SE n = 2,726 s 1 ' ' 1 1 1 I ’ i i ' I o 5 10 15 20 25 30 Age (yr) Figure 2 Comparison of male and female yellowfin sole mean length at age between northwest (NW) and southeast (SE) areas of the eastern Bering Sea (1982-94). Error bars indicate 95% confidence intervals. sizes at age in the northwest than in the southeast area of the eastern Bering Sea shelf (Fig. 2). Male and female yellowfin sole were on average 1.22 and 1.02 cm larger at age, respectively, in the northwest area than in the southeast area. Average length-at- age differences between areas (northwest-southeast) increased with increasing age to more than 2 cm for both males and females (Fig. 3). Total lengths were generally greater at age in deeper (>50 m) waters than in shallow waters (<50 m) for males and females less than 8 and 9 years of age, respectively (Fig. 3). Length at maturity Female yellowfin sole lengths corresponding to 50% maturity (L50) were greater in the northwest than in the southeast area, and L50 increased with increas- ing bottom depth (Fig. 4). Area accounted for a 2.3 cm female length-at-maturity difference (P=0.0001) in 1993 and a 0.91 cm difference (P=0.049) in 1994 (Fig. 4; Table 4). Female L50 increased with increas- ing bottom depth (P<0.013), varying by as much as 4 cm between shallow and deeper waters (Fig. 4; Table 4). Annual variation ( 1992-94) in L50 appeared to be approximately 1 cm (Fig. 4). Age at maturity In contrast to length-at-maturity results, no signifi- cant age-at-maturity difference (P=0.080) was found for females between the two areas (Fig. 5; Table 5). A similar increase in female age at maturity, however, did occur with increasing bottom depth (P=0.0001), with age at 50% maturity increasing by 3 years from shallow to deeper waters (Fig. 5). Discussion Larger northwest fish lengths corresponding to a particular age or percent maturity, combined with no apparent age-at-maturity difference between ar- eas, indicate faster yellowfin sole growth in the north- west area than in the southeast area. Bottom depth effects on length-at-age, length-at-maturity, and age- at-maturity estimates, however, were more the re- sult of sampling uneven fish distributions, as dis- cussed below. Area effects The length-at-age and length-at-maturity differences found between areas support the hypothesis that northwest and southeast complexes are functionally allopatric during the summer spawning period. Tag- 498 Fishery Bulletin 95(3), 1997 ging studies (Wakabayashi, 1989) indicated only lim- ited movements of yellowfin sole between northwest and southeast areas as they migrate inshore, and Kashkina (1965) supported a two-stock concept, cit- ing differences in egg-stage advancement and abun- dance between northern and southern regions of the eastern Bering Sea. Despite the lack of genetic evi- dence supporting the coexistence of two independent Table 3 Factorial analysis of variance of male and female yellowfin sole length (cm) at age (years), by area (northwest and southeast) and bottom depth (<30, 30-49, and >50 m), pooled across years 1982-94. Note that nonsignificant (P>0.05) main effects were retained because they were included in one or more significant interaction terms. SS = sum of squares. Source df SS Mean square E-value P > F Males Age 29 68,007.6 2,345.1 360.0 0.0001 Area 1 295.5 295.5 45.4 0.0001 Depth 2 5.7 2.8 0.4 0.6457 Age x depth 49 1,906.9 38.9 6.0 0.0001 Age x area 24 160.4 6.7 1.0 0.4268 Area x depth 2 51.8 25.9 4.0 0.0188 Age x area x depth 34 366.9 10.8 1.7 0.0098 Error 3,749 24,421.2 6.5 Females Age 29 152,746.0 5,267.1 667.9 0.0001 Area 1 408.1 408.1 51.8 0.0001 Depth 2 2.1 1.1 0.1 0.8738 Age x depth 49 3,219.2 65.7 8.3 0.0001 Age x area 24 411.2 17.1 2.2 0.0008 Area x depth 2 44.7 22.3 2.8 0.0589 Age x area x depth 44 593.1 13.5 1.7 0.0025 Error 5,057 39,880.8 7.9 Table 4 Logistic regression coefficients for the equation MAT=l/( 1+e-'^ + P'L + a area + S D + AL D)) relating female yellowfin maturity status to fish length (L), binary variable area (northwest or southeast), and continuous variable depth ( D ). L xD denotes the interaction between length and depth. SE= standard error of the estimate. L x D coefficients were considered nonsignificant (P>0.05) and were therefore removed from the model. Remaining coefficients and estimates result from model runs with L xD removed (under- lined values). Year n Variable Coefficient Symbol Estimate SE P > chi-square 19927 1,478 Length P -0.82 0.047 0.0001 Depth 8 0.046 0.0048 0.0001 LxD A 0.0046 0.0024 0.057 Intercept P 22.22 1.36 0.0001 1993 1,028 Length P -0.83 0.063 0.0001 Area a 2.29 0.30 0.0001 Depth 8 0.042 0.0078 0.0001 LxD A -0.0065 0.0035 0.063 Intercept P 21.70 1.81 0.0001 1994 291 Length P -0.69 0.085 0.0001 Area a 0.91 0.46 0.049 Depth 8 0.042 0.012 0.0004 LxD A 0.0046 0.0048 0.34 Intercept P 17.92 2.52 0.0001 1 Southeast area only. Small sample size in the northwest area (n=107) limited a comparison by area. Nichol: Effects of geography and bathymetry on growth and maturity of Pteuronectes asper 499 stocks (Grant et al., 1983), the persistence of area- based differences suggests that mixing of adults be- tween northwest and southeast complexes may be minimal. Growth-rate differences may be associated with geographic differences in yellowfin sole density or bottom temperature (or both). Yellowfin sole mean density (1982-94; author, unpubl. data), measured in catch per unit of effort (kg/hectare) during spring- summer, has been consistently higher in the south- east area (88.9 kg/ha) than in the northwest area (27.8 kg/ha). Spring-summer bottom temperatures have also been consistently higher in the southeast area than in the northwest area (Fig. 6). The higher yellowfin sole growth rate in the northwest areas is consistent with density-dependent hypotheses (Beverton and Holt, 1957; Cushing, 1975; Rijnsdorp, 1994) that suggest a negative correlation between fish growth and fish density. Reasons why fish growth ap- pears faster in cooler northwest waters are less clear. Age-composition data used in stock assessments for yellowfin sole in the eastern Bering Sea (Wakabayashi et al., 1985) have been based upon age-length keys generated from annual age-length collections (Armi- stead and Nichol, 1993). Al- though independent age (oto- lith) collections have been made for the southeast and northwest areas, age-length keys are cur- rently pooled across areas. The resulting estimates of growth are considered accurate because Alaska Fisheries Science Center age-structure collections have been spread fairly evenly be- tween areas. However, given the spatial patterns described here, separate northwest-southeast age-length keys might improve the precision of these estimates. Implications of depth- related sampling biases Immature yellowfin sole, like many other fish species ( Hunter et al., 1990; Macpherson and Duarte, 1991; Jacobson and Hunter, 1993), undergo an on- togenetic migration, distribut- ing themselves along a size- depth continuum where smaller individuals inhabit shallow wa- ters and larger individuals in- habit deeper waters. A single cohort can also distribute itself along this size-depth continu- um. In doing so, faster-growing individuals can be found at Males Females 40 35 30 25 20 15 10 5 0 NW y«0 r °o # V >50 m • 30- 49 m o <30 m 1 1 1 1 1 1 1 1 ■ ■ ■ i ■ ■ ■ 1 1 1 1 1 1 1 1 1 1 1 1 5 10 15 20 25 30 40 — i 35 30 - 25 - 20 15 10 5 0 NW •o v >50 m • 30-49 m o <30 m 0 5 10 15 20 25 30 40 35 30 25 20 15 - 10 - 5 0 SE qO O a Q • «• * • v >50 m • 30-49 m O <30 m 0 1 I 1 1 1 1 I 1 ' ‘ 1 i 1 ’ 1 1 i * 1 1 1 i 1 1 1 1 i 5 10 15 20 25 30 7 6 5 4 3 2 1 0 -1 -2 -3 -4 NW-SE • o * V w ° 5 v V qO 1~Tt 1 1 1 | 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 5 10 15 20 25 30 7 6 5 4 3 4 2 1 - 0 - -1 - -2 - -3 - -4 NW-SE ■ . .. v;! S3”* * “ «• v* -• CO O V 0 1 1 1 1 1 1 1 1 1 1 1 i 11 1 1 i 1 1 1 1 1 1 1 1 1 i 5 10 15 20 25 30 Age (yr) Figure 3 Factorial length at age model results comparing length at age of male and female yellowfin sole across bottom depths (m) in both northwest (NW) and southeast (SE) areas of the eastern Bering Sea. Age x Area x Depth interactions are included with data pooled across 1982-94 age-length collections. 500 Fishery Bulletin 95(3), 1997 35 -| 34 - 33 - 32 - 31 - 30 - 29 - 28 - 27 - 26 - 1992 SE area 25 i i -i — i — i — i — i — i — i — i 0 10 20 30 40 50 60 70 80 90 100 .I' 35 JJ 03 34 - t 33 - & O 32 - IT) 31 - 03 30 - E 29 - O^ _c 28 - o> 27 - c 0 26 - _J 25 - 0 1993 “i i — i — i — i — i — i — i — i — i 10 20 30 40 50 60 70 80 90 100 35 34 - 33 - 32 - 31 - 30 - 29 - 28 - 27 - 26 - 25 -- 0 1994 NW area SE area “T I I I I 1 1 1 1 1 10 20 30 40 50 60 70 80 90100 Bottom depth (m) Figure 4 Female yellowfin sole length at 50% maturity, with respect to bottom depth (D) and geographic area (NW and SE) for years 1992-94. Curves illustrate predicted values of fish length (L) from the logistic equation: MAT = 1/(1 + e ~(n+P L+a area+8 D)^ t where the pro- portion mature (MAT) is 0.5. deeper depths than can slower-growing individuals. Length-at-age estimates for immature yellowfin sole (<8 years of age) from shallower waters, therefore, are biased low in comparison with those from deeper wa- ters (Fig. 3). The cessation of this depth effect with in- creasing bottom depth after 8 years of age (Fig. 3) may indicate the approximate age at which immature yel- lowfin sole leave the size-depth continuum and become migratory (i.e. to shelf-slope waters in winter and back to nearshore waters in spring-summer). The timing of first maturity may very well coincide with the initia- tion of a “spawning” migration. ' ~ ] 1 992-94 14 - 13 - 12 - 1 1 - 10 - 9 - 8 H — i — i — i — i — i — i — i — i — i — i 0 10 20 30 40 50 60 70 80 90 100 Bottom depth (m) Figure 5 Female yellowfin sole age at 50% matu- rity, with respect to bottom depth (D), pooled among years 1992-94. Curves il- lustrate predicted values of fish age (A) from the logistic equation: MAT = 1/ (l+e~(H+P A+8 D)-. j where the propor- tion mature (MAT) is 0.5. o in < The increase in female length and age at maturity with increasing bottom depth was largely due to the variation of immature female length distributions as depth increased. In spring-summer, when yellowfin sole have migrated nearshore for spawning, distri- butions of mature migratory individuals and imma- ture “ontogenetically driven” individuals overlap. The differences between these two migration patterns act to separate immature and mature fish of similar lengths (and ages) along a bottom depth gradient. Immature females were distributed unevenly by size across depth; larger sizes (25-32 cm TL) were more common at deeper depths (Fig. 7). In contrast, ma- ture females were distributed similarly by length between shallow (<30 m) and deeper (>30 m) bottom depths (Fig. 7). The absence of these larger immature females in shallow water resulted in lower values of length at maturity for shallow waters and higher values for deeper waters (Fig. 7). Trippel and Harvey (1991) demonstrated how age at maturity of white suckers ( Catostomus commersoni) could be affected if year classes falling within the progression from immatu- rity to maturity are missing. Missing length classes, similarly, affect estimates of length at maturity. Sam- pling of female yellowfin sole in shallow waters (<30 m) misses critical length and age classes of imma- ture individuals on the verge of maturity. Length or age at maturity, as with growth rela- tions, are often measured simultaneously across multiple cohorts and are therefore approximations Nichol: Effects of geography and bathymetry on growth and maturity of Pleuronectes asper 501 Table 5 Logistic regression coefficients for the equation MAT=l/( 1 + + P A + “ area + s-d + aa-D)^ relating female yellowfin maturity status to fish age (A), binary variable area (northwest or southeast), and continuous variable depth (D). Ax D denotes the interaction between age and depth. SE= standard error of the estimate. A x D and age coefficients were considered nonsignificant (P>0.05) and were therefore removed from the model. Remaining coefficients and estimates result from model runs with A x D and age removed (underlined values). Year2 n Variable Coefficient Symbol Estimate SE P > chi-square 1992-94 929 Age P -0.70 0.048 0.0001 Area a 0.38 0.21 0.080 Depth 5 0.032 0.0056 0.0001 A x D A -0.0040 0.0021 0.054 Intercept P 6.25 0.50 0.0001 1 Data were pooled among years 1992-94. Bottom depth (m) Figure 6 Mean bottom temperatures of the southeast and north- west areas of the eastern Bering Sea shelf, averaged across years 1982-94 at 10 m bottom depth intervals. Error bars indicate 95% confidence intervals. of individual growth and length or age at maturity. This estimation assumes that there is no between year-class growth or maturity differences with re- spect to a given age and that samples within each age class are random with respect to the population 502 Fishery Bulletin 95(3), 1997 (Ricker, 1975). Obtaining representative samples from each age class becomes difficult when fish size and age vary with bottom depth. Considering also that during spring-summer, yellowfin sole abundance increases with decreasing bottom depth (Nichol, 1995), population estimates of yellowfin sole growth and length or age at maturity should be weighted accordingly. Because groundfish assessment surveys in the eastern Bering Sea do not cover the shallow- est areas of yellowfin sole distribution during spring- summer, current estimates of yellowfin sole growth, as well as size and age at maturity, are inherently biased high. Given that most demersal fish distrib- ute themselves along a size-depth continuum (Macpherson and Duarte, 1991), the potential for similar depth-related sampling biases in other dem- ersal fish species appears probable. Acknowledgments I thank Steve Syrjala for his comprehensive statisti- cal advice, and the Age and Growth Unit of the Alaska Fisheries Science Center for the age determinations. Claire Armistead, Pam Goddard, Gary Walters, and Terry Sample helped tremendously with the data collection. Frank Morado, Nick Hodges, and Lisa Mooney performed much of the needed histology work. I thank Dave Somerton, Bob McConnaughey, David Sampson, Gary Walters, Dan Kimura, Jim Ianelli, Gary Duker, and James Lee for their con- structive reviews. Suggestions from three anonymous reviewers helped improve the final manuscript. Literature cited Armistead, C. E., and D. G. Nichol. 1993. 1990 bottom trawl survey of the eastern Bering Sea continental shelf. U.S. Dep. Commer., NOAATech. Memo. NMFS-AFSC-7, 190 p. Bakkala, R. G. 1981. Population characteristics and ecology of yellowfin sole. In D. W. Hood and J. A. Calder (eds.), The eastern Bering Sea shelf: oceanography and resources, Vol. 1, p. 553-574. U.S. Dep. Commer., NOAA, Off. Mar. Pollut. Assess., U.S. Gov. Print Off., Wash., D.C. 1993. Structure and historical changes in the groundfish complex of the eastern Bering Sea. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 114, 91 p. Bevertom, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish. Invest. Ser. II Mar. Fish. G.B. Minist. Agric. Fish. Food 19:1-533. Chilton, D. E., and R. J. Beamish. 1982. Age determination methods for fishes studied by the Groundfish Program at the Pacific Biological Station. Can. Spec. Publ. Fish. Aquat. Sci. 60, 102 p. Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge Univ. Press, New York, NY., 278 p. Fadeev, N. W. 1970. The fishery and biological characteristics of yellow- fin soles in the eastern part of the Bering Sea. Translated by Isr. Prog. Sci. Transl., 1972. In PA. Moiseev (ed.), So- viet fisheries investigations in the northeastern Pacific, part 5, p. 332-396. [Available from U.S. Dep. Commer., Natl. Tech. Inf. Serv., Springfield, VA, as TT 71-50127.] Grant, W. S., R. Bakkala, F. M. Utter, D. J. Teel, and T. Kobayashi. 1983. Biochemical genetic population structure of yellow- fin sole, Limanda aspera, of the North Pacific Ocean and Bering Sea. Fish. Bull. 81(4):667-677. Jacobson, L. D., and J. R. Hunter. 1993. Bathymetric demography and management of Dover sole. N. Am. J. Fish. Manage. 13(3):405-420. Hunter, J. R., J. L. Butler, C. Kimbrell, and E. A. Lynn. 1990. Bathymetric patterns in size, age, sexual maturity, water content, and caloric density of Dover sole, Microsto- mus pacificus. Calif. Coop. Oceanic Fish. Invest. Rep. 31:132-144. Kashkina, A. A. 1965. Reproduction of yellowfin sole (Limanda aspera, Pallas) and changes in its spawning stocks in the eastern Bering Sea. Translated by Isr. Prog. Sci. Transl., 1968. In P. A. Moiseev (ed.), Soviet fisheries investigations in the northeastern Pacific, part 4, p. 182-190. [Available from U.S. Dep. Commer., Natl. Tech. Inf. Serv., Springfield, VA, as TT 67-51206.] Krivobok, M. N., and O. I. Tarkovskaya. 1964. Chemical characteristics of yellowfin sole, cod and Alaska pollock of the southeastern part of the Bering Sea. Translated by Isr. Prog. Sci. Transl., 1968. In P. A. Moiseev (ed.), Soviet fisheries investigations in the northeastern Pa- cific, part 2, p 271-287. [Available from U.S. Dep. Commer., Natl. Tech. Inf. Serv., Springfield, VA, as TT 67-51204.] Macpherson, E., and C. M. Duarte. 1991. Bathymetric trends in demersal fish size: Is there a general relationship? Mar. Ecol. Prog. Ser. 71(2): 103-112. Nichol, D. G. 1995. Spawning and maturation of female yellowfin sole in the eastern Bering Sea. In Proceedings of the interna- tional flatfish symposium; October 1994, Anchorage, Alaska, p. 35-50. Univ. Alaska, Alaska Sea Grant Rep. 95-04. Reish, R. L., R. B. Deriso, D. Ruppert, and R. J. Carol!. 1985. An investigation of the population dynamics of At- lantic menhaden (Brevoortia tyrannus). Can. J. Fish. Aquat. Sci. 42(Suppl. 1):147-157. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics of fish populations. Bull. Fish. Res. Board Can. 191, 382 p. Rijnsdorp, A. D. 1994. Population-regulating processes during the adult phase in flatfish. Neth. J. Sea Res. 32(2):207-223. SAS Institute Inc. 1989. SAS/STAT user’s guide, version 6, fourth ed., vol. 2. SAS Institute, Cary, NC, 846 p. Trippel, E. A., and H. H. Harvey. 1991. Comparison of methods used to estimate age and length of fishes at sexual maturity using populations of white sucker ( Catostomus commersoni ). Can. J. Fish. Aquat. Sci. 48:1446-1459. Nichol: Effects of geography and bathymetry on growth and maturity of Pleuronectes asper 503 Wakabayashi, K. 1989. Studies on the fishery biology of yellowfin sole in the eastern Bering Sea. [In Jpn., Engl. summ.]. Bull. Far Seas Fish. Res. Lab. 26:21-152. Wakabayashi, K., R. G. Bakkala, and M. S. Alton. 1985. Methods of the U.S. -Japan demersal trawl surveys. In R. G. Bakkala and K. Wakabayashi (eds.). Results of cooperative U.S. -Japan groundfish investigations in the Bering Sea during May-August 1979, p. 7-29. Int. N. Pac. Fish. Comm. Bull. 44. Walters, G. E. 1983. An atlas of demersal fish and invertebrate commu- nity structure in the eastern Bering Sea: Part 2, 1971- 77. U.S. Dep. Commer, NOAA Tech. Memo. NMFS-F/ NWC-40, 152 p. Walters, G. E., and M. J. McPhail. 1982. An atlas of demersal fish and invertebrate commu- nity structure in the eastern Bering Sea: Part 1, 1978- 81. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-F/ NWC-35, 122 p. Wilderbuer, T. K., G. E. Walters, and R. G. Bakkala. 1992. Yellowfin sole, Pleuronectes asper , of the eastern Bering Sea: biological characteristics, history of exploita- tion, and management. Mar. Fish. Rev. 54(4): 1-18. 504 Abstract .—Juveniles of four species of pleuronectid flatfishes were abun- dant in bays and nearshore areas around Kodiak Island, Alaska, during August 1991 and 1992. The four most abundant species of juvenile (age-0 or age- 1) flatfishes were rock sole ( Pleuro - nectes bilineatus), flathead sole (Hip- poglossoides elassodon ), Pacific halibut ( Hippoglossus stenolepis), and yellow- fin sole ( Pleuronectes asper ). These spe- cies appeared to share nursery areas; however, physical characteristics of the nursery areas occupied by each species limited the amount of true overlap among species. Tree-based regression of catch-per-unit-of-effort data on physical parameters was used to refine conceptual models of species distribu- tion, which were originally based only on 1991 data. Threshold values of the physical pa- rameters were specified that best dis- criminated among stations with differ- ent abundances. Highest abundances of age-0 rock sole were found on sand or muddy sand at temperatures greater than 8.7°C, as well as on other mixed sand stations less than 28 m deep. Age- 0 flathead sole were most abundant at temperatures less than 8.9°C and on mixed mud substrates. At warmer tem- peratures, abundances were high only if the depth was greater than 48 m, re- gardless of sediment type. Age-0 Pacific halibut were most abundant in depths less than 40 m at sites more than 2.9 km outside the mouths of bays. Inside bays, halibut were found in lower abun- dances in water over 9.0°C and on sedi- ments containing both sand and mud. Age-1 yellowfin sole were always found in depths less than 28 m on mixed mud substrates. They were usually found within bays, with highest abundances at heads of large bays more than 32 km from the bay mouth. These four most abundant flatfishes therefore appeared to partition the available habitat in ways that minimized resource competition. Manuscript accepted 26 February 1997 Fishery Bulletin 95:504-520 (1997). Habitat models for juvenile pleuronectids around Kodiak Island, Alaska* Brenda L. Norcross** Franz-Josef Muter Brenda A. Holfaday Institute of Marine Science School of Fisheries and Ocean Sciences University of Alaska Fairbanks Fairbanks, Alaska 99775-7220 **E-mail address: norcross@ims.alaska.edu In the Gulf of Alaska, there are di- rected fisheries for deep-water and shallow-water complexes of fishes. The deep-water complex is made up of rex sole ( Errex zachirus), Dover sole ( Microstomus pacificus), Green- land halibut ( Reinhardtius hippo- glossoides), arrowtooth flounder (Atherestes stomias ), and rockfishes ( Sebastes spp.). The shallow water complex incorporates all other flat- fishes found in the area, including rock sole ( Pleuronectes bilineatus), flathead sole ( Hippoglossoides elassodon), yellowfin sole ( Pleuro- nectes asper), English sole (Pleuro- nectes vetulus), starry flounder ( Platichthys stellatus), Alaska pla- ice (Pleuronectes quadritubercu- latus), butter sole ( Pleuronectes isolepis), and sand sole (Psettichthys melanostictus), in addition to Pacific cod ( Gadus macrocephalus) and walleye pollock (Theragra chalco- gramma). The groundfish harvest from the Gulf of Alaska has been over 190,000 metric tons (t) annu- ally from 1990 through 1995, for a total of 1,320,000 t, not including Pacific halibut (Hippoglossus steno- lepis) or discards. Of that, in 1995, 716,000 t were landed in Kodiak, Alaska, for a value of $34 million (NMFS, Fisheries Management Biv., P.O. Box 21668, Juneau, AK 99802-1668). When Pacific halibut, a species regulated separately from other groundfishes, is included, the total landed at Kodiak in 1995 was 75,000 t at $49 million. Although rockfishes (Carlson and Straty, 1981; Krieger, 1992, 1993), cod (Wespestad et al., 1986; Dunn and Matarese, 1987), and pollock ( Janusz, 1986; Dunn and Matarese, 1987; Kendall et al., 1994; Muter and Norcross, 1994; Swartzman et al., 1994) have been studied in the Gulf of Alaska, very little is known about flatfishes (Parker, 1989; Moles and Norcross, 1995). The large abun- dance and value of these commer- cially important flatfishes and lack of knowledge of their early life his- tory led us to investigate distribu- tion of juvenile flatfishes around Kodiak Island. In general, recently metamor- phosed flatfishes recruit to shallow, nearshore nursery areas with fine- grained sediments (Edwards and Steele, 1968; Gibson, 1973; Toole, 1980; Hogue and Carey, 1982; de Ben et al., 1990). Intertidal zones, estuaries, and shallow protected bays are nursery areas for flatfishes in the continental United States (Krygier and Pearcy, 1986; Allen, 1988; Rogers et al., 1988; Wyanski, 1990), Canada (Tyler, 1971), Europe * Contribution 1627, Institute of Marine Sci- ence, University of Alaska, Fairbanks, Alaska 99775-7220. Norcross etal.: Habitat models for juvenile pleuronectids 505 (McIntyre and Eleftheriou, 1968; Gibson, 1973; Lockwood, 1974; Poxton et al., 1982; Poxton and Nasir, 1985; van der Veer and Bergman, 1986), and Japan (Tanaka et al., 1989). Abundance and size dis- tributions have been related to water depth (Edwards and Steele, 1968; McIntyre and Eleftheriou, 1968; Lockwood, 1974; Riley et al., 1981; Poxton et al., 1982; Wyanski, 1990), sediment size (Poxton et al., 1982; Poxton and Nasir, 1985; Wyanski, 1990; Jager et al., 1993; Keefe and Able, 1994; Moles and Norcross, 1995), and food availability (McIntyre and Elef- theriou, 1968; Allen, 1988; Jager et al., 1993). The generally accepted rationale for juvenile recruitment to shallow, fine-grained nursery areas includes es- cape from predation, increased cover and food avail- ability, and decreased intraspecific food competition (Toole, 1980; de Ben et al., 1990; Minami and Tanaka, 1992). The examination of diet diversity among a subset of the fishes in the present study showed a reduction of both interspecific and intraspecific di- etary overlap when flatfishes coexisted in large abun- dances (Holladay and Norcross, 1995). The coastline of Kodiak Island, Alaska, encom- passes a variety of habitats from shallow, fine-grained tidal flats to deep and rocky areas. Kodiak Island is mountainous and cut by many fjords and open bays with shallow waters (<10 m) usually within 0.5 km of the beach. The tidal range is 3 to 4 m. The region is characterized by deep bays, rough bottom topo- graphy, strong currents, and bottom characteristics that change rapidly over relatively short distances. Around Kodiak Island, juvenile flatfishes occupy fine- grained sediments in bays and nearshore waters, as do flatfishes in other locations, but waters less than 10 m in depth are only a minor component of the area that is used (Norcross et al., 1995). A nursery may be partitioned into areas dominated by individual species or intraspecific age groups (Edwards and Steele, 1968; Zhang, 1988; Harris and Hartt1; Smith et al.2). Habitats occupied by juvenile rock sole, flathead sole, Pacific halibut, and yellow- fin sole collected on the east and south sides of Kodiak Island in August 1991 can be differentiated on the basis of depth, substrate, and within-bay distribu- tion (Norcross et al., 1995). 1 Harris, C. K., and A. C. Hartt. 1977. Assessment of pelagic and nearshore fish in three bays on the east and south coasts of Kodiak Island, Alaska: final report. In Volume 1: Environmen- tal assessment of the Alaskan continental shelf, p. 483- 688. U.S. Dep. Commer., and U.S. Dep. Interior Quarterly Reports of Principal Investigators, Anchorage, AK. 2 Smith, R. L., A. C. Paulson, and J. R. Rose. 1976. Food and feeding relationships in the benthic and demersal fishes of the Gulf of Alaska and Bering Sea. In Volume 7: Environmental assessment of the Alaskan continental shelf, p. 471-508. An- nual Report RU 0284.7. U.S. Dep. Commer. and U.S. Dep. Inter- ior Environmental Research Laboratories, Boulder, CO. We used linear discriminant functions to identify tentively the habitat characteristics of juvenile flat- fishes with data collected in August 1991 along east and south Kodiak Island (Norcross et al., 1995). In this study, we repeated the linear discriminant func- tion analysis with combined 1991-92 data to include observations from a much wider geographic area around the entire island of Kodiak collected in Au- gust 1992. We refined our previous habitat models by using tree-based regression methods (Venables and Ripley, 1994) on catch-per-unit-of-effort data. Materials and methods Sample collections Two cruises were conducted in the nearshore waters of Kodiak Island, Alaska, during August 1992 (Fig. 1). These cruises were similar to, but covered more area than, two cruises conducted in August 1991 (Norcross et al., 1995; Norcross et al.3). Cruise KI9201 consisted of collections taken with a 7.3-m skiff from Kalsin, Middle, and Womens Bays near the town of Kodiak during 9-14 August 1992. Because these bays were sampled with a skiff, extremely shallow collec- tions could be made. Collections ranged in depth from 1 to 60 m. Ten stations were occupied in Kalsin Bay, six stations in Middle Bay, and five stations were occupied within Womens Bay. Kalsin and Middle Bays were also sampled during August 1991. Immediately following the sampling from the skiff, a counterclockwise circuit of Kodiak Island was com- pleted aboard a 24.7-m chartered trawling vessel (FV Big Valley , cruise KI9202). Collections during KI9202 were made from 16 to 29 August 1992 and ranged in depth from 5 to 180 m. Areas sampled in 1992, but not sampled in 1991, included 52 stations in bays on the north and west sides of the island. Collections were also made at 41 stations off south Kodiak, Sitkalidak Strait, and in Ugak Bay, which were sampled during August 1991. Sampling gears, vessels, and vessel operators were the same in both 1991 and 1992 (Norcross et al., 1995; Norcross et al.3; Norcross et al.4). At each station one sediment sample was collected with a 0.06-m3 Ponar grab for analysis of grain size, and a portable con- ductivity, temperature, and depth (CTD) profiler was deployed to measure temperature and salinity. Fishes 3 Norcross, B. L., B. A. Holladay, and M. Frandsen. 1993. Re- cruitment of juvenile flatfish in Alaska, phase 1. Final Con- tract Report, NOAA NA-16FD0216-01, 504 p. 4 Norcross, B. L., B. A. Holladay, F.-J. Muter, and M. Frandsen. 1994. Recruitment of juvenile flatfish in Alaska, phase 2. Fi- nal Contract Report, NOAA NA-26FDQ156-01, 653 p. 506 Fishery Bulletin 95(3), 1997 were collected on rising tides during daylight hours by using a modified 3.7-m plumb staff beam trawl with a double tickler chain (Gunderson and Ellis, 1986). Tows of 10-min duration were made at the ap- proximate speed of 0. 5-1.0 kn from both the skiff and the trawler. Sample processing Substrate type, water depth, bottom temperature, bottom salinity, and distance from the mouth of the nearest bay were evaluated for each station. Sedi- ment samples were analyzed, as in 1991, by means A Rock sole B Flathead sole c Pacific halibut D Yellowfin sole Figure 1 Distribution (CPUE) of (A) age-0 rock sole, (B) age-0 flathead sole, (C) age-0 Pacific halibut, and (D) age-1 yellowfin sole in August 1991 and 1992. Norcross et a I.: Habitat models for juvenile pleuronectids 507 of a simplified sieve and pipette procedure to obtain the percents of gravel, sand, and mud (Norcross et al., 1995). Distance from the mouth of the bay was used as a relative index of fish distribution with respect to sta- tion position within or outside the bay. Distance from each station to the nearest position at the mouth of a bay was calculated by drawing a line on a chart across the bay mouth between the two outermost capes. The shortest distance from the station to any position on this line was measured. Stations inside the mouth were designated as positive distances, and stations outside of bays were assigned negative dis- tances. The narrowest point of Sitkinak Strait was considered the “mouth” of the bay; stations to the west of that point were considered within the bay and the exposed stations in the open ocean on the east side of Sitkinak Strait were considered outside the mouth. Flatfishes were identified, and total length (mm) was measured in the field with a Limnoterra elec- tronic, digital fish-measuring board. Ages of flatfishes captured in August 1992 were estimated with 1) length-frequency plots of fishes collected August 1992 (Norcross et al.4), 2) length-frequency plots (Norcross et al.3) and analysis of regional differences in total lengths (Norcross et al., 1995) of fish caught during August 1991, and 3) available literature (Southward, 1967; Best, 1974, 1977; Walters et al., 1985; Harris and Hartt1; Blackburn and Jackson5). Fish lengths were used to separate age classes of juvenile flat- fishes. Catch per unit of effort (CPUE) based on a 10-min tow time was calculated for age-0 and age-1 individuals of each species. Habitat models were de- veloped for the most abundant species and age-class combinations. Statistical analyses Linear discriminant function analysis of combined 1991-92 data included the broad range of conditions sampled around Kodiak Island. Canonical loadings of each variable and misclassification rates based on cross-validation were evaluated as outlined in Norcross et al. (1995) to test whether the same pa- rameters had been selected as the best discrimina- tors as those that had been selected solely on 1991 data. The magnitude of the canonical loading of each variable in the discriminant analysis is a measure of the importance of that variable in separating the sta- 5 Blackburn, J. E., and P. B. Jackson. 1982. Seasonal compo- sition and abundance of juvenile and adult marine finfish and crab species in the nearshore zone of Kodiak Island’s east side during April 1978 through March 1979. In Outer continental shelf environmental assessment program, p. 377-570. U.S. Dep. Commer., Final Reports of Principal Investigators 54. tions with (presence) and without (absence) the fish species under consideration. The success of each com- bination of variables in assigning a new station to the presence or absence group can be evaluated by using misclassification rates from cross-validation. The combined 1991 and 1992 data were further used to calculate Spearman’s rank correlation (rho) between the abundance of each fish species and each physical parameter. The significance of rank corre- lations was evaluated at the 95% level. The nonpara- metric test with Spearman’s rho was chosen because of non-normality of the CPUE data (even after trans- formation) and because of the high sensitivity of the parametric correlation coefficient (Pearson’s r) to outliers. To maintain an overall confidence level of 95%, a Bonferroni-adjusted critical level of a = 0.025/ 28 = 0.001 was used for the two-tailed test and for 28 comparisons (4 species x 7 variables). To refine our previous habitat models, which were based primarily on presence or absence data (Norcross et ah, 1995), we used regression trees to model CPUE as a function of habitat parameters. We used the same parameters as in the discriminant analysis, except instead of percentages of gravel, sand, and mud in the substrate, we used a categori- cal description of sediment type based on Folk (1980), i.e. sand (S), mud (M), gravel (G), and the modifiers of these substrates, such as sandy mud (sM), sandy gravelly mud (sgM), etc., 12 categories in all. This categorical classification avoided problems with high correlations among the three sediment variables. Both continuous and categorical predictor variables can easily be accommodated in regression trees. The regression tree used the logarithm of CPUE (log(CPUE+D) as the response variable and depth, distance from mouth of bay, bottom temperature, bottom salinity, and sediment type as predictor vari- ables. A regression tree progressively splits stations on the basis of their values for one of the predictor variables until a leaf or terminal node is reached. Each leaf gives a predicted value of the response variable for the stations assigned to the leaf. The fit of the model is measured by the deviance, which is defined as D = I.(y.-^.])2, or the sum of the squared differences between y; = log(CPUE-fl) at each station i and H[,j = the mean for all stations i at a leaf. The deviance is defined for the entire tree, as well as for each leaf, and is the analogue of the sum of squares in regression mod- els. Each successive partitioning of the data reduces the deviance. For noisy data, the regression tree may overfit the data, resulting in an overly complex tree 508 Fishery Bulletin 95(3), 1997 (Venables and Ripley, 1994). Therefore the initial tree was pruned to an optimum number of terminal nodes as determined by cross-validation. Cross-validation as implemented in S-Plus (Ven- ables and Ripley, 1994) uses 90% of the data as a training set to grow the tree and test it on the re- maining 10%. This procedure is repeated 10 times with nonoverlapping test sets. Predictions on the test set are done for the initial tree as well as for trees pruned to smaller sizes. The resulting deviances are computed and plotted against tree size. Deviances typically are minimized at an intermediate tree size. We chose as optimum tree size the largest size be- fore a marked increase in deviance occurred. Results Rock sole was the most abundant flatfish captured in our 1992 sampling (67% of flatfish), as in 1991 (51% of flatfish). In 1992, a total of 4,625 age-0 rock sole (17-60 mm TL) were collected across almost all locations, with the highest CPUE in the Sitkinak Strait region (Fig. 1A). Age-0 rock sole were found mainly near the mouths of bays ± 8-10 km, except for a single large catch at the head of Uyak Bay. Age-0 rock sole were somewhat more abundant with in- creasing depth between 0 and 30 m, and were col- lected in high numbers to 70 m, although they were also found deeper than 70 m. Age-0 rock sole were collected in large numbers between 7.5°C and 9.5°C and were most often found at salinities of 32.5-33.0 psu (Norcross et al.4). Rock sole were predominantly distributed on sand and mixed sand substrates. Al- though found at almost all combinations of depth and sand, rock sole were somewhat more concentrated in shallow, sandy locations (Fig. 2A). Spearman’s rank correlation coefficients (Table 1) indicated that rock sole abundance was positively correlated with percent sand in the substrate and negatively corre- lated with depth, distance from mouth of bay, gravel, and mud. Rank correlation was highest with percent sand in the substrate. Flathead sole increased from 12% of the 1991 catch to 18% of the 1992 catch. We captured 1,079 age-0 flathead sole (23-52 mm TL) during 1992. The dis- tribution of flathead sole was more restricted than that for rock sole. Age-0 flathead sole were found al- most everywhere around the island but were found in reduced numbers in Southeast Kodiak (Fig. IB). A 0 04 £ ° CL <9 <1> Q o o \ % O o 0 (8 ' ® ^,.991^1 . : , o • . © % © o cm •*. o ° o° • ? . x>° ® cP-* • • nW'o# • o o co . o • ’ * ° ° .98, -o *0 B O 04 O <9 0 0 v ° o°9 0 0 . *' • . O X 0 . 0 O 00°- g? -oo 80 a, „ 0 0 0 ^ ~ 00 0 0 0 o° n 8 § . 0 0 ^ 0 q 0 0 O 0 20 40 60 80 100 0 20 40 60 80 100 Percent sand Percent mud c o 04 ? °-'v' •••'. ' • D O 04 • Q)Qd 0 R 0 0 °o .•9^0 • 80o ^ «9n|& -C o Q. <9 Q) o O O . * A • . • ° ° . " • * O <9 0 0 r • • • • -10 0 10 20 30 40 0 20 40 60 80 100 Distance (km) Percent sand Figure 2 Presence (circles) and absence (dots) of (A) age-0 rock sole, (B) age-0 flathead sole, (C) age-0 Pacific halibut, and (D) age-1 yellowfin sole plotted against the two “best” discriminator variables. The depth axis is plotted with a 120-m limit, and data points occurring between 120 and 180 m are plotted at 120 m. Norcross et al.: Habitat models for juvenile pleuronectids 509 They were concentrated mainly in central, deep ar- eas of bays at depths of 80-120 m, 6.0-9.0°C, 31.5- 33.5 psu, on mud or mixed mud substrates (Norcross et al.4). High abundances of flathead sole were asso- ciated with deep stations, low temperatures, high salinities, low sand content, and high mud content; the highest rank correlations for flathead sole were obtained for depth and mud (Table 1). Flathead sole were predominantly collected in depths > 40 m, ex- cept on substrates with a high mud content (Fig. 2B). Pacific halibut composed 5% of the catch in 1991 and 7% in 1992. During 1992, 627 age-0 halibut (22 — 84 mm TL) were found in exposed sites at all loca- tions on the east and south sides of Kodiak Island (Fig. 1C). In northwestern Kodiak, halibut were col- lected only at the mouth of Uyak Bay. Age-0 halibut were found mainly at 10-70 m depth, 7.0-10.5°C, 32.0-33.0 psu, on mixed sand substrates, outside of or within 7 km of bay mouths (Norcross et al., 1995). Pacific halibut abundances were positively correlated with temperature and sand content and negatively correlated with depth, distance from mouth of bay, and mud content in the substrate. The highest rank correlations were with sand and mud (Table 1). Un- like rock sole, halibut were seldom found in water deeper than 50 m. Halibut juveniles, like rock sole, were concentrated most often in shallow waters with sandy substrate, near or outside mouths of bays (Fig. 20. Yellowfin sole was very abundant in 1991, com- posing 28% of captured flatfishes, but this species represented only 4% of the 1992 total catch. Unlike the other three species examined, in which age-0 fish predominated, age-1 yellowfin sole (41-105 mm TL) were analyzed in both 1991 and 1992 because of the small number (n= 4) and size (15-20 mm TL) of age-0 Table 1 Spearman’s rank correlation coefficients between CPUE of four flatfish species and environmental parameters with 1991 and 1992 data combined. * indicates significance at an overall 5% confidence level. Parameter Rock sole Flathead sole Pacific halibut Yellowfin sole Depth -0.258* 0.644* -0.284* -0.369* Distance -0.308* -0.074 -0.314* 0.204 Temperature 0.193 -0.467* 0.346* 0.212 Salinity -0.083 0.246* -0.163 -0.192 Gravel -0.240* -0.219 -0.078 -0.168 Sand 0.583* -0.219 0.449* 0.113 Mud -0.310* 0.540* -0.417* 0.188 yellowfin sole collected during the second year. Dur- ing 1992, 268 age-1 yellowfin sole were collected at depths less than 40 m, mainly between 5 and 30 m. Age-1 yellowfin sole were found near the heads of bays, in warm (9.0-11.5°C) saline (31.0-33.5 psu) water (Norcross et al., 1995). They were collected on sandy mud, gravelly muddy sand, and muddy sand. Unlike rock sole and halibut, yellowfin sole were col- lected in the inner reaches of bays around Kodiak Island (Fig. ID). The only significant correlation be- tween yellowfin sole abundance and an environmen- tal variable was a negative rank correlation with depth (Table 1). Yellowfin sole were never found deeper than 50 m and were always on mixed substrate, i.e. not predominantly on one grain size (Fig. 2D). Linear discriminant function analysis for the com- bined 1991-92 data resulted in depth having the highest correlation with discriminant scores (canoni- cal loadings) for flathead sole and yellowfin sole (Table 2). Sand was most highly correlated with the discriminant scores for rock sole and Pacific halibut. For all species, except flathead sole, the three high- est canonical loadings were obtained for depth, tem- perature, and sand. In the case of flathead sole, mud was more highly correlated with the discriminant score than was sand. Sand was clearly a good predictor for rock sole pres- ence and was included in the habitat model for rock sole. Depth and temperature performed equally well in the discrimination owing to their high (negative) correlation. However, although rock sole abundance was significantly correlated with depth, the correla- tion with temperature was not significant. Therefore, sand and depth seemed to be the most important variables determining rock sole distribution (Fig. 2A). The three best predictor variables for flathead sole were depth, gravel, and mud. Of these, depth and mud resulted in the lowest total error rates. Because Table 2 Canonical loadings from linear discriminant function analysis for combined 1991 and 1992 flatfish data. Parameter Rock sole Flathead sole Pacific halibut Yellowfin sole Depth -0.557 -0.776 -0.620 -0.696 Distance -0.379 -0.011 -0.501 0.234 Temperature 0.474 -0.597 0.647 0.545 Salinity 0.180 -0.225 -0.026 -0.005 Gravel -0.453 0.321 -0.249 -0.377 Sand 0.783 0.220 0.655 0.406 Mud -0.391 -0.624 -0.473 -0.099 510 Fishery Bulletin 95(3), 1 997 these variables also had the largest rank correlations with abundance (Table 1), they were likely to be the most important parameters for flathead sole distri- bution. The error rates for predicting absence of flat- head sole were consistently much lower than those for predicting presence. Pacific halibut presence or absence could be most accurately predicted by using either depth or tem- perature with either distance or sand. Halibut abun- dance had a higher rank correlation with tempera- ture than with depth and a higher correlation with sand than with distance from the mouth of the bay (Table 1). It is difficult to evaluate the relative im- portance of depth and temperature and of sand and distance owing to high correlations among these vari- ables (Fig. 3). The depth-temperature factor ex- plained most of the observed distribution. The error rates for predicting presence or absence changed sig- nificantly only if both depth and temperature were excluded. Error rates for stations where Pacific hali- but were present were consistently much lower than those for stations where no halibut were found, thus this species appears to be strongly associated with specific habitat characteristics. The three best predictors for yellowfin sole were depth and gravel combined with either sand or tem- perature. Of these, depth and gravel resulted in the lowest total error rates. Only depth was significantly correlated with yellowfin sole abundance (Table 1). The sediment parameters added very little informa- tion because yellowfin sole occurred over a wide range of substrate types. Error rates for stations where yellowfin sole were present were much lower than those for stations where this species was absent, re- flecting the restricted depth range within which yel- lowfin sole were encountered. Presence and absence patterns for all four species are plotted against the two best discriminator variables (Fig. 2). Regression trees were constructed by using CPUE for each species to refine our habitat models. The initial trees were allowed to grow, provided the num- ber of stations in a node was five or greater. The re- sultant regression trees had sizes of 22 terminal nodes for rock sole, 16 for flathead sole, 19 for Pa- cific halibut, and 18 for yellowfin sole. The total deviances for the initial trees were 1.24, 0.57, 0.38, and 0.44 respectively, indicating that the model fit- ted for rock sole was much poorer than that for the other species and that the tree for Pacific halibut had the best fit. The trees for all species seemed to overfit the data, as indicated by cross-validation. Plots of deviance against tree size (number of terminal nodes) for the four flatfish species indicated that deviance was usu- ally at a minimum at very small tree sizes, consist- ing of only two or three nodes (Fig. 4). The deviance for each species tended to increase steeply at a tree size between 4 and 6 nodes, and we chose the largest size before a steep increase as optimum size for the tree. The initial tree was pruned to six terminal nodes for rock sole and halibut and to four terminal nodes for flathead sole and yellowfin sole. The pruned regression tree for rock sole indicated that sediment, depth, and temperature were the best predictor variables for rock sole CPUE. The deviance of the pruned tree increased to 1.852 from 1.242 for the initial tree. This relatively poor fit may again be due to the widespread distribution of rock sole, a species that does not seem to be limited to any par- ticular habitat type. Stations were first separated by sediment type into 89 stations on sand or muddy sand with a high mean CPUE (18 fish/10-min tow) and 80 stations on other sediment types that had a much lower mean CPUE (1.6 fish/10-min tow) (Fig. 5). The highest mean CPUE (25 fish/10-min tow) oc- curred at stations on sand or muddy sand which had a bottom temperature of more than 8.7°C. The colder stations on sand and muddy sand were separated into seven low salinity stations with low mean CPUE (0.58 fish/10-min tow) and 10 high salinity stations with medium to high CPUE (11 fish/10-min tow). Most stations on other sediment types, which in- cluded gravel, mud, gravelly mud, gravelly sand, gravelly muddy sand, gravelly sandy mud, muddy gravel, muddy sandy gravel, sandy gravel, and sandy mud, had low CPUE values except for a group in shallow water (<27.5 m) on gravelly muddy sand, sandy gravel, or sandy mud (13 fish/10-min tow). Thus, by combining results from the correlation analysis, presence and absence patterns, and regres- sion trees, rock sole were found to be most common on sand or mixed sand substrates and most abun- dant in shallow and relatively warm water. The regression tree for flathead sole indicated that temperature, sediment type, and depth were the best predictors of flathead sole abundance. The deviance of the pruned tree was 0.774 compared with 0.569 for the initial tree. Highest CPUE values tended to occur at stations where bottom temperature was less than 8.9°C (Fig. 6). At warmer stations, mean CPUE of flathead sole was very low (0.17 fish/10-min tow) if stations were less than 48 m deep, which was the case for the majority (n=109) of the stations. Mean CPUE at warm stations was higher, however, for the six stations located in water deeper than 48 m (4.6 fish/10-min tow). Stations with bottom temperatures below 8.9°C had a low flathead sole CPUE if the sedi- ment was categorized as gravel, sand, muddy sandy gravel, or sandy gravel (1.6 fish/10-min tow). The CPUE was much higher on pure mud or mixed mud Norcross et al.: Habitat models for juvenile pleuronectids 51 I sediments at low temperatures (8.0 fish/10-min tow). Thus, the highest CPUE values for flathead sole were on mixed mud sediments at stations with a bottom temperature of less than 8.9°C, as well as at warmer stations if they were deeper than 48 m. This sug- gests that temperature should be used in addition to 512 Fishery Bulletin 95(3), 1 997 C aJ > a> O Rock sole Flathead sole 19.0 8 1 5.2 2.4 7.20 1.70 0.95 Pacific halibut 870 4.10 094 Yellowfin sole 89 5.1 2.0 Tree size (number of terminal nodes) Figure 4 Cross-validation plots (deviance versus tree size) used to prune regression trees of catch per unit of effort on five habitat parameters. See text for further explanation. sediment and depth selected by linear discriminant analysis as an important factor in determining the distribution of juvenile flathead sole. Distance from the mouth of the bay and depth were the best predictors of halibut CPUE (Fig. 7) The de- viance of the pruned tree was 0.587 compared with 0.384 for the initial tree. Highest CPUE values oc- curred at stations less than 40 m deep and more than 2.9 km outside the mouth of bays (10 fish/10-min tow). Very low abundances or no halibut were found at stations more than 7.9 km up the bay (0.13 fish/ 10-min tow). Intermediate CPUE values were found at 61 stations near the mouth of bays (-2.9 km to 7.9 km) which had high bottom temperature (>9.0°C) on sand or mixed sand substrates (2.9 fish/10-min tow). This confirmed our earlier finding that halibut tend to remain outside or near the mouth of bays in water less than 40 m deep on sandy substrates. The most important variables used in predicting yellowfin sole abundance were depth, sediment, and distance. Deviance was increased from 0.442 for the initial tree to 0.895 for the pruned tree. The first split separated 100 stations less than 28 m deep from 69 stations deeper than 28 m (Fig. 8). The deeper sta- tions had a very low mean CPUE (0.17 fish/10-min tow), and yellowfin sole were absent at 64 of the 69 stations that were deeper than 28 m. The shallow stations had a low mean CPUE (0.82 fish/10-min tow) if the substrate type was pure gravel, sand, or mud, or had mixed gravel sediment, whereas stations on mixed mud sediments had medium to high abun- dances of yellowfin sole (9.0 fish/10-min tow). The 53 shallow stations on mixed mud substrates were further split by distance, indicating that the highest CPUE values occurred near the heads of long bays. Thus yellowfin sole tended to be concentrated in very shallow locations on mixed mud sediments near the head of bays. This finding agreed with results of the linear discriminant function in its identification of depth and sediment as important factors, but fur- ther added distance from the bay mouth as a third important factor. Norcross et a I.: Habitat models for juvenile pleuronectids 513 Discussion Our results show that relations among flatfish dis- tributions and habitats found within the geographic restrictions of eastern Kodiak Island in 1991 can be applied more broadly to other areas around Kodiak Rock sole Figure 5 Pruned regression tree for age-0 rock sole with predictions of log(CPUE+l) at each terminal node, and number of stations in parentheses (top figure). Bottom figure contains box plots of log(CPUE+l) by terminal node. Each box plot summarizes distribution of CPUE at stations rep- resented by the terminal node directly above it. Box plots indicate median (filled circle), interquartile range (box height), and outliers (open circles). Whiskers indicate upper quartile plus 1.5 times interquartile range and lower quartile minus 1.5 times interquartile range. Width of boxes is proportional to square root of number of stations at that node. See text for sediment abbreviations. 514 Fishery Bulletin 95(3), 1 997 Island, i.e. to those areas sampled during 1992. Two groups of variables explain much of the observed dis- tribution. These variables are substrate composition and a depth-temperature factor. The relative impor- tance of depth and temperature or of gravel, sand, and mud is difficult to assess because each group is Flathead sole temperature <8.9 deg.C > 8 9 deg.C in + LU O a. o cn o ■sf co CM O 1 2 3 4 Node Figure 6 Pruned regression tree for age-0 flathead sole with predictions of log(CPUE+l) at each terminal node, and number of stations in parentheses (top figure). Bottom figure contains box plots of log(CPUE+l) by terminal node. Each box plot summarizes distribution of CPUE at stations rep- resented by the terminal node directly above it. Box plots indicate median (filled circle), interquartile range (box height), and outliers (open circles). Whiskers indicate upper quartile plus 1.5 times interquartile range and lower quartile minus 1.5 times interquartile range. Width of boxes is proportional to square root of number of stations at that node. See text for sediment abbreviations. Norcross et a I.: Habitat models for juvenile pleuronectids 515 highly intercorrelated (Norcross et al., 1995; Fig. 3). These parameters have been linked to the habitat quality of juvenile flatfishes in many other locations (Gibson, 1994). Larvae of many flatfish species are known to settle either in shallow water (Edwards and Steele, 1968; Lockwood, 1974) or offshore water Pacific halibut distance <7.9 km >7.9 km Node Figure 7 Pruned regression tree for age-0 Pacific halibut with predictions of log(CPUE+l) at each termi- nal node, and number of stations in parentheses (top figure). Bottom figure contains box plots of log(CPUE-t-l) by terminal node. Each box plot summarizes distribution of CPUE at stations rep- resented by the terminal node directly above it. Box plots indicate median (filled circle), interquartile range (box height), and outliers (open circles). Whiskers indicate upper quartile plus 1.5 times interquartile range and lower quartile minus 1.5 times interquartile range. Width of boxes is proportional to square root of number of stations at that node. See text for sediment abbreviations. 516 Fishery Bulletin 95(3), 1997 and then to move into shallow water as age-0 juve- niles (Gibson, 1973; Lockwood, 1974; Tanaka et al., 1989). In prior studies (Gibson, 1994) as well as this one, depth and its effect on water temperature may play an important part in determining distribution of juveniles. Water temperature affects growth and Yellowfin sole depth <28 m G,M,S,gS,msG,sG gM,gmS,mG,mS,sWI >28 m 0.16 (69) 0.60 (47) distance <32 km >32 km 2.05 4.25 (47) (6) Node Figure 8 Pruned regression tree for age-1 yellowfin sole with predictions of log(CPUE+l) at each terminal node, and number of stations in parentheses (top figure). Bottom figure contains box plots of log(CPUE+l) by terminal node. Each box plot summarizes distribution of CPIJE at stations rep- resented by the terminal node directly above it. Box plots indicate median (filled circle), interquartile range (box height), and outliers (open circles). Whiskers indicate upper quartile plus 1.5 times interquartile range and lower quartile minus 1.5 times interquartile range. Width of boxes is proportional to square root of number of stations at that node. See text for sediment abbreviations. Norcross et al. : Habitat models for juvenile pleuronectids 517 feeding rates, and shallow, warm waters promote faster growth (Malloy and Targett, 1991; van der Veer et al., 1994). Distribution of juvenile flatfishes has been linked to substrate type (Tanda, 1990; Kramer, 1991; Gibson and Robb, 1992). Juvenile flatfishes appear to avoid coarse sediments (Moles and Norcross, 1995) and choose fine-grained sediments (Rogers, 1992; Keefe and Able, 1994) which vary in size from mud (Wyanski, 1990; van der Veer et al., 1991) to sand ( Jager et al., 1993). In laboratory tests, rock sole pre- fer sand and mixed sand substrates, halibut prefer a combination of mud and fine sand, and yellowfin sole prefer mud and mixed mud sediments (Moles and Norcross, 1995); these findings are in agreement with the classification and regression trees of our study. Choice of settlement location is affected by the abil- ity of a fish to bury itself in the sediment (Gibson and Robb, 1992) as well as by the availability of prey in the substrate (Burke et al., 1991). When diets of juveniles of the four species were examined from the same collections in 1991 that were used in these models, it was found that epibenthic crustacean taxa composed most of the diets (Holladay and Norcross, 1995). Stomach contents were related to physical parameters of capture, including location, depth, and substrate. When distribution of juveniles overlapped, dietary overlap was sometimes reduced, in that one or more groups of flatfishes appeared to alter their feeding (Holladay and Norcross, 1995), i.e. pref- erence for specific prey types did not appear to be a primary factor governing distribution of these species. A discriminant analysis was employed in this study to test whether stations could be accurately classi- fied into groups defined by the presence or absence of a given flatfish species. The classification based on the observed parameters resulted in relatively high error rates for all species; between one-sixth and one-third of the stations were incorrectly classified. Although no discrimination method is able to pre- dict perfectly the presence or absence of populations that have a gradation in abundance in marginal habi- tats, there are several possible reasons for the ob- served high error rates found in this study. For rock sole, halibut, and yellowfin sole, error rates for pre- dicting presence were generally much lower than error rates for predicting absence. This finding may indicate that these species were mostly confined to relatively well-defined depth-substrate characteris- tics. The high misclassification rate for predicting absence of rock sole, halibut, and yellowfin sole sug- gests that many stations may offer suitable depth, temperature, and substrate conditions for these spe- cies but that the species are not collected there be- cause their physical habitat preferences may be dif- ferent. The situation is different for flathead sole; their presence is not as predictable as their absence. Flathead sole are generally absent from shallow ar- eas with little mud, whereas they are usually, but not always, present in deep, muddy places. The classification results suggest that although the seven environmental variables (%sand, %mud, %gravel, depth, temperature, salinity, and distance from bay mouth) used in our discriminant analysis do not account fully for observed flatfish distribu- tions, they do provide a useful first step at defining juvenile flatfish habitat near Kodiak. The initial lin- ear discriminant function models developed with the 1991 data (Norcross et al., 1995) are still applicable after incorporating 1992 data. Similar linear dis- criminant methods have been used to examine nurs- ery grounds of Solea solea (Rogers, 1992). Regression trees of CPUE for each species gener- ally agree with the results of the linear discriminant analyses. They determine specific values of the physi- cal parameters as related to the abundance of juve- nile flatfishes and, as easily comprehensible dia- grams, can be used to predict species abundance based on habitat parameters. This detailed analysis, based on CPUE and incor- porating both 1991 and 1992 data, does not disagree with the original models that we were testing (Norcross et al., 1995) but rather refines those mod- els and incorporates actual abundances (CPUE) in the multivariate analysis. The previous models char- acterized nursery areas of age-0 rock sole, flathead sole, Pacific halibut, and age-1 yellowfin sole on the basis of correlations and discriminant analyses by using presence or absence for 1991 data. Depth and substrate were statistically significant variables pre- viously, and a measure of distance in relation to mouth of the bay was included qualitatively for each species. Depth, temperature, sediment composition, and distance from bay mouth were all found to be important predictors of the abundance of juvenile pleuronectids with regression trees for the combined 1991 and 1992 data. Additional factors influence the presence or ab- sence of these flatfish species at any given site. Pos- sible factors that were not included in this study are additional measures of location (such as position around the island or distance from shore), abun- dances of prey or predators, and a substrate or habi- tat parameter that would account for microhabitat features not reflected in sediment composition. A location parameter may be a categorical vari- able that assigns each station to a well-defined geo- graphical area. For example, we observed large dif- ferences in the abundance of halibut and rock sole 518 Fishery Bulletin 95(3), 1997 between the east and west sides of Kodiak Island and among different bays. These differences possi- bly reflect oceanographic conditions that lead to vari- able levels of recruitment into different nearshore areas around Kodiak Island. Habitat models incor- porating geographical and oceanographic infor- mation may help to reveal these mechanisms but would require larger sample sizes than are presently available. The abundance of prey (McIntyre and Eleftheriou, 1968; Minami, 1986; Allen, 1988) and predators (van der Veer et al., 1991; Seikai et al., 1993) may influ- ence the distribution and abundance of flatfish spe- cies but cannot be quantified without extensive sur- veys. Incorporating prey or predator abundance into a general habitat model is therefore probably of little practical use in applying the model to other areas. Postmetamorphic flatfishes in southeastern Alaska (Sturdevant, 1987) and juvenile flatfishes near Kodiak (Holladay and Norcross, 1995) feed prima- rily on small meiofaunal, benthic, and epibenthic crustaceans, including mysids, amphipods, cuma- ceans, and copepods. The diets of flathead sole, Pa- cific halibut, yellowfin sole, and rock sole were dif- ferent in different capture sites, when region, depth, and substrate were the parameters used for the sites. This finding suggests that these species are oppor- tunistic and feed on the prey available in their lo- cale, rather than that they are discriminating, de- termining locale on the basis of prey availability. Additional information is desirable to describe the microhabitat at each station more precisely. During our sampling, we obtained qualitative descriptions of the benthic flora and fauna that were collected at each station and a very broad quantification of the dominant invertebrates that were caught together with the fishes. In the future, we will attempt to con- solidate this information into a categorical “commu- nity descriptor” for each station. This “community descriptor” can then be used as an additional ex- planatory variable in future models. Acknowledgments We wish to thank all those who helped us collect data: Gary Edwards, Jane Eisemann, Bruce Short, Waldo Wakefield, Dave Doudna, and Charlene Zabriskie; and Michele Frandsen for the graphics. Thanks to David Welch and three anonymous reviewers for critically reviewing this manuscript. This research was funded by Saltonstall-Kennedy funds through NOAA grant #NA-16FD0216-01, and the Coastal Marine Institute of the University of Alaska through grant #14-35-0001-30661. Literature cited Allen, L. G. 1988. Recruitment, distribution and feeding habits of young-of-the-year California halibut (Paralichthys californicus) in the vicinity of Alamitos Bay-Long Beach Harbor, California, 1983-1985. Bull. South. Calif. Acad. Sci. 87:19-30. Best, E. A. 1974. Juvenile halibut in the eastern Bering Sea: trawl surveys, 1970-1972. Int. Pac. Halibut Comm. Tech. Rep. 11, 32 p. 1977. Distribution and abundance of juvenile halibut in the southeastern Bering Sea. Int. Pac. Halibut Comm. Sci. Rep. 62, 23 p. Burke, J. S., J. M. Miller, and D. E. Hoss. 1991. Immigration and settlement pattern of Paralichthys dentatus and P. lethostigma in an estuarine nursery ground. North Carolina, U.S.A. Neth. J. Sea Res. 27:393-405. Carlson, H. R., and R. R. Straty. 1981. Habitat and nursery grounds of Pacific rockfish, ‘ Sebastes ’ sp., in rocky coastal areas of southeastern Alaska. Mar. Fish. Rev. 43(4):13-19. de Ben, W. A., W. D. Clothier, G. R. Ditsworth, and D. J. Baumgartner. 1990. Spatio-temporal fluctuations in the distribution and abundance of demersal fish and epibenthic crustaceans in Yaquina Bay, Oregon. Estuaries 13:469-478. Dunn, J. R., and A. C. Matarese. 1987. A review of the early life history of Northeast Pacific gadoid fishes. In D. R. Gunderson and S. Sundby (eds.), Comparative biology, assessment and management of ga- doids from the North Pacific and Atlantic Oceans, p. 163- 184. Fish. Res. (Amst.) 5. Edwards, R., and J. H. Steele. 1968. The ecology of 0-group plaice and common dabs at Loch Ewe. J. Exp. Mar. Biol. Ecol. 2:215-238. Folk, R. L. 1980. Petrology of sedimentary rocks. Hemphill Publish- ing Co., Austin, TX. 182 p. Gibson, R. N. 1973. The intertidal movements and distribution of young fish on a sand beach with special reference to the plaice ( Pleuronectes platessa L.). J. Exp. Mar. Biol. Ecol. 12: 79-103. 1994. Impact of habitat quality and quantity on the recruit- ment of juvenile flatfishes. Neth. J. Sea Res. 32: 191-206. Gibson, R. N., and L. Robb. 1992. The relationship between body size, sediment grain size and the burying ability of juvenile plaice, Pleuronectes platessa L. J. Fish Biol. 40:771-778. Gunderson, D. R., and I. E. Ellis. 1986. Development of a plumb staff beam trawl for sam- pling demersal fauna. Fish. Res. (Amst.) 4:35-41. Hogue, E. W., and A. G. Carey Jr. 1982. Feeding ecology of 0-age flatfishes at a nursery ground on the Oregon coast. Fish. Bull. 80:555-565. Holladay, B. A., and B. L. Norcross. 1995. Diet diversity as a mechanism for partitioning nurs- ery grounds of pleuronectids. In Proceedings of the in- ternational symposium on North Pacific flatfishes, p. 177- 203. Alaska Sea Grant AK-SG-95-04. Jager, Z., H. L. Kleef, and P. Tydeman.. 1993. The distribution of 0-group flatfish in relation to abi- Norcross et a I.: Habitat models for juvenile pleuronectids 519 otic factors on the tidal flats in the brackish Dollard (Ems Estuary, Wadden Sea). J. Fish Biol. 43(A):31-43. Janusz, J. 1986. Biology of walleye pollock (Theragra chalcogramma ) from the Gulf of Alaska. Int. North Pac. Fish. Comm. Bull. 45:247-261. Keefe, M., and K. W. Able. 1994. Contributions of abiotic and biotic factors to settle- ment in summer flounder, Paralichthys dentatus. Copeia 2:458-465. Kendall, A. W., Jr., L. S. Incze, P. B. Ortner, S. R. Cummings, and P. K. Brown. 1994. The vertical distribution of eggs and larvae of wall- eye pollock, Theragra chalcogramma, in Shelikof Strait, Gulf of Alaska. Fish. Bull. 92:540-554. Kramer, S. H. 1991. Growth, mortality, and movements of juvenile Cali- fornia halibut Paralichthys californicus in shallow coastal and bay habitats of San Diego County, California. Fish. Bull. 89:195-207. Krieger, K. J. 1992. Shortraker rockfish, Sebastes borealis, observed from a manned submersible. Mar. Fish. Rev. 54(4):34-37. 1993. Distribution and abundance of rockfish determined from a submersible and by bottom trawling. Fish. Bull. 91:87-96. Krygier, E. E., and W. G. Pearcy. 1986. The role of estuarine and offshore nursery areas for young English sole, Parophrys vetulus Girard, off Oregon. Fish. Bull. 84:119-132. Lockwood, S. J. 1974. The settlement, distribution and movements of 0-group plaice Pleuronectes platessa (L.) in Filey Bay, Yorkshire. J. Fish. Biol. 6:465-477. Malloy, K. D., and T. E. Targett. 1991. Feeding, growth and survival of juvenile summer flounder Paralichthys dentatus : experimental analysis of the effects of temperature and salinity. Mar. Ecol. Prog. Ser. 72:213-223. McIntyre, A. D., and A. Eleftheriou. 1968. The bottom fauna of a flatfish nursery ground. J. Mar. Biol. Assoc. U.K. 48:113-142. Minami, T. 1986. Comparative studies on the early life history of Japa- nese pleuronectine flatfishes. In Indo-Pacific fish biology: proceedings of the second international conference on Indo- Pacific fishes; Tokyo, 29 July-3 August 1986, p. 950. Minami, T., and M. Tanaka. 1992. Life history cycles in flatfish from the northwestern Pacific, with particular reference to their early life histories. Neth. J. Sea Res. 29:35-48. Moles, A., and B. L. Norcross. 1995. Sediment preference in juvenile Pacific flat- fishes. Neth. J. Sea Res. 34:177-182. Miiter, F.-J., and B. L. Norcross 1994. Distribution, abundance, and growth of larval wall- eye pollock, Theragra chalcogramma , in an Alaskan Fjord. Fish. Bull. 92:579-590. Norcross, B. L., B. A. Holladay, and F.-J. Miiter. 1995. Nursery area characteristics of pleuronectids in coastal Alaska, USA. Neth. J. Sea Res. 34:161-175. Parker, K. S. 1989. Influence of oceanographic and meteorological processes on the recruitment of Pacific halibut, Hippoglossus stenolepis, in the Gulf of Alaska. In R. J. Beamish and G.A. McFarlane ( eds. ), Effects of ocean variability on recruitment and an evalu- ation of parameters used in stock assessment models, p. 221- 237. Can. Spec. Publ. Fish. Aquat. Sci. 108. Poxton, M. G., A. Eleftheriou, and A. D. McIntyre. 1982. The population dynamics of 0-group flatfish on nurs- ery grounds in the Clyde Sea area. Estuarine Coastal Shelf Sci. 14:265-282. Poxton, M. B., and N. A. Nasir. 1985. The distribution and population dynamics of 0-group plaice ( Pleuronectes platessa L.) on nursery grounds in the Firth of Forth. Estuarine Coastal Shelf Sci. 21:845-857. Riley, J. D., D. J. Symonds, and L. Woolner. 1981. On the factors influencing the distribution of 0-group demersal fish in coastal waters. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 178:223-228. Rogers, C. W., D. R. Gunderson, and D. A. Armstrong. 1988. Utilization of a Washington estuary by juvenile En- glish sole, Parophrys vetulus. Fish. Bull. 86:823-831. Rogers, S. I. 1992. Environmental factors affecting the distribution of sole (Solea solea) within a nursery area. Neth. J. Sea Res. 29:153-161. Seikai, T., I. Kinoshita, and M. Tanaka. 1993. Predation by crangonid shrimp on juvenile Japanese flounder under laboratory conditions. Nippon Suisan Gakkaishi 59:321-326. Southward, G. M. 1967. Growth of Pacific halibut. Int. Pac. Halibut Comm. Rep. 43, 40 p. Sturdevant, M. V. 1987. The role of meiofauna in the diets and feeding ecol- ogy of postmetamorphic flatfish. M.S. thesis, University of Alaska Juneau, Juneau, AK, 194 p. Swartzman, G., W. Stuetzle, K. Kulman, and M. Powolowski. 1994. Relating the distribution of pollock schools in the Bering Sea to environmental factors. ICES J. Mar. Sci. 51:481-492. Tanaka, M., T. Goto, M. Tomiyama, and H. Sudo. 1989. Immigration, settlement and mortality of flounder ( Paralichthys olivaceus) larvae and juveniles in a nursery ground, Shijiki Bay, Japan. Neth. J. Sea Res. 24:57-67. Tanda, M. 1990 Studies on burying ability in sand and selection to the grain size for hatchery-reared marbled sole and Japa- nese flounder. Nippon Suisan Gakkaishi 56:1543-1548. Toole, C. L. 1980. Intertidal recruitment and feeding in relation to op- timal utilization of nursery areas by juvenile English sole (Parophrys vetulus: Pleuronectidae). Environ. Biol. Fishes 5:383-390. Tyler, A. V. 1971. Surges of winter flounder, Pseudopleuronectes ameri- canus, into the intertidal zone. J. Fish. Res. Board Can. 28:1727-1732. van der Veer, H. W., R. Berghahn, and A. D. Rijndorp. 1994. Impact of juvenile growth on recruitment in flat- fish. Neth. J. Sea Res. 32:153-173. van der Veer, H. W., and M. J. N. Bergman. 1986. Development of tidally related behaviour of a newly settled 0-group plaice (Pleuronectes platessa) population in the western Wadden Sea. Mar. Ecol. Prog. Ser. 31:121-129. van der Veer, H. W., M. J. N. Bergman, R. Dapper, and J. I. Witte. 1991. Population dynamics of an intertidal 0-group floun- der Platichthys flesus population in the western Dutch Wadden Sea. Mar. Ecol. Prog. Ser. 73:141-148. 520 Fishery Bulletin 95(3), 1997 Venables, W. N., and B. D. Ripley. 1994. Modern applied statistics with S-Plus. Springer- Verlag, New York, 462 p. Walters, G. E., G. B. Smith, P. B. Raymore, and W. Hischberger. 1985. Studies of the distribution and abundance of juve- nile groundfish in the northwest Gulf of Alaska, 1980-82, II: Biological characteristics in the extended region. U.S. Dep. Commer., NOAATech. Memo NMFS F/NWC-77, 95 p. Wespestad, V. G., A. M. Shimada, and J. A. June. 1986. Recruitment variations in Pacific cod and its effects in the Bering Sea and Gulf of Alaska. Int. North Pac. Fish. Comm. Bull. 45:287-294. Wyanski, D. M. 1990. Patterns of habitat utilization in 0-age summer floun- der ( Paralichthys dentatus). MS thesis, College of Will- iam and Mary, Gloucester Point, VA, 54 p. Zhang, C. I. 1988. Food habits and ecological interactions of Alaska pla- ice, Pleuronectes quadrituberculatus , with other flatfish species in the eastern Bering Sea. Bull. Korean Fish. Soc. 21:150-160. 521 Daily growth increments in otoliths of juvenile weakfish, Cynoscion regalis: experimental assessment of changes in increment width with changes in feeding rate, growth rate, and condition factor Richard Paperno University of Delaware Graduate College of Marine Studies, Lewes, Delaware 1 9958 Present address: Florida Department of Environmental Protection Florida Marine Research Institute 1220 Prospect St., Suite 285, Melbourne, Florida 32901 E-mail address: rpaperno@winnie.fit.edu Timothy E. Targett University of Delaware Graduate College of Marine Studies, Lewes, Delaware 1 9958 Paul A. Grecay University of Delaware Graduate College of Marine Studies, Lewes, Delaware 1 9958 Present address: Department of Biological Sciences Salisbury State University, Salisbury, Maryland 21801 Abstract .—Laboratory analyses were conducted on age-0 weakfish, Cynoscion regalis , to determine if depo- sition rate of otolith increments was daily and to examine the relation among otolith increment growth, daily feeding rate, specific growth rate, and condition factor. Tetracycline-marked juveniles (n= 58) had a mean deposition rate of 0.98 (0.03 SE) increments/d. Feeding rate significantly affected in- crement width and was positively cor- related with somatic growth rate and condition factor. Increment width re- sponse to changes in ration level was immediate, significant differences oc- curring between day 7 and 14. Mean increment width and specific growth rate were positively correlated (r=0.86). The continuation of otolith growth dur- ing periods of negative fish growth re- flects the conservative nature of otolith growth and the lack of otolith resorp- tion. An established relation between known growth rates of juvenile weak- fish in the laboratory and otolith incre- ment width will allow otolith increment widths to be applied to field samples. Such analyses could be used to exam- ine closely factors affecting growth, survival, and recruitment. Manuscript accepted 27 February 1997 Fishery Bulletin 95:521-529 ( 1997). Growth rates in fishes can be deter- mined by using scales, otoliths, modal analysis, RNA-DNA ratios, and assorted skeletal structures (Bagenal, 1978; Campana and Neilson, 1985; Summerfelt and Hall, 1987). Interpretations from these structures are based upon assumptions that increments within otoliths are added periodically and that the change in thickness of con- secutive rings is proportional to fish length (Campana and Neilson, 1985). Otoliths have also proven to provide an accurate record of fish growth because there has been no evidence of resorption (Degens et ah, 1969; Dunkelberger et al., 1980; Watabe et ah, 1982; Mugiya, 1987), except under extreme physiological stress (Mugiya and Uchimura, 1989). The periodicity of increment ad- dition has been shown to be daily in larval and juvenile fishes (Cam- pana and Neilson, 1982; Hettler, 1984; Schmitt, 1984; Tsukamoto, 1985; Wilson et al., 1987; Tzeng and Yu, 1989; Monaghan, 1993). Otolith microstructure has more recently been used to compare growth of dif- ferent cohorts within a year class (Townsend and Graham, 1981; Warlen, 1982; Methot, 1983; Jones, 1985), examine life history transi- tions (Brothers and McFarland, 1981; Laroche et ah, 1982; Powell, 1982; Miller and Storck, 1982; Vic- tor, 1986; Thresher and Brothers, 1989), estimate mortality and sur- vival (Crecco et ah, 1983; Graham and Townsend, 1985; Neilson and Geen, 1986; Essig and Cole, 1986; Dalzell et ah, 1987; Post and Pranke- vicius, 1987; Rice et ah, 1987), and determine the effects of biotic and abiotic factors on microstructure 522 Fishery Bulletin 95(3), 1997 (Taubert and Coble, 1977; Tanaka et al., 1981; Campana and Neilson, 1982; Neilson and Geen, 1982, 1985; Neilson et al., 1985; Maillet and Checkley, 1991; Moksness, 1992). Growth histories of fish are determined primarily through back calculation by using either a direct proportion or some nonlinear relation between otolith size and fish age (Maciena et al., 1987; Thorrold and Williams, 1989). Recent studies have used the rela- tion between increment width (IW) and growth rate to show growth histories (Maillet and Checkley, 1990; Molony and Choat, 1990; Wright et al., 1990). Key assumptions for this use are that distance between increments is proportional to growth rate and that the increments are produced daily (Beamish and McFarland, 1983). The objectives of this study were to validate the daily periodicity of otolith increments in juvenile weakflsh, Cy noscion regalis, and to describe the re- lation between otolith IW and changes in feeding rate, specific growth rate, and condition factor. Materials and methods Experiment 1 (daily otolith increment validation) Juvenile weakflsh were captured in Delaware Bay in August, 1987, maintained in the laboratory in a recirculating seawater system under a photoperiod of 14 h light/10 h dark at 22°C (0.20 SE), 20 %c, and fed ad libitum on squid. Fifty-eight fish were injected with a 200-mg oxytetracycline hydrochloride/0. 1 mL saline solution and held in recirculating seawater for 26 days. Throughout the 26-d period, between 1 and 5 fish were removed and measured (SL), and their otoliths were removed for analysis. Fish sizes ranged from 68 to 150 mm SL. Experiment 2 (effect of ration level on increment width and specific growth rate) Weakflsh were reared from eggs fertilized in the labo- ratory and raised to the juvenile stage in recirculat- ing seawater (photoperiod=14L/10d at 22°C, 20 %c). Individual fish were held in 20-L circular containers and fed ad libitum for two days to determine maxi- mum ration (pretreatment period). Each day, fish were fed a known weight of live mysid shrimp ( Neomysis americana) in excess of what they could consume. Fish were allowed to feed for 24 hours whereupon uneaten mysids were collected and weighed. Maximum ration was determined to be ap- proximately 20% body weight/d. Experimental treatment rations were randomly assigned on the third day of the experiment. Fish were weighed to the nearest 0. 1 g and randomly as- signed one of six daily rations: 100% maximum ra- tion (MR, n- 5), 90% MR (n= 4), 80% MR (n= 4), 60% MR (n= 4), 40% MR (n= 4), 20% MR (n= 4, Table 1). Feeding levels were also calculated as percentages of body weight for individual fishes. For 14 days, fish were fed daily at these assigned levels; that is to say, they were allowed to feed for 24 h whereupon un- eaten mysids were removed and collectively weighed. Fish were re weighed on day 7, and final weights and lengths were measured on day 14. The absolute weight of the daily feeding level offered (as a per- centage body weight) was adjusted on day 7 to ac- count for growth and maintain ration levels as a func- tion of fish weight. Specific growth rate (SGR) was calculated for each fish as SGR = [(In Wj4 - In W0)/ 14] x 100, where W14 = the wet weight (g) on day 14; W0 = the initial wet weight (g); and 14 = the duration of the treatment period in days. Mean specific growth rates were calculated for each treatment. Fulton’s condition factor (K) at the end of the experiment was calculated for each fish as K = W / L3 x 10,000, where W = the wet weight (g); and L = the standard length (mm). Daily ration (percentage body weight/d) was cal- culated for each fish for each day on the basis of the Table 1 Summary of ration levels. Actual treatment feeding levels and daily ration calculated based on calculated daily fish weights. Feeding levels (% of maximum ration) Daily ration (% body weight/day) Estimated feeding level Actual feeding level Mean Week 1 Week 2 100 100 23.8 26.4 21.3 90 66 15.6 16.1 15.0 80 58 13.9 14.6 13.3 60 46 11.0 11.1 10.9 40 32 7.6 7.7 7.5 20 17 4.1 4.1 4.1 Paperno et a I.: Daily growth increments in otoliths of juvenile Cynoscion regalis 523 weight of mysids consumed (weight of mysids offered minus weight of mysids not eaten) and estimated fish weights (assuming exponential growth between weighing). Mean daily ration was calculated for each feeding level treatment (Table 1). Henceforth, feed- ing will refer to daily ration, whereas treatment lev- els will continue to be referred to as percentage of maximum ration (MR). Because measurements of feeding depended upon the reliability with which uneaten mysids were col- lected, retrieval efficiency was determined. Live mysids, in amounts comparable to the feeding levels described above, were weighed to the nearest mg and placed in all ten containers with recirculating sea- water. After 24 hours, the mysids were retrieved and reweighed. Mean weight of retrieved mysids was 89% (0.017 SE) of initial weight. Therefore, differences between the weight of mysids provided and the weight retrieved was considered to be a useful esti- mate of feeding. Otolith preparation and analysis Otoliths were ground by hand following modified procedures of Neilson and Geen (1981) and Volk et al. (1984). Both sagittal otoliths were embedded in EPON resin. Otoliths were attached to a glass slide with thermoplastic and ground to half their thick- ness across a transverse plane by using a series of 400-600 grit carborundum paper. Otoliths were pol- ished with 0.3 pm alumina oxide paste, reattached to a glass slide with the polished side down, then ground and polished to produce a thin section through the nucleus. All counts and measures were made from the origin along the dorsal edge of the neural groove to the otolith margin. All other transects lacked precision. Tetracycline-marked otoliths were examined with UV light at 400x magnification. Increment counts were made from the fluorescent mark to the edge of the otolith. Each otolith was counted twice, without knowledge of the previous measurement, and con- firmed by an independent counter. Otoliths from ex- periment 2 were examined under 400x magnifica- tion with transmitted light. Mean IW was calculated from three “blind” measurements. We made all counts and measurements with an Olympus Cue 2 Image Analysis System. One pair of otoliths from the 66% ration treatment was not readable and was subse- quently discarded. Statistical analyses Experiment 1 — daily otolith increment valida- tion To validate the daily nature of otolith incre- ment, the regression slope of increment count on day was tested to determine if it differed from one (Stu- dents’ Atest). Outliers were detected by calculating the leverage coefficients and by computing the stan- dard residuals from the regression equation line. Only 2% of the otoliths were reexamined because the leverage coefficient was greater than 4 In and the standard residual was greater than the Avalue for a sample size of n (Sokal and Rohlf, 1981). Experiment 2 — effect of ration level on increment width and specific growth rate Mean IW among ration treatments for increments formed during the pretreatment period was compared with ANOVA (a =0.05), followed by Tukey’s multiple comparison tests (Zar, 1984). Mean IW among ration treatments for weeks 1 and 2 and between weeks within ration treatments was analyzed with two-way ANOVA (a =0.05). Mean specific growth rate for each treat- ment was regressed against mean increment width following confirmation of normality ( Kolmogorov - Smirnov test) and homogeneity of variances (Coch- ran’s C test) with a=Q.Q5 for all treatment levels. Increment width was compared with SGR, daily ra- tion, and Fulton’s K at the end of the experiment for each fish (Pearson product-moment). Regression analysis was used to determine the relation between IW and SGR Oog{G+l}) for each fish. Regression lines were fitted by using a second-order regression against daily ration (Zar, 1984). Results Experiment 1 — daily otolith increment validation The slope of the regression increments on days after injection was not significantly different from one (y=0.975x + 0.825, P>0.05), thus supporting the daily periodicity of otolith increment formation in this spe- cies (Fig. 1). Experiment 2 — effect of ration level on increment width and specific growth rate No differences were found in mean IW among treat- ments during the 2-d pretreatment period. Initially, IW narrowed in the lower feeding levels: 17%, 32%, and 46% maximum MR (Fig. 2). Mean daily IW ranged from a low of 2.3 pm (17% MR or 4.1% body weight/d) during week 2 to a high of 4.5 pm (66% MR or 15.6% body weight/d) during week 1 (Table 2). Mean IW was significantly lower for the 17%, 32%, and 46% MR treatments during the entire 14-d pe- riod as compared with the higher ration treatment 524 Fishery Bulletin 95(3), 1997 (Table 2; Fig. 3). Increments in the higher ra- tion treatments remained relatively wide throughout the experimental period. Narrow- ing of increment width in the 17% and 32% MR treatments ensued immediately and continued to decrease after the first week of the experi- ment (Fig. 2). For all treatments, mean IW was lower during week 2 than week 1; significant differences occurred between weeks in the 17%, 32%, and 66% MR treatments (Table 2). By week 2, there were significant among- treatment differences in mean IW among the 46% MR and the 32% and 17% MR treatments. Mean IW among the higher feeding levels did not differ throughout the entire experiment except for the 66% MR level. During several days of the first week, this group had a significantly higher mean IW compared with other treatments (Table 2). Daily variability in IW was high in all treatments (Fig. 2). There was a positive correlation (IW = 2.58 + 1.49(log{G+l|), r=0.86, P<0.05) between mean daily IW and mean specific growth rate for each treatment (Fig. 4). Although some fish lost weight at the lowest ration level, daily increments continued to be produced. There was a positive correlation between mean IW and mean daily Table 2 Results of two-way analysis of variance and multiple-range tests from comparisons of mean weekly increment width from differ- ent ration treatments and one-way analysis of variance between weeks. MR is the maximum ration. * = P<0.05, ns = not signifi- cant; letters indicate the results of Tukey’s HSD comparison among treatments. SE is the standard error associated with treat- ment means. Treatment feeding Treatment levels (% MR) Week 1 SE Week 2 SE means 17 3.042 0.0243 2.310 0.0378* 2.68“ 32 3.021 0.0465 2.616 0.1203* 2.82“6 46 3.121 0.1402 3.027 0.1495ns 3.076 58 3.823 0.7427 3.817 0.5864ns 3.82“ 66 4.514 0.2835 3.766 0.8348* 4.14“ 100 3.864 1.2942 3.755 0.6493ns 3.81“ Overall 3.538 3.215* 3.38 feeding rate and between mean specific growth rate and K at day 14 (Table 3; Figs. 3 and 4). This relation (P<0.05) developed during week 1 of the experiment and strengthened during week 2 (Table 3). Discussion Figure 1 Relation between otolith increments distal to the fluorescent mark and days after tetracycline injection in juvenile weakfish, Cynoscion regalis. Symbols may represent more than one observation. Injection of oxytetracy cline hydrochloride solu- tion produced clear fluorescent bands in the otoliths of juvenile weakfish, and increment deposition occurred daily. Several other juve- nile sciaenids have shown daily increments: spot ( Leiostomus xanthurus ), red drum ( Sciaenops ocellatus), spotted seatrout (Cynoscion nebulosus), and silver perch ( Bairdiella chrysoura ) ( Gjosaeter et al., 1984; Hettler, 1984; Peters and McMichael, 1987; McMichael and Peters, 1989; Hales and Hurley, 1991). The present study provides vali- dation for ageing juvenile weakfish, thus en- abling estimates of growth and providing, in combination with abundance data, a means of estimating accurate age specific mortality rates during this life history stage. The rapid response of IW to changes in ra- tion and the strong relation with SGR suggest that IW’s may be used to infer growth history. Furthermore, significant differences in IW be- tween high (58-100% MR) and low (17—46% MR) feeding levels suggests that IW may be used to approximate feeding history. Because of approximately a one-week lag time prior to stabilization of IW among treatments, the full magnitude of the change in IW cannot be as- sessed by examining just a few increments. At Paperno et al.: Daily growth increments in otoliths of juvenile Cynoscion regalis 525 least one week is required before IW responses to feeding and growth are statistically detectable al- though the physiological processes that result in IW differences begin acting sooner (Molony and Choat, 1990). Therefore, mean IW taken over several con- secutive days would be most useful for making in- ferences regarding recent feeding and growth his- tory for small sample sizes. The magnitude of variability in IW observed in this study, particularly for the higher rations, has been documented for other species (Volk et al., 1984; Neilson and Geen, 1985; Maillet and Checkley, 1991). The reduced variability under the stress of lower ration may be related to reduced growth and utiliza- tion of food and stored reserves for maintenance (Molony and Choat, 1990). Continuation of otolith increment formation during periods of negative fish growth suggests that otolith growth is conservative and otolith resorption is not likely (Campana and Neilson, 1985; Secor et al., 1989). Table 3 Results of Pearson product-moment correlation analysis (r) between daily increment width, daily feeding rate, and specific growth rate. Increment width r n P< 0.05 Daily feeding rate Week 1 0.5100 168 0.000 Week 2 0.6552 168 0.000 Specific growth rate Week 1 0.3961 168 0.000 Week 2 0.6232 168 0.000 Condition factor 0.3576 336 0.000 Otolith increments of spot, like those of weakfish, were found to have an immediate response to changes in ration (Govoni et al., 1985), although deposition 526 Fishery Bulletin 95(3), 1997 5 : 4 17 32 46 58 66 l 100 4 I i { 3 E _c £ 1 t i Week 1 c 2 § Q) 5 10 15 20 25 30 g « 5 a> 17 32 46 58 66 100 4 i\ 3 \ i1 Week 2 i t t i , i , T , , — i 2 0 5 10 15 20 25 30 Daily feeding rate (%body wt/day) Figure 3 Relation between mean daily otolith increment width and daily feeding rate in juvenile weakfish, Cynoscion regalis, during week 1 and week 2. Error bars represent 95% Tukey’s multiple comparison intervals. Values along the top of each panel are treatment feeding levels expressed as a percentage of maximum daily ration. was found to be less than daily under low ration con- ditions (Siegfried and Weinstein, 1989). However, for the bloater ( Coregonus hoyi) there was a loss of con- trast between the hyaline and opaque bands and no effect on relative increment width (Rice et al., 1985). Maillet and Checkley (1990) found that starved At- lantic menhaden ( Brevoortia tyrannus ) larvae pro- duced narrower increments with IW, increasing dur- ing a 3-6 day recovery period. In contrast to these examples of rapid otolith response to ration, changes in IW in juvenile chum salmon ( Oncorhynchus tshawytscha) and the tropical glass fish ( Ambassis vachelli ) do not become discernible for three weeks and two weeks, respectively (Neilson and Geen, 1985; Molony and Choat, 1990). Recent experimental data suggest that the IW-growth relation may be more complex than originally thought (Reznick et al., 1989; Secor et al., 1989; Francis et al., 1993; Jenkins et al., 1993). The result of these studies suggests that the IW response to feeding and growth is variable, may be of limited use in some species, and needs to be evaluated on a species by species basis. For species in which the relation of IW to growth has been established, otolith increment analysis can provide a means by which an investigator may re- late recent environmental conditions to recent growth history during the important early life stages. Thus, a more complete understanding of the role of the environmental conditions relating to feeding, growth, and ultimately survival may be obtained. Paperno et al.: Daily growth increments in otoliths of juvenile Cynoscion regalis 527 Acknowledgments This study was supported by a grant to T. E. Targett from the U. S. Fish and Wildlife Service, Federal Aid in Fisheries Restoration, Wallop-Breaux Fund (project F-36-R:l-3), through the Delaware Department of Natural Resources and Environmental Control. Literature cited Bagenal, T. 1978. Methods for assessment of fish production in fresh waters. IBP Handbook No. 3, Blackwell Scientific Publ., Oxford, 365p. Beamish, R. J., and G. A. McFarland. 1983. The forgotten requirements for age validation in fish- eries biology. Trans. Am. Fish. Soc. 112:735-743. Brothers, E. B., and W. N. McFarland. 1981. Correlations between otolith microstructure, growth and life history transitions in newly recruited French grunts ( Haemulon flavolineatum). Rapp. P.-V. Reun. Cons. Int. Explor. Mer 178:369-374. Campana, S. E., and J. D. Neilson. 1982. Daily growth increments in otoliths of starry floun- der (Platichthys stellatus) and the influence of some envi- ronmental variables in their production. Can. J. Fish. Aquat. Sci. 39:937-942. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1032. Crecco, V., T. Savoy, and L. Gunn. 1983. Daily mortality rates of larval and juvenile Ameri- can shad ( Alosa sapidissima) in the Connecticut River with changes in year-class strength. Can. J. Fish. Aquat. Sci. 40:1719-728. Dalzell. P., S. Sharma, and J. Prakash. 1987. Preliminary estimates of the growth and mortality of three tuna baitfish, Heklotsichthys quadrimaculatus and Spratelloides delicatulus (Clupeidae) and Rhabdamia gra- cilis (Apogonidael from Fijian waters. Tuna and Billfish Assoc. Prog. Tech. Rep. 20, 15 p. Degens, E. T., W. G. Deuser, and R. L. Haedrich. 1969. Molecular composition and structure of fish otoliths. Mar. Biol. (N.Y.) 2:105-113. Dunkelberger, D. G., J. M. Dean, and N. Watabe. 1980. The ultrastructure of the otolithic membrane and the otolith in the juvenile mummichog, Fundulus hetero- clitus. J. Morphol. 163:367—377. 528 Fishery Bulletin 95(3), 1997 Essig, R. J., and R. F. Cole. 1988. Methods of estimating larval fish mortality by daily increments in otoliths. Trans. Am. Fish. Soc. 115:34-40. Francis, M. P., M. W. Williams, A. C. Pryce, S. Pollard, and S. G. Scott. 1993. Uncoupling of otolith and somatic growth in Pagrus auratus (Sparidae). Fish. Bull. 91:159-164. Gjosaeter, J., P. Dayaratne, O. A. Begstad, H. Gjosaeter, M. I. Sousa, and I. M. Beck. 1984. Ageing tropical fish by growth rings in the otoliths. FAO Fish. Circ. 776, 54 p. Govoni, J., J. Alexander, J. Chester, D. E. Hoss, and P. B. Ortner. 1985. An observation of episodic feeding and growth of lar- val Leiostomus xanthurus in the northern Gulf of Mex- ico. J. Plankton Res. 7:137-146. Graham, J., and D. W. Townsend. 1985. Mortality, growth, and transport of larval Atlantic herring, Clupea harengus, in Maine coastal waters. Trans. Am. Fish. Soc. 114:490-498. Hales, L. S., Jr., and D. H. Hurley. 1991. Validation of daily increment formation in the otoliths of juvenile silver perch, Bairdiella chrysoura. Estuaries 14:199-206. Hettler, W. F. 1984. Marking otoliths by immersion of marine fish larvae in tetracycline. Trans. Am. Fish. Soc. 113:370-373. Jenkins, G. P., M. Shaw, and B. D. Stewart. 1993. Spatial variation in food-limited growth of juvenile greenback flounder, Rhombosolea tapirina: evidence from otolith daily increments and otolith scaling. Can. J. Fish. Aquat. Sci. 50:2558-2567. Jones, C. M. 1985. Within-season differences in growth of larval Atlan- tic herring, Clupea harengus. Fish. Bull. 83:289-298. 1986. Determining age of larval fish with the otolith incre- ment technique. Fish. Bull. 84:91-103. Laroche, J. L., S. L. Richardson, and A. A. Rosenberg. 1982. Age and growth of a plueronectid, Parophrys vetulus, during the pelagic larval period in Oregon coastal waters. Fish. Bull. 80:93-104. Maceina, M. J., D. N. Hata, T. L. Linton, and A. M. Landry Jr. 1987. Age and growth analysis of spotted seatrout from Gavelston Bay, Texas. Trans. Am. Fish. Soc. 116:54-59. Maillet, G. L., and D. M. Checkley Jr. 1990. Effects of starvation on the frequency and width of growth increments in sagittae of laboratory reared Atlan- tic menhaden Brevoortia tyrannus larvae. Fish. Bull. 88:155-165. 1991. Storm-related variation in the growth rate of otoliths of larval Atlantic menhaden, Brevoortia tyrannus : a time series analysis of biological and physical variables and implications for larva growth and mortality. Mar. Ecol. Prog. Ser. 79:1-16. McMichael, R. H., Jr., and K. M. Peters. 1989. Early life history of spotted seatrout, Cy noscion nebulosus (Pisces: Sciaenidae), in Tampa Bay, Florida. Estuaries 12:98-110. Methot, R. D. 1983. Seasonal variation in survival of larval northern an- chovy, Engraulis mordax , estimated from the age distri- bution of juveniles. Fish. Bull. 81:741-750. Miller, S. T., and T. Storck. 1982. Daily growth rings in otoliths of young-of-the-year large mouth bass. Trans. Am. Fish. Soc. 111:527-530. Moksness, E. 1992. Differences in otolith microstructure and body growth rate of North Sea herring (Clupea harengus L.) larvae in the period 1987-1989. ICES J. Mar. Sci. 49:223-230. Molony, B. W., and J. H. Choat. 1990. Otolith increment widths and somatic growth rate: the presence of a time-lag. J. Fish. Biol. 37:541-551. Monaghan, J. P., Jr. 1993. Comparison of calcein and tetracycline as chemical markers in summer flounder. Trans. Am. Fish. Soc. 122:298-301. Mugiya, Y. 1987. Phase difference between calcification and organic matrix formation in the diurnal growth of otoliths in the rainbow trout, Salmo gairdneri. Fish. Bull. 85:395-401. Mugiya, Y., and T. Uchimura. 1989. Otolith resorption induced by anaerobic stress in goldfish, Carassius auratus. J. Fish Biol. 35(61:813-818. Neilson, J. D., and G. H. Geen. 1981. Method for preparing otoliths for microstructure examination. Prog. Fish. Cult. 43:90-91. 1982. Otoliths of chinook salmon ( Oncorhynchus tshawy- tscha):dai\y growth increments and factors influencing their production. Can. J. Fish. Aquat. Sci. 39:1340-1347. 1985. Effects of feeding regimes and diel temperature cycles on otolith increment formation in juvenile chinook salmon, Oncorhynchus tshawytscha. Fish. Bull. 83:91—101. 1 986. First-year growth rate of Sixes River chinook salmon as inferred from otoliths: effects on mortality and age at maturity. Trans. Am. Fish. Soc. 115:28-33. Neilson, J. D., G. H. Geen, and B. Chan. 1985. Variability in dimensions of salmonid nuclei:impli- cations for stock identification and microstructure interpretation. Fish. Bull. 83:81-90. Peters, K. M., and R. H. McMichael Jr. 1987. Early life history of the red drum, Sciaenops ocellatus (Pi- sces: Sciaenidae), in Tampa Bay, Florida. Estuaries 10:92-107. Post, J. R., and A. B. Prankevicius. 1987. Size-selective mortality in young-of-the-year yellow perch (Perea flavescens ): evidence from otolith micro- structure. Can. J. Fish. Aquat. Sci. 44:1840-1847. Powell, A. B. 1982. Annulus formation on otoliths and growth of young summer flounder from Pamlico Sound, North Caro- lina. Trans. Am. Fish. Soc. 111:688-693. Reznick, D., E. Lindbeck, and H. Bryga. 1989. Slower growth results in larger otoliths: an experi- mental test with guppies (Poecilia reticulata). Can. J. Fish. Aquat. Sci. 46:108-112. Rice, J. A., L. B. Crowder, and M. E. Holey. 1987. Exploration of mechanisms regulating larval survival in Lake Michigan bloater: a recruitment analysis based on characteristics of individual larvae. Trans. Am. Fish. Soc. 116:703-718. Schmitt, P. D. 1984. Marking growth increments in otoliths of larval and juvenile fish by immersion in tetracycline to examine the rate of increment formation. Fish. Bull. 82:237-242. Secor, D. H., J. M. Dean, and R. B. Baldevarona. 1989. Comparison of otolith growth and somatic growth in larval and juvenile fishes based on otolith length/fish length relationships. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 191:431-438. Siegfried, R. C., and M. P. Weinstein. 1989. Validation of daily increment deposition in the otoliths of spot (Leiostomus xanthurus). Estuaries 12:180-185. Paperno et al.: Daily growth increments in otoliths of juvenile Cynoscion regalis 529 Sokal, R. R., and F. J. Rohlf. 1981. Biometry: The principles and practice of statistics in biological research. W. H. Freeman and Co., San Fran- cisco, CA, 776 p. Summerfelt, R. C., and G. E. Hall. 1987. Age and growth of fish. Iowa State Univ. Press, 544 p. Tanaka, K., Y. Mugiya, and J. Yamada. 1981. Effects of photoperiod and feeding on daily growth patterns in otoliths ofjuvenile Tilapia nilotica. Fish. Bull. 79:459-466. Taubert, B., and D. W. Coble. 1977. Daily rings in otoliths of three species of Lepomis and Tilapia mossambica. J. Fish. Res. Board Can. 34:332-340. Thorrold, S. R., and D. McB. Williams. 1989. Analysis of otolith microstructure to determine growth histories in larval cohorts of a tropical herring ( Herkotsichthys castelnaui). Can J. Fish. Aquat. Sci. 46:1615-1624! Thresher, R. E., and E. B. Brothers. 1989. Evidence of intra- and inter-oceanic regional differ- ences in the early life history of reef-associated fishes. Mar. Ecol. Prog. Ser. 57:187-205. Townsend, D.W., and J. J. Graham. 1981. Growth and age structure of larval Atlantic herring, Clupea harengus, in Sheepscot River Estuary, Maine, as determined by daily growth increments in otoliths. Fish. Bull. 79:123-130. Tsukamoto, K. 1985. Mass-marking of ayu eggs and larvae by tetracycline- tagging of otoliths. Bull. Jpn. Soc. Sci. Fish. 51:903-911. Tzeng, W.-N., and S.-Y. Yu. 1989. Validation of daily growth increments in otoliths of milkfish larvae by oxytetracycline labeling. Trans. Am. Fish. Soc. 118:168-174. Victor, B. C. 1986. Duration of the planktonic larval stage of one hun- dred species of Pacific and Atlantic Wrasses (family Labridae). Mar. Biol. (Berl.) 90:317-326. Volk, E. C., R. C. Wissmar, C. A. Simenstad, and D. M. Eggers. 1984. Relationship between otolith microstructure and the growth ofjuvenile chum salmon ( Oncorhynchus keta) under different prey rations. Can. J. Fish. Aquat. Sci. 41:126—133. Warlen, M. S. 1982. Age and growth of larvae and spawning time of At- lantic croaker in North Carolina. Proc. Ann. Conf. South- east. Assoc. Fish. Wild]. Agencies 34:204-214. Watabe, N., K. Tanaka, J. Yamada, and J. M. Dean. 1982. Scanning electron microscope observations of the or- ganic matrix in the otolith of the teleost fish Fundulus heteroclitus and Tilapia nilotica. J. Exp. Mar. Biol. Ecol. 58:127-134. Wilson, C. A., D. W. Beckman, and J. M. Dean. 1987. Calcein as a fluorescent marker of otoliths of larval and juvenile fish. Trans. Am. Fish. Soc. 116:668-670. 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. Zar, J. H. 1984. Biostatistical analysis, 2nd ed. Prentice-Hall, Inc. Englewood Cliffs, NJ, 718 p. 530 Daily age and growth of larval and early juvenile Spanish mackerel, Scomberomorus maculatus, from the South Atlantic Bight* John S. Peters Department of Biology, College of Charleston 66 George St., Charleston, South Carolina 29424 E-mail address: petersj@cofc.edu David J. Schmidt Smartech 1725 Signal Pt. Rd., Charleston, South Carolina 29412 Abstract.-Age and growth of lar- val and juvenile Spanish mackerel, Scomberomorus maculatus, were deter- mined by examining increments of daily growth on the otoliths (lapilli) of specimens collected along the south- eastern Atlantic coast, 1983-89. Mar- ginal increment analysis was per- formed on 152 fish (7.4-97.0 mm SL) to validate the deposition of daily rings. A mean standardized marginal incre- ment (SMI) was calculated by compar- ing the width of the marginal increment to the adjacent increment on the lapilli of fish captured over a diel cycle. The distribution of mean SMI was uni- modal. A nonlinear equation was used to model growth (In SL - 6.2 - 55.1/ Age). Based on this growth equation, predicted absolute growth rates for the first 23 days of life were approximately 1.9 mm/day, followed by a surge of rapid growth approaching 5.0 mm/day over the next 17 days. Absolute growth rates subsequent to 40 days of age were 2.1 mm/day. Manuscript accepted 22 November 1996 Fishery Bulletin: 530-539 (1997). The Spanish mackerel, Scombero- morus maculatus (Mitchill), is an inhabitant of the Gulf of Mexico and the Atlantic coast of the United States. During winter months, Spanish mackerel are concentrated in waters off southern Florida. In late spring and summer, however, they are widely distributed along the Atlantic coast to the Gulf of Maine (Klima, 1959; MacEachran et al., 1980; Finucane and Collins, 1986). Most life history studies on Span- ish mackerel have focused on adults from southern Florida and the Gulf of Mexico (Klima, 1959; Powell, 1975; Finucane and Collins, 1986; Fable et al., 1987; Schmidt et al., 1993 ). Except for work by DeVries et al. (1990) on growth rates of larval and early juvenile Spanish and king mackerel (2.8-22.0 mm SL), very little has been done on the early life history of S. maculatus, particularly in the South Atlantic Bight (SAB). Daily growth increments on oto- liths of juvenile scombrids (skipjack, Euthynnus pelamis, and yellowfin tuna, Thunnus albacares, bluefin tuna, T. thynnus, black skipjack, Euthynnus lineatus, Atlantic mack- erel, Scomber scombrus, and south- ern bluefin tuna, Thunnus maccoyii) have been tentatively validated (Uchiyama and Struhsaker, 1981; Radtke, 1983; Wild and Foreman, 1980; Brothers et al., 1983; D’Amours et al., 1990; Jenkins and Davis, 1990; Wexler, 1993). However, no published study has been directed at the vali- dation of daily growth increments on the otoliths of Spanish mackerel. The validation of the consistent periodic deposition of growth rings generally requires that fishes be held in captivity under conditions that approximate the natural envi- ronment. However, Spanish mack- erel larvae and juveniles are diffi- cult to rear in the laboratory. An- other method that has moderate reliability involves demonstrating that initiation of increment forma- tion is synchronous throughout the population (Tanaka et al., 1981; Geffen, 1987; Jenkins and Davis, 1990). If fishes deposit increments in response to external environmen- tal cues of diel periodicity, or an en- dogenous daily rhythm, then indi- viduals experiencing the same envi- ronmental conditions (light, tempera- ture, feeding activity) would be ex- pected to initiate increment deposi- * Contribution 377 from the South Carolina Department of Natural Resources, Charles- ton, South Carolina 29422 and contribu- tion 135 from the University of Charles- ton’s Grice Marine Laboratory, Charleston, South Carolina 29412. Peters and Schmidt: Age and growth of larval and early juvenile Scomberomorus maculatus 531 tion at approximately the same time of day. A review of this approach is presented in Tanaka et al., 1981; Broth- ers and MacFarland, 1981; and Geffen, 1987. The age and growth data used in this paper came from two separate studies, one dealing primarily with larvae and small juveniles less than 100 mm SL, the other with larger young-of-the-year (YOY) juveniles. The primary objectives of this paper are to combine these studies to present a more comprehensive analy- sis of age and growth of larval and juvenile (7-353 mm SL) Spanish mackerel and to validate the daily deposition of increments on their otoliths. Methods Collection and treatment of specimens Past studies attempting to describe the age, growth, and distribution of Spanish mackerel have resulted in the collection of a relatively small number of speci- mens over a limited size range (MacEachran et ah, 1980; Collins and Stender, 1987; DeVries et ah, 1990). Because of the apparent difficulty in capturing Span- ish mackerel larvae and juveniles, we attempted to increase our sample size by pooling ancillary collec- tions of Spanish mackerel from unrelated studies when they became available. This allowed us to use a wide size range of specimens collected over an en- tire diel cycle. Most of the Spanish mackerel larvae and juveniles (7.4-353.8 mm SL) were collected with a 1 m x 2 m neuston net (2.0 mm mesh) from Breach Inlet bridge, near Charleston, SC, during the entire nighttime flood tide (Fig. 1). The sampling effort was designed by the South Carolina Department of Natural Re- sources (SCDNR) to capture larval gag that enter the estuaries during the spring of each year. Span- ish mackerel were obtained from samples that were taken during the month of June from 1986 through Figure 1 Locations of nearshore ichthyoplankton stations sampled during 1988 and 1989 and location of Breach Inlet, off Charleston, SC, where sampling was done for this study. 532 Fishery Bulletin 95(3), 1997 1988. In addition, 25 larval and juvenile Spanish mackerel were collected in ichthyoplankton samples from coastal waters off Charleston, SC, during May- October 1988 and 1989 (Fig. 1). During 1988, samples were obtained with a 0.5 m x 1 m (0.505-mm mesh) side-towing neuston net. The 1989 samples were taken with a 1 m x 2 m (0.947 mm mesh) neuston net. In addition, larger juvenile Spanish mackerel (> 60 mm SL) were obtained during 1983-89 along the coast of North Carolina, South Carolina, and Georgia from SCDNR research cruises aboard the RV Oregon and RV Lady Lisa with trawls, gill nets, seines, and from commercial shrimp trawling bycatch (Collins et al., 1988; Beatty et al.1). Larvae and juveniles (<100 mm SL) were preserved in 95% ethanol and measured (standard length [SL], fork length [FL], and total length [TL] ) to the near- est 0.1 mm with dial calipers or ocular micrometer (Wild, M5 dissecting scope). Owing to the poor con- dition of the caudal fin on many of the smaller fish, standard length was used in the age and growth analysis. A factor of 3% was added to the length of each fish to account for shrinkage in ethanol (Schmidt, unpubl. data, 1988). All fish were identi- fied following Wollam (1970) and Richardson and MacEachran ( 1981). Sagittae and lapilli were excised from larvae and small juveniles by immersing the head region in 5% sodium hypochlorite solution for no more than 30 minutes (Brothers, 1987). Otoliths were separated from undissolved tissue and bone under a dissecting microscope with transmitted crossed polarized light. Otoliths from larger juveniles were removed by dissecting out the entire otic cap- sule and by separating the otoliths from their respec- tive ampullae. Excess tissue was dissolved in sodium hypochlorite solution. Otoliths were then rinsed in water, mounted whole (concave side down, unpolished) in immersion oil on a microscope slide and examined on a video-enhanced (Hitachi, MOS) compound micro- scope (Nikon, Labophot). Lapilli were used to estimate age in Spanish mackerel larvae and juveniles because increments were more discernible in the lapilli than in the sagittae. Young-of-year juveniles (>100 mm SL) were treated according to the same procedures used for YOY king mackerel by Collins et al., 1988. Marginal increment analysis To confirm the hypothesis of daily increment deposi- tion, a marginal increment analysis was performed. 1 Beatty, H. R., J. W. Hall, and E. L. Wenner. 1988. Results of trawling efforts in the coastal habitat of the South Atlantic Bight 1987-1988. South Carolina Division of Natural Re- sources, P.O. Box 12559, Charleston, SC 29422. SEAMAP Report, 94 p. In this analysis, the stage of completion of the mar- ginal increment was compared with the adjacent fully formed increment on the lapilli from fish captured over a daily cycle (Fig. 2). Because Breach Inlet speci- mens were captured over an entire flood tide, it was impossible to know their precise time of capture. Therefore, the mean stage of completion of the mar- ginal increment of several specimens, captured over 5-6 hour periods that progressed throughout the day and night, was compared. Large collections were subsampled by selecting as many as 35 individuals representing the size range of fish captured in the sample. Additional mackerel taken in SCDNR trawls and nearshore ichthyoplankton samples were also used. The time of capture of these specimens was known to within 30 minutes. A total of 165 larval and juvenile Spanish mackerel (7-97 mm SL) were examined. Attempts to find evidence for the daily nature of otolith rings in larger juveniles by measur- ing diel variation in marginal increments with SEM were not successful. Measurements of the marginal increment and the adjacent increment were made along each of three separate axes on each otolith. These axes were cho- sen because their optical properties allowed accept- able ring resolution. Occasionally, it was not possible to measure all three axes owing to opacity or dam- age to the otolith. Increments were displayed on a video monitor at l,000x and measured to the near- est 0.1 mm with dial calipers. Care was taken in ob- serving the opaque and transparent zones because different focal planes may invert their appearance. Consistent counts and marginal increment measure- ments were obtained at a “high” focal point (the dis- tance [with the highest lens power] to object that will produce a well-defined image). We were unaware of time of capture while performing the measurements. A standardized marginal increment (SMI) for each axis of measurement was calculated as W SMI = — W VV(n-l) where Wn = width of marginal increment; and W(n_1)= width of complete adjacent increment. The SMI’s for each of the axes were averaged to ob- tain a mean SMI for each otolith. Two independent mean SMI’s were calculated for each otolith from separate measurements. Although there was no sig- nificant difference between the two measurements (paired Atest, P=0.153), the second measurement was used in the analysis because we were more experienced at locating and measuring the marginal increment. Peters and Schmidt: Age and growth of larval and early juvenile Scomberomorus maculatus 533 •/*/ Figure 2 (A) Lapillus from a Spanish mackerel juvenile (SMI=0. 3) captured at Breach Inlet between 0309 h and 0830 h EDT. (B) Lapillus from a Spanish mackerel juvenile (SMI=0.9) captured at Breach Inlet between 1613 h and 2211 h. Incomplete marginal increment (mi), and adjacent fully formed increment (ai) are indicated. Age and growth analysis Whole lapilli from 415 larval and juvenile Spanish mackerel were examined. For larvae and juveniles <100 mm SL, otolith radius was measured from the center of the primordium to the margin of the otolith along a consistent axis. Measurements were made with an ocular micrometer at magnifications of lOOx or 400x depending upon the size of the otolith. Pre- sumed daily increments on the lapilli were counted on a video monitor under l,000x magnification. Two independent counts of presumed daily increments 534 Fishery Bulletin 95(3), 1997 were made; we were unaware of fish length and any prior age determination during counting. Incomplete marginal increments were not counted. Furthermore, counts of right and left otoliths were conducted sepa- rately. In situations where the first two counts differed, a third independent count was performed. The assigned age corresponded to the two counts that were in agree- ment. If agreement could not be reached on two of the three counts, the otolith was considered unreadable and was not used. Otoliths in YOY juveniles (>100 mm SL) were counted according to the procedures used for YOY king mackerel in Collins et al. (1988). Nonlinear regression analysis was used to describe the relation between age and length. Statistical analyses were performed with SYSTAT software (Wilkinson, 1988) and Table Curve ( Jandel Scientific) and were based on a significance level of 0.05. Results Several features of increment deposition were ob- served to be consistent among the otoliths examined. Two diffuse and poorly defined increments (core in- crements) surrounded the primordium (Fig. 3). Mean core width was 11.4 mm and there was little varia- tion with fish length (SB=0.54 mm, n=40, length range=9. 0-300.1 mm). Although these increments were counted as daily, the nature of their deposition was clearly different from that of subsequent rings. This finding indicated that they were formed during a separate developmental stage. The absence of fish younger than 9 days precluded precise determina- tion of the time period represented by these two in- crements. Subsequent increments were clearly de- fined on most lapilli and were easily discernible in whole otoliths examined under a light microscope without any special preparation (grinding or polish- ing). Subdaily increments occurred, particularly in older juveniles, and were discernible from the daily increments (Fig. 4). Marginal increment analysis Of 165 fish examined for marginal increments, 13 were not used in the final analysis owing to damage to the otoliths or to uncertainties in distinguishing the marginal or adjacent increments (or both). No significant difference in SMI was found between left and right lapilli (paired Ptest, P=0.191). Examina- tion of fish captured during the 1613-2330 h time period revealed an obvious split in the stage of mar- ginal increment completion (Table 1). Aunimodal dis- tribution of mean SMI, for fish captured over a 24-h period, was obtained if the mean SMI of those otoliths whose margin was bordered by a translucent zone Peters and Schmidt: Age and growth of larval and early juvenile Scomberomorus maculatus 535 (late stage of increment formation) was plotted separately from the mean SMI of otoliths whose margin was bordered by an opaque zone (early stage of increment formation) (Fig. 5). The ob- served separation in the stage of increment for- mation would be expected if initiation of incre- ment formation occurred between 1613 h and 2330 h. Age and growth Otolith radii measurements revealed no signifi- cant difference between right and left lapilli (paired sample Gtest, P=0.127). The relation between SL versus lapillus radius was described by the following regression equation: SL = -4.78 + 0.68 (Radius) [r2=0.99, n=364]. No significant difference was found between increment counts for left and right lapilli (paired Gtest, P=0.190). Therefore, if left and right counts differed, the otolith whose increments were more clearly defined, or which was in better condition, was used to assign an age to the fish. Growth in young Spanish mackerel was quite vari- able within age classes, particularly in juveniles older Table 1 Mean standardized marginal increment (MSMI), range, standard deviation for left (L) and right (R) lapilli, and size range for S. maculatus captured from 1613 to 2211 h and 1755 to 2330 h. Lapilli with opaque margins are considered separately from lapilli with translucent margins. ( n refers to the number of specimens, no. refers to the number of right or left lapilli.) Hours are those of eastern daylight time. Lapilli from fish Lapilli from fish collected 1613-2211 h collected 1755-2300 h Opaque Translucent Opaque Translucent R L R L R L R L n 34 10 No. 18 18 16 16 6 6 2 4 MSMI 0.19 0.22 0.88 0.86 0.25 0.23 0.75 0.75 SD 0.05 0.07 0.08 0.15 0.06 0.05 0.07 0.13 Range Size range 0.3 0.2 0.3 0.5 0.1 0.1 0.1 0.3 (SLMmm) 17.1-97.0 22.6- -79.0 than 23 days (Fig. 6). Nonlinear regression analysis provided the following growth equation: In SL = 6.2 - 55.1/Age. Figure 4 Lapillus of an 18-day-old juvenile Spanish mackerel (18.7 mm SL). Note occurrence of subdaily rings as the fish ages. A daily growth increment (d) and a presumed subdaily increment (sd) are indicated. 536 Fishery Bulletin 95(3), 1997 Based on this growth equation, predicted absolute day for the first 150 days of life. Early growth was growth rate (predicted SL/age in days) was 2.4 mm/ characterized by relatively slow growth for the first 23 days of life (1.9 mm/day) followed by a surge of rapid growth from 23 to 40 days, during which growth rates approached 5.0 mm/day. Pre- dicted absolute growth of older ju- veniles (40-150 days) was 2.0 mm/ day. Discussion Validation of daily formation of the microstructural increment is a nec- essary prerequisite to using otoliths for ageing larval and juvenile fishes. Determination of the stage of com- pletion of the most recently formed increment over a daily light cycle does not directly validate daily in- crement formation but lends strong support to the hypothesis that in- crements are deposited daily. Several studies have shown that increment deposition is most likely controlled by an endogenous rhythm that can be modified by physical or behavioral parameters (or both), such as light and dark periodicity, temperature regimes, feeding frequency, food avail- ability, activity patterns (such as daily vertical mi- grations), or a combination of these and other fac- tors (Jones, 1986; Campana and Neilson, 1985, for review). There is presently no information available on the effects of changes in environmental factors on the periodicity or pattern of increment formation in larval or juvenile scombrids. However, work done with other teleosts (Taubert and Coble, 1977; Tanaka et al., 1981; Campana, 1984; Neilson and Geen, 1985; Jenkins and Davis, 1990) suggests that an internal diel clock alone is not responsible for daily increment formation but that it is entrained by some external en- vironmental cue that can vary between species of fishes. Observations on the seasonal occurrence and dis- tribution of larval Spanish mackerel in the northern Gulf of Mexico and the South Atlantic Bight suggest that they are restricted to middle and inner conti- nental shelf waters (Dwinell and Futch, 1973; MacEachran et al., 1980; Collins and Stender, 1987). Since daily fluctuations in salinity and turbidity are minimal in shelf waters outside estuarine influence, they are not likely to modify cyclic daily deposition of increments in larval Spanish mackerel. It seems more likely that feeding periodicity or diel vertical + Right otolith • Lett otolith Figure 5 Mean standardized marginal increment and standard deviation for left and right lapilli of Spanish mackerel larvae and juveniles collected throughout the day and night. Mean SMI for opaque bordered marginal increments are plotted separately from translucent bordered marginal increments in samples taken from 1613-2330 h. Capture time ranges (dashed lines) and sample sizes are indicated. Peters and Schmidt: Age and growth of larval and early juvenile Scomberomorus maculatus 537 migrations, which are often strongly associated with light cycles, serve to increase daily increment defi- nition (Campana and Neilson, 1985). Other species of scombrid larvae and early juve- niles ( E . pelamis, T. albacares, E. alletteratus, Auxis thazard, and Scomber scombrus) are known to un- dergo vertical diel migration feeding primarily at the surface at night (Matsumoto, 1958; Grave, 1981). Collins and Stender (1987) found statistical evidence for vertical migration to the surface at night in both S. maculatus and S. cavalla. Spanish mackerel and other species of Scomberomorus are known to feed on ichthyoplankton during the larval stage and are almost completely piscivorous as juveniles (Naughton and Saloman, 1981; Jenkins et al., 1984). The large eyes of scombrids, even during the larval stage, sug- gest that they are visual predators. Therefore, light cycles probably have a strong influence on prey de- tection. Moreover, feeding opportunities related to diurnal vertical migrations of prey organisms, along with fluctuations in temperature associated with diel vertical migrations, may further serve to entrain this endogenous rhythm of calcium carbonate deposition. Age and growth Marginal increment analysis indicated that otolith increments are deposited daily in larvae and juve- niles from 7 to 95 mm SL. However, because we were unsuccessful at capturing preflexion larvae, it was impossible to determine if increment counts truly reflected age from fertilization. Very little informa- tion is available on otolith formation in scombrids, although otoliths are among the first calcified struc- tures and are present in scombrid embryonic stages (Matsumoto, 1958; Radtke, 1983; Brothers et ah, 1983). In E. pelamis larvae reared from hatching (Radtke, 1983), the core region of the otolith (the primordium and two diffuse increments), along with the pattern of subsequent increment formation, is very similar in appearance to otoliths of S. maculatus (Fig. 3). Radtke (1983) observed that the two core increments were present at hatching. The two core increments in S. maculatus , because of their atypical pattern of deposition, are likely to have been formed during the egg stage or prior to yolk-sac absorption. However, it is not known whether these increments are deposited daily. Be- cause hatching and yolk-sac absorption of Spanish mackerel larvae usually occurs five days after fer- tilization at temperatures experienced during the spawning season in South Carolina waters (Berrien and Finan, 1977; Fritzche, 1978), errors in age esti- mation are likely to be consistent among most fish, and growth rate calculations would remain unaffected. Considerable variation was observed in the growth rates of individual fish, particularly among juveniles older than 23 days. The use of specimens collected over a wide spatial and temporal range was prob- ably responsible for much of this variation. However, the overall predicted mean absolute growth rate of 2.4 mm/d ay is within the range of growth rates ob- served in other scombrids during the first few months of life (1 mm/day-6 mm/day) (see Brothers et al., 1983, for review). The regression lines estimating the relationship between age and length appeared to be a good approximation (r2=0.97, P<0.0001) of growth in young Spanish mackerel. Acknowledgments We would like to express sincere thanks to Mark Collins, Robert Johnson, George Sedberry, and Bruce Stender for their assistance and advice. For their assistance in field sampling we would like to thank Shirelle Brown, Scott Van Sant, Byron White, Paula Keener-Chavis, Pete Richards, and Vince Taylor. We would also like to thank Bruce Stender, Charles Barans, William Roumillat, and SEAMAP for pro- viding many of the specimens used in this study. Fi- nancial support and equipment for this study were provided by the South Carolina Department of Natu- ral Resources, the University of Charleston’s Grice Marine Laboratory, and the Slocum-Lunz Foundation. Literature cited Berrien, P., and D. Finan. 1977. Biological and fisheries data on Spanish mackerel, Scomberomorus maculatus (Mitchill). U.S. Dep. Commer., NOAA/NMFS Sandy Hook Lab. Tech. Ser. Rep. 9, 58 p. Brothers, E. B. 1987. Methodological approaches to the examination of otoliths in ageing studies. In R. C. Summerfelt and G. E. Hall (eds.), The age and growth of fishes, p. 319-330. Iowa State University Press, Ames, Iowa. Brothers, E. B., and W. N. MacFarland. 1981. Correlations between otolith microstructure, growth, and life history transitions in newly recruited french grunts [Haemulon flavolineatum (Desmarest), Haemulidae]. In R. Lasker and K. Sherman (eds.), The early life history of fish: recent studies, p. 369-374. ICES Mar. Sci. Symp., Woods Hole. Brothers, E. B., E. D. Prince, and D. W. Lee. 1983. Age and growth of young-of-the-year bluefin tuna, Thunnus thynnus, from otolith microstructure. In E. D. Prince and L. M. Pulos (eds ), Proceedings of the interna- tional workshop on age determination of oceanic pelagic fishes: tunas, billfishes, and sharks, p. 49-59. U.S. Dep. Commer., NOAATech. Rep. NMFS 8. Campana, S. E. 1984. Interactive effects of age and environmental modifi- 538 Fishery Bulletin 95(3), 1 997 ers on the production of daily growth increments in otoliths of plainfin midshipman, Porichthys notatus. Fish. Bull. 82( 1 ): 165—177. Campana, S. E., and J. D. Neilson. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1030. Collins, M. R., and B. W. Slender. 1987. Larval king mackerel ( Scomberomorus cavalla), Spanish mackerel ( Scomberomorus maculatus) and blue- fish ( Pomatomus saltatrix) of the south east coast of the U.S., 1973-1980. Bull. Mar. Sci. 41(3):822-824. Collins, M. R., D. J. Schmidt, C. W. Waltz, and J. L. Pickney. 1988. Age and growth of king mackerel, ( Scomberomorus cavalla), from the Atlantic coast of the United States. Fish. Bull. 87(11:49-61. D’ Amours, D., J. G. Landry, and T. C. Lambert. 1990. Growth of juvenile (0-group) Atlantic mackerel ( Scomber scombrus ) in the Gulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 47:2212-2218. DeVries, D. A., C. B. Grimes, K. L. Lang, and D. B. White. 1990. Age and growth of king and Spanish mackerel lar- vae and juveniles from the Gulf of Mexico and the South Atlantic Bight. Environ. Biol. Fish. 29:135-143. Dwinell, S. E., and C. R. Futch. 1973. Spanish and king mackerel larvae and juveniles in the northeastern Gulf of Mexico June through October 1969. Fla. Dep. Nat. Resour. Mar. Res. Lab. Leaflet Ser. 4, part. 1, no. 24, 14 p. Fable, W. A., A. G. Johnson, and L. E. Barger. 1987. Age and growth of Spanish mackerel ( Scomberomorus maculatus) from Florida and the Gulf of Mexico. Fish. Bull. 85(41:777-783. Finucane, J. H., and L. A. Collins. 1986. Reproduction of Spanish mackerel, Scomberomorus maculatus, from the southeastern United States. North- east Gulf Sci. 8(21:97-106. Fritzche, R. A. 1978. Development of fishes of the Mid-Atlantic Bight. Vol. 5: Chaetodontidae through Ophidiidae. U.S. Fish. Wildl. Serv., Biol. Serv. Prog. FWS/OBS-78-12, 340 p. Geffen, A. J. 1987. Methods of validating daily increment deposition in otoliths of larval fish. In R. C. Summerfelt and G. E. Hall (eds.l, The age and growth of fishes, p. 433-442. Iowa State Univ. Press, Ames, Iowa. Grave, H. 1981. Food and feeding of mackerel larvae and early juve- niles in the North Sea. In R. Lasker and K. Sherman (eds.l, The early life history of fish: recent studies, p. 454- 459. ICES Symposium, Woods Hole. Jandel Scientific. 1996. Table curve 2D, version 4.0: automated curve fitting and equation discovery. San Rafael, CA, 472 p. Jenkins, G. P., N. E. Milward, and R. F. Hartwick. 1984. Food of larvae of Spanish mackerels, genus Scomber- omorus (Teleostei: Scombridae), in shelf waters of the Great Barrier Reef. Aust. J. Mar. Freshwater Res. 35: 477-482. Jenkins, G. P„ and T. L. O. Davis. 1990. Age, growth rate, and growth trajectory determined from otolith microstructure of southern bluefin tuna, Thun- nus maccoyii, larvae. Mar. Ecol. Prog. Ser. 63:93-104. Jones, C. 1986. Determining age of larval fish with the otolith incre- ment technique. Fish. Bull. 84(11:91-103. Klima, E. F. 1959. Aspects of the biology and the fishery for Spanish mackerel, Scomberomorus maculatus (Mitchill), of south- ern Florida. Fla. Board Conserv. Mar. Res. Lab. Tech. Ser. 27, 39p. Matsumoto, W. M. 1958. Description and distribution of larvae of four species of tuna in central Pacific waters. Fish. Bull. 58:31-72. MacEachran, J. D., J. H. Finucane, and L. S. Hall. 1980. Distribution, seasonality and abundance of king and Spanish mackerel larvae in the northwestern Gulf of Mexico (Pisces: Scombridae). Northeast Gulf Sci. 4(1): 1-16. Naughton, S. P., and C. H. Saloman. 1981. Stomach contents of juveniles of king mackerel ( Scomberomorus cavalla) and Spanish mackerel (S. maculatus). Northeast Gulf Sci. 5(11:71-74. Neilson, J. D., and G. H. Geen. 1985. Effects of Feeding regimes and diel temperature cycles on otolith increment formation in juvenile chinook salmon, Oncorhynchus tshawytscha. Fish. Bull. 83(1): 91-101. Powell, D. 1975. Age, growth and reproduction in Florida stocks of Spanish mackerel, ( Scomberomorus maculatus). Fla. Mar. Res. Publ. 5, 21 p. Radtke, R. L. 1983. Otolith formation and increment deposition in labo- ratory-reared skipjack tuna, Euthynnus pelamis, larvae. In E. D. Prince and L. M. Pulos (eds.l, Proceedings of the international workshop on age determination of oceanic pelagic fishes: tunas, billfishes, and sharks, p. 99- 103. U.S. Dep. Commer., NOAATech. Rep. NMFS 8. Richardson, S. A., and J. D. MacEachran. 1981. Identification of small (< 3mm) larvae of king and Spanish mackerel, Scomberomorus cavalla and S. maculatus. Northeast Gulf Sci. 5(11:75-79. Schmidt, D. J., M. R. Collins, and D. M. Wyanski. 1993. Age, growth, maturity and spawning of Spanish mackerel, Scomberomorus maculatus (Mitchill), from the Atlantic coast of the southeastern United States. Fish. Bull. 91:526-533. Tanaka, K., Y. Mugiya, and J. Yamada. 1981. Effect of photoperiod and feeding on daily growth patterns in otoliths of juvenile Tilapia nilotica. Fish. Bull. 79(31:459-466. Taubert, B. D., and D. W. Coble. 1977. Daily rings of otoliths of three species of Lepomis and Tilapia mossambica. J. Fish. Res. Board Can. 34:332-340. Uchiyama, J. H., and P. Struhsaker. 1981. Age and growth of skipjack tuna, Katsuwonas pelamis, and yellowfin tuna, Thunnus albacares, as indi- cated by daily growth increments of sagittae. Fish. Bull. 79(11:151-162. Wexler, J. B. 1993. Validation of daily growth increments and estima- tion of growth rates of larval and early-juvenile black skip- jack, Euthynnus lineatus, using otoliths. Inter-Am. Trop. Tuna Comm. Bull. 20(7), 40 p. Wild, A., and T. J. Foreman. 1980. The relationship between otolith increments and time for yellowfin and skipjack tuna marked with tetra- cycline. Inter-Am. Trop. Tuna Comm. Bull. 17(71:509-541. Wilkinson, L. 1988. SYSTAT: the system for statistics. SYSTAT, Inc. Evanston, IL, 822 p. Peters and Schmidt: Age and growth of larval and early juvenile Scomberomorus macutatus 539 Wollam, M. B. 1970. Description and distribution of larvae and early juve- niles of king mackerel, Scomberomorus cavalla (Cuvier), and Spanish mackerel, Scomberomorus maculatus (Mitchill); (Pi- sces: Scombridae); in the western North Atlantic. Fla. Dep. Nat. Resour. Mar. Res. Lab. Tech. Ser. 61, 31 p. 540 Abstract.— Stable nitrogen (515N) and carbon (513C) isotope measure- ments were used to differentiate groups of king mackerel, Scomberomorus cav- alla, in the northwestern Gulf of Mexico and off the southeastern coast of Florida, as well as off the coast of Mexico. Northwestern (+13.1%e) and southeastern (Mexico=+10.8%e and Florida=+10.8% 71=1 where Wn = weight of the segment in milligrams; and bn = isotopic value for the segment. Multivariate analysis of covariance (MANCOVA) was used to determine which independent variables had significant effects in the general linear models (GLM) (Eqs. 3 and 4) (SAS, 1990). S15N or <513C = collection site + season + sex; [length was used as a covariate.] (3) 815N or <513C = region + season + sex ; [length was used as a covariate.] (4) If an independent variable did not have a significant effect on the model, the variable was eliminated and the GLM was conducted again. Least squared means (LSmeans) and a pairwise comparison, with a 95% confidence interval (P=0.05), were performed to de- termine significant differences in nitrogen and car- bon isotope data between sample collection sites and regions. As king mackerel increased in fork length, an increase in 15N was observed (Fig. 2B); therefore, analysis of covariance was used to control for differ- ences in fish size. Fork length was used as the covariate. Results No significant correlation between weight of the fin spine sample segment and 515N (r2=0.03) or 813C (r2=0.06) was detected for any of the samples, sug- gesting that the mineral phase had been completely removed, and 100% collagen carbon and nitrogen as C09 and N2 had been recovered, respectively. Isotopic variations within individual spines were examined to try to determine the life history of indi- viduals. Spines were delineated into three portions (tip, mid, and base) (Fig. 3). The base of the spine is believed to contain more recently acquired material. Isotopic trends in carbon, along the length of the spine, were observed for many sites. Isotopic values for carbon became lighter as the fish aged (from tip to base); however, few trends existed for nitrogen. In general, the isotopic difference within the spine was generally less for nitrogen compared with carbon. Nitrogen and carbon isotopic differences were ob- served between various sites and regions (Table 1; Fig. 4). In the pairwise comparison, more significant differences were found between individual sites for nitrogen isotope ratios than for carbon isotopic ra- tios (Fig. 4). Nitrogen isotopic data displayed a geo- graphical pattern (Fig. 5). In general, king mackerel from Mississippi, Louisiana, and Texas were 15N- enriched in contrast with those from the Mexican Roelke and Cifuentes: Use of isotopes to assess groups of Scomberomorus cavalla 545 and Florida sites. Spines of individuals from Florida were typically 13C-depleted in contrast with those from the Mexico, Mississippi, Louisi- ana, and Texas sites. After the nonsignificant variable (sex) was eliminated from the GLM (Eq. 3), the season of sample collection (P=0.Q001), fork length (P=0.Q31), and collection site (P=0.0001 ) influenced the varia- tion in nitrogen isotope data (F= 9.27; P=Q.00Q1 for the overall model). Only collection site (P=0.Q28) significantly influenced the revised GLM (Eq. 3) for 513C (F= 2.38; P=0.0281 for the overall model). A GLM was also constructed for the three regions in this study: Florida, Mexico, and northwestern Gulf of Mexico (Eq. 4). These regions were determined on the basis of previous king mackerel stock structure studies (Baughman, 1941; Trent et ah, 1987; Johnson et ah, 1994; May2) and isotopic patterns observed in previous stud- ies (Fry, 1983; Macko et al., 1984) as well as in this study. The GLM (Eq. 4) for 815N (P=26.42; P=0.0001) showed that collection site (P=0.0001) and fork length (P=0.0023) were statistically sig- nificant regionally. Collection site (P=0.Q23) and fork length (P=0.047) also had a significant re- gional influence on 813C (P=4.04; P=0.011) (Eq. 4). Discussion Although the dorsal fin spines were divided into multiple sections, and isotopic trends were ob- served along a spine (Fig. 3, A and B), life history could not be determined from isotopic data because it was not possible to assign accurately an age to a particular portion of the spine. The length-at- age relation varies regionally and shows large in- dividual variation (DeVries and Grimes3). For ex- ample, male king mackerel from the eastern Gulf of Mexico, with a fork length of 105-110 cm, ranged from 4 to 22 years of age (DeVries and Grimes3). Additionally, female king mackerel are larger than males at a given age (Beaumariage, 1973; Johnson et al., 1983; DeVries and Grimes3) and sex was not known for the majority of the fish analyzed (Table 2). Isotopic differences within individual dorsal spines were studied. One would generally expect the king mackerel with the greater fork length to have a larger range of isotopic values within its dorsal spine ow- ing to variation in trophic-level feeding with size; however, no clear trends were found (Table 2). For example one 113-cm-FL female king mackerel from GulfPort, MS, exhibited little variation among the segments analyzed. The §15N varied by only 1.1 and (1) C Q_ o CD O) CO CD > < -12 1 A -14 -16- % • -18 - ® - , . * Wn -20 • V -22 1 40 60 80 100 120 140 o> c CL SS o O) (0 1 7 1 6 1 5 14 1 3 12 1 1 1 0 9 B • ® v* *4 ' • •• • ® ® • *® ® « ^ • • »• 1 1 1 1 0 60 80 100 120 140 Fork length (cm) Figure 2 King mackerel fork length versus (A) stable carbon and (B) nitrogen isotopes. King mackerel fork length showed a posi- tive relation to nitrogen isotopic values, but not to carbon iso- topic values. the isotopic ratio of carbon varied by only 2.5 %o. Con- versely, a 76-cm-FL female from Celestun, Mexico differed by 4.3%c in nitrogen and 3.3%c in carbon among the spine segments analyzed. Additionally, an isotopic trend of an individual spine becoming heavier over time (from tip to base) would be expected because of an increase in trophic level feeding; however, this trend was not generally observed in the sites or regions for either carbon or nitrogen (Fig. 3). A greater enrichment in nitrogen, compared with carbon, would be expected for an in- crease in trophic level feeding (DeNiro and Epstein, 1978; DeNiro and Epstein, 1981); however this ten- dency was not observed. In fact, carbon isotopic trends were contradictory to this assumption, and no clear nitrogen isotopic trends could be detected 546 Fishery Bulletin 95(3), 1997 ■ Fort Pierce-Palm Beach, FL (0.80, 0.34) O Veracruz, Mexico (0.99, 0.84) i □ Panama City, FL (0.68, 0.21) • Celestun, Mexico (0.69, 0.65) ♦ GulfPort, MS (0.68, 0.21) O Dzilam DeBravo, Mexico (0.98, 0.86) o Grand Isle, LA (0.62, 0.98) Q Florida Region (0.74, 0.03 1 ) ▲ Galveston, TX (0.91, 0.08) 1 Northwestern GOM Region (0.90. 0.68) T Port Aransas, TX (0.91, 0.02) O Mexico Region (0.95, 0.0031) o co 'oo -15 -16 -17 -18 -19 o § § -20“ -21 8 d a T ■ o ♦ O A 0 □ B Figure 3 (A) Stable carbon and ( B ) nitrogen isotopic results along the length of a spine for each king mackerel site and region. Legend contains r 2 values or carbon and nitrogen isotopes respectively. Whole spines were delineated into three sections each (tip, mid, and base). If the spines had more than three divisions for analysis, the middle sections were averaged together to create one value. If the spine was divided into two sections for analysis, the middle section was excluded for the figure. along the individual spines for the sites or regions. The data suggest a factor other than change in trophic level determines isotopic values found within the spines. Numerous factors could influence the range and trend of isotopic values found within a dorsal spine, such as feeding region, trophic- level status of the individual, and the manner in which the spine was segmented. Stable carbon and nitrogen isotopic values at the base of the food chain vary within the Gulf of Mexico (Table 1), particularly among waters off southern Florida and the northwestern Gulf of Mexico. We Roelke and Cifuentes: Use of isotopes to assess groups of Scomberomorus cavalla 547 6 s m 8 < 0) s N 8 c/D 1 S c o < jy on s o u cd § s 3 f~ < C/D e o > 3 <4-j "5 § a u > CL O a a CL Fort Pierce-Palm Beach, FL Panama City, FL Gulf Port, MS Grand Isle, LA Galveston, TX Port Aransas, TX Veracruz Mexico Celestun Mexico Significantly different for nitrogen Significantly different for carbon Significantly different for both nitrogen and carbon Figure 4 Results of MANCOVA, LSmeans, and pairwise comparison for all mack- erel sites. Fork length is a covariate. observed similar differences for mean 815N values of king mackerel spines collected in these regions. The 815N data of Macko et al. (1984), Fry (1983), and this study all showed an 15N-enriched in the northwest- ern Gulf of Mexico relative to samples col- lected in Florida and Mexico. Enriched nitrogen values are often observed off the mouths of estuaries (e.g. Cifuentes et al., 1989). This enrichment often reflects the assimilation of isotopically altered inor- ganic nitrogen from riverine sources by algae. The influence of the Mississippi River could account for the more positive 515N values (Lopez-Veneroni4) detected in the northwestern Gulf of Mexico. In contrast to the 515N data for the king mackerel, 813C measurements were not as discriminating between sites (Fig. 4) or regions (Table 1). Although not significant, the king mackerel 513C values for the northwestern Gulf of Mexico region were more negative than those for the Mexico region. More negative 813C values were also detected at the base of the food chain in the northwestern Gulf of Mexico region in comparison with those for Florida (Table 1). The influence of the Mississippi River on the northwestern Gulf of Mexico area is most likely the primary reason for 813C values being more negative. Although C02 depletion resulting from enhanced pri- mary production can increase 813C values (Raven et al., 1993), the primary impact of the Mississippi River is the large terrestrial input of particulate organic matter (Trefry et al., 1994) lead- ing to more negative 813C values. Commonly, less variability is observed in carbon isotopes than with nitrogen. This trend was not ob- served in this study. Our results, however, are con- sistent with some previous studies that reported that 815N data could be more discriminating than 813C data. For example, Sholto-Douglas et al. (1991) used carbon and nitrogen isotopes to study food web rela- tions among plankton and pelagic fish and found greater variability in 813C data than in 815N mea- surements. Perhaps these systems have numerous carbon sources that create greater than expected variation in stable carbon isotope values, thereby rendering them ineffective. 4 Lopez-Veneroni, D. 1997. Oceanography Department, Texas A&M University, College Station, TX 77843. Manuscript in prep. Numerous studies have observed seasonal migra- tions of king mackerel. King mackerel migrate along the eastern coast of the Gulf of Mexico and into the northern Gulf of Mexico from southeastern Florida (wintering grounds) in the summer (Trent et al., 1987; Sutter et al., 1991; see also Johnson et al., 1994). Migrations may extend as far as Galveston and Port Aransas, TX (Williams and Sutherland, 1978). A re- turn migration from the northern Gulf of Mexico into southeast Florida occurs in late summer and early fall (Williams and Sutherland, 1978). While the king mackerel that winter in southeast Florida are mi- grating into the northern Gulf of Mexico, a simulta- neous migration from the Yucatan area (wintering grounds) occurs along the western coast of the Gulf of Mexico into the northern Gulf of Mexico (Trent et al., 1987; see also Johnson et al., 1994). Wind circulation along the Mexican and south Texas coast during the late spring and early sum- mer may cause upwelling off the Texas-Mexico bor- der (Dagg et al., 1991). Consequently, coastal bound- 548 Fishery Bulletin 95(3), 1997 & o -16 •17 -18 -19 -20 10 1 1 12 13 1 4 15 515N (%o) Figure 5 Weighted mean 615N versus 813C values for all king mackerel sample collection sites and regions. Number of samples per site can be seen in Table 2. Standard error bars are shown for sites. Regions were defined according to stable isotopic data and stock structure studies (Baughman, 1941; Fry, 1983; Macko et al., 1984; Trent et ah, 1987; Fable et ah, 1990; Johnson et ah, 1994; May2). ary water masses off Mexico and south Texas may collide and form a convergence zone that directs low-salinity wa- ters offshore, near Browns- ville, TX (Vastano5 *) (Fig. 1). This convergence zone may act as a temporary boundary between northwestern Gulf of Mexico fish and Mexico fish. The innate migratory pat- terns of the king mackerel can influence the isotopic val- ues observed in their dorsal spines. The location in which food is assimilated should directly influence the isotopic value recorded in the spine. Consequently, the area in which the individual fed, rather than collection site of the individual, would be de- tected in the spine. There- fore, determination of groups of king mackerel from collec- tion site alone, may be inap- propriate. In addition, re- gional groupings of the mack- erel that were based upon isotopic data and on previous king mackerel studies may be more suitable for drawing conclusions. Likewise, the migratory nature of king mackerel complicates the use of season of collection as a vari- able (Table 2). For example, the GLM (Eq. 3) indicated that the season in which the specimens were collected, fork length, and collection site influenced the nitrogen isotope results. However, when the data were divided into regions, fork length and region were the only vari- ables that had a significant effect on the GLM (Eq. 4). Again, consideration of the data by region, as opposed to individual site, may be more appropriate, particu- larly because, in the former case, time of collection did not bias the isotopic findings. Our method of using stable isotopes is a new ap- proach in trying to determine the number and loca- tion of king mackerel groups. An advantage of isoto- pic analysis over that of genetics is that stable iso- topes enable researchers to view significant changes in an individual, whereas genetic methods require generations to see significant variations. The disad- vantage to stable isotopes is that the signal is ac- 5 Vastano, D 1995. Oceanography Department, Texas A&M Univ., College Station, TX 77843. quired only from areas in which food is assimilated, which may or may not represent the location and number of king mackerel groups. Although the num- ber and location of king mackerel stocks have been researched previously by using genetic techniques (see below), several scenarios exist. Research by DeVries and Grimes (1991) has suggested the possi- bility of three stocks: a western Gulf of Mexico, an eastern Gulf of Mexico, and an Atlantic stock. From mitochondrial DNA data, Gold et al. (in press) found weak genetic differences between Atlantic and Gulf of Mexico king mackerel that implied more than one stock. Additionally, Johnson et al. (1994), using elec- trophoretic data, suggested the existence of two stocks, eastern and western, within the Gulf of Mexico. The idea of separate eastern and western stocks of king mackerel within the Gulf of Mexico has also been supported by Baughman ( 1941), May,2 and Trent et al. (1987) with observational, electro- phoretic, and catch results, respectively. Our 815N data showed significant differences be- tween king mackerel caught in Mexican and Florida waters in contrast to those collected in the north- western Gulf of Mexico. Thus, our isotopic results suggest that at least two distinct groups exist within Roelke and Cifuentes: Use of isotopes to assess groups of Scomberomorus cavalla 549 the Gulf of Mexico (Fig. 5). Statistically, the Mexico and Florida regions are significantly different in car- bon isotopes; however, neither region differs signifi- cantly from the northwestern Gulf of Mexico. It is conceivable that a separate Mexico and Florida group of king mackerel exists. Possibly neither site dif- fered significantly from the northwestern Gulf of Mexico owing to individuals from both the Mexico and Florida regions being contained in the catch from the northwestern Gulf of Mexico. Recall, the north- western Gulf of Mexico individuals were collected in the summer when migrations to the northwestern Gulf of Mexico from Florida and Mexico have been documented (Trent et al. 1987; Sutter et al. 1991). However, the similarity in nitrogen isotopic compo- sition indicates that the Florida and Mexico regions are related. A year-round sustained population in the north- west Gulf of Mexico would contribute to their isoto- pically different nitrogen values compared with Mexi- can and Florida fish. Other studies have surmised that Louisiana may have a resident population (Fisher, 1980; Fable et al., 1987) along a broad area from the Mississippi delta westward to regions off Texas, which are adjacent to oil rigs (Trent et al., 1983). These artificial structures may attract bait fish (Wickham et al., 1973). Northwestern Gulf of Mexico fish, being significantly 15N-enriched, might be a nonmigrating or a separate group of king mack- erel that feed on an isotopically enriched food source compared with king mackerel from Mexico and Florida. Alternatively, it is conceivable that the indi- viduals are migratory and that the isotopic signal is due to assimilation of material from the northwest- ern Gulf of Mexico region although they are not per- manent inhabitants of the region. Physical dynamics within the Gulf of Mexico may influence mixing between sites and therefore the iso- topic values in mackerel found at different sites. The primary current in the Gulf of Mexico is the Loop Current, which enters the Gulf of Mexico through the Yucatan Channel and exits through the Florida Straits (Leipper, 1970; Cooper et al., 1990) (Fig. 1). This current is formed by waters from the western, north, and south Atlantic and the Mediterranean Sea that flow into the Caribbean Sea (Koch et al., 1991). It has a mean position of 88° and 89°W and 27°N (Auer, 1987). Although the Loop Current reaches into the northern Gulf of Mexico, its influence is to the east of the Mississippi Delta. Thus, Mexican and Florida fish could be linked by the Loop Current to the extent that they consume isotopically similar food sources. In contrast, fish in the northwestern Gulf of Mexico are most likely minimally affected by the Loop Current. Northwestern Gulf of Mexico fish may also be strongly influenced by runoff from the Mississippi River system (Dagg et al., 1991). The majority of this runoff (two thirds) is westward and contains high concentrations of dissolved nutrients in relation to the open Gulf of Mexico (Dagg et al., 1991). Dagg et al. (1991) also suggested that the Mississippi River system is the ultimate source of much of the biologi- cal productivity on the Louisiana and Texas shelf. The flow of the Mississippi River into the northwest- ern area influences the isotopic differences within these Gulf of Mexico sites ( Lopez- Veneroni4). Al- though GulfPort, MS, is east of the Mississippi river, specimens collected from this area could conceivably be feeding in or near the Mississippi River Plume region. Discharge from the Mississippi River is trans- ported west along the shore (Dagg et al., 1991), and consumption of prey from this region would be heavily influenced by the Mississippi River leading to 15N-enriched values found in this study. Further- more, Dagg et al. (1991) stated that king mackerel from the northern Gulf of Mexico generally consumed prey that were estuarine dependent and are, there- fore, most likely influenced by runoff. Conclusions Stable nitrogen isotope values of spines of king mack- erel varied geographically. The northwestern Gulf of Mexico ( + 13.1%e) was isotopically distinct from the Mexican and Florida ( + 10.8%c and +10.8 %c) regions. We interpret these results to mean that there are, at least, two distinct groups of king mackerel within the Gulf of Mexico. Our results contrast with certain previous stock-structure assessments that distin- guish only between Gulf of Mexico and Atlantic stocks. Stable carbon isotopes were able to distin- guish between Mexico and Florida regions, although, not the northwestern Gulf of Mexico region. Although carbon isotopes were expected to be less variable than nitrogen, owing to the enrichment from trophic level to trophic level, they were found to be more variable within individual spines. The variability and perplex- ing isotopic trends within individual spines create difficulties in drawing conclusions from the data for stable carbon isotopes. In addition, fewer significant differences were detected between sites for stable carbon isotopes than for nitrogen isotopes. Stable carbon isotopes may be more useful when the isoto- pic discrimination among food resources is greater, which may be found when individuals also feed in coastal habitats. King mackerel, being of great com- mercial and recreational value, need to be managed with a clearer understanding of the number of groups 550 Fishery Bulletin 95(3), 1997 that exist. The isotopic data we have generated in conjunction with genetic research and tagging stud- ies may be able to answer questions pertaining to location and number of king mackerel groups. Acknowledgments This work was funded in part by the Environmental Protection Agency (cooperative agreement CR- 818222). We are grateful for the useful discussions with John McEachran, Churchill Grimes, and John Gold, whose knowledge of the species was critical in formulating this manuscript. Barbara Palko (NMFS, Panama City, FL) and John Gold (Texas A&M Uni- versity) obtained the fish used in this study. We thank Alan Abend, John Pohlman, Dan Roelke, and Beth Trust who reviewed the manuscript. We would also like to acknowledge the reviewers of this manuscript who did an exceptional job. Literature cited Auer, S. J. 1987. Five-year climatological survey of the Gulf Stream system and its associated rings. J. Geophys. Res. 84(C12):7749-7768. Baughman, J. L. 1941. Scombriformes, new, rare, or little known in Texas waters with notes on their natural history or distri- bution. Texas Acad. Sci. Trans. 24:14-26. Beaumariage, D. S. 1973. Age, growth, and reproduction of king mackerel, Scomberomorus cavalla, in Florida. Florida Marine Re- search Publications 1, 45p. Cifuentes, L. A., M. L. Fogel, J. R. Pennock, and J. H. Sharp. 1989. Biogeochemical factors that influence the stable ni- trogen isotope ratio of dissolved ammonium in the Dela- ware estuary. Geochim. Cosmochim. Acta 53:2713-2721. Cooper, C., G. Z. Forristall, and T. M. Joyce. 1990. Velocity and hydrographic structure of two Gulf of Mexico warm core rings. J. Geophys. Res. 95(C2):1663- 1679. Dagg, M., C. Grimes, S. Lohrenz, B. McKee, R. Twilley, and W. Wiseman Jr. 1991. Continental shelf food chains of the northern Gulf of Mexico. In K. Sherman, L. M. Alexander, and B. D. Gold (eds.). Food chains, yields, models, and management of large marine ecosystems, p. 329-345. Westview Press, Boulder, CO. DeNiro, N., and S. Epstein. 1978. Influence of diet on the distribution of carbon iso- topes in animals. Geochim. Cosmochim. Acta 42:495-506. 1981. Influence of diet on the distribution of nitrogen iso- topes in animals. Geochim. Cosmochim. Acta 45:341-351. Deuel, D. G. 1973. The 1970 salt-water angling survey. U.S. Natl. Mar. Fish. Serv. Curr. Fish. Stat. 6200, 54 p. Deuel, D. G., and J. R. Clark. 1968. The 1965 salt-water angling survey. U.S. Fish Wildl. Serv., Resour. Publ. 67, 51 p. Dwinell, S. E, and C. R. Futch. 1973. Spanish and king mackerel larvae and juveniles in the northeastern Gulf of Mexico, June through October 1969. FI. Dep. Nat. Resour., Mar. Res. Lab., Leafl. Ser. 4 (part 1, no. 24), 14 p. Estep, M. L. F., and S. Vigg. 1985. Stable carbon and nitrogen isotope tracers of trophic dynamics in natural populations and fisheries of the Lahontan Lake system, Nevada. Can. J. Fish. Aquat. Sci. 42:1712-1719. Fable, W.A., Jr., L. Trent, G. W. Bane, and S. W. Ellsworth. 1987. Movements of king mackerel, Scomberomorus cav- alla, tagged in southeast Louisiana, 1983-85. Mar. Fish. Rev. 49( 2 ):98— 10 1 . Finucane, J. H., L. A. Collins, H. A. Brusher, and C. H. Saloman. 1986. Reproductive biology of king mackerel, Scombero- morus cavalla, from the southeastern United States. Fish. Bull. 84(4):841-850. Fisher, M. 1980. Size distribution, length-weight relationships, sex ratios, and seasonal occurrence of king mackerel (Scom- beromorus cavalla ) off the Southeast Louisiana coast. La. Dep. Wildl. Fish., Contrib. Mar. Res. Lab. 31:1-21. Fry, B. 1983. Fish and shrimp migrations in the northern Gulf of Mexico analyzed using stable C, N, and S isotope ratios. Fish. Bull. 81:789-801. Gold, J. R., A. Y. Kristmundsdottir, and L. R. Richardson. In press. Population genetic studies of king mackerel in the gulf of Mexico. Mar. Biol. (Berl.). Grimes, C. B., J. H. Finucane, L. A. Ceilings, and D. A. DeVries. 1990. Young mackerel, Scomberomorus cavalla, in the Gulf of Mexico, a summary of the distribution and occurrence of larvae and juveniles, and spawning dates for Mexican juveniles. Bull. Mar. Sci. 46:640-654. Grimes, C. B., A. G. Johnson, and W. A. Fable Jr. 1987. Delineation of king mackerel (Scomberomorus cav- alla) groups along the U.S. east coast and in the Gulf of Mexico. In H. E. Kumpf, R. N. Vaught, C. B. Grimes, A. G. Johnson, and E. L. Nakamura (eds.), Proceedings of the stock identification workshop, p. 186-187. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFC-199. Johnson, A. G., W. A. Fable Jr., M. L. Williams, and L. E. Barger. 1983. Age, growth, and mortality of king mackerel, Scomberomorus cavalla, from the southeastern United States. Fish. Bull. 81(1):97-106. Johnson, A. G., W. A. Fable Jr., C. B. Grimes, L. Trent, and J. V. Perez. 1994. Evidence for distinct stocks of king mackerel, Scomberomorus cavalla, in the Gulf of Mexico. Fish. Bull. 92:91-101. Koch, S. P., J. W. Barker, and J. A. Vermersch. 1991. The Gulf of Mexico Loop Current and deepwater drilling. J. Petrol. Technol. 43:1046-1050 and 1118-1119. Koch, P. L., J. Heisinger, C. Moss, R. W. Carlson, M. L. Fogel, and A. K. Behrensmeyer. 1995. Isotopic tracking of change in diet and habitat use in African elephants. Science (Wash. D.C.) 267:1340-1343. Leipper, D. F. 1970. A sequence of current patterns in the Gulf of Mexico. J. Geophys. Res. 75:637-657. Roelke and Cifuentes: Use of isotopes to assess groups of Scomberomorus cavalla 551 Libby, W. F., R. Berger, J. Mead, G. Alexander, and J. Ross. 1964. Replacement rates for human tissue from atmo- spheric radiocarbon. Science (Wash. D.C.) 146:1170-1172. Macko, S. A. 1981. Stable nitrogen isotope ratios as tracers of organic geochemical processes. Ph.D. thesis, Univ. Texas, Austin, 181 p. Macko, S. A., L. Entzeroth, and P. L. Parker. 1984. Regional differences in nitrogen and carbon isotopes on the continental shelf of the Gulf of Mexico. Naturwis- senschaften 71:374-375. Manooch, C. S., III. 1979. Recreational and commercial fisheries for king mack- erel, Scomberomorus cavalla in the South Atlantic Bight and Gulf of Mexico, USA. In E. L. Nakamura and H. R. Bullis Jr. (eds.), Proceedings: colloquium on the Spanish and king mackerel resources of the Gulf of Mexico, p. 33- 41. Gulf States Mar. Fish. Commisson 4. Manooch, C. S., Ill, E. L. Nakamura, and A. B. Hall. 1978. Annotated bibliography of four Atlantic scombrids: Scomberomorus brasiliensis, S. cavalla , S. maculatus, and S. regalis. U.S. Dep. Commer., NOAATech. Rep. NMFS Circ. 418, 166 p. Minagawa, M., and E. Wada. 1984. Stepwise enrichment of 15N along food chains: fur- ther evidence and the relation between 815N and animal age. Geochim. Cosmochim. Acta 48:1135-1140. Peterson, B. J., and B. Fry. 1987. Stable isotopes in ecosystem studies. Annu. Rev. Ecol. Syst. 18:293-320. Raven, J. A., A. M. Johnston, and D. H. Turpin. 1993. Influence of changes in C02 concentration and tem- perature on marine phytoplankton 13C/12C ratios: an analy- sis of possible mechanisms. Global Planet. Change 8: 1-12. SAS Institute, Inc. 1990. SAS user’s guide: statistics, 4th ed., ver. 6. SAS Institute, Inc. Cary, NC, 890 p. Schoeninger, M. J., and M. J. DeNiro 1984 Nitrogen and carbon isotopic composition of bone col- lagen from marine and terrestrial animals. Geochim. Cosmochim. Acta 48:625-639. Sholto-Douglas, A. D., J. G. Field, A. G. James, and N. J. van der Merwe. 1991. 13C/12C and 15N/14N isotope ratios in the southern Benguela ecosystem: indicators of food web relationships among different size-classes of plankton and pelagic fish; differences between fish muscle and bone collagen tissues. Mar. Ecol. Prog. Ser. 78:23-31. Sutter, F. C., Ill, R. O. Williams, and M. F. Godcharles. 1991. Movement patterns and stock affinities of king mack- erel in the southeastern United States. Fish. Bull. 89:315-324. Trefry, J. H., S. Metz, T. A. Nelsen, R. P. Trocine, and B. J. Eadie. 1994. Transport of particulate organic carbon by the Mis- sissippi River and its fate in the Gulf of Mexico. Estuaries 17(41:839-849. Trent, L., R. O. Williams, R. G. Taylor, C. H. Saloman, and C. H. Manooch III. 1983. Size, sex ratio, and recruitment in various fisheries of king mackerel, Scomberomorus cavalla , in the South- eastern United States. Fish. Bull. 81(41:709-721. Trent, L., B. A. Palko, M. L. Williams, and H.A. Brusher. 1987. Abundance of king mackerel, Scomberomorus cavalla , in the southeastern United States based on CPUE data from charter boats, 1982-85. Mar. Fish. Rev. 49(2): 78-90. Tsimenides, N., and G. Tserpes. 1989. Age determination and growth of swordfish Xiphias gladium L., 1758 in the Aegean Sea. Fish. Res. (Amst.) 8:159-168. Tuross, N., M. L. Fogel, and P. E. Hare. 1988. Variability in the preservation of the isotopic compo- sition of collagen from fossil bone. Geochim. Cosmochim. Acta 52: 929-935. U.S. Dep. Commer. 1985-1987. Marine recreational fishery statistics survey, Atlantic and Gulf coasts, NOAA, NMFS, 1981 to 1986. U.S. Dep. Commer., NOAA, Natl. Mar. Fish. Serv. Curr. Fish. Stat. 8324, 8326, 8327, 8329, 8392. 1992. Marine recreational fishery statistics survey, Atlan- tic and 1990-1991. U.S. Dep. Commer., NOAA, Natl. Mar. Fish. Serv., Curr. Fish. Stat. 9204, 275 p. 1995. Fisheries of the United States, 1994. U.S. Dep. Commer., NOAA, Natl. Mar. Fish. Serv. Curr. Fish. Stat. 9400, 113 p. Wickham, D. A., J. W. Watson Jr., and L. H. Ogren. 1973. The efficacy of mid-water artificial structures for at- tracting pelagic sport fish. Trans. Am. Fish. Soc. 102(31:563-572. Williams, R. O., and D. F. Sutherland. 1978. King mackerel migrations. Proceedings: Colloquium on the Spanish and king mackerel resources of the Gulf of Mexico. Gulf States Mar. Commiss. publ. 4, 57 p. lAbstr.] 552 Effectiveness of four industry- developed bycatch reduction devices in Louisiana's inshore waters Donna R. Rogers* Barton D. Rogers Janaka A. de Silva Vernon L. Wright School of Forestry, Wildlife, and Fisheries Louisiana State University Agricultural Center Baton Rouge, Louisiana 70803-6202 *Present address: Department of Oceanography and Coastal Sciences Coastal Fisheries Institute, Wetland Resources Bldg. Louisiana State University, Baton Rouge, LA 70803-7503 E-mail address (for Donna Rogers): DRogers991@aol.com AbStraCt.-Trawling was conducted in three areas of coastal Louisiana dur- ing the two inshore shrimp seasons of 1992 to evaluate the effectiveness of four industry-developed bycatch reduc- tion devices (BRD’s). Each BRD (Authe- ment-Ledet excluder, Cameron shooter, Lake Arthur excluder, and Eymard ac- celerator) was towed alongside a con- trol net 72 times; tows were equally di- vided between areas and seasons. The Authement-Ledet excluder, Cameron shooter, and Lake Arthur excluder BRD’s caught fewer fish (-36%, -51%, and -21%, respectively), but also fewer shrimp (-18%, -16%, and -24%) than corresponding control nets. Biomass catch differences were -42%, -33%, and -21% for fish and -14%, -14%, and -17% for shrimp. The Eymard accelera- tor caught 26% more fish numerically, 19% less fish biomass, and more shrimp (38% in numbers, 26% in biomass) than control nets. Differences between catches obtained with BRD nets and those with control nets depended upon the organisms present in an area. Abun- dances and size distributions of many species differed between areas; thus BRD’s may have to be selected for the area where they are intended to be used. Manuscript accepted 28 January 1997. Fishery Bulletin 95:552-565 (1997). The inshore shrimping area of Loui- siana is typically the waters land- ward of the barrier islands and the general Gulf of Mexico shoreline. The Louisiana inshore shrimp fish- ery is managed as three geographic zones (Fig. 1) and has two inshore shrimping seasons. Brown shrimp, Penaeus aztecus, dominate spring catches, whereas white shrimp, P. setiferus, dominate fall catches, al- though both species are caught dur- ing each season. In 1992, nearly 101 million kg of shrimp, valued at about $389 million, were landed commercially in the Gulf of Mexico (National Marine Fisheries Service, 1993). From 1986 to 1989, 40% of the total commercial catch in Loui- siana was caught inshore (Baron- Mounce et al.1 ). Although fishing gears and areas fished have varied, a survey in 1987 (Keithly and Baron-Mounce2 ) char- acterized Louisiana’s commercial inshore shrimpers as follows. The average vessel size was 10.2 m for full-time shrimpers, 6.1 m for part- time shrimpers. Smaller boats tended to be constructed of fiber- glass and powered by outboard mo- tors. About 75% to 80% of the com- mercial inshore shrimpers partici- pated in the fishery on a part-time basis. State law limited the size of each trawl to a headrope of 7.6 m length when two trawls were towed in inshore waters, except for Breton and Chandaleur Sounds. The in- shore shrimp fleet was not highly mobile between management zones; only about 10% of the full-time shrimpers with boats in the 20-30 ft range and 2% of the part-time shrimpers fished in more than one zone during either season. The es- timated inshore shrimping effort in 1987 by management zone was 18% (zone 1), 73% (zone 2), and 9% (zone 3)(Keithly and Baron-Mounce2). The otter trawl has been the pri- mary gear used by the inshore shrimp fishery in Louisiana (Keithly and Baron-Mounce2), although but- terfly (wing) nets, cast nets, and skimmer (bay sweepers) nets have 1 Baron-Mounce, E., W. Keithly, and K. J. Roberts. 1991. Shrimp facts. La. Sea Grant Coll. Prog., Communications Office, Louisiana State Univ., Baton Rouge, LA, 22 p. 2 Keithly, W. R., Jr., and E. Baron-Mounce. 1990. An economic assessment of the Louisiana shrimp fishery. Final report to NMFS NA88WC-H-MF179. Coastal Fish- eries Institute, Louisiana State Univ., Ba- ton Rouge, LA, 129 p. Rogers et at: Effectiveness of bycatch reduction devices in Louisiana inshore waters 553 also been used. The minimum legal stretch mesh size at the time of the present study was 3.2 cm; how- ever, shrimpers often use larger mesh to reduce the catch of small shrimp and nontarge ted (by catch) or- ganisms. Some methods that shrimpers have used to reduce bycatch have included relocating to areas of lower fish concentrations, cutting openings in nets, reduc- ing tow speeds before haulback, and modifying nets in various ways. Heightened pressure by environ- mental organizations and pending legislation to re- duce bycatch has furthered the development of shrimp trawls equipped with bycatch reduction de- vices (BRD’s) to reduce the catch of nontargeted or- ganisms. Previous research on BRD designs tested in the United States has been summarized by Watson and Taylor.3 Some of the BRD designs used successfully in other shrimp fisheries have proven ineffective in Gulf of Mexico waters. For example, a horizontal separator panel yielded a 75% reduction in bycatch but lost 30% of the shrimp (Seidel, 1975). Seidel (1975) tested six modifications of the Pacific Northwest shrimp 3 Watson, J. W., and C. W. Taylor. 1990. Research on selective shrimp trawl designs for penaeid shrimp in the United States: a review of selective shrimp trawl research in the United States since 1973. Proceedings ASMFC Fisheries Conservation En- gineering Workshop, Narragansett, RI, April 1990, 21 p. separator trawl, which has a vertical separator panel and several chutes for fish escapement. Shrimp losses ranged from 9.1% to 63.5%, and fish reduction ranged from 37% to 83.5%; however, the modification with the best fish reduction had a shrimp loss of 63.5%. The lowest attainable shrimp loss (6%) from a trawl with vertical separator panels of varying mesh had a 45% bycatch reduction (Watson and McVea, 1977). The Gulf has a high diversity of bycatch species, many of which are similar in size to shrimp; shrimp, however, may represent as little as 10% of the total catch (Seidel, 1975). Prior to this study, most evalu- ations of BRD’s in the Gulf had been conducted in offshore waters. Inshore organisms are often smaller than those caught offshore, inshore trawls and ves- sels are typically smaller, and trawling conditions, such as water depth and turbidity, may differ. Be- cause of these differences, the present study was designed to determine the performance of four BRD’s in inshore waters of Louisiana. Materials and methods Bycatch reduction devices To gather regional expertise on trawling and BRD design, an advisory committee of shrimpers, net makers, and fishery-related agency personnel was 554 Fishery Bulletin 95(3), 1997 & & & Top oblique view Side view Hoop Figure 2 Diagrams of the bycatch reduction devices used in this study: (A) Authement- Ledet excluder; (B) Lake Arthur excluder; (C) Cameron shooter; and (D) Eymard accelerator. organized. The industry commit- tee members recommended basic trawl specifications and sampling areas. BRD’s were selected from a pool of 14 industry- and NMFS- developed BRD and turtle ex- cluder device (TED) designs sug- gested by members of the com- mittee. Four industry-developed BRD designs were selected: Au- thement-Ledet excluder, Lake Arthur excluder, Cameron shooter, and Eymard accelerator (Fig. 2). The Cameron shooter, and very similar designs, such as the fisheye and Florida fish exclud- ers, have had the widest use among commercial shrimpers along the U.S. Gulf of Mexico and Atlantic coasts. The other devices have been used on a limited ba- sis in inshore and offshore waters of Louisiana, primarily in certain management zones: Eymard (zone 1), Authement-Ledet (zone 2), and Lake Arthur (zone 3). Seven identical four-seam, ny- lon semiballoon trawls with 6.1-m headrope length were construct- ed; three were used as control nets and four were randomly se- lected for BRD installation. Each net had 3.5-cm stretch mesh in the body (no. 7 twine) and in the codend (no. 15 twine). The Authement-Ledet excluder (Fig. 2A), constructed of 3.5-cm stretch mesh polyethylene web- bing, was 35 meshes long and contained an inclined plane that was angled 20° from the net bot- tom to guide the catch upwards. The inclined plane was 18 meshes wide at the front and 30 meshes wide at the back; the back of the inclined plane was attached 18 meshes from the top seam of the trawl net. Fishes swimming for- ward from the codend were guided by the inclined plane to exit through the 18-mesh-wide, 40- mesh-long bottom opening. The Lake Arthur excluder (Fig. 2B) was constructed by cutting 22 meshes across the top of the trawl Rogers et al.: Effectiveness of bycatch reduction devices in Louisiana inshore waters 555 net beginning 30 meshes from the tail. A3-mm chain was attached to the forward edge of this opening and a 10-mesh x 12-mesh long cover was attached to the rear edge of this opening. Half of a 3.8 x 7.6 cm float (“small float” in Fig. 2B) was attached under, and a 4.5-mm rope was threaded along, the forward edge of this cover. A 13 x 28 cm conical float (“large float” in Fig. 2B) was attached 10 meshes behind the open- ing. The chain and floats created the escape opening. The Cameron shooter (Fig. 2C) was a 30-cm wide, 45-cm long, 15-cm deep half cone of 12.7-mm alumi- num round stock. A 30-mesh opening was cut in the top of the trawl net, and the forward edge of the cone was inserted into the codend, 20 meshes back from the body of the trawl. The semicircular frame open- ing faced the codend and protruded inside the trawl net. The Eymard accelerator (Fig. 2D) had a polyethyl- ene-webbing accelerator funnel, 45 meshes in diam- eter and 24 meshes long surrounded by six 10 x 10 x 10 mesh triangular openings. A 60-cm hoop of rub- ber coated cable was attached to pull the trawl net away from the funnel after initial dive tests indicated that the funnel blocked the escape openings. Personnel from the National Marine Fisheries Service (NMFS) Harvesting Systems Branch exam- ined the trawls several times by using scuba equip- ment, and adjustments were made to the trawl rig- ging and the BRD’s. Dye was injected into various parts of the nets around the devices to observe wa- ter flow, and the behavior of escaping fish was docu- mented. Sampling A6.7-m Boston whaler, powered by twin 115-hp out- board motors and equipped with a single-drum winch and double boom, was used to tow twin trawls off the stern (Harrington et al., 1988); one trawl con- tained a BRD, the other a bare control net. The nets were equipped with tickler chains, connected to an aluminum dummy door, and spread by two 0.6 x 1.07 m pinewood trawl doors. Although twin trawls are not typically towed behind commercial vessels in Louisiana, the use of twin trawls to replace single large trawls off outriggers is increasing, particularly in the Gulf of Mexico (Watson et al., 1984). Smaller inshore vessels in Louisiana, without outriggers, typi- cally tow a single larger net behind the boat. Our twin-trawl rigging configuration was approved by the committee to ensure that the nets sampled an area as closely as possible, given the patchy nature of many species. Nets were towed for 20 minutes (time at towing speed); the speed over ground was main- tained between 2.0 and 2.5 knots (2.2 kn, average) by using a Global Positioning System, as recom- mended by the committee. Tows were made during daylight hours, near commercial shrimp boats when- ever possible. Average water depth and the salinity were recorded for each trawl tow. The four BRD’s were evaluated in each inshore shrimp zone (Fig. 1). Eighteen two-day sampling trips were made, three trips to each area during each sea- son. Each BRD net was towed 72 times over the year. Tows were divided among the areas and seasons. The towing order and trawl side for each BRD were ini- tially selected randomly, although the same nets were not used on consecutive tows owing to the time taken to empty the nets. The three control nets were num- bered and alternated to ensure equal pairing with a particular BRD net throughout the study. Sampling was conducted during the 1992 inshore shrimp sea- sons, although the short spring season necessitated sampling the week before the season opened in zone 1 and a few days after the season in zone 3. This sched- ule was approved by the advisory committee. Samples were tagged and placed in mesh bags in ice and water. In the laboratory, organisms in each sample were identified, counted, and the biomass of each species in a sample was weighed to the nearest 0.1 g. When numerous, individuals of a species within a sample were subsampled and the total number es- timated by weight. Standard lengths of most fishes, carapace widths of crabs, and total lengths of penaeid shrimp were measured. Organisms were measured in 5-mm length increments, designated by the lower end of the length range (e.g. 10-mm class=10.0 to 14.9 mm). Statistical analysis Residuals were examined for univariate normality and homogeneity of variances prior to accepting the analysis of variance model. Normality was tested with the Wilk-Shapiro test, and a modified Levene test was used to test for homogeneity of variances. These tests indicated that the raw data were not dis- tributed normally and variances were not homoge- neous. The transformation ln(catch+l) was used to create a new variable that met the criteria of being approximately normally distributed with homoge- neous variances. This transformed variable was used in the analysis of variance (ANOVA). Statistical analysis was performed by using the Statistical Analysis System (SAS). Control nets The transformed catch ( both numbers and biomass of abundant species) of the three con- trol nets was used as the dependent variable in an ANOVA with season, area, and season-by-area terms 556 Fishery Bulletin 95(3), 1 997 and with the interaction of these terms with the con- trol net number, with tow as the experimental unit. Length-frequency distributions of abundant organ- isms collected by the control nets in each area were visually examined. IBRD's versus control nets Catches of shrimp, fish, and the nine most abundant species were analyzed. An AN OVA model was used to compare the differ- ence between control and BRD net catches between areas and seasons. ANOVA was also used to detect differences caused by towing a BRD on the port or starboard side of the twin trawl. The number of individuals and biomass of each species (or group) caught in a BRD net was compared with the number caught by the control net by using a univariate paired t-test. The univariate procedure is appropriate if one can assume that the probability of one species being retained within the net is inde- pendent of another species being retained. The short tow duration contributes to the chance that this as- sumption is valid. Paired Ctests were conducted on untransformed and log-transformed differences be- tween BRD- and control-net catches. Percent catch differences of untransformed data were calculated to compare device nets: Percentage catch Device net - Control net = x 100. difference Control net Percent catch difference values could range from -100 to infinity. Differences between areas and seasons A univariate paired t-test was also used to compare differences between a BRD net and the control net within each area. This test had less power because the sample size was reduced by two-thirds and be- cause the test was not able to detect as small a dif- ference as the test with the areas combined. A simi- lar analysis was conducted to examine device per- formance in each season. For all analyses, differences between means with an alpha of 0.05 or less were considered significant. How- ever, the exact probabilities are presented in the tables. Results Control nets The control nets collected 88 species of fishes and invertebrates; fewer species were collected in the spring than in the fall in all areas (Table 1). More than 64% of the 84,919 organisms collected in the control nets were caught during the spring. Nine species represented nearly 89% of the total control- net catch. Bay anchovy, white shrimp, and hardhead catfish catches were higher in the fall, but the other Table 1 Numbers of most abundant organisms collected in the control nets in inshore waters of Louisiana during the spring and fall of 1992. Spring Fall Area Area Combined Species Borgne Bar re Calcasieu Total Borgne Barre Calcasieu Total total Brown shrimp Penaeus aztecus 2,842 5,375 10,593 18,810 400 481 245 1,126 19,936 Atlantic croaker Micropogonias undulatus 597 8,346 5,162 14,105 240 207 2,015 2,462 16,567 Bay anchovy Anchoa mitchilli 712 976 1,106 2,794 1,011 2,387 2,166 5,564 8,358 White shrimp Penaeus setiferus 47 51 732 830 1,954 3,425 1,752 7,131 7,961 Hardhead catfish Arius felis 423 363 2,638 3,424 1,087 1,432 1,864 4,383 7,807 Spot Leiostomus xanthurus 846 2,059 1,603 4,508 180 161 339 680 5,188 Sand seatrout Cynoscion arenarius 353 196 2,097 2,646 153 689 408 1,250 3,896 Blue crab Callinectes sapidus 149 1,610 154 1,913 111 907 13 1,031 2,944 Gulf menhaden Brevoortia patronus 86 278 1,533 1,897 54 68 623 745 2,642 Other species 273 2,108 1,152 3,533 1,263 2,273 2,551 6,087 9,620 Total 6,328 21,362 26,770 54,460 6,453 12,030 11,976 30,459 84,919 Number of species 38 51 53 66 47 65 57 82 88 Rogers et ai : Effectiveness of bycatch reduction devices in Louisiana inshore waters 557 six species were much more abundant in the spring. The catches from Lake Borgne were typically much smaller than catches from the other areas. Brown shrimp, hardhead catfish, sand seatrout, and gulf men- haden were most abundant in Calcasieu Lake, whereas Atlantic croaker and blue crab were most abundant in Lake Barre. Penaeid shrimp constituted 36% of the catch in the spring and 27% of the catch in the fall. The numbers of organisms collected by the three control nets did not differ significantly. Control-net catches did not differ significantly with respect to the excluder with which they were paired. Length- frequency distributions of the abundant species dif- fered between areas (Fig. 3). The side of the trawl on which the control or BRD net was towed did not significantly affect catches of the abundant species. Each BRB was towed equally on each side of the twin trawl. BRD's versus control nets Fish All BRD nets had significantly different catches of fish from those of the control nets. Nu- merically, the Cameron BRD had the highest overall reduction of fish (-51%) compared with the catch of the control nets (Table 2). The Eymard BRD caught 26% more fish than the control nets. In terms of bio- mass, the Authement-Ledet BRD had the highest re- duction (-42%), and the Eymard BRD had a 19% lower catch than the control nets. The Authement-Ledet, Lake Arthur, and Cameron BRD nets caught fewer fish than the control nets in all size categories (Fig. 4). The Cameron BRD, in particular, had the highest reduction of small fish (<75 mm). The Eymard BRD caught more small fish (<85 mm) and fewer large fish than the control net. The Cameron BRD had the best reduction in num- bers of Atlantic croaker (49%), and the Authement- Ledet the best reduction in biomass (39%) (Table 3). The Authement-Ledet and Eymard BRD’s caught 50% or fewer spot in terms of numbers and biomass; in contrast, the Cameron had very poor reductions for spot. Both the Cameron and Authement-Ledet BRD’s caught 50% or fewer hardhead catfish than the control nets. The Cameron BRD caught 75% fewer bay anchovy than the control net, and the Authement-Ledet and Lake Arthur reduced bay an- chovy by 37%. For most bycatch species, the Eymard BRD caught more than the control nets, although catches of the bay anchovy were markedly higher (83% numbers, 86% biomass). Shrimp The catch of shrimp with all BRD nets dif- fered significantly from the control net catch. The Cameron, Authement-Ledet, and Lake Arthur BRD’s caught fewer shrimp than the control nets; shrimp catch with the Eymard was higher (38% numbers, 25% biomass) (Table 2). Numerically, the Cameron BRD had 16% fewer shrimp, and both the Cameron and Authement-Ledet had 14% lower shrimp bio- mass than the control nets. Most of the catch difference between the BRD nets and corresponding control nets appeared to be smaller (<85-90 mm) shrimp (Fig. 5). Catch differ- Table 2 Comparison of numbers and biomass of fish and shrimp collected in bycatch reduction nets and corresponding control nets in selected inshore waters of Louisiana in 1992. SDis the standard deviation of the difference. Significance levels are 0.01 (**) and 0.05 (*). n= 72. Superscripted letters denote significance levels of paired t-tests on log-transformed data: 0.01 (“). BRD = bycatch reduction device. Numbers Biomass (g) Mean catch/tow Percent Mean catch/tow Percent Type of catch and BRD Control Device catch difference SD P>t-value Control Device catch difference SD P>t-value Fish Authement-Ledet 170.5 109.0 -36 102.6 0.01**“ 2,541.8 1,464.1 -42 1,363.1 0.01**“ Lake Arthur 171.4 134.7 -21 124.9 0.01**“ 2,904.4 2,282.6 -21 1,616.5 0.01**“ Cameron 181.7 88.3 -51 109.8 0.01**“ 3,068.5 2,068.1 -33 1,233.3 0.01**“ Eymard 190.4 239.7 26 167.6 0.01**“ 2,980.3 2,406.3 -19 2,059.8 0.02*“ Shrimp Authement-Ledet 84.8 69.3 -18 58.1 0.03*“ 483.1 417.5 -14 240.7 0.02* Lake Arthur 93.2 70.9 -24 57.5 0.01**“ 514.0 425.0 -17 218.1 0.01**“ Cameron 110.9 93.0 -16 77.7 0.05*“ 579.3 500.3 -14 220.5 0.01**“ Eymard 98.9 136.4 38 97.4 0.01**“ 517.2 645.3 25 340.1 0.01**“ 558 Fishery Bulletin 95(3), 1997 Jjjiian 500 blue crab 250 - Length class (mm) Length class (mm) Figure 3 Length-frequency distributions of abundant species collected in control nets for the different areas. Note that length-class data are in mm, except those for catfish, which are in cm. ences of brown shrimp and white shrimp differed for the BRD’s (Table 3); fewer brown shrimp tended to be caught with the BRD nets. Differences between areas and seasons Although the Authement-Ledet BRB caught 18% fewer shrimp than the control nets overall, losses in Lake Barre were low and statistically nonsignificant (Table 4). Fish reduction was consistent across the areas for this BRD. The Lake Arthur BRD had a fairly consistent reduction of shrimp and fish across all areas but had the lowest shrimp catch difference in Lake Borgne and had poorer fish reductions in Calcasieu Lake. The Cameron BRD had the highest fish reduction in Lake Borgne; however, this was accompanied by the highest shrimp loss. This BRD lost the fewest shrimp in Lake Barre. The Eymard BRD caught more shrimp than the control net in all areas and reduced fish biomass in all areas, by as much as 35% in Lake Borgne. Mean water depths and salinities differed between Lake Borgne and the other two areas. Lake Borgne (2.7 ±0.64 m) was slightly deeper than Lake Barre (1.9 ±0.34 m) and Calcasieu Lake (1.5 ±0.27 m). Mean salinities during sampling were 9.6 ±2.9%e (Lake Borgne), 22.5 ±2.8%c (Lake Barre), and 20.6 ±5.2%£> (Calcasieu Lake). There were some slight differences in BRD perfor- mance between seasons (Table 5), reflecting differ- Rogers et al.: Effectiveness of bycatch reduction devices in Louisiana inshore waters 559 ences in species composition and length-frequency distributions. Discussion The Authement-Ledet, Lake Arthur, and Cameron BRD’s significantly reduced bycatch, but also the catch of shrimp. Excluded shrimp were primarily in the smaller size classes. The Eymard BRD caught significantly more fish and shrimp than the corre- sponding control nets. The Lake Arthur and Cameron BRD’s were de- signed with a similar opening, but the Lake Arthur BRD did not release as many fish. Because the weight of the aluminum frame of the Cameron BRD caused the device to sink slightly, the bottom of the Cameron opening was only about 15 cm from the bottom of the trawl net. In contrast, the floats of the Lake Arthur BRD raised the opening to about 30 cm above the trawl bottom. A design somewhat similar to the Lake Arthur BRD and placed 1.7 m from the end of the net did not significantly reduce bycatch in inshore waters of Alabama (Wallace and Robinson, 1994). Reductions with the Cameron BRD (16% shrimp, 51% fish) were similar to those found by Watson et al. (1993) for the fisheye top position excluder in off- shore waters (17% shrimp, 70% fish). Inshore fish are typically smaller and less able to escape by swim- ming; this may account for the lower reductions with the Cameron BRD. This BRD also released shrimp of most size classes; Watson and McVea ( 1977 ) found that the fish escape device, a somewhat similar de- vice, also lost shrimp over the entire size range. Changing the location of the Cameron shooter may affect performance, although Watson et al. (1993) noted that the top position appeared to have the best effectiveness for fish reduction and shrimp retention. A bottom-mounted Florida fish shooter, placed 1.7 m from the end of a 4.9-m trawl, reduced bycatch 26% by weight and 46% by number and caught 14% fewer shrimp than an unmodified net (Wallace and Robinson, 1994). McKenna and Monaghan4 reported that the efficiency of the Florida fish excluder de- pended on the size of the escape opening, placement of the excluder in the net, and the number of devices installed. 4 McKenna, A., and J. P. Monaghan Jr. 1993. Gear development to reduce bycatch in the North Carolina trawl fisheries. Completion report to Gulf and South Atlantic Fisheries Develop- ment Foundation Cooperative Agreement No. NA90AA-H-SK052. North Carolina Div. Mar. Fish., Morehead City, NC, 79 p. 560 Fishery Bulletin 95(3), 1997 The Eymard BRD was developed to reduce the catch of larger hardhead catfish, particularly in Loui- siana waters east of the Mississippi River. Hardhead catfish reductions were 6% in terms of numbers, but 51% in biomass, resulting from the loss of larger cat- fish. Overall, the Eymard BRD caught 26% more fish, but fish biomass was 19% less than that caught by the control net. This finding was the result of the BRD catching more fish smaller than 80 mm and fewer large fish than the control nets. The Eymard BRD caught significantly more numbers and bio- mass of shrimp, particularly smaller shrimp. The Eymard BRD design contained a webbing funnel, designed to carry shrimp and fish into the codend with accelerating water flow (Watson5 ). Because swimming speed of a fish is a function of size (Blaxter and Dickson, 1958), smaller fishes may not be able to swim in increased water flow, with the result that fewer shrimp and small fish can escape from the Eymard BRD than from a control net. However, dye released into the Eymard indicated that the water flow was not perceptibly increased by the funnel, probably because the funnel diam- eter was only slightly smaller than the net diameter. However, the Eymard BRD had a 21.6-cm greater net spread than that of the control net; this greater net opening may have resulted in the higher catches of many species. Because the nets were otherwise constructed identically, we suspect this difference was most likely due to the presence of the hoop. The polyethylene webbing may have increased the incidence of anchovy be- ing gilled, particularly during haul- back. Numerous small bay anchovies were found, upon retrieval, to be gilled in the polyethylene webbing of the Eymard BRD, and the device caught 83% more bay anchovy than the con- trol net. In a subsequent study, a poly- ethylene net caught 245% more ancho- vies than a nylon net (Rogers et al.6 ). Fish were observed escaping from several of the BRD’s during diver evaluations in Florida. Divers ob- served several large juvenile pinfish (Lagodon rhomboides ) escaping from the bottom openings of the Eymard BRD and numerous juvenile pinfish escaping the Authement-Ledet BRD 5 Watson, J. W. 1988. Fish behaviour and trawl design: potential for selective trawl development. In S. G. Fox and J. Hunting- ton (eds.), Proceedings of the world sympo- sium on fishing gear and fishing vessel de- sign, p. 25-29. Newfoundland and Labra- dor Institute of Fisheries and Marine Tech- nology, St. John’s, Newfoundland. 6 Rogers, D. R., B. D. Rogers, J. A. de Silva, and V. L. Wright. 1994. Evaluation of shrimp trawls designed to reduce bycatch in inshore waters of Louisiana. School of For- estry, Wildlife, and Fisheries, Louisiana State Univ. Agricultural Center. Final re- port submitted to NMFS, St. Petersburg, FL. NOAA Award No. NA17FF0375-01, 230 p. Available from LSU library. Rogers et al.: Effectiveness of bycatch reduction devices in Louisiana inshore waters 561 Table 3 Comparison of numbers and biomass of the most abundant species collected in bycatch reduction device (BRD) nets and corre- sponding control nets. SD is the standard deviation of the difference. Significance levels are 0.01 (”) and 0.05 (*). n= 72. Superscripted letters denote significance levels of paired t-tests on log-transformed data: 0.01 (“), 0.05 ( * ). Numbers Biomass (g) Species and BRD Mean catch/tow Percent catch difference Mean catch/tow Percent catch difference Control BRD SD P>t-value Control BRD SD P>t-value Atlantic croaker Authement-Ledet 58.0 38.0 -34 67.9 0.01**“ 832.9 509.8 -39 711.1 0.01** Lake Arthur 56.8 41.6 -27 63.4 0.05*6 783.6 627.8 -20 450.9 0.01**“ Cameron 58.5 29.6 -49 64.8 0.01** 859.8 570.2 -34 524.7 0.01** Eymard 56.9 83.6 47 117.6 0.06* 789.7 871.3 10 907.3 0.45 Spot Authement-Ledet 12.2 5.6 -54 12.4 0.01**° 294.4 110.7 -62 318.0 0.01**“ Lake Arthur 25.3 15.2 -40 52.3 0.10 575.5 372.7 -35 1,121.1 0.13 Cameron 12.0 9.8 -18 6.7 0.01** 322.5 269.4 -16 221.4 0.05* Eymard 22.6 11.2 -50 49.4 0.05*° 562.0 234.0 -58 1,137.9 0.02*“ Sand seatrout Authement-Ledet 14.1 8.8 -38 26.3 0.09* 101.0 63.5 -37 114.9 0.01** Lake Arthur 9.8 8.5 -13 9.6 0.27 132.1 110.5 -16 138.8 0.19 Cameron 13.3 4.8 -64 14.5 0.01**“ 119.8 69.5 -42 105.5 0.01**“ Eymard 17.0 20.1 19 21.7 0.22* 165.5 160.1 -3 209.1 0.83 Hardhead catfish Authement-Ledet 22.3 11.0 -51 24.0 0.01**“ 625.5 244.9 -61 596.6 0.01**“ Lake Arthur 26.2 20.8 -21 31.2 0.14“ 729.6 548.4 -25 380.4 0.01**“ Cameron 35.3 14.4 -59 61.7 0.01**“ 960.0 549.1 -43 784.2 0.01**“ Eymard 24.6 23.2 -6 51.3 0.82 802.0 393.7 -51 1,045.0 0.01** Bay anchovy Authement-Ledet 29.3 18.3 -37 34.6 0.01** 40.1 22.1 -45 45.1 0.01*** Lake Arthur 23.3 14.7 -37 30.2 0.02** 35.3 25.0 -29 49.3 0.08* Cameron 27.8 7.0 -75 31.2 0.01**“ 36.7 9.5 -74 41.8 0.01**“ Eymard 35.8 65.4 83 67.4 0.01**“ 43.4 80.8 86 79.8 0.01**“ Gulf menhaden Authement-Ledet 12.8 10.0 -22 18.5 0.20 135.2 110.8 -18 170.0 0.23 Lake Arthur 7.8 8.5 9 14.2 0.67 96.0 106.7 11 161.7 0.57 Cameron 7.6 7.1 -7 12.5 0.71 82.1 82.0 0 126.3 0.99 Eymard 8.5 12.1 42 22.8 0.19 96.0 109.7 14 173.0 0.50 Blue crab Authement-Ledet 9.9 9.0 -9 6.9 0.25 430.3 352.1 -18 353.3 0.06 Lake Arthur 9.0 6.4 -28 7.6 0.01** 378.6 295.9 -22 326.0 0.03* Cameron 11.4 9.3 -19 15.8 0.24 571.0 457.2 -20 888.7 0.28 Eymard 10.5 9.3 -12 6.3 0.09 574.4 410.0 -29 647.9 0.03* Brown shrimp Authement-Ledet 60.0 46.7 -22 52.0 0.03** 332.4 270.7 -19 197.7 0.01** Lake Arthur 64.3 48.2 -25 54.6 0.01**“ 333.8 268.8 -19 181.7 0 oi*** Cameron 78.5 64.0 -19 72.4 0.09 375.6 315.6 -16 170.2 0.01** Eymard 73.4 99.1 35 85.8 0.01** 356.3 438.6 23 273.4 0.01** White shrimp Authement-Ledet 24.6 22.3 -9 22.9 0.40 150.3 146.3 -3 116.5 0.77 Lake Arthur 28.7 22.7 -21 20.9 0.02*“ 179.6 156.2 -13 137.4 0.15“ Cameron 32.0 28.8 -10 24.6 0.28 202.9 184.5 -9 124.1 0.21 Eymard 25.4 37.0 46 47.5 0.04** 160.8 206.2 28 209.1 0.07 opening while the nets were being towed. Few fish were observed escaping from the Cameron and Lake Arthur BRD openings during these tests. Although each BRD net contained escape openings, smaller species, such as the bay anchovy, could have escaped through the codend meshes. The devices may 562 Fishery Bulletin 95(3), 1997 have affected escape rates through the meshes by altering the shape of the codend. The percentage of fish and shrimp that escaped during trawling, as opposed to escaping during haulback, is also un- known. Watson et al. (1993) reported that most spe- cies escaped through escape openings during trawl haulback or when fish were crowded near the open- ings. Further diver evaluations are necessary to iden- tify methods by which fish and shrimp escape. Differences in catch rates of brown and white shrimp observed for many of the BRD’s may have been due to species-specific behavior or size differ- ences (or both). White shrimp swim more actively than brown shrimp during the day (Wickham and Minkler, 1975). The white shrimp caught by the con- trol nets were larger, on average, than the brown shrimp, a finding that is typical for Louisiana catches (Keithly and Baron-Mounce2). In addition, the white shrimp caught in the spring were substantially larger than those in the fall. These data are derived from a fishery-independent study; results from commercial shrimping could dif- fer. Had the Eymard BRD been used in a larger trawl and without a hoop, the results might have been quite different. Many fish and shrimp may have been lost during haulback and although we had mechanical re- trieval, a larger commercial vessel may have had faster retrieval. Al- though we trawled near shrimp boats whenever possible, at times no shrimp boats were present in a sam- pling area. When this was the case, we began trawling in an area where shrimp had been caught previously; if few or no shrimp were caught, we moved to another area. Moving short distances (one or two km) could re- sult in very different catches. Be- cause of time and fuel limitations, however, movements of very long dis- tances were not feasible. Provided that shrimp were being caught, we did not relocate if large quantities of fishes or crabs were also present. In this situation, a shrimper would most likely relocate in an attempt to find more shrimp or cease shrimping un- til conditions in the area become more favorable. We found higher ra- tios of fish to shrimp when shrimp catches were low. Other studies have reported that bycatch ratios depend on shrimp abundance; when few shrimp are present, fishing times are longer and result in high catches of bycatch species (Adkins7 ). The 20- minute tows used in our study were three to six times shorter than those typically used in commercial opera- tions. Longer tows would have neces- sitated decreasing the number of 7 Adkins, G. 1989. A comprehensive as- sessment of bycatch in the Louisiana shrimp fishery. Final report to NMFS NA89WC- H-MF006. La. Dep. Wildlife Fish., Bourg, LA, 75 p. 2,700 1,800 900 0 2,700 1,800 900 V////. Control Device Authement - Ledet Lake Arthur Length class (mm) Figure 5 Length-frequency distribution of shrimp collected by the four devices and cor- responding control nets. Rogers et a\. : Effectiveness of bycatch reduction devices in Louisiana inshore waters 563 Table 4 Comparisons of numbers and biomass of fish and shrimp collected by the four bycatch reduction device (BRD) nets and corre- sponding control nets in the three areas. SD is the standard deviation of the difference. Significance levels are 0.01 (**) and 0.05 (*). n= 24. Superscripted letters denote significance levels of paired t-tests on log-transformed data: 0.01 (“), 0.05 t b ). Numbers Biomass (g) BRD and type of catch Mean catch/tow Percent catch difference Mean catch/tow Percent catch difference Area Control Device SD P>t-value Control Device SD P>t-value Authement-Ledet Fish L. Borgne 64.6 38.4 -41 45.7 0 Ql**“ 1,119.9 732.7 -35 588.1 0.01**“ L. Barre 202.7 134.8 -34 122.6 0.01**“ 2,425.2 1,458.9 -40 1,302.6 0 .01**“ Calcasieu L. 244.0 153.8 -37 114.7 0.01**“ 4,080.2 2,200.8 -46 1,584.7 0.01**“ Shrimp L. Borgne 49.0 35.8 -27 21.3 0.01** 381.7 285.4 -25 178.1 0.01** L. Barre 90.5 86.7 -4 40.1 0.64 613.6 579.7 -6 233.9 0.49 Calcasieu L. 115.1 85.5 -26 89.4 0.126 454.1 387.5 -15 300.6 0.29 Lake Arthur Fish L. Borgne 65.4 47.0 -28 36.4 0.02*ft 1,365.2 1,072.9 -21 638.6 0.03* L. Barre 215.3 151.3 -30 149.3 0.05* 3,046.8 2,263.0 -26 2,441.9 0.13 Calcasieu L. 233.4 205.6 -12 152.9 0.38 4,301.0 3,511.9 -18 1,235.3 0.01**“ Shrimp L. Borgne 57.0 47.0 -18 20.1 0.02*“ 389.1 350.6 -10 142.5 0.20“ L. Barre 95.3 70.3 -26 39.5 0 oi**“ 649.4 509.4 -22 239.9 0.01**“ Calcasieu L. 127.4 95.4 -25 89.4 0.09 503.5 415.1 -18 252.1 0.10 Cameron Fish L. Borgne 89.8 32.5 -64 113.0 0.02*“ 1,987.7 1,170.2 -41 1,361.9 0.01**“ L. Barre 188.9 98.6 -48 75.1 0 oi**“ 2,693.3 1,776.4 -34 869.6 0 .01**“ Calcasieu L. 266.5 133.9 -50 125.9 0.01**“ 4,524.6 3,257.7 -28 1,402.8 0.01**“ Shrimp L. Borgne 59.8 45.0 -25 23.6 0 oi**“ 419.8 338.2 -19 145.7 0 oi*** L. Barre 108.2 97.5 -10 42.8 0.23 719.6 655.1 -9 225.8 0.18 Calcasieu L. 164.8 136.6 -17 126.7 0.296 598.4 507.7 -15 278.4 0.12 Eymard Fish L. Borgne 81.6 114.4 40 83.9 0.07“ 1,970.0 1,288.0 -35 1,394.5 0.03* L. Barre 201.8 301.0 49 193.1 0.02*“ 2,724.3 2,554.9 -6 2,158.8 0.70 Calcasieu L. 287.8 303.7 6 195.7 0.69 4,246.5 3,376.1 -20 2,493.5 0.106 Shrimp L. Borgne 52.7 84.1 60 79.2 0.06“ 365.2 510.4 40 343.3 0.05* L. Barre 97.5 120.6 24 34.1 0 oi**“ 672.8 730.8 9 166.2 0.106 Calcasieu L. 146.5 204.5 40 145.4 0.06“ 513.5 694.8 35 450.7 0.06“ trips or evaluating fewer BRD’s. Increasing the trawl- tow duration decreases the ability of a fish to main- tain swimming speed (Bainbridge, 1960). Reductions over longer tow periods may differ; if most reduction occurs during haulback, fish may be too exhausted to escape. Longer tows also increase the chances of catching large quantities of fish and shrimp that may clog the net and cause organisms to be released from the BRD. In terms of abundances and size distributions, bycatch varied between the areas and seasons; some species were very abundant in one or two areas. The capability of a BRD to reduce fish or shrimp depends on the species assemblage present in an area. A BRD may work well in one area under certain conditions but perform poorly in another area owing to assem- blage differences. Species-specific size selectivity has been reported in other studies (e.g. Rulifson et al., 1992). Of the four BRD’s, the Cameron had the best overall fish reduction. However, if spot and gulf men- haden were the most abundant species in an area, the Authement-Ledet may be a better choice. By catch reduction devices may have to be selected for par- ticular areas or seasons, depending on the type and size distributions of predominant bycatch species, because a particular device may not be as effective in all areas or at all times of the year. The high shrimp losses from the BRD’s evaluated in this study would most likely be unacceptable for commercial operations. However, further modifica- tions to these devices, such as altering the size or location of escape openings, could reduce these losses. 564 Fishery Bulletin 95(3), 1 997 Table 5 Comparison of numbers and biomass of fish and shrimp collected by the four bycatch reduction device (BRD) nets and correspond- ing control nets for the different seasons. SD is the standard deviation of the difference. Significance levels are 0.01 (**) and 0.05 (*). n= 36. Superscripted letters denote significance levels of paired i-tests on log-transformed data: 0.01 (a), 0.05 (*). Numbers Biomass (g) BRD and type of catch Mean catch/tow Percent catch difference Mean catch/tow Percent catch difference Season Control Device SD P>£-value Control Device SD P>£-value Authement-Ledet Fish spring 218.3 137.8 -37 127.7 0.01**“ 2,895.3 1,667.1 -42 1,374.2 0.01**“ fall 122.6 80.2 -35 65.5 0.01**“ 2,188.3 1,261.2 -42 1,354.2 0.01**“ Shrimp spring 116.6 92.1 -21 73.2 0.05** 650.3 556.6 -14 288.7 0.06* fall 53.1 46.5 -12 36.4 0.29* 316.0 278.5 -12 180.4 0.22 Lake Arthur Fish spring 216.1 165.2 -24 157.1 0.06 3,459.1 2,761.7 -20 2,119.1 0.06* fall 126.7 104.2 -18 80.9 0.10* 2,349.6 1,803.6 -23 893.6 0 oi*** Shrimp spring 124.8 96.5 -23 75.4 0.03** 642.3 541.1 -16 244.8 0.02** fall 61.7 45.3 -27 31.0 0.01**“ 385.7 308.9 -20 190.3 0.02*“ Cameron Fish spring 203.1 106.5 -48 101.5 0.01**“ 3,296.6 2,229.1 -32 1,187.8 0.01**“ fall 160.3 70.2 -56 118.9 0.01**“ 2,840.4 1,907.1 -33 1,290.5 0.01**“ Shrimp spring 157.9 134.9 -15 104.4 0.19“ 773.1 677.9 -12 252.8 0.03*“ fall 64.0 51.2 -20 35.8 0.04* 385.4 322.8 -16 184.8 0.05* Eymard Fish spring 233.8 301.1 29 197.3 0.05*“ 3,386.9 2,773.0 -18 2,029.4 0.08* fall 147.1 178.3 21 131.7 0.16* 2,573.8 2,039.7 -21 2,117.8 0.14 Shrimp spring 145.1 199.8 38 119.5 0.01**“ 716.1 883.0 23 380.7 0.01**“ fall 52.6 73.0 39 65.9 0.07“ 318.2 407.7 28 294.3 0.08 The BRD nets tested here did not appear to slow water flow in the trawl net. Other studies, however, have indicated that flow rate around and through the BRD may be a key factor in fish and shrimp es- capement. Watson et al. (1993) found that juvenile fish could exit a BRD at flow rates between 0.2 and 0.5 m/sec. However, shrimp accumulated in areas of reduced flow and crawled along the webbing against the flow to escape some devices (Watson et al., 1993). Devices can be designed to create a 0.2 to 0.5 m/sec flow rate, but debris can alter the flow rate and af- fect BRD performance. The ability to sustain swim- ming appears to be related to length, but this rela- tionship often differs for each species (Bainbridge 1960). Further testing is necessary to acquire escape flow rates for the major species of concern. Reduction rates for numbers and biomass of many species differed for the four BRD’s, reflecting size- dependent selectivity. Escape rates of different spe- cies also varied considerably owing to differences in size and behavior. These differences, coupled with the high variability in organisms between areas, in- dicate that the performance of BRD’s should be evalu- ated at the species and size level. Future studies should continue to involve mem- bers of the industry. The advisory committee provided suggestions and valuable insight that greatly en- hanced the success of this project and the accept- ability of the results. Other studies have reported successful industry involvement (Rulifson et al., 1992; McKenna and Monaghan4). The design and construction of BRD’s should be a dynamic process which will benefit from the cooperation of industry, research, and management personnel. Acknowledgments This paper is funded in part by a cooperative agree- ment with the National Oceanic and Atmospheric Administration (NOAAAward No. NA17FF0375-01). Additional funding was provided by the Louisiana State University Agricultural Center. We are espe- cially grateful to J. Watson, I. Workman, W. Taylor, Rogers et al.: Effectiveness of bycatch reduction devices in Louisiana inshore waters 565 and J. Barbour of the NMFS Harvesting Systems Branch in Pascagoula, MS, for technical guidance and diver evaluations. We thank V. Hebert, T. McGuff, and P. Williams for technical assistance and W. Herke, J. Watson, P Williams, and C. Wilson for criti- cal reviews. We also extend our thanks to the Indus- try Advisory Committee: L. Authement, D. Bankston, J. Black, C. Boudreaux, P. Bowman, L. Brunet Jr., P. Cantrelle, S. Charpentier, C. Cheramie, P. Coreil, S. Corkern, W. Delacroix, J. Duhon, P. Gisclair, C. Guidry, J. Horst, C. J. Kiffe, D. Kiffe, C. Ledet, D. Lirette, G. Maggiore, A. Matherne, T. J. Mialjevich, L. Pelas, K. Savoie, C. Terrebonne, T. Terrebonne, P. Thibodeaux, and J. Zeringue. Literature cited Bainbridge, R. 1960. Speed and stamina in three fish. J. Exp. Biol. 37:129-153. Blaxter, J. H. S., and W. Dickson. 1958. Observations on the swimming speeds of fish. J. Cons. Cons. Int. Explor. Mer 24:472-479. Harrington, D. L., J. W. Watson, L. G. Parker, J. B. Rivers, and C. W. Taylor. 1988. Shrimp trawl design and performance. Ga. Sea Grant Coll. Prog., Univ. of Georgia, Athens, Mar. Ext. Bull. No. 12, 41 p. National Marine Fisheries Service. 1993. Fisheries of the United States, 1992. U.S. Dep. Commer., NOAA, NMFS, Curr. Fish. Stat. 9200, 115 p. Rulifson, R. A., J. D. Murray, and J. J. Bahen. 1992. Finfish catch reduction in South Atlantic shrimp trawls using three designs of by-catch reduction devices. Fisheries (Bethesda) 17(0:9-20. Seidel, W. R. 1975. A shrimp separator trawl for the southeast fisheries. Proc. Gulf Caribb. Fish. Inst. 27:66-76. Wallace, R. K., and C. L. Robinson. 1994. Bycatch and bycatch reduction in recreational shrimping. Northeast Gulf Sci. 13(2): 139—144. Watson, J. W., and C. McVea Jr. 1977. Development of a selective shrimp trawl for the south- eastern United States penaeid shrimp fisheries. Mar. Fish. Rev. 39(10): 18-24. Watson, J. W., I. K. Workman, C. W. Taylor, and A. F. Serra. 1984. Configurations and relative efficiencies of shrimp trawls employed in southeastern United States waters. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 3, 12 p. 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 south- eastern United States 1990-1992. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-327, 131 p. Wickham, D. A., and F. C. Minkler III. 1975. Laboratory observations on daily patterns of burrow- ing and locomotor activity of pink shrimp, Penaeus duorarum, brown shrimp, Penaeus aztecus, and white shrimp, Penaeus setiferus. Contrib. Mar. Sci. 19:21-35. Global population structure of yellowfin tuna, Thunnus albacares, inferred from allozyme and mitochondrial DIMA variation Robert D. Ward Nicholas G. Elliott Bronwyn H. Innes Adam J. Smolenski Peter M. Grewe CSIRO Division of Marine Research GPO Box 1538, Hobart, Tasmania 7001, Australia E-mail address (for Robert Ward): Bob.Ward@marine.csiro.au Abstract .—Yellowfin tuna, Thun- nus albacares, were sampled from one region of the Atlantic Ocean, two re- gions of the Indian Ocean, and six re- gions of the Pacific Ocean. One of the Indian Ocean collections could not be allozymically analyzed; the remaining eight collections were examined for four polymorphic allozyme loci (ADA*, FH*, GPI- A*, and GPI-B*, rc=540 to 677). All nine collections were examined for mi- tochondrial DNA variation (n= 767), with two restriction enzymes (Bel I and Eco RI ) that detect polymorphic restric- tion sites in yellowfin tuna. Allele fre- quencies at three of the allozyme loci were homogeneous across collections, whereas GPI-A* showed highly signifi- cant differentiation (PcO.OOl). The GPI-A* data, taken together with the geographic location of the collections, suggested the existence of at least four yellowfin tuna stocks: Atlantic Ocean, Indian Ocean, west-central Pacific Ocean, and east Pacific Ocean. Mito- chondrial DNA differentiation was more limited, but spatial heterogene- ity of the 24 observed haplotypes over the nine regions (P=0.048) and three oceans (P=0.009) was significant. The mtDNA data did not differentiate west- central Pacific Ocean collections from east Pacific Ocean collections but did support the separation of Atlantic Ocean, Indian Ocean, and Pacific Ocean stocks. Manuscript accepted 10 February 1997 Fishery Bulletin 95:566-575 (1997). The yellowfin tuna, Thunnus alba- cares (Bonnaterre), supports impor- tant fisheries in tropical and sub- tropical oceans. Catches have in- creased from about 600,000 metric tons (t) in 1982 and 1983 to about 1,100,000 t in 1993 and 1994; in 1994, about 63% of the catch came from the Pacific Ocean, about 24% from the Indian Ocean, and 14% from the Atlantic Ocean (FAO, 1996). Given the size and circum- global nature of the resource, there is considerable management inter- est in determining stock structures. It is only comparatively recently that yellowfin tuna has been recog- nized as a single species (Gibbs and Collette, 1967); its high degree of morphological variation led Jordan and Evermann (1926) to recognize seven yellowfin tuna species. How- ever, a major morphometric study by Royce (1964) revealed that intra- oceanic differences could be greater than interoceanic differences and that several characters showed cli- nal variation. He concluded that the morphometric data are best ex- plained by a single worldwide pan- tropical species, a conclusion con- firmed by Gibbs and Collette ( 1967). Most stock structure studies of yellowfin tuna have focused on the large Pacific Ocean component of the catch. Here, tagging experi- ments indicated that yellowfin tuna usually migrate hundreds rather than thousands of kilometers and that their movements do not range far both east-west or north-south (Joseph et al., 1964; Bayliff, 1979; Hunter et al., 1986; Lewis, 1992). Morphometric studies have pro- vided commensurate results, with Mexico and Ecuador fish being much more similar to one another than to fish from the central (Ha- waii) and western (Australia, Ja- pan) Pacific (Schaefer, 1991). Stud- ies of the microchemical composi- tion of larval portions of otoliths in West Pacific fish (Indonesia, Phil- ippines, Coral Sea, Hawaii) have shown some differences, indicating that such analyses may be useful in determination of spawning origins (Gunn and Ward1 ; Gunn2 ). Genetic studies of four to five polymorphic 1 Gunn, J. S„ and R. D. Ward. 1994. The discrimination of yellowfin tuna sub-popu- lations within the AFZ. Phase 1: a pilot study to determine the extent of genetic and otolith microchemical variability in populations from different parts of the Pa- cific and Indian Oceans. Final Report (91/ 27) to Fisheries Research and Development Corporation, Deakin, ACT, Australia. 2 Gunn, J. S. 1996. CSIRO Division of Marine Research, Hobart, Tasmania, Australia. Unpubl. data. Ward et al.: Population structure of Thunnus albacares 567 allozyme loci in Pacific Ocean collections have shown significant spatial heterogeneity at one locus (GPI- A*)\ the common allele in western and central regions differed from that in the east (Sharp, 1978; Ward et al., 1994). This finding either indicates the existence of two reproductively isolated groups of yellowfin tuna in the Pacific Ocean or suggests that selection pressures are different in the two regions. There is no evidence of mitochondrial DNA (mtDNA) differentiation between eastern Pacific and western Pacific yellowfin tuna (Scoles and Graves, 1993; Ward et al., 1994). In the Atlantic Ocean, where it was once assumed that there were separate eastern and western stocks, recent taggings of large yellowfin tuna have resulted in 15 trans-Atlantic recoveries (ICCAT, 1992b); a single stock is now assumed (ICCAT, 1995). There have been no genetic comparisons of eastern and western Atlantic yellowfin tuna. The extent of genetic differentiation of yellowfin tuna from different oceans has been little studied. Suzuki (1962) found no differences in the incidence of the Tg2 blood group antigen in fish from the equa- torial Pacific and Indian Oceans. Scoles and Graves (1993) found no significant differentiation in mtDNA from one west Atlantic collection and five Pacific col- lections (each of 20 fish). Here we compare genetic variation in collections from the Pacific, Indian, and Atlantic oceans. We used larger sample sizes than those used in the study by Scoles and Graves (1993) and examined both allozyme and mtDNA variation to see if the increased statistical power would en- able us to reject the null hypothesis of no interoce- anic genetic differentiation. Materials and methods Samples were collected from one region of the Atlan- tic Ocean, two regions of the Indian Ocean, and six regions of the Pacific Ocean. Details of most of the Pacific collections (Philippines, Coral Sea, Kiribati, Hawaii, California, and Mexico) are given in Ward et al. (1994). For the present paper, the 1991 and 1992 Hawaii collections were pooled. A second Phil- ippines collection was collected in October-Decem- ber 1994. This showed no significant genetic differ- entiation from the earlier collection; therefore the two collections were pooled for our study. The Atlan- tic collection was taken from the Caribbean Sea (Gulf of Mexico, approx. 28°N, 88°W) in September 1993. The Indian Ocean collections were taken from off Sri Lanka (approx. 6°N, 80°E) and off the Seychelles (approx. 7°S, 52°E) in December 1994. White muscle samples were flown (airfreight) frozen to Hobart and stored at -80°C. Specimens were studied by allozyme and mtDNA analysis. The experimental methods are given in Ward et al. (1994). Four allozyme loci known to be polymorphic in white muscle were examined: ADA* (adenosine deaminase, EC 3. 5. 4. 4), FH* (fumarate hydratase, EC 4.2. 1.2), GPI-A*, and GPI-B* (glucose- 6-phosphate isomerase, EC 5.3. 1.9). MtDNA varia- tion was examined by using two restriction enzymes ( Bel I and Eco RI) known to discriminate most of the mtDNA haplotypes revealed by eight restriction en- zymes in an earlier survey (Bam HI, Ban I, Bel I, Eco RI, Hind III, Pvu II, Sal I, and Xho I — see Ward et al., 1994). The homogeneity of allele and haplotype frequen- cies of the collections was tested by the randomized Monte Carlo chi-square procedure of Roff and Bentzen (1989). For each test, 2,000 randomizations of the data were carried out, each giving a randomized chi-square value ( X2nuii )• The probability that the null hypothesis of genetic homogeneity was correct was given by P = n/2,000, where n is the number of ran- domizations that generate X2 null - X2 and where %2 is the chi-square value given by the actual observations. The extent of genetic differentiation among collec- tions was quantified by Nei’s gene diversity statistic Gst (Nei, 1987), which estimates the proportion of total genetic variation attributable to differentiation between populations. For each allozyme locus, Gsr was estimated as (HT- Hg) / HT, where HT represents total heterozygosity and Hs is average (Hardy- Weinberg expected) population heterozygosity. The mtDNA data were analyzed in a similar way, treat- ing haplotypes as alleles and Hr and Hs as diversity estimates. The proportion or magnitude of GST gen- erated by sampling error, which we have termed G sT.nuii > was estimated with a bootstrapping program, with the observed allele or haplotype frequencies and collection sizes. Simulations were run 1,000 times to provide a mean value of GSTnull and a standard de- viation. The probability of obtaining a value of GSTnull as large or larger than that obtained from the actual observations of GgT was given by P = n/1,000, where n is the number of randomizations that generate G sr.nuii - GSt- Values of P less than 0.05 indicated significant differentiation between areas that could not be explained by sampling error alone. Bonferroni adjustments of significance levels, to correct for multiple tests, were carried out with the sequential procedure advocated by Hochberg ( 1988). Tests are ordered according to their probability value. The highest probability value, P , is compared with the significance value a. Here we initially set a = 0.05. If P ^ oc, that test is judged to be nonsignifi- cant, and comparisons continue with subsequent probabilities, each compared with a modified signifi- 568 Fishery Bulletin 95(3), 1997 cance level = aJ(l+i), where i is the number of tests already performed. When a test is significant, it and all subsequent tests are deemed significant. Cluster analysis of the allozyme allele frequency data and the mtBNA haplotype frequency data used the UPGMA (unweighted pair-group method using averages) algorithm with Nei’s (1978) unbiased ge- netic distance measure, as implemented in BIOSYS-1 (Swafford and Selander, 1981). Estimates of mtDNA nucleotide sequence diversity and divergence (Nei and Tajima, 1981; Nei, 1987) were made with REAP vers. 4.0 (see McElroy et al., 1992), and population divergences were clustered by using UPGMA. Results The Seychelles muscle samples were partially de- graded on arrival, and could not be confidently screened for allozyme determinations, although mtBNA analysis presented no problems. Because it is sometimes difficult to distinguish tuna species, we usually find a small percentage (3-5%) of non-yel- lowfin tunas among nominal yellowfin tuna collec- tions. These misidentified fish can be recognized by aberrant allozyme (Graves et al., 1988; Elliott and Ward, 1995) and mtBNA patterns (Grewe3 ). For ex- ample, five (3.7%) of the 135 Philippines samples collected in 1994 proved to be bigeye tuna, Thunnus obesus. However, at times, the proportion of mis- identified fish can be much higher: 18 (46.2%) of the 39 “yellowfin tuna” from Sri Lanka proved to be big- eye tuna. The misidentified fish were excluded from the following analyses. Allozyme allele frequencies at four polymorphic loci (ADA*, FH*, GPI-A*, GPI-B *) for eight collections (Table 1) and mtBNA haplotype frequencies for nine collections (Table 2) were determined. No significant deviations from Hardy-Weinberg expectations were recorded for any allozyme locus. Heterogeneity chi-square analyses (Table 3) of allele frequencies revealed no significant differentiation for three loci (ADA*, FH*, and GPI-B*), but highly sig- nificant heterogeneity at the fourth locus, GPI-A* (P<0.001, a=0.0125). Genetic diversity ( GST) analyses (Table 3) indicated that for ADA*, FH*, and GPI-B*, 3 Grewe, P. M. 1993. CSIRO Division of Marine Research, Hobart, Tasmania, Australia. Unpubl. data. Table 1 Allozyme allele frequencies and sample sizes ( n ). GOM = Gulf of Mexico, S.Lan. = Sri Lanka, Philipp. = Philippines, Cl. Sea = Coral Sea, Calif. = California. Locus Allele Atlantic Indian Pacific GOM S. Lan. Philipp. Cl. Sea Kiribati Hawaii Calif. Mexico ADA* 125 0.005 0.003 115 0.414 0.310 0.330 0.306 0.399 0.361 0.317 0.359 100 0.548 0.643 0.622 0.638 0.567 0.609 0.671 0.628 85 0.033 0.048 0.045 0.056 0.034 0.030 0.012 0.013 n 105 21 176 98 89 115 41 39 FH* 130 0.118 0.091 0.081 0.117 0.086 0.076 0.051 0.075 100 0.875 0.909 0.910 0.878 0.900 0.920 0.949 0.925 75 0.007 — 0.009 0.005 0.014 0.004 — — n 68 11 111 98 70 112 39 29 GPI-B* -20 0.015 0.004 -60 0.180 0.167 0.176 0.163 0.233 0.113 0.187 0.231 -100 0.806 0.833 0.824 0.837 0.767 0.878 0.813 0.769 -125 — — — — — 0.004 — — n 103 21 176 98 88 115 40 39 GPI-A * 145 0.003 135 0.045 — 0.015 0.036 0.011 0.035 0.122 0.077 100 0.624 0.286 0.651 0.683 0.673 0.609 0.305 0.231 75 0.332 0.714 0.328 0.281 0.316 0.357 0.573 0.692 40 — — 0.003 — — — — — n 101 21 175 98 87 115 41 39 Ward et a I.: Population structure of Thunnus albacares 569 slightly less than 1% of the observed diversity arose from differences between collections and could be attributed to sampling error alone (GSTnu[l). For GPI- A*, the observed value of GST, at 12%, was much larger than the value attributable to sampling error (about 1%). The “true” GSTe stimate of GPI-A* — the difference between Gsrand GSTnull — is thus around 11%, indicating that about 11% of the observed di- versity at the GPI-A* locus comes from differences between collections. The GPI-A* heterogeneity (Fig. 1) was further ex- plored by comparing all collections pairwise (Fig. 2; Table 4). This comparison essentially revealed two groups of collections: 1) the west-central Pacific Ocean and the Atlantic Ocean (Gulf of Mexico) col- lections; and 2) the Indian Ocean (Sri Lankan) and eastern Pacific Ocean (Californian and Mexican) col- lections. Within each of these two groups there was no significant differentiation, but between them dif- ferentiation was marked. This conclusion holds af- ter Bonferroni corrections of a levels for multiple tests. The GPI-A* 100 allele was the most frequent allele in the west-central Pacific Ocean and the At- lantic Ocean group, whereas the GPI-A* 75 allele was the more frequent allele in the Indian Ocean and the eastern Pacific Ocean group. The genetic differen- tiation of the Atlantic Ocean collection from the In- dian Ocean collection suggests that fish from these areas constitute separate stocks; the separation of the Indian Ocean collection from the west-central Pacific Ocean collections suggests that fish from these areas constitute separate stocks; the separation of the west-central Pacific Ocean collections from the eastern Pacific Ocean collections suggests that fish from these areas constitute separate stocks; and the separation of the eastern Pacific Ocean collections Table 2 Mitochondrial DNA haplotype frequencies (Bel I and Eco RI haplotypes respectively), sample sizes (n), haplotype diversities (h) and percent nucleotide diversities (% n.d.). Abbreviations are defined in Table 1. Seych. = Seychelles. Locus Haplotype Atlantic Indian Pacific GOM Seych. S. Lan. Philipp. Cl. Sea Kiribati Hawaii Calif. Mexico mtDNA AA 0.266 0.319 0.381 0.286 0.340 0.443 0.276 0.294 0.325 AB 0.543 0.407 0.333 0.509 0.402 0.364 0.537 0.463 0.425 AC — — — — — 0.011 0.015 — — AF — 0.011 — — — — 0.007 0.049 0.025 AG — — — 0.006 — — 0.007 — — BA 0.011 0.011 — 0.025 0.010 — 0.007 0.049 — BB 0.064 0.066 — 0.031 0.052 0.068 0.060 0.073 0.025 CA 0.032 0.011 — 0.031 0.062 0.011 0.015 0.024 0.050 CB 0.021 0.088 0.286 0.068 0.103 0.057 0.045 0.024 0.125 CO 0.021 — — — — — — — — DB — — — — 0.010 — — — — EB — — — — 0.021 — — — — LB — — — — — 0.011 — — — MB — — — — — 0.011 — — — NB — — — 0.019 — — 0.007 — — PB — — — — — — 0.015 0.024 — OA — — — 0.006 — — — — 0.025 OB — — — 0.006 — 0.023 — — — QA — 0.011 — — — — — — — QB 0.011 0.011 — — — — 0.007 — — WA 0.032 0.033 — — — — — — — ZB — 0.022 — 0.006 — — — — — A2B — — — 0.006 — — — — — Q2B — 0.011 — — — — — — — n 94 91 21 161 97 88 134 41 40 h 0.634 0.727 0.695 0.655 0.712 0.670 0.633 0.705 0.712 % n.d. 0.998 1.263 1.017 1.017 1.174 1.027 0.901 1.099 1.047 570 Fishery Bulletin 95(3), 1997 Figure 1 Map of sample sites showing GPI-A* gene frequencies in yellowfin tuna, Thunnus albacares. Larger circles represent our data (Table 1), the three smaller circles (Bismark Sea, Roca Partido, and Ecuador) data are from Sharp (1978). The location of the Seychelles sample, examined for mtDNA variation but not for GPI-A* variation, is identified. The shaded area represents the approximate global distribu- tion of yellowfin tuna. from the Atlantic Ocean collection suggests that these fish constitute separate stocks. Thus the GPI-A* data, taken together with the spatial orientation of these collections, indicate the existence of at least four yellowfin tuna stocks: Atlantic Ocean, Indian Ocean, west-central Pacific Ocean, and east Pacific Ocean. Six of the 24 mtDNA haplotypes (CO, QA, WA, ZB, A2B, Q2B, see Table 2) were not recorded in the ear- lier survey of Ward et al. (1994) but were rare (fre- quencies less than 3.5%). Fragment sizes for most haplotypes are given in Ward et al. (1994), but a full list is available on request. Haplotype (nucleon) di- versities per collection ranged from 0.633 to 0.727 (mean estimate of 0.683) (Table 2). Percent nucle- otide diversities per collection ranged from 0.998 to 1.263 (mean estimate of 1.061) (Table 2). Table 3 Analyses of genetic differentiation among the samples. Locus Number of fish Number of alleles/haplotypes Heterogeneity x2 analysis Genetic diversity analysis t P Gst ^ ST.nulA P ADA* 684 4 17.326 0.666 0.006 0.008 ±0.004 0.569 FH* 538 3 9.241 0.821 0.006 0.011 ±0.008 0.736 GPI-B* 680 4 29.256 0.128 0.008 0.008 ±0.005 0.370 GPI-A* 677 5 131.416 <0.001 0.118 0.008 ±0.004 <0.001 mtDNA 767 24 227.743 0.048 0.023 0.015 ±0.005 0.071 Ward et a I.: Population structure of Thunnus albacares 571 Pairwise comparisons of GPI-A Table 4 * allele frequencies (P above, chi square below) . GOM = Gulf of Mexico. GOM Sri Lanka Philippines Coral Sea Kiribati Hawaii California Mexico GOM <0.001 0.156 0.468 0.129 0.769 <0.001 <0.001 Sri Lanka 21.700 — 0.001 <0.001 <0.001 0.001 0.031 0.214 Philippines 5.938 24.115 — 0.264 0.965 0.291 <0.001 <0.001 Coral Sea 1.586 28.644 4.870 — 0.270 0.237 <0.001 <0.001 Kiribati 3.905 22.586 1.196 2.630 — 0.203 <0.001 <0.001 Hawaii 0.493 19.130 4.635 2.824 3.280 — <0.001 <0.001 California 24.850 6.047 46.871 35.021 37.324 25.374 — 0.289 Mexico 34.934 3.579 51.193 46.401 44.495 33.366 2.526 — Table 5 Pairwise comparisons of mtDNA haplotype frequencies ( P above, chi square below). GOM = Gulf of Mexico. GOM Seychelles Sri Lanka Philippines Coral Sea Kiribati Hawaii California Mexico GOM 0.277 0.009 0.089 0.025 0.009 0.384 0.285 0.076 Seychelles 14.079 — 0.549 0.062 0.188 0.212 0.200 0.478 0.707 Sri Lanka 23.110 9.697 — 0.269 0.369 0.171 0.111 0.025 0.612 Philippines 19.823 22.529 14.013 — 0.223 0.025 0.428 0.207 0.597 Coral Sea 18.085 16.020 7.785 15.432 — 0.087 0.053 0.092 0.683 Kiribati 21.523 17.480 11.862 22.368 15.117 — 0.086 0.093 0.253 Hawaii 13.801 18.574 18.760 15.533 19.300 18.754 — 0.667 0.378 California 11.888 11.682 13.905 17.165 14.055 16.316 8.888 — 0.383 Mexico 15.622 9.718 5.017 10.188 7.203 12.214 13.080 8.473 — A chi-square test (Table 3) showed that the mtDNA haplotype variation across all nine regions was just signifi- cant (a=0.05, P=0.048, with the stan- dard 2,000 replicates, and P=0.045, with 10,000 replicates). Genetic diver- sity analysis (Table 3) gave a result bor- dering on significance (P=0.071, with a “true” Gst of about 1%). All collections were compared pairwise with chi- square tests (Table 5) to determine which collections contributed most to the marginal chi-square differentiation. Although some pairs appeared signifi- cantly different (e.g. Gulf of Mexico ver- sus Sri Lanka, P=0.009; Gulf of Mexico versus Kiribati, P=0.009), none was sig- nificant after Bonferroni adjustments for table-wide comparisons. Two UPGMA dendrograms were es- timated. One, based on mtDNA haplo- type frequencies alone (Fig. 3A), showed the maximal genetic-distance estimates among col- based on percent sequence divergence (Fig. 3B), con- lections to be about 0.05 — much less than the ma- firmed the high degree of similarity among the col- jor GPI-A* split of nearly 0.30 (Fig. 2). The second, lections. After correcting for within-collection nucle- 0.30 0.20 0.10 0.00 Genetic distance Figure 2 UPGMA phenogram for the GPI-A* locus constructed from Nei’s (1978) unbiased genetic distance. 572 Fishery Bulletin 95(3), I 997 A Seychelles Gulf of Mexico California Mexico California Philippines Philippines - Coral Sea Hawaii Seychelles Hawaii Kiribati Gulf of Mexico Sri Lanka Sri Lanka Mexico Kiribati J 1 Coral Sea L 1 _l 0.050 0.025 0.000 0.014 0.007 0.000 Genetic distance Percentage divergence Figure 3 UPGMA phenograms for the mtDNA data with (A) unbiased genetic dis- tance (Nei 1978), and (B) percentage nucleotide divergence (Nei, 1987). was significant following Bonferroni correction to a levels (west-central Pacific versus east Pacific, P=0.456, a=0.05; Indian versus east Pacific, P=0.316, a=0. 025; Atlantic versus east Pacific, P=0. 119, a=0.017; Atlantic ver- sus Indian, P=0. 047, a=0. 0125; Atlan- tic versus west-central Pacific, P=0.032, a=0.010; Indian versus west-central Pacific, P=0.0135, a=0.008), the three pairwise comparisons of the Atlantic Ocean, Indian Ocean, and west-central Pacific all showed P-values less than 0.05. Clearly, the mtDNA data do not dif- ferentiate west-central Pacific Ocean collections from east Pacific Ocean col- lections but, considering the inter- ocean analyses alone, do provide some support for the delineation of Atlantic Ocean, Indian Ocean, and Pacific Ocean stocks. otide divergence, pairwise nucleotide divergence ranged from 0.040% to -0.025% (mean 0.004%). There was little correspondence between these two mtDNA dendrograms, and this lack of correspon- dence, together with the low distances observed, sug- gests that the tree topologies are unreliable. Because there is no significant mtDNA differen- tiation between the six Pacific Ocean collections (Table 5; and Ward et al., 1994) nor between the two Indian Ocean collections (Table 5), the collections within each ocean were pooled to test for interoce- anic differences. A comparison of the three oceans yielded a chi-square analysis that was significant (P=0.009, a=0.05) and a genetic diversity analysis bordering on significance (observed GST=0.010, G sT.nuii =0 005 ±0.003, P=0.059). A pairwise compari- son of the oceans showed that all pairs were signifi- cant (Indian versus Atlantic, P=0.047, a=0.05; Pa- cific versus Atlantic, P=0. 017; Pacific versus Indian, P=0.009). Finally, the mtDNA data were analyzed to see whether they offered any support to the conclusion from the GPI-A* data that there are (at least) four yellowfln tuna stocks. The four putative stocks con- sisted of the following units: Atlantic (Gulf of Mexico), Indian (Seychelles and Sri Lanka), west-central Pa- cific (Coral Sea, Kiribati, Philippines, Hawaii), and east Pacific (California and Mexico). Chi-square analysis of mtDNA data from these four regions in- dicated limited but significant (P=0.Q24) heteroge- neity. Although none of the six pairwise comparisons Discussion Samples of yellowfin tuna from the Pacific, Indian, and Atlantic oceans were compared with respect to four polymorphic allozyme loci and with respect to mtDNA variants. No significant allele frequency differences were observed for three of the allozyme loci, but the fourth locus, GPI-A*, showed considerable differentiation. Across all collections, the “true” GgT indicated that about 11% of the variation at this locus was attrib- utable to differences between collections. Two geneti- cally distinguishable groups were apparent. One con- sists of eastern Pacific Ocean and Indian Ocean fish, with a high frequency of the GPI-A*lb allele, the other of Atlantic Ocean and west-central Pacific Ocean fish, with a high frequency of the GPI-A* 100 allele. Because there are no migration routes between the eastern Pacific Ocean (California and Mexico) and the Indian Ocean that avoid the west-central Pacific Ocean, and between the Atlantic Ocean and west- central Pacific Ocean that avoid the Indian Ocean, there is reason to believe that there are at least four stocks of yellowfin tuna: Atlantic Ocean, Indian Ocean, west-central Pacific Ocean, and eastern Pa- cific Ocean. Sharp ( 1978) also examined GPI-A* allele frequen- cies in western and eastern Pacific populations. His GPI-A* allele frequencies for collections from Ecua- dor and Mexico were very similar to our California and Mexico frequencies, and his GPI-A* frequencies Ward et al.: Population structure of Thunnus albacares 573 from the Bismarck Sea in the western Pacific were very similar to our western Pacific Ocean frequen- cies (Ward et al., 1994), supporting the separation of western and eastern Pacific stocks. Another allozyme study (Fujino, 1970) failed to find differences between Hawaiian and eastern Pacific fish for an esterase and for transferrin, although the esterase was nearly monomorphic. We interpret the GPI-A* differentiation as being indicative of stock differences, resulting from re- stricted gene exchange between the four identified regions. However, the alternative explanation, that of differential selection in the presence of gene flow, cannot be ruled out. Indeed, the very limited mtDNA differentiation observed could be held to support this interpretation. Microsatellite analysis, currently underway, may help to resolve this question. Selec- tion acting on these noncoding genetic markers is presumed to be minimal or nonexistent; therefore microsatellite differentiation paralleling the GPI-A* differentiation would suggest drift of neutral GPI- A* alleles, whereas lack of microsatellite differen- tiation would indicate significant gene flow and thereby implicate selection as the cause of the GPI- A* differentiation. Pogson et al. ( 1995) have recently suggested that the highly heterogeneous distribution of anonymous nuclear RFLP markers among popu- lations of cod, Gadus morhua, reflects limited gene flow and that the much more homogeneous distribu- tion of allozyme alleles reflects stabilizing selection rather than extensive gene flow. Such an argument applied to yellowfin tuna data would interpret the GPI-A* heterogeneity as indicative of limited gene flow, and the ADA*, FH*, and GPI-B* homogeneity as in- dicative of stabilizing selection at these three loci. Differences between collections in mtDNA was only just significant (P=0.048), with a “true” GST value across all nine collections of around 1%. When col- lections were pooled within oceans, i.e. the three groups (Atlantic, Indian, and Pacific), significant dif- ferentiation was detected (P=0.009), although the “true” Gsr was only of the order of 0.5%. All three pairwise ocean comparisons were statistically signifi- cant. However, because collections within oceans did not always pool together in the distance dendrograms (Fig. 3), possibly because of limited sample sizes, it would clearly be useful to have more data to confirm (or refute) this evidence of interoceanic differentia- tion. When collections were pooled into the four pu- tative stocks indicated by the GPI-A* data, limited but significant heterogeneity in mtDNA haplotype frequencies was apparent (P=0.024), but there were no significant pairwise comparisons. Scoles and Graves (1993) were unable to detect significant mtDNA differentiation between Pacific and Atlantic yellowfin tuna, whereas the probability of homogeneity in our tests of these two oceans was only 0.017. However, they adopted a different test strategy. Instead of examining relatively large num- bers of fish (our study: Pacific fish, n=561; Atlantic fish, n- 94) with relatively few restriction enzymes (n= 2, but known to detect polymorphic sites), they chose to examine relatively few fish (Pacific fish, re=100; Atlantic fish, t?=20) with a relatively large number of restriction enzymes (n- 12, which included the two enzymes we used). Given that the common 12-enzyme haplotype in Scoles and Graves’ study comprised 52 fragments or 304 bp and that the com- mon 2-enzyme haplotype in our study comprised 7 fragments or 42 bp (see Ward et al., 1994) and that the mean size of the yellowfin tuna mtDNA genome is about 16,702 bp (Scoles and Graves (1993] esti- mate= 16,549; Ward et al. [ 1994]= 16,856), Scoles and Graves surveyed about 1.8% of the mtDNA genome, whereas we surveyed only about 0.3%. However, al- though it is of course true that had we surveyed more restriction enzymes, we would have uncovered many additional haplotypes, the two enzymes that we did select revealed most of the mtDNA diversity shown by Scoles and Graves (1993). For example, the (pooled) 12-enzyme haplotype diversity of 0.840 of Scoles and Graves was not much larger than our (pooled) 2-enzyme diversity of 0.677. Four of the en- zymes used by Scoles and Graves showed no varia- tion at all in the 120 fish and therefore were of no use for population discrimination. Twenty of the 34 12-enzyme haplotypes detected by Scoles and Graves (1993) among their 120 fish were seen only once, whereas only four of the 22 2-enzyme haplotypes in our 655 Atlantic and Pacific fish were seen only once: such rare haplotypes are of extremely limited use in population studies. Given that mtDNA heterogene- ity among regions is very limited, it is not surprising that the approach of screening large numbers of fish for a small number of sequences known to be vari- able should be more powerful than screening small numbers of fish for a larger number of sequences, many of which are relatively invariant. MtDNA data from another tuna, the albacore, Thunnus alalunga, showed a somewhat more pro- nounced separation of Atlantic Ocean and Pacific Ocean collections than did data for yellowfin tuna, but again no intraoceanic heterogeneity was detected (Chow and Ushiama, 1995). The limited mtDNA differentiation among yellow- fin tuna sampled throughout their range contrasts with the marked population subdivision revealed by the GPI-A* locus. Mitochondrial DNA has an effec- tive population size only one quarter that of nuclear DNA (Birky et al., 1989) and evolves more rapidly 574 Fishery Bulletin 95(3), 1997 (Brown et al., 1979); in principle it should be a more effective indicator of population substructure than nuclear loci. Given that more mtDNA than nuclear BNA divergence is expected, how can an allozyme locus show differentiation when mtDNA haplotypes do not? The lack of mtDNA differentiation in yellow- fin tuna does not appear to be the result of a lack of variation nor of a small sample size: although in- creasing haplotype diversities and sample sizes will increase statistical power, the mtDNA haplotype di- versities of our populations, assayed for just two re- striction enzymes, were quite high at around 0.65- 0.70, and sample sizes were similar to those used in the allozyme analyses. Nuclear DNA differentiation can exceed mtDNA differentiation when either the migration rate or the breeding sex ratio is strongly biased towards females (because mtDNA is mater- nally inherited), but there is no evidence that either of these conditions holds for yellowfin tuna (e.g. IATTC, 1992). The explanation for the seeming dis- crepancy may be that several independent polymor- phic allozyme loci were screened, whereas haplotypes of mtDNA are best treated as alleles at a single, nonrecombining locus. In a situation of low overall genetic divergence (resulting from gene flow or re- cent separation), the stochastic nature of genetic drift means that if several allozyme loci are screened, and notwithstanding the expected higher rate of mtDNA evolution, divergence might be first detected at an allozyme locus before it is detected for mtDNA. An alternative explanation, as intimated earlier, is that the GPI-A* differentiation results from selection. The delineation of the four stocks of yellowfin tuna does not seem unreasonable given what we know of their distribution and movements. Yellowfin are found circumglobally, but only in tropical and sub- tropical oceanic waters, approximately between the latitudes 40°N and 40°S (Collette and Nauen, 1983). Spawning occurs throughout the year in all core ar- eas of distribution, peaking in the warmer months (Collette and Nauen, 1983). Waters off the southern regions of South America (approximately 55°S) are too cold for Atlantic Ocean and Pacific Ocean fish to migrate around Cape Horn. Furthermore, direct con- nections between the tropical Atlantic Ocean and the eastern Pacific Ocean were severed after the Isth- mus of Panama closed about 3.5 million years ago (e.g. Keigwin, 1982; Coates et al., 1992), a closure likely to have predated the origin of yellowfin tuna (estimated by Elliott and Ward (1995) to have oc- curred within the last two million years). Thus Pa- cific Ocean and Atlantic Ocean fish could not mix. In contrast, Atlantic Ocean and Indian Ocean fish could mix (through southern Africa waters), as could In- dian Ocean and Pacific Ocean fish (through Indone- sian waters), but tagging experiments indicate that most yellowfin tuna move on a scale of hundreds rather than thousands of kilometers (Joseph et al., 1964; Bayliff, 1979; Hunter et al., 1986; Lewis, 1992). The extent of migration between ocean basins is therefore likely to be low, with intraoceanic recruit- ment predominating. Nonetheless, interoceanic movements are possible and could account for the low degree of genetic differentiation among areas. Further discussion of the genetic and other biologi- cal data with respect to Pacific Ocean fish is given in Ward et al. (1994). At present, these suggestions on the global stock structure of yellowfin tuna are essentially based on gene frequencies at a single polymorphic allozyme locus, GPI-A*, because no significant genetic hetero- geneity was detected for three other polymorphic allozymes and the mitochondrial DNA variants showed little interpopulation differentiation. It may well be that the stock structure of yellowfin tuna, in management terms, is more complex than these present findings suggest: very limited migration be- tween areas can effectively homogenise gene frequen- cies, and thus dispersal between areas can still be low even between populations that cannot be geneti- cally discriminated. Future genetic work should include the examina- tion of more fish from the Indian Ocean because the identification of these fish as a separate stock is based primarily on the analysis of just 21 fish for a single allozyme locus. Further clarification of genetic stock structure issues in yellowfin tuna will require larger sample sizes, examination of more areas (especially from the Indian and Atlantic Oceans), and the de- ployment of genetic techniques, such as microsatellite analysis, with enhanced resolving power and less concern over neutrality and selection issues. Acknowledgments This work was supported by grant 91/27 from the Fisheries Research and Development Corporation. We thank the following who assisted us by collecting samples: Noel Barut, Barbara Block, Chris Boggs, Thor Carter, Tim Davis, Ed Everett, John Gunn, Aubrey Harris, Nishi Karunasinghe, Theresa O’Leary, Dennis Lee, Dave Milton, Pianet Renaud, and Toki Takenaka. Shirlena Soh helped with some of the allozyme analyses. Chris Bolch, Rene Vaillancourt, Vivienne Mawson, Peter Rothlisberg, and two anonymous referees made useful comments on a draft of this manuscript. Ward et al.: Population structure of Thunnus albacares 575 Literature cited Bayliff, W. H. 1979. Migrations of yellowfin tuna in the eastern Pacific Ocean as determined from tagging experiments initiated during 1968-1974. Inter-Am. Trop. Tuna Comm., Bull. 17:445-506. Birky, C. W., P. Fuerst, and T. Maruyama. 1989. Organelle gene diversity under migration, mutation, and drift: equilibrium expectations, approach to equilib- rium, effects of heteroplasmic cells, and comparison to nuclear genes. Genetics 121:613-627. Brown, W. M., M. George, and A. C. Wilson. 1979. Rapid evolution of animal mitochondrial DNA. Proc. Natl. Acad. Sci. U.S.A. 76:1967-1971. Chow, S., and H. Ushiama. 1995. Global population structure of albacore (Thunnus alalunga) inferred by RFLP analysis of the mitochondrial ATPase gene. Mar. Biol. (Berl.) 123:39-45. Coates, A. B., J. B. C. Jackson, L. S. Collins, T. M. Cronin, H. J. Dowsett, L. M. Bybell, P. Jung, and J. A. Obando. 1992. Closure of the Isthmus of Panama: the near-shore marine record of Costa Rica and western Panama. Geol. Soc. Am. Bull. 104:814-828. Collette, B. B., and C. E. Nauen. 1983. FAO species catalogue. Vol. 2: scombrids of the world. United Nations Development Programme, Food and Agricul- ture Organization of the United Nations, Rome, 137 p. Elliott, N. G., and R. D. Ward. 1995. Genetic relationships of eight species of Pacific tu- nas (Teleostei, Scombridae) inferred from allozyme analysis. Mar. Freshw. Res. 46:1021-1032. FAO (Food and Agriculture Organization of the United Nations). 1996. FAO yearbook of fishery statistics: catches and land- ings 1994. Volume 78. FAO, Rome, 700 p. Fujino, K. 1970. Immunological and biochemical genetics of tuna. Trans. Am. Fish. Soc. 99:52-178. Gibbs, R. H., Jr., and B. B. Collette. 1967. Comparative anatomy and systematics of the tunas, genus Thunnus. Fish. Bull. 66:65-130. Graves, J. E., M. A. Simovich, and K. M. Schaefer. 1988. Electrophoretic identification of early juvenile yel- lowfin tuna, Thunnus albacares. Fish. Bull. 86:835-838. Hochberg, Y. 1988. A sharper Bonferroni procedure for multiple tests of significance. Biometrika 75:800-802. 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. LATTC (Inter- American Tropical Tuna Commission). 1992. 1990 Annual report of the Inter-American Tropical Tuna Commission. IATTC, La Jolla, CA, 261 p. ICCAT (International Commission for the Conservation of Atlantic Tunas). 1992. Comments on the stock structure of Atlantic yellow- fin tuna. International Commission for the Conservation of Atlantic Tunas. Report for 1990-91, Part II, p. 217- 220. 1995. Yellowfin tuna: state of the stocks. International Com- mission for the Conservation of Atlantic Tunas. Report for 1994-95, part I, vol. 2, p. 18-22. Jordan, D. S., and B. W. Evermann. 1926. A review of the giant mackerel-like fishes, tunnies, spearfishes and swordfishes. Calif. Acad. Sci., Occasional Papers 12:1-113. Joseph, J., F. G. Alverson, B. D. Fink, and E. B. Davidoff. 1964. A review of the population structure of yellowfin tuna, Thunnus albacares, in the eastern Pacific Ocean. Inter- Am. Trop. Tuna Comm. Bull. 9:53-112. Keigwin, L. D. 1982. Pliocene closing of the Isthmus of Panama, based on stratigraphic evidence from nearby Pacific Ocean and Car- ibbean Sea cores. Geology (Boulder) 6:630-634. Lewis, A. D. 1992. Stock structure of Pacific yellowfin — a review. South Pacific Commission, Noumea, New Caledonia, 14 p. McElroy, D., P. Moran, E. Bermingham, and I. Kornfield. 1992. REAP: an integrated environment for the manipula- tion and phylogenetic analysis of restriction data. J. Hered. 83:157-158. Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590. 1987. Molecular evolutionary genetics. Columbia Univ. Press, New York, NY, 512 p. Nei, M., and F. Tajima. 1981. DNA polymorphism detectable by restriction endo- nucleases. Genetics 97:145-163. Pogson, G. H., K. A. Mesa, and R. G. Boutilier. 1995. Genetic population structure and gene flow in the Atlantic cod Gadus morhua : a comparison of allozyme and nuclear RFLP loci. Genetics 139:375-385. Hoff, 1) A., and P. Bentzen. 1989. The statistical analysis of mitochondrial DNA poly- morphisms: x2 and the problem of small samples. Mol. Biol. Evol. 6:539-545. Royce, W. F. 1964. A morphometric study of yellowfin Thunnus albacares (Bonnaterre). Fish. Bull. 63:395-443. Schaefer, K. M. 1991 . Geographic variation in morphometric characters and gill-raker counts of yellowfin tuna Thunnus albacares from the Pacific Ocean. Fish. Bull. 89:289-297. Scoles, D. R., and J. E. Graves. 1993. Genetic analysis of the population structure of yel- lowfin tuna, Thunnus albacares , from the Pacific Ocean. Fish. Bull. 91:690-698. Sharp, G. D. 1978. Behavioral and physiological properties of tunas and their effects on vulnerability to fishing gear. In G. D Sharp and A. E. Dizon (eds.), The physiological ecology of tunas, p. 397-449. Academic Press, New York. 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. Suzuki, A. 1962. On the blood types of yellowfin and bigeye tuna. Am. Nat. 96:239-246. Ward, R. D„ N. G. Elliott, P. M. Grewe, and A. J. Smolenski. 1994. Allozyme and mitochondrial DNA variation in yel- lowfin tuna (Thunnus albacares) from the Pacific Ocean. Mar. Biol. (Berl.) 118:531-539. 576 Abstract .—The recruited biomass of orange roughy, Hoplostethus atlanti- cus, was estimated for the New Zealand mid-east coast orange roughy stock with the daily fecundity reduction method (DFRM). These fish migrate to Ritchie Bank and spawn between 850 and 900 m for about one month in win- ter. The biomass of spawning females was estimated by dividing mean daily planktonic egg production, N0 (eggs/ day), by mean daily fecundity, D (eggs/ kg per day). The stock biomass was then estimated by multiplying the spawning female biomass by the ratio of all recruited fish to females that would spawn that year, estimated with a wide-area trawl survey made over the stock area two months before the spawning season. The mean daily planktonic egg pro- duction was sampled near the peak of the spawning season, by using a strati- fied-random plankton survey. Eggs were staged and aged after accounting for their thermal history as they as- cended the water column. Because young eggs were damaged by the net and older eggs were affected by advec- tion out of the plankton survey area, relatively few egg stages were available for estimating N0 ( 10.9 x 109 eggs/day), and the estimate was somewhat impre- cise (CV=0.46). Mean daily fecundity (787 eggs/(kg x day), CV=0.11) was es- timated from the daily rate of decline in population fecundity per mature fe- male weight (/?,). Fecundity per female weight was estimated from a trawl sur- vey made in the spawning area during the spawning season and was calcu- lated as the mature eggs/kg of active spawners multiplied by the proportion of active spawners in each trawl. Spawning female biomass was 14,000 t (CV=0.50), and stock biomass was 26,000 t (CV=0.50). Mean daily fecun- dity was probably under-estimated be- cause spent fish appeared to migrate from the spawning area during the fe- cundity reduction measurement period and reduce stock biomass to about 18,200 t. The DFRM biomass estimate was of central importance in the intro- duction of greatly reduced total allow- able catch levels in this fishery. Manuscript accepted 26 February 1997. Fishery Bulletin 95:576-597(1997). An estimate of orange roughy, Hoplostethus atlanticus, biomass using the daily fecundity reduction method John R. Zeldis* R. I. Chris Francis Malcolm R. Clark Jonathan K. V Ingerson Paul J. Grimes Marianne Vignaux National Institute of Water and Atmospheric Research (NIWA) RO. Box 14-901, Kilbirnie, Wellington, New Zealand *E-mail address: j.zeldis@niwa.cri.nz The orange roughy ( Hoplostethus atlanticus , Trachichthyidae) fishery on Ritchie Bank on the eastern New Zealand continental slope (Fig. 1) was the second largest orange roughy fishery in New Zealand dur- ing the late 1980’s and early 1990’s, with a total allowable commercial catch (TACC) of about 10,000 met- ric tons (t) per year. Trends in catch per unit of effort (CPUE) indicated that the stock size was diminishing rapidly under this management re- gime, although the stock reduction analysis for the fishery did not esti- mate stock size precisely (Field et al.1 ). Experience with other major orange roughy fisheries in New Zealand (the “Box” fishery on north- ern Chatham Rise and the Chal- lenger Plateau fishery [Fig. 1; Clark, 1995]) indicated that over- fishing of the Ritchie Bank stock was likely because of low productiv- ity. However, without adequate knowledge of stock size, it was dif- ficult to set a TACC that would al- low a sustainable fishery. Zeldis ( 1993) concluded that both the annual egg-production method ( AEPM; Saville, 1964; Picquelle and Megrey, 1993; Koslow et al., 1995) and the daily fecundity reduction method (DFRM; Lo et al., 1992; Lo et al., 1993) would be feasible for the estimation of absolute spawning biomass of orange roughy. A voyage was made to Ritchie Bank from early June to early July, 1993 ( Tangaroa voyage TAN9306), with the intention of using both types of egg-production survey method. With the AEPM, annual egg produc- tion is the sum of daily planktonic egg production estimates made over the entire spawning season from separate subsurveys. Unfortu- nately, the voyage failed to sample annual egg production for two rea- sons. First, although the voyage was executed as a series of five sub- surveys, two of the subsurveys were not completed because of ship and equipment breakdowns and be- cause of a lack of time at the end of the voyage. Second, the voyage pe- riod ended before spawning had fin- ished for the season, so that the annual egg production was not com- pletely sampled. 1 Field, K. D., R. I. C. C. Francis, and J. H. Annala. 1993. Assessment of the Cape Runaway to Banks Peninsula (ORH 2A, 2B, and 3A) orange roughy fishery for the 1993-94 fishing year. MAF Fisheries, Fisheries Assessment Research Document 93/8. NIWA, Greta Point, Wellington, New Zealand, 17 p. Zeldis et al.: An estimate of biomass of Hoplostethus atlanticus 577 Figure 1 Map of New Zealand and location of spawning grounds (ovals) of orange roughy, Hoplostethus atlanticus. Also shown are the areas (2A, 3B, and 3A) of the east coast stock. Unlike sampling for the AEPM, sampling for the DFRM did not need to cover the entire spawning sea- son, allowing an estimate of spawning female bio- mass on Ritchie Bank to be produced from the por- tion of the voyage that did not suffer from ship and equipment breakdowns and that provided adequate data on planktonic egg production and fecundity. The DFRM was used to estimate the biomass of spawn- ing females by dividing the daily planktonic egg pro- duction in the survey area (eggs/day) by the daily fecundity of females (eggs/(kg x day). The biomass of spawning females was scaled by the maturity ogive, sex ratio, and spawning proportion to estimate the re- cruited biomass (>32 cm standard length) of orange roughy in the stock (where the stock is defined as those fish assumed to spawn on Ritchie Bank). The latter data were taken from a wide-area trawl survey in March-April 1993 (Tangaroa voyage TAN9303; Field et al.2 ), that covered the area inhabited by the stock (Banks Peninsula to East Cape, Fig. 1). 2 Field, K. D., R. I. C. C. Francis, J. R. Zeldis, and J. H. Annala. 1994. Assessment of the Cape Runaway to Banks Peninsula (ORH 2A, 2B, and 3 A) orange roughy fishery for the 1994-95 fishing year. MAF Fisheries, Fisheries Assessment Research Document 94/20. NIWA, Greta Point, Wellington, New Zealand, 24 p. A necessary biological prerequisite for using the DFRM is that the target species has determinate annual fecundity ( Hunter and Lo, 1993). This enables total seasonal fecundity to be determined before fi- nal maturation so that the fecundity reduction rate can be monitored through the spawning season ( Lo et al., 1993). Orange roughy have determinate an- nual fecundity (Pankhurst et al., 1987; Bell et al., 1992; Zeldis, 1993). Application of egg production methods to orange roughy also is possible, but only if the age of planktonic eggs at morphological stage can be estimated. Ageing of eggs was achieved by developing a model for ageing the eggs as they tra- versed the thermal gradient in the water column (Zeldis et al., 1995) and by describing the morpho- logical stages of the eggs (Grimes et al.3 ). DFRM model To calculate the biomass of Ritchie Bank spawning females, the daily planktonic egg production in the 3 Grimes, P. J., A. C. Hart, and J. R. Zeldis. 1997. Embryology and early larval development of orange roughy ( Hoplostethus atlanticus Collett). Unpubl. data. 578 Fishery Bulletin 95(3), 1997 survey area was divided by the daily fecundity/kg of the females: Bspf - Af0 /( 1,000.0), where B ^ = biomass of spawning females (tons); N0 = daily egg production, (eggs/day); D = mean daily fecundity (eggs/(kg x day)) for mature fish; and the factor 1,000 converts kg to tons. The recruited biomass, Brec (defined as the biomass of fish of length >32 cm) was calculated from the bio- mass of Ritchie Bank spawning females, Bgp^ as Brec = BspfS, where the scalar (S) was an estimate of the ratio B / Bsp^ This ratio allows for recruited females that did not spawn, females that did spawn but were <32 cm, as well as the sex ratio. To estimate the parameters of the above biomass model, the data analysis deals with three distinct data sets: • daily planktonic egg production; • daily fecundity per female weight; and • proportions spawning, recruited, and female. Daily planktonic egg production Survey design The timing of plankton sampling coincided with the period when orange roughy fe- males on the Ritchie Bank were in late maturation or spawning stages (mid-June to the end of the first week of July; Pankhurst, 1988). The location of plank- ton sampling was determined from research trawl catch rates (Fincham et al.4), which showed adult biomass to be highly aggregated on Ritchie Hill5 (Fig. 2, A and B) at the northern end of Ritchie Bank. Ritchie Hill catch rates accounted for 84% of the rela- tive orange roughy biomass over the Ritchie Bank survey area in July 1986 (Fincham et al.4). In plank- ton sampling during the spawning season on Ritchie Bank in early July 1986, 6 orange roughy eggs were 4 Fincham, D. J., D. A. Banks, and P. J. McMillan. 1987. Orange roughy trawl survey, Tolaga Bay to Cape Turnagain, 14 June to 11 July 1976: cruise report. Fisheries Research Divi- sion Internal Report 60. NIWA, Greta Point, Wellington, New Zealand, 38 p. 5 The names “Ritchie Bank,” “Ritchie Hill,” and “North Hill” used in this paper refer to the features called “Ritchie Ridge,” “Calyptogena Bank,” and “Pantin Bank,” respectively, in the following: Arron, E. S., and K. B. Lewis. 1992. Mahia, 2nd ed. N.Z. Oceanogr. Inst. Chart, Coastal Series, 1:200,000. NIWA Greta Point, Wellington, New Zealand. caught only at stations near Ritchie Hill, and samples taken >20 km away contained no eggs, indicating that eggs were aggregated near the spawners (Zeldis, 1993) and that plankton sampling would need to be highly concentrated near Ritchie Hill. During their first 36 hours of development, orange roughy eggs ascend the water column at 300-350 m/ day from a spawning depth of about 850 m (Zeldis et al., 1995). Geostrophic currents over Ritchie Bank during July 19866 were to the south and averaged about 12 cm/sec between 700 m and 250 m; these currents would displace these eggs at least 15.5 km to the south of Ritchie Hill by the time the eggs had reached 36 h of development (this is a minimum es- timate because there was probably some residual flow at the postulated level of no motion at 700 m). Con- sidering that drift would probably vary in direction and speed but would lie predominantly along isobaths, we designed the survey area with eight strata, elongated alongshore and arranged symmetri- cally around a central stratum centred over the top of Ritchie Hill (Fig. 2B). This central stratum (10.0 x 7.6 km) was about the size of the area of high fish density observed during a trawl survey done in the area in June— July 1987 (Grimes7 ). The middle layer of four strata surrounding the central stratum had outer boundaries of 18.5 xl3.7 km. These dimensions were chosen to approximate the distance at which catch rates of 1-day-old orange roughy eggs on the St. Helen’s seamount in eastern Tasmania were re- duced to half (9.3 km from the spawners [Koslow8 ]). The outer layer of four strata had an outer boundary 40.0 x 30.0 km long, to allow for maximal drift of 36- h-old eggs. The St. Helen’s data were used to estimate opti- mal allocation of stations to Ritchie Hill strata. The St. Helen’s data were collected over an entire spawn- ing season, in six separate subsurveys, each of which had two spatial strata. The counts in each St. Helen’s subsurvey and stratum were standardized such that their means equalled the mean across all of the subsurveys for each stratum. This procedure removed the within-season variation in mean egg density in each stratum. These standardized data were then laid over the Ritchie Bank stratum layers (central, middle, and outer), and mean egg densities (M) were estimated for each stratum layer, j. To allocate sta- 6 Zeldis, J. R. 1986. Cruise report J08/86 (second half). MAF Fisheries unpublished cruise report held in NIWA Library, Greta Point, Wellington, New Zealand, 7 p. 7 Grimes, P. 1987. NIWA, P.O. Box, 14-901, Kilbirnie, Wellington, New Zealand. Unpubl. data. 8 Koslow, T. 1992. CSIRO Division of Fisheries, GPO Box 1538, Hobart, Tasmania 7001, Australia. Personal commun. Zeldis et al.: An estimate of biomass of Hoplostethus atlanticus 579 tions optimally in a survey of size N stations, n/ sta- tions were allocated to each stratum layer such that rij c< A;M; and where A • = the area of each stratum layer. Allocation was done in proportion to the strata means (Mj) because they were highly correlated with the strata standard deviations and were probably esti- mated more reliably than the standard deviations (Francis, 1984). To estimate values of N that would yield a desired coefficient of variation for egg abun- dance estimates, the standardized counts in each stratum layer were randomly sampled with replace- ment (bootstrapped) to estimate M-, where the num- ber of samples taken from each stratum was n. .. The survey mean egg abundance combined across all strata, E, was estimated as This procedure was repeated 500 times and the mean of the 500 survey estimates was taken as the egg abundance estimate. The standard deviation of the 500 estimates divided by the mean was taken as the coefficient of variation of the egg abundance estimate. This analysis suggested that the optimal alloca- tion would have stations allocated to the central, middle, and outer strata in ratios of 1.5:0.38:0.25, respectively. It also suggested that 400 stations would provide adequate precision (CV=0.15) in the egg abundance estimate. Therefore, for the AEPM design, five 80-station subsurveys were planned with sta- tions allocated to strata in the above ratios. Using simulations, we found that by occupying sta- tions within each stratum in an order which mini- mized steaming distance between stations, before moving to a new, randomly chosen stratum, about 50% less steaming time would be involved than by occupying stations in completely random sequence in each subsurvey. This procedure would be done, how- ever, at the cost of variable (and possibly long) periods of no coverage of each stratum between subsurveys and could be a serious drawback if spawning intensity var- ies significantly and rapidly (over a few days) during 580 Fishery Bulletin 95(3), 1997 the spawning season, especially for coverage of the central stratum. To counteract this, the stations in the central stratum were divided randomly between two time strata to reduce the time between occupation of this high-density stratum between subsurveys. In the DFRM analysis, the plankton samples used for the estimation of egg production were from two consecutive AEPM subsurveys that were occupied near the peak of the spawning season and that were not subject to “downtime” from ship and equipment failures. These two subsurveys had no time break between them and were treated as a single survey, in which the central stratum was occupied four times and each of the surrounding strata was occupied twice. The plankton sampling in the entire survey was done from 14 June to 7 July, and the two subsurveys used in the DFRM analysis were done from 28 June to 6 July (subsurveys 3 and 4). Egg sampling and staging, count standardization, and production estimation The plankton net used in sampling had a cylinder-cone design with 900-pm mesh, a mouth area of 2 m2, and was fitted to a 125- kg flat-steel ring. It was designed to be efficient, with the ratio of the open area in the mesh to mouth area being >5:1 (Tranter and Smith, 1968). The net was deployed from a starboard crane while the ship was stationary (i.e. not under power) and its starboard side faced the wind. The winch had dynamic tensioning, to minimize surging of the net as a re- sult of the rolling motion of the ship. Tow depths were within 30 m of the bottom to the surface if the bot- tom was less than 950 m and from 850 m if the bot- tom was deeper than 950 m. A conductivity-tempera- ture-depth (CTD) probe or a net sonde within the mouth of the net was used to measure net depth. Warp payout was measured with winch instrumen- tation. Warp payout and recovery rates were 1 m / sec, also measured with winch instrumentation. Eggs were staged (Grimes et al.3) on board, prior to preservation, and generally within 0.5 h of landing the net, except for two samples with many eggs. For these, staging was done partly on board and partly in the labo- ratory on 4% formaldehyde-preserved eggs. All eggs < stage 7 (32-cell) were grouped, because it was not pos- sible to identify with confidence the stages from germi- nal disk through 32 cells within the plankton samples because most of these younger eggs were damaged (76% of the 8,293 eggs < stage 7), which caused the cell walls of the embryos to rupture, the cells to fuse, and the perivitelline space to collapse. Justification for assum- ing that the damaged eggs were all < stage 7 are given in Zeldis et al. (1995) and Grimes et al.3 The standardization from egg count to egg density (eggs/m2) was based on the formula density = count x correction factor, where the correction factor takes into account the mouth area of the net and the vol- ume of water filtered by it. With a vertical haul, the correction factor is 0.5 (because the net mouth area=2 m2). However, because the vessel almost always drifted (owing to wind and current) during shooting and hauling, hauls were not vertical and therefore the correction factors were almost always <0.5. Dis- tance towed was not estimated by using flowmeters because flowmeters were found to record spurious revolutions during the deployment (descent) phases of tows in subsequent tests (Grimes9). Instead, to calculate the volume of water filtered by the net, it was necessary to use global positioning satellite (GPS) vessel positions, warp length, depth, and cur- rent velocities to infer the path of the net (which, because of the ship’s drift, would be curved in the vertical dimension during hauling; Appendix 1). With a curved net trajectory, there was a different correction factor for each combination of plankton tow and egg stage because the different egg stages occupied different depths as they ascended the wa- ter column and because the net sampled more water in layers of equal thickness in the upper water col- umn than in the lower column. To calculate egg age and depth range (Table 1), data on egg development rate as a function of temperature, buoyancy by egg stage, and temperature as a function of depth were 9 Grimes P. 1996. NIWA, P.O. Box 14-901, Kilbirnie, Welling- ton, New Zealand. Unpubl. data. Table 1 Egg age (h) and depth (m) ranges used in calculating cor- rection factors for count standardization for Ritchie Bank plankton samples. Egg stages are described in Grimes et al. (Footnote 3 in the text). Stage Min. age Max. age Max. depth Min. depth 0 0.0 0.0 850 725 1 0.0 5.1 850 784 2 5.1 8.2 784 743 3 8.2 11.2 743 704 4 11.2 14.0 704 667 5 14.0 16.7 667 631 6 16.7 19.3 631 596 7 19.3 21.8 596 563 8 21.8 24.1 563 531 9 24.1 26.3 531 500 10 26.3 28.4 500 470 11 28.4 33.4 470 400 12 33.4 40.0 400 301 13 40.0 45.5 301 216 14 45.5 50.2 216 143 15 50.2 54.3 143 78 Zeldis et al.: An estimate of biomass of Hoplostethus atlanticus 581 used with the methods of Zeldis et al. (1995 ) and the Ritchie Bank CTD temperature profiles described below. Ritchie Bank temperatures were within the range of those observed in Zeldis et al. (1995). Cur- rent velocities as a function of depth were calculated from geostrophic velocity profiles (Pond and Pickard, 1978) by using Guildline CTD profiles taken over Ritchie Bank. A reference depth of no motion at 1,000 m was assumed, and geostrophic velocities at 100 m were corrected to match the 100-m velocities mea- sured by a buoy drogued to that depth. This buoy was deployed over Ritchie Hill on 28 June 1993, re- located 20 hours later, and subsequently lost. Correction factors were to convert egg counts to densities were also calculated (assuming the trajec- tory of the net was straight) for comparison with those where a curved trajectory was assumed. For this, the volume filtered v (m3) was estimated by using v = 2 Jp2 + z2 , where p = the distance the ship drifted from deploy- ment to recovery of the net, determined with GPS; = the maximum depth of the net; and factor 2 = the area (m2) of the net mouth. It was assumed that, at the start of hauling, the net was at the position of the vessel when deployment commenced, i.e. the net dropped vertically through the water during deployment. Estimates of egg abundance by age group in the survey area (Na) were calculated by multiplying the mean egg density at age in each stratum by stratum area and by summing across strata. In the case of the time strata in the center of the survey area, the egg abundances were averaged before summing with the other strata. The CV of N was calculated by us- ing the standard deviation of egg density at age and stratum, weighted by stratum area. The average of the standard deviations was used in the case of the central time strata. The maximum age of eggs that could be used in estimating daily egg production was the maximum age for which there appeared to be no significant advection out of the survey area owing to water move- ments (a loss of eggs by advection would cause a nega- tive bias in N ). Advection was examined by plotting centroids of each age group (Appendix 2) and by us- ing the buoy and CTD data described above. The daily egg production, N0 (eggs/day), and in- stantaneous mortality for eggs, Z (per day), were estimated by maximum likelihood with the mortal- ity model Na=N0e^t where t = the mean age (days) of age group a (Ap- pendix 3). The precision of these estimates was estimated by a bootstrap procedure (Appendix 3). Daily fecundity per female weight Survey design Female fecundity and ovarian stage samples were taken from trawls from the RV Tangaroa on Ritchie Bank from 7 June-6 July and from commercial vessels on 22 June, 11 July, and 13 July (Table 2). Trawling was done during the week before spawning started (8-11 June), to sample to- tal annual fecundity for the AEPM, and from the start of spawning until spent fish were common (20 June- 13 July), to sample fecundity reduction for the DFRM. Twenty five of the trawls were from Ritchie Hill (within the area of the central stratum in Fig. 2B) and three were from North Hill, a spur off the north end of Ritchie Bank, about 12 km north of Ritchie Hill (Fig. 2B) where fish were spawning. Oocyte sampling, ovarian staging, and daily fecun- dity estimation The oocytes of 569 mature fish (about 35 fish/trawl) in macroscopic ovarian stages 3 (late vitellogenic, prespawning), 4 (hydrated), 5 (ovulated), or 8 (late vitellogenic, partially spent) were counted. Of these, 218 fish were prespawning and used for total annual fecundity analysis. The remainder were used for DFRM analysis. Oocytes were counted by using the automated system de- scribed in Appendix 4. The proportions of females in ovarian stages 3, 4, 5, and 8, and stage 6 (spent) in the trawl samples were estimated by using macroscopic ovarian stag- ing of about 100 randomly chosen females per trawl. The ovarian stages were further grouped because it was observed that stages 3 and 4 (group 1) had higher fecundities/ kg than stages 5 and 8 (group 2); this difference was due to fish in group 2 having started spawning. Therefore, the estimate of fecun- dity per female weight, Rp was stratified by these groupings to minimize error. Stage-6 fish (spent) were placed in group 3. Thus, Rt was estimated as the mean number of eggs/kg of all females that would spawn, were spawning, or were spent, for trawl i: tv. j= i 582 Fishery Bulletin 95(3), 1997 Table 2 Trawl stations used for estimation of total annual fecundity for the AEPM and fecundity reduction for the DFRM June-July 1993 on Ritchie Hill (R. Hill) and North Hill (N. Hill) (Fig. 2B). Catches taken on 22 June and 11 and 13 July were from commercial tows and catch sizes were unknown. No. staged = number of fish staged; Immat. = immature; Prop, active = proportion of active fish. Date Site Catch (kg) No. staged Immat. Stage 3 Stage 4 Stage 5 Stage 8 Stage 6 Prop. active AEPM 8 Jun R.Hill 67 26 0 24 2 0 0 0 1.00 8 Jun R.Hill 30 9 0 8 1 0 0 0 1.00 9 Jun R.Hill 980 95 0 78 15 0 0 2 0.98 9 Jun R.Hill 980 95 0 78 15 0 0 2 0.98 11 Jun R.Hill 35 10 0 9 1 0 0 0 1.00 11 Jun R.Hill 239 88 2 70 11 3 2 0 1.00 11 Jun N.Hill 7,494 165 2 125 33 4 1 0 1.00 11 Jun R.Hill 549 81 1 67 13 0 0 0 1.00 14 Jun R.Hill 122 41 1 28 7 4 1 0 1.00 14 Jun R.Hill 103 31 3 22 5 0 1 0 1.00 14 Jun R.Hill 573 75 0 51 17 3 4 0 1.00 15 Jun R.Hill 1,195 77 0 51 21 1 4 0 1.00 15 Jun R.Hill 82 23 0 10 8 2 3 0 1.00 16 Jun R.Hill 47 14 1 9 2 0 2 0 1.00 16 Jun N.Hill 3,634 63 1 25 23 10 3 1 0.98 17 Jun R.Hill 21,478 214 0 90 96 24 4 0 1.00 17 Jun R.Hill 215 63 2 30 21 9 1 0 1.00 DFRM 20 Jun R.Hill 23,535 139 0 31 85 22 1 0 1.00 22 Jun R.Hill unknown 107 0 19 83 4 1 0 1.00 27 Jun R.Hill 7,218 83 0 6 26 39 10 2 0.98 27 Jun R.Hill 33,975 227 0 10 128 78 8 3 0.99 30 Jun R.Hill 19,405 135 0 3 50 69 8 5 0.96 2 Jul R.Hill 7,797 45 0 3 12 22 5 3 0.93 4 Jul R.Hill 2,936 147 0 1 67 50 13 16 0.89 6 Jul N.Hill 25,400 56 0 0 7 27 10 12 0.79 6 Jul R.Hill 402 83 0 0 2 57 19 5 0.94 11 Jul R.Hill unknown 31 0 0 2 15 3 11 0.65 13 Jul R.Hill unknown 77 0 0 6 31 11 29 0.62 where n; . = number of fish of ovarian group j at sta- tion i; and rif = mean fecundity (eggs/kg) of fish of ova- rian group j at station i. The mean fecundities/kg of fish of ovarian groups 1 and 2 (r. ,) were estimated as f J m *=1 where e( jk = total fecundity of the £th fish of group j in the fecundity sample from station i (adjusted by the gonad wall proportions of total ovary weight in Appendix 4); wijk= body weight (kg) of the kth. fish of group j in the fecundity sample from station i\ and mi f = number of fish of group j in the fecun- dity sample from station i. The R’ s from the 11 DFRM trawls were fitted with a linear regression against time (weighted by the total number of fish in the ovarian-stage sample for each Rt) to estimate D, which is the mean daily fe- cundity (eggs/(kg x day)) for mature fish. The CV of D was estimated as the standard error of the slope of the regression, divided by the slope. Biomass of spawning females The biomass of spawning females 5 , was calculated by using the DFRM model given above. The CV of Zeldis et al. : An estimate of biomass of Hoplostethus atlsnticus 583 B t -was determined from the standard error of 1,000 estimates of B ^ , formed by dividing the 1,000 esti- mates of Nq (Appendix 2) by 1,000 normally distrib- uted estimates of D, formed from the mean and stan- dard error of D. Proportions spawning, recruited, and female The scaling factors needed for converting the bio- mass estimate of Ritchie Bank spawning females to one for recruited fish (>32 cm SL for both sexes) for the entire mid-east coast stock were estimated from the March- April 1993 wide-area east coast trawl sur- vey (Field et al.2), instead of from the trawl data gath- ered at the time of the egg survey, for two reasons. First, not all recruited fish spawn each year (and thus may not migrate to Ritchie Hill). Second, a more pre- cise estimate of sex ratio is available from the trawl survey than from the relatively few trawls carried out during spawning (when sex ratios are most vari- able; Zeldis, 1993). The trawls from the wide-area survey used for the scaling factors were those from over the entire mid-east coast survey area, from just north of Ritchie Bank, south to Banks Peninsula (Fig. 1; quota management areas 2A South, 2B, and 3A, respectively). This is the likely distribution of the stock that migrates to the Ritchie Bank to spawn.10 No spawning orange roughy have been located in areas 2B or 3A, and genetic data show that orange roughy from these three areas cannot be separated, whereas they are genetically distinct from orange roughy on Chatham Rise (Fig. 1). Stage-3 (late vitellogenic) females in each trawl in the 1993 wide-area trawl survey were assumed to be those that would spawn that year (Bell et al., 1992). Because ovaries of stage-3 females are indistinguish- able macroscopically from ovaries of fish in which massive atresia has occurred (Bell et al., 1992), the macroscopic staging was checked by histological ex- amination of about 20 stage-3 fish collected on each day of the 28-day trawl survey. Recruited biomass The scalar (S) was calculated as 10 Annala, J. H., and K. J. Sullivan (compilers). 1996. Report from the Fishery Assessment Plenary, April-May 1996: stock assessments and yield estimates, 308 p. Unpubl. report held in NIWA Library, Wellington, New Zealand. where Preci and Pspfi Ci A l the proportions (by weight) of recruited fish and spawn- ing (stage-3) females, re- spectively, in the catch from the ith trawl; the catch rate (t/nmi) at the zth trawl; the stratum area for the stratum containing the ith trawl; and the number of trawls for the stratum containing the i th trawl. The precision of the estimates of S was estimated by using the following bootstrap procedure. For each trawl, the triplet (X, , Prec i , Pspfi ) was calculated, where X = CAi/nj. One thousand simulated data sets were generated by drawing triplets at random with re- placement. Each simulated survey contained the same number of trawls as the area of the original trawl survey. For each simulated survey a bootstrap estimate of S was calculated as X.'fV.A)’ The bootstrap estimates of S were used to calculate a CV for S. The recruited biomass, B , was calculated by us- ing the DFRM biomass model given above. The boot- strap estimates for S were combined with the boot- strap estimates of B t to obtain a CV for Brec. Results Daily planktonic egg production Planktonic eggs were first captured in very low num- bers on 15 June during subsurvey 1 (Fig. 3). Egg abundance remained low until the end of subsurvey 2 (25 June). This subsurvey was prolonged by ship and sampling-equipment breakdowns, and only the central strata, two of the middle strata, and none of the outer strata were sampled. The breakdowns pre- vented further sampling until the start of subsurvey 3 on 28 June, when large quantities of eggs were captured. Catches then decreased during subsurvey 4 (ending 6 July). Sampling of subsurveys 3 and 4 was completed. In subsurvey 5, only the central stra- tum and one middle stratum were occupied before the scheduled survey period ended on 7 July. Egg production continued at this time. 584 Fishery Bulletin 95(3), 1997 Stages 1-7 0-21.8 h old Stages 8-15 Stages 16-hatching 21.8-54.3 h old 54.3-240 h old A Subsurvey 1 14-20 June B Subsurvey 20-25 June C Subsurvey 28 June-2 July D Subsurvey 2-6 July E Subsurvey 6-7 July Figure 3 Positions and egg catch rates (m2) for plankton tows by subsurvey and egg age group. In all panels (A-E) , catch rates are proportional to circle area (zero catches are filled dots). Only six eggs were caught in subsurvey 1 (A); therefore only one panel was used to show station posi- tions. The extra stations (diamonds) and strata were added after completion of subsurvey 4 shown in (D). The largest symbol in the figure = 310 eggs/m2 sea surface area. Refer to Figure 2 for the plankton survey location and bathymetry. Zeldis et a I.: An estimate of biomass of Hoplostethus atlanticus 585 Data from the completed subsurveys 3 and 4 showed that egg catches were highest in the central and middle strata (Fig. 3, C and D). These eggs were predominately in the young, grouped-age category (stage 7 or less, <21.8 h old; Table 1), but many middle-aged and some older eggs were also caught in these strata. A few orange roughy yolk-sac larvae were also caught in the central and middle strata. These high catch rates for eggs were spatially corre- lated with high research trawl catch rates for adults (Fig. 2, A and B; Table 2 ). Catches in the outer strata were usually <10 eggs/tow or zero eggs/tow. However, there was one large catch of young eggs near North Hill (Fig. 3D), on which relatively large catches of spawning adults were made with research trawling (Table 2) and commercial trawling. All of the undam- aged eggs in this sample (19% of all eggs) were at the 1-cell stage, indicating that these eggs arose from localized spawning on North Hill. Four additional random tows were then made within an extra stra- tum created in this area (Fig. 3D) at the end of subsurvey 4. Only 5 additional eggs in total were caught in these tows, indicating that egg production on North Hill was low compared with that on Ritchie Hill. A few moderate egg catches were also made in the southwestern corner of the survey area (Fig. 3, C and D). Most of these eggs (84%) were > stage 7. Four additional tows were then made within an extra stra- tum created in this area (Fig. 3D) at the end of subsurvey 4, and in three of these tows, all eggs were > stage 7. The fourth sample had many damaged eggs (< stage 7), but it was likely that these were at the older end of the age range of “young” eggs, judging by the stages of undamaged young eggs in that sample (all were > stage 5). The bottom in the area where these samples were taken (Fig. 2B) is deeper (>1,000 m) than depths at which orange roughy nor- mally spawn (850-900 m), and trawl catch rates in the area were very low in this survey and in previ- ous surveys (Fig. 2, A and B). This finding indicated that these eggs had not been produced locally, but rather had been advected from the main spawning center on Ritchie Hill. The positions of the centroids (centers of gravity) of successive egg age groups suggested that advec- tion was initially to the southwest out of the survey area (Fig. 4 A) but that older eggs (those of the very- broad-age-group stage 16+, >54.3 h old; Table 1) re- entered the area, possibly from the east. In inter- preting Figure 4, it is important to realize that cen- troids close to the boundary of the survey area were unlikely because only eggs from within the survey area were used in the calculation of centroid posi- tion. An estimate of the average rate of advection, at the depths of these egg stages, was calculated by di- viding the distance between the centroids of stages <7 and 11 by the difference between the mean ages in these two age groups (Appendix 3) and was found to be 7.4 cm/sec. The distribution of eggs by age within the survey area (Fig. 4, B and C) confirmed the southwesterly drift pattern. Eggs > stage 11 (> 28.4 h old) became increasingly centered in the southwestern corner of the survey area (region 1, Fig. 4, B and C). However, the old eggs in the stage 16+ group were most abun- dant in the eastern and central regions (Fig. 4, B and C). The advection inferred from egg distributions can be compared with hydrographic results. The drogued buoy was relocated 11.2 km south-southwest (bearing=207°) of the release site after 20 h at lib- erty (Fig. 5A), indicating that advection (at 100 m depth) was to the south-southwest, at a rate of 16 cm/sec. Geostrophic analysis of the CTD data (Fig. 5, A-D) indicated that in the northern part of the station grid (in the vicinity of the Ritchie Hill spawn- ing site), current directions turned from south-south- east through south to southwest as depth increased from 100 through 400 to 800 m. The southwestern component of current velocity between 800 and 400 m (the approximate depth range of eggs < stage 11; Table 1) in the vicinity of Ritchie Hill averaged about 7 cm/sec (Fig. 5D). Geostrophic current speeds aver- aged about 12-13 cm/sec in the upper 100 m of the water column, in the vicinity of Ritchie Hill. These speeds were probably underestimates because the velocity profiles showed little evidence of reaching asymptotically low values as the postulated level of no motion (1,000 m) was approached (Fig. 5D), sug- gesting that some residual velocity existed at that level. This may explain the greater buoy speed than geostrophic speed at 100 m. Current speeds toward the southwest (Fig. 5D) were higher on the northern side of the grid than on the southern side through the upper water column; a significant easterly com- ponent was observed in the upper 200 m. Thus, the advection pattern indicated by egg cen- troids and stage distributions (Fig. 4, A and C) was consistent with results from the drogued buoy ( Fig. 5A) and the geostrophic analysis (Fig. 5, A-D), which indicated that the direction of drift of young eggs in the lower half of the water column was toward the southwest. The geostrophic velocities in the lower half of the water column in the vicinity of Ritchie Hill (at least 7 cm/sec) were consistent with the ve- locity of egg advection (7.4 cm/sec) from our calcula- tions. The older eggs, which would have spent most of their time in the mixed layer (Zeldis et al., 1995), might have been conveyed into the survey area from 586 Fishery Bulletin 95(3), 1997 the east by a return flow in the upper water column to the south and east of Ritchie Hill. Egg abundances over the survey area during subsurveys 3 and 4 were calculated by using the curved and straight trajectory assumptions (Table 3). The ratios of the curved and straight abundances generally decreased with egg stage. This decrease was expected because the curved trajectory is nearer to vertical deeper in the water col- umn (where the younger eggs are) than shallower in the column (where the older eggs are). The curved and straight trajectories tended to be nearly parallel in shallower water. Because eggs were advecting into the southwestern corner of the survey area (region 1, Fig. 4C) by the time they reached stage 11, only stages <10 were used in the estimation of daily egg production. The criterion used for this cutoff stage was that all stages > the first stage to have >20% of their abun- dance in region 1 would be ex- cluded from the analysis. This cri- terion was reached at stage 11. Using the abundance data calcu- lated by assuming a curved net tra- jectory (Table 3), we estimated that egg production was N0 = 10.9 x 109 eggs/day (CV=0.46) and that the mortality-rate estimate was Z = 0.70/day (CV=0.69) (Fig. 6; Table 4). These calculations used the ex- tra strata at North Hill and in the southwest, and all stations in subsurveys 3 and 4 falling within these strata were considered to have been originally selected within them. The N0 estimated by assuming a straight trajectory was 8.0 x 109 eggs/day (CV=0.49) with Z = 0.56 (CV=0.88). In the remain- ing analyses, the NQ value calcu- lated by assuming a curved trajec- tory was used because this value was likely to be a better approxima- tion to the truth than that obtained by assuming a straight trajectory. Daily fecundity of females A \ — 1 \ 1 1 j — / 1 / 1 — 1( H 1 — 16 1 y \ | \ 9 \ \ \ \ \ \ B C 16 15 14 13 s, 12 03 <55 11 10 9 8 <7 Figure 4 Studies of egg-stage distributions derived from subsurveys 3 and 4. (A) The posi- tions of centroids of the egg stages and the 36 subareas of the plankton survey area (solid lines) used in calculating centroid positions (Appendix 2). Centroid 7 represents the combined stage group 1-7, and centroid 16 represents stages 16 to 29 (hatching). The stratum boundaries of the survey area (dotted lines, slightly offset) are shown for reference. (B) Six regions and stations used in the distribu- tion analysis (region 3 comprised the inner and middle strata of the survey area), including the extra stations allocated in the southwestern corner. These latter stations were assumed to lie within region 1 for this analysis. (C) Proportions (circle areas) of eggs within the regions in B, by egg stage. -H-+ + + + + _P+H- ' + ++ +++ + ^ + -F + ,+ + it. T+ f +++ ++4l ■F + ++ + -HF+ + -F ++ _|F++3+ -F_|_ ^ 1 ++++ + -KP" + -fb+4+ H- + O 0 0 0 0 0 O 0 0 000 O 0 0 000 O 0 0 000 O O 0 000 O 0 0 0 0 0 O 0 0 0 0 0 0 0 0 000 O 0 0 000 O 0 0 0 0 1 2 3 4 5 6 Region The proportions of females in each ovarian stage in each DFRM trawl (Fig. 7) showed that the female population was, at first, nearly all in the prespawning state (stage 3). Maxima of hydrated, ovulated, and Zeldis et al.: An estimate of biomass of Hoplostethus atlanticus 587 A B Figure 5 Contours of dynamic height (dynamic meters) at 100, 400, and 800 m in the hydrographic survey area (A-C). The arrows on the contours show inferred current directions. The large filled circles indicate stations used in the analy- sis (where cast depths were > the 1,000 m reference depth). Panel (D) shows depth profiles of current velocity (cm/sec) at the symbols in A-C, flowing in the direction of the arrows through each symbol. In A, the plankton survey area is shown for reference and the start and end positions of the buoy deployment are shown by the beginning and end of the open-headed arrow. spent proportions followed at ap- proximately 10-15 day intervals. Serial spawning was evident in that partially spent, hydrated, and ovu- lated proportions became fairly con- stant during a 10-day period (roughly 25 June-5 July) when ova- ries of the fish were developing among these stages. Hydration did not appear to be associated with imminent spawning, because sig- nificant proportions (0.15) of hy- drated fish were present about 5 days before planktonic eggs were first caught on 15 June. However, the first appearance of significant proportions (>0.05) of ovulated fish (14 June) and planktonic eggs (15 June) nearly coincided. The decline in Rt (Fig. 8) or the daily fecundity per female weight, D, was 787 eggs/(kg x day) (CV= 0.11; Table 4). Ritchie Bank spawning female biomass When N0 was divided by D , the es- timate ofB ,was 14,000 tons (Table 4), with C v = 0.50. Biomass of recruited orange roughy in the mid-east coast stock The factor S was estimated to be 1.77, with CV = 0.03. However, his- tology showed that for 4.5% of the females identified as spawners in the wide area trawl survey, their entire exogenous vitellogenic oocyte complement was actually in the process of atresia (Bell et al., 1992). These fish were indistinguishable macro- scopically from fish that would spawn successfully and did not appear to cluster in any particular area of the trawl survey. The factor S was scaled upward to ac- count for these fish, resulting in S = 1.85 (Table 4). The resulting estimate of Brec was 26,000 t (Table 4), with CV = 0.50. The bootstrap procedure used may have slightly overestimated the CV of S because it did not fully take into account the stratum structure of the wide-area survey. However, this overestima- tion is of little importance because the CV of S was so much smaller than that of the other components that made up the CV of Brec (Table 4). Bias due to turnover An important potential bias in the DFRM arises from the fact that the method is sensitive to turnover of fish on the spawning ground. For example, if fish arrive, complete spawning, and leave the trawl sur- vey area before or after the trawl survey period, fe- cundity will be undetected and biomass will be un- derestimated. These effects were unlikely, however. Almost no eggs were caught in the first subsurvey (Fig. 3A), and very few spent fish were detected in trawls on Ritchie Hill until 2 July (Fig. 7). This find- ing suggested that no spawning was completed be- fore the trawl survey began (8 June). In addition, 588 Fishery Bulletin 95(3), 1997 Table 3 Abundances of eggs (x 10-6) and coefficients of variation (CV) at stage, during DFRM planktonic egg survey, calcu- lated with curved and straight net trajectory assumptions. Also given are the ratios of curved and straight abundances. Stage Curved abund. CV Straight abund. CV Ratios <7 7,036 0.26 5,453 0.28 1.29 8 487 0.40 424 0.43 1.15 9 673 0.60 451 0.51 1.49 10 316 0.43 276 0.49 1.15 11 400 0.24 351 0.26 1.14 12 1,420 0.32 1,276 0.32 1.11 13 740 0.23 716 0.24 1.03 14 276 0.38 260 0.37 1.06 15 34 0.42 38 0.38 0.90 proportions of macroscopic stage-3 (vitellogenic) fish declined to 5% by the midpoint of the trawl survey period and to 0% by the end; stage-4 (hydrated) fish showed a similar decreasing trend. Thus no prespawning fish were present at the end of the trawl survey (13 July). Turnover during the trawl survey period may have biased the biomass estimate if prespawning fish ar- rived late to the trawl survey area (after trawling for R/ had started), at the beginning of the season. This means that not all prespawning fish would have been sampled by trawls in the spawning area, which would cause an underestimate of R-, because fish that had started spawning would be over-represented. Similarly, if spent fish departed the trawl survey area early (before trawling for Rt had ended) toward the end of the season, spent fish would be under-repre- sented by trawls in the spawning area. In this case, i?; would be overestimated because fish which had not finished spawning would be over-represented. Both of these effects (late arrivals and early depar- tures) would cause an underestimate of D, which, in turn, would cause an overestimate of biomass, NQ/D. Was it likely that late arrivals or early departures (or both) of spawners occurred in the present study? The annual fecundity/kg of prespawners, estimated from fish sampled before the i?; sampling period and before any eggs were caught in the plankton (Fig. 8), was 27, 271 eggs/(kg x yr). If this estimate is divided by the estimated daily fecundity/kg (787 eggs/(kg x day); Table 4), the period required for the average fish to spawn completely is 35 days. However, the time lag between the first appearance of significant proportions (>0.05) of ovulated and spent fish was about 19 days (from 14 June to 2 July; Fig. 7). If this Table 4 Parameter estimates for DFRM for Ritchie Hill spawning female biomass and mid-east coast recruited biomass, June-July 1993 (with coefficients of variation in paren- theses). N0 = daily egg production for Ritchie Hill survey area (estimated with curved net trajectory); D = weight specific daily fecundity of females; Bsp^ = biomass of spawn- ing females in Ritchie Hill survey area; S = ratio of re- cruited biomass to that of spawning females; B = biom- ass of recruited fish. Parameter estimates with subscripts marked “ turn ” have turnover incorporated (see text); CV’s were not estimated in these cases. Parameter Estimate D B S Bi D spf J spf turn rec,turn 10.9 x 109 eggs/day (0.46) 787 eggs/(kg x day) (0.11) 14.000 t (0.50) 1.85 (0.03) 26.000 (0.50) 1,106 eggs/( kg x day) 9,900 t 18,200 t Table 5 Mean abundances (per m2) of all eggs < stage 7 and dates of sampling in the central strata for each subsurvey. Subsurvey Date Mean CV 1 15-17 June 0.0 0.0 2 20-25 June 3.9 0.34 3 29 June-1 July 49.0 0.29 4 3-5 July 22.6 0.46 5 6-7 July 22.2 0.50 4 and 5 3-7 July 22.4 0.34 lag is interpreted as the duration of spawning in in- dividual fish, D was underestimated by the fecun- dity reduction trawling. It would appear, however, that late arrival of prespawners to the trawling area did not contribute greatly to the underestimation of D. The Rt sampling period began on 20 June when prespawner (stage-3) proportions had become low (0.20; Fig. 7) and were decreasing rapidly. At this time the estimated Rt was only about 10% below the prespawning level (Fig. 8). Eggs first appeared in the plankton on 15 June, but catches of young eggs in the central strata were still relatively small (3.9 eggs/m2; Table 5) during 20-25 June (subsurvey 2). Therefore, only a relatively small reduction in Rt would have been expected by Zeldis et al. : An estimate of biomass of Hoplostethus atlanticus 589 20 June, in agreement with the reduc- tion actually observed by that date (Fig. 8). It was likely, however, that early departures caused spent fish to be un- der-represented in the trawl samples toward the end of the Rt time series. The abundance of young planktonic eggs in the central strata was reduced by more than half between subsurvey 3 (mid-date 30 June) and subsurvey 4 (mid-date 4 July; Table 5). Because subsurvey 3 was done when the spent proportion was virtually zero (Fig. 7), this reduction of planktonic egg abun- dance implied that about half of the fish had ceased spawning by 4 July (assuming that the spawning rate of remaining active fish was constant). However, only 0.11 of fish were re- corded as spent on 4 July (Table 2; Fig. 7); thus this proportion appeared to be underestimated by 0.50 - 0.11 = 0.39 on 4 July. Commercial catch rates on Ritchie Hill (Fig. 9) also decreased con- siderably during the last week of June and first week of July, suggesting that fish abundance in the trawling area had declined. To account for the error in D that the underestimation of spent fish would cause, the proportion of active fish on 4 July was adjusted downward to 0.89 - 0.39 = 0.50 to estimate a new D value of 1,106 eggs/ (kg x day) (assuming a linear decline be- tween 20 June and 4 July; Fig. 8; Table 4). This adjustment resulted in a period of 25 days for an average fish to spawn com- pletely, which is not greatly different from the 19 days estimated from the time lag between ovulated and spent fish propor- tions (shown above). This re-estimation of D, to account for turnover, had propor- tional effects on Bsp^ and Brec (Table 4). Figure 6 The daily production of eggs as a function of age, calculated by dividing the curved trajectory egg-abundance estimates (Table 3) by the duration of each age group and by plotting against the mean age of eggs in each age group ( Appendix 3 ). The line was fitted by using N0 and Z ( Appendix 3 ), calculated with only egg stages < 7 to 10 (crosses). Egg stages 11 to 15 (triangles), were not used in the production estimate because they were subject to advection out of the survey area (see text). 5 Jun 10 Jun 15 Jun 20 Jun 25 Jun 30 Jun 5 Jul 10 Jul 15 Ju I Date Figure 7 Discussion Time series of macroscopic ovarian stages over the survey period. Each point is the proportion of females at that stage, as a proportion of all mature females, averaged for all trawls on that date. Only trawls on Ritchie and North Hills were used (Table 2). This study has used the DFRM to estimate the biomass of recruited individuals in the mid-east coast orange roughy stock, by combining estimates of daily egg production rate, the rate of fecundity reduction, and the proportion of the stock that was spawning females. In this section vari- ous sources of error are discussed, and the methods and results of this study are compared with those of other studies. Also, a brief description is made of how the results from this survey have been used in as- sessing the stock. 590 Fishery Bulletin 95(3), 1997 Sources of error In egg-production surveys, the trajectory of the plank- ton net during deployment and retrieval is usually assumed to be straight (and often, vertical). The analyses presented here (Appendix 1) suggest that, for the present survey, 1) there was significant hori- 30000 25000 20000 t 55 15000 - ao PJ 10000 5000 - 05 Jun 10 Jun 15 Jun 20 Jun 25 Jun 30 Jun 05 Jul 10 Jul 15 Jul Date Figure 8 Weight-specific fecundity (eggs/kg spawning female: Ri ) from trawls prior and during the DFRM sampling period (circles and squares, respectively, Table 2). Solid line is a weighted linear regression fitted for DFRM trawls. Dotted line is a regression fitted when it was assumed that proportion active was overestimated by 39% on 4 July (see text). zontal movement of the net during deployment ow- ing to drag from the warp (which was caused by ship drift), and 2) the plankton net followed a curved tra- jectory during hauling. The curved trajectory meant that less water was filtered in deeper than in shal- lower ocean layers of equal thickness and that larger correction factors were required for younger (deeper) eggs in order to standardize the egg counts to eggs/m2. The effect of using curved, rather than straight, trajectories was to increase the estimated production rate by 36% (from 8.0 to 10.9 billion eggs/day). Recent egg-production survey work, using a digitally recording flow- meter system developed at NIWA (Grimes9), has shown that 1) there often are spurious revolutions of the flowmeter on the downcast, negating the use of conventional flowmeters and that 2) more water is filtered at shallower depths than at deeper depths, justifying the assumption of a curved net trajectory. There- fore, it is likely that by allowing depth varia- tion in the estimates of the amount of water filtered, a major improvement was made over the assumption of a straight net trajectory. The precision of the planktonic egg produc- tion estimate was influenced by damage to early stage eggs (Zeldis et al., 1995; Grimes et al.3). In the present study, the inability to stage most eggs less than 21.8 h old caused a greater reliance on older egg stages to estimate pro- duction. However, the number of older egg stages that could be used for the esti- mation was limited because eggs older than stage 10 (28.4 h) were subjected to advection toward and through the southwestern boundary of the survey area as they aged. Thus, the original strategy of making the survey area large enough to retain all of the eggs up to 36 h old was defeated, and rela- tively few stages were available for a mortality estimate. Clearly, the com- bination of nearly concurrent informa- tion on ocean circulation (from CTD and drogued buoy) and on the drift pat- terns of the eggs themselves provided useful corroborative information for quantitative decisions about the egg- age range available for mortality esti- mation, when eggs were subject to ad- vection. Precision estimates for the egg- abundance data may have been biased by potential autocorrelation among the data, which would not bias the esti- Zeldis et a I.: An estimate of biomass of Hoplostethus atlanticus 591 mate of egg production but could cause its precision to be overestimated. No attempt was made to cor- rect for this bias because the data were judged to be inadequate to estimate an autocorrelation structure (which would need to have both spatial and tempo- ral terms). Picquelle and Megrey (1993) drew the same conclusion for pollock egg surveys. For the present study, a minimum spacing of 1,000 m was imposed when allocating the stations to the strata. This spacing should have gone some way toward minimizing the effects of spatial autocorrelation. In an explicit study of autocorrelation in anchovy egg catches, Smith and Hewitt (1985) found that spatial autocorrelation diminished rapidly at spatial scales of 2,000 m and that temporal autocorrelation dimin- ished at 0.5 h for 1-day-old eggs. Given that the mini- mum distances and times between samples for the egg stages used in this study were of this order and that the egg catch rates were highly variable, over- estimation of precision was likely to be minor. Use of the DFRM required the assumption that the 11 trawls used to estimate D representatively sampled spawning females on Ritchie and North Hills. Although it is clear that spawning female bio- mass is highly aggregated in the Ritchie Hill area (Fig. 2), it is not known to what extent females are randomly distributed within this small area, with respect to ovarian stage. Therefore, it was useful to look at a much larger set of fecundity data collected on these hills in 1995 by N.Z. Ministry of Fisheries (MOF) scientific observers working on commercial vessels (Zeldis, unpubl. data). Samples were taken from 47 trawls made with four vessels (range of 7— 16 trawls per vessel) throughout the entire spawn- ing season. These tows were long, typically travers- ing much of the ridge between North Hill and the south side of Ritchie Hill (Fig. 2) and covering the 850-900 m depth range of orange roughy spawning. The fecundity reduction rate from these 1995 data (uncorrected for turnover) was 965 eggs/(kg x day) (CV=0. 11 ), which was not significantly different from the uncorrected estimate from 1993 in the present study, 787 eggs/(kg x day) (CV=0.11; Table 4). This rate showed that widespread and intensive trawling in this area would yield an estimate of D similar to that from the less intensive research trawling of the present study, indicating that the less intensive trawling accurately represented the spawning dy- namics of the population. The reliance of the DFRM on within-spawning sea- son trawl data makes it susceptible to bias due to turnover, which causes an underestimate of fecun- dity reduction rate. In the present study, it appeared that spent females left the survey area during the period when fecundity reduction was measured, such that only 0.11 of remaining females were spent 21 days after the start of spawning. A similar pattern of low proportion spent was seen in the 1995 Ritchie Bank scientific observer data described above, with proportion spent < 0.20 about 23 days after the on- set of spawning. In contrast, in recent NIWA orange roughy egg production surveys done at East Cape in 199510 and at the “Graveyard” area on northern Chatham Rise in 199611 (Fig. 1), spent proportions were between 0.60 and 0.70 about 20 days after the onset of spawning. This suggested that turnover was less prevalent at these latter sites, an effect which may be related to the fact that although commercial fishing was intensive on Ritchie Bank during the 1993 and 1995 spawning seasons, there was no com- mercial fishing during the 1995 and 1996 spawning seasons at East Cape and the “Graveyard.” Signifi- cantly, the value of D at East Cape was 1,036 eggs/ (kg x day), with no correction for turnover, which was similar to the turnover-corrected value for the present study at Ritchie Bank of 1,106 eggs(kg x day) ( an estimate is not yet available for the “Graveyard”). For this reason, the turnover-corrected fecundity reduction rate estimated for Ritchie Bank in 1993 is considered to be more reliable than the uncorrected estimate and also a reasonably reliable estimate of the true rate of decline in Ritchie Bank population seasonal fecundity. Comparision with other studies The first use of the DFRM was by Lo et al. (1992; 1993) to estimate the biomass of a deepwater pleuronectid flatfish, Dover sole ( Microstomus pacificus), which spawns between 600 and 1,500 m depth on the continental slope of western North America. The model used in the present study to describe the daily fecundity reduction was simpler than that of Lo et al. (1993). It assumed that the fecundity per fish weight was a linear function of time only (fish weight was considered as an addi- tional predictor but did not significantly improve the fit). Lo et al. (1993) assumed that both total fecun- dity of active females and the fraction of active fe- males (their Ef and G() were linear functions of time and fish weight. The former model was used for two reasons. First, in calculating fecundity reduction, there seemed no need to treat active and inactive females separately. Second, in the absence of any evidence of lack of fit, the rule of Occam’s razor sug- gested using the simpler model. 11 Grimes, P. J. 1996. Voyage Report, TAN9608 (Part II). NIWA unpublished voyage report held in NIWA Library, Greta Point, Wellington, New Zealand, 4 p. 592 Fishery Bulletin 95(3), 1997 A second difference between the DFRM used here and that of Lo et al. (1993) is the way sex ratio and active female proportion were estimated and brought into the biomass model. In the present study, the proportion of all recruited fish that were spawning females was estimated with parameter S, whereas this proportion was estimated within the R and D parameters in the model of Lo et al. (1993). Orange roughy spawning takes place in dense aggregations that form after the spawners have migrated hun- dreds of km from the nonspawners. Therefore, it is not possible to estimate the proportion of active fe- males to all recruited fish with the trawls done on the spawning ground during the spawning season (which are used to monitor fecundity reduction). This estimation must be done with a separate trawl sur- vey over the whole stock area, preferably before the spawners have aggregated significantly. In contrast, Dover sole spawning appears to be dispersed over a wide latitudinal area of the North American west coast slope (Lo et al., 1993) and takes place over a long spawn- ing season (6 months; Hunter et al., 1992). In Lo et al. (1993), the spawners were assumed to be dispersed in the same geographic region as the nonspawners dur- ing spawning, and therefore sex ratio and proportion active components were estimated from the same trawls that were used to estimate fecundity reduction. There are few other estimates of planktonic egg mortality (Z) in the literature for deepwater spawn- ers that can be compared with the value (Z=0.7) ob- tained for orange roughy in this study. Lo et al. ( 1993) fitted a Pareto decay function to Dover sole egg abun- dances, in which mortality was allowed to decrease with egg age. Initial mortality was 0.63 which halved by the age of 1 day. Another western North Ameri- can slope species, sableflsh ( Anoplopoma fimbria), has been the subject of egg-production studies (Moser et al., 1994) and for this species, Z varied between 0.25 and 0.48, depending on region (it should be noted that counts of 1-day-old and 2-day-old eggs were ex- cluded because these eggs were undersampled). Thus, the few Z estimates that exist for other deepwater species indicate variable mortality rates, but that mortalities can be quite high. Orange roughy egg abundance was estimated by Koslow et al. (1995) in their application of the AEPM to orange roughy biomass estimation on St. Helen’s Seamount, Tas- mania. These authors assumed that mortality in the first 28 h after spawning was zero and that the mean abundance of 0 to 28 h old eggs equalled N(y This assumption was based on their finding no differences in abundance among fertilization or 1-cell eggs (stages 1 and 2 of the present study) and their sub- sequent two stages which were 2 cell and 4-128 cells (stages 3 and 4-9, respectively, of the present study; see also Grimes et al.3). As mentioned, the differentia- tion of stages 1-7 was not possible for the great major- ity of the samples in the 1993 Ritchie Bank survey be- cause of egg damage. However, the probability that the mortality rate estimated from the grouped stages 1-7 and stages 8, 9, and 10 was zero was low (P=0.10). The procedure used to estimate S, for converting spawning female biomass to recruited biomass, is easily adapted to estimate the proportion of females in the stock area that will spawn in the current year. For the mid-east coast stock in 1993, S was estimated as 0.52 of all mature females (by weight), a result similar to that for orange roughy stocks in southern Australia (0.45, by numbers; Bell et al., 1992). Wide- area trawl surveys of the east coast were also done in March-April of 1992 and 1994, 10 and these yielded similar values for these proportions (0.49 and 0.42 by weight). Also, a low proportion of females with fully atretic ovaries was found (0.045) over the mid- east coast survey area in 1993, again in common with the results of Bell et al. (1992). A B0 ('000 t) B Figure 10 Probability distributions of (A) virgin biomass (B0) and (B11994 biomass as a percentage of B0, from the 1994 stock reduction analysis for the mid-east coast orange roughy, Hoplostethus atlanticus, fish- ery (see Footnote 2 in the text). Each panel shows (solid line) the distribution when the analysis was done without the DFRM estimate (with only CPUE, trawl survey indices, and mean fish length data) and (dotted line) the distribution when the DFRM estimate was included. Zeldis et al. : An estimate of biomass of Hoplostethus atlanticus 593 Use in stock assessment Although the biomass estimates derived from this DFRM survey were rather imprecise (CV=0.50), they were capable of having a dramatic effect on the as- sessment of the mid-east coast stock. In 1994, an estimate of 45,000 t recruited biomass (CV=0.40) from a preliminary analysis of these data was used in assessing this stock (Field et al.2: this estimate differs from those given above because some data errors have subsequently been corrected and the analyses have been refined). The inclusion of this estimate greatly improved the precision of the as- sessment (Fig. 10), which resulted in a 2-year stepped reduction in TACC from 10,333 t in 1993-94 to 2500 t in 1995-96. Acknowledgments The authors would like to thank Kevin Sullivan for statistical advice and revision of the manuscript, Steve Chiswell for assistance with geostrophic analy- sis, Karen Field for laboratory assistance, and Alan Hart, Jack Fenaughty, and the officers and crew of Tangaroa for assistance at sea (all at NIWA). We thank Tony Koslow (CSIRO) for provision of St. Helen’s seamount orange roughy egg data, and the external reviewers for their comments. This work was partly funded by the New Zealand Ministry of Fish- eries under project DOR609. Literature cited Bell, J. D., J. M. Lyle, C. M. Bulman, K. J. Graham, G. M. Newton, and D. C. Smith. 1992. Spatial variation in reproduction, and occurrence on non-reproductive adults, in orange roughy Hoplostethus atlanticus Collett (Trachichthyidae), from south-eastern Australia. J. Fish Biol. 40:107-122. Bycroft, B. L. 1986. A technique for separating and counting rock lobster eggs. N.Z. J. Mar. Freshwater Res. 20:623-626. Clark, M. R. 1995. Experience with management of orange roughy ( Hoplostethus atlanticus) in New Zealand waters, and the effects of commercial fishing on stocks over the period 1980-1993. In A. G. Hopper (ed.), Deep-water fisheries of the North Atlantic oceanic slope, p. 251-266. Kluwer Academic Publishers. Francis, R. I. C. C. 1984. An adaptive strategy for stratified random trawl surveys. N.Z. J. Mar. Freshwater Res. 18:59-71. Hunter, J. R., B. J. Macewicz, N. C. H. Lo, and C. A. Kimbrell. 1992. Fecundity, spawning, and maturity of Dover sole Microstomus pacificus, with an evaluation of assumptions and precision. Fish. Bull. 90:101-128. Hunter, J. R., and N. C.-H. Lo. 1993. Ichthyoplankton methods for estimating fish biom- ass introduction and terminology. Bull. Mar. Sci. 53(2):723-727. Koslow, J. A., C. M. Bulman, J. M. Lyle, and K. A. Haskard. 1995. Biomass assessment of a deep-water fish, the orange roughy ( Hoplostethus atlanticus), based on an egg survey. Mar. Freshwater Res. 46:819-830. Lo, N. C. H., J. R. Hunter, H. G. Moser, and P. E. Smith. 1992. The daily fecundity reduction method: a new proce- dure for estimating adult fish biomass. ICES J. Mar. Sci. 49:209-215. Lo, N. C.-H., J. R. Hunter, H. G. Moser, and P. E. Smith. 1993. A daily fecundity reduction method of biomass esti- mation with application to Dover sole Microstomus paci- ficus. Bull. Mar. Sci. 53(21:842-863. Moser, H. G., R. I. Charter, P. E. Smith, N. C. H. Lo, D. A. Ambrose, C. A. Meyer, D. M. Sandknop, and W. Watson. 1994. Early life history of sablefish, Anoplopoma fimbria, off Washington, Oregon, and California, with application to biomass estimation. CALCOFI Rep. 35:144-159. Pankhurst, N. W. 1988. Spawning dynamics of orange roughy, Hoplostethus atlanticus , in mid slope waters of New Zealand. Environ. Biol. Fishes 21(21:101-116. Pankhurst, N. W., P. J. McMillan, and D. M. Tracey. 1987. Seasonal reproductive cycles in three commercially exploited fishes from the slope waters off New Zealand. J. Fish. Biol. 30:193-211. Picquelle, S. J., and B. A. Megrey. 1993. A preliminary spawning biomass estimate of wall- eye pollock, Theragra chalcogramma , in the Shelikof Strait, Alaska, based on the annual egg production method. Bull. Mar. Sci. 53(21:728-749. Pond, G. L., and M. A. Pickard. 1978. Introductory dynamic oceanography. Pergamon Press, Oxford, 241 p. Saville, A. 1964. Estimation of the abundance of a fish stock from egg and larval surveys. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 155:164-173. Smith, P. E., and R. P. Hewitt. 1985. Anchovy egg dispersal and mortality as inferred from close-interval observations. CalCOFI Rep. 26:97-108. Tranter, D. J., and P. E. Smith. 1968. Filtration performance. In Zooplankton sampling, p. 27-56. UNESCO monographs on oceanographic method- ology 2. Zeldis, J. 1993. The applicability of egg surveys for spawning stock biomass estimation of snapper, orange roughy and hoki in New Zealand. Bull. Mar. Sci. 53(21:864-890. Zeldis, J., P. J. Grimes, and J. K. V. Ingerson. 1995. Ascent rates, vertical distribution, and a thermal history model of development of orange roughy ( Hoplo- stethus atlanticus) eggs in the water column. Fish. Bull. 93(21:373-385. 594 Fishery Bulletin 95(3), 1997 Appendix 1 : Calculation of "curved" correction factors This appendix describes, for the case where the path of the net during hauling is assumed to be curved, the calcu- lation of the correction factors that convert egg counts to egg density (eggs/m2). For each plankton tow, we calculated the following: 2n(i1), and P(;(f2) may written as Pnl, znl, and Pu2, respectively. From the above assumptions, P_vV P .„ z v and wx are known, as are C(z) and C'(z) for all 2. Also, of course, Pn2 = Calculating Pn] 1 the position of the net at the start of hauling; 2 the position of the net at a series of equally spaced times during hauling; 3 the flow of water through the net at each of these times; and 4 a correction factor for each of these times. Finally, interpolation was used to calculate the correction factor when the net was at the mid-point of the depth layer associated with each egg stage. This was taken as the cor- rection factor for that egg stage at that plankton tow. First, the assumptions behind these calculations and some notation are defined. Assumptions and notation The calculations required the following assumptions: 1 the vessel drifted at a constant velocity while shooting and hauling; 2 the net dropped at a constant speed during shooting; 3 the warp was always straight and decreased in length at a constant rate during hauling; 4 the net mouth was always perpendicular to the warp; 5 the water velocity varied only with depth (not with lon- gitude, latitude, or time); and 6 the following are known exactly: a) the vessel position at the time of shooting and at the start and finish of hauling, b) the net depth and warp length at the start of haul- ing, and c) the water velocity profile. Two further assumptions are described in sections “Cal- culating P d” and “Calculating the net path during hauling.” Let Pn(f) and P()(£) be 3-dimensional vectors describing the position of the net and the vessel at time t, where time and position are measured in relation to the time and the vessel position when the net was shot, and where the three coordinates give distances, in meters, to the east, north, and downwards, respectively. (In this Appendix, vectors are underlined to distinguished them from scalars.) Let t1 and 1 2 stand for the times at the start and finish of hauling. Let the water velocity at depth 2 be described by the vector C(z), and the mean water velocity between the sur- face and depth 2 by C'(z). Denote the net depth and the warp length at time t by zn(t) and w(t), respectively (note that zn(t) is the depth coordinate of Pn(t)). Where it is convenient, the symbols t1 and t2 will be re- placed by subscripts 1 and 2, so that, for example, Pn{tx), In calculating the position of the net at the start of haul- ing, it was assumed that, while the net was sinking prior to hauling, its horizontal velocity was the sum of the wa- ter velocity and some fraction c of the vessel velocity (the latter component being caused by drag from the warp). Thus, Pnl — cPvl + txQ\Znl) + Znl , (Al) where 0 < c < 1, and Znl is the 3-dimensional vector, (0,0, znl). Also, from assumptions 3 and 6, \Enl-E.vl\ = Wl- (A2> The value of c (and thus Pnl) can be calculated by solv- ing the simultaneous equations A 1 andA2. Geometrically, this is equivalent to finding a point of intersection of the horizontal line defined by Equation Al and the horizontal circle defined by Equation A2 (both are at depth znl). Substituting for Pnl from Equation Al in Equation A2 and expanding, we get the quadratic equation in (c — 1 ) (c-l)2|P„1f + 2(c-l)t1Pvl> (A3) C"( Zn j ) + 1 2 |C_(2n 1 )| + 22j — uv\ — 0 , which is solved with the usual formula ( • denotes the vec- tor dot product). Where there were two solutions for c (at all but a few stations), the one using the negative square root was cho- sen because it was usually the one that satisifed the con- dition 0 < c < 1. For the few stations where Equation A3 had no real so- lution (i.e. the line and circle did not intersect), Pnl was taken to be the point on the circle closest to the line. It may be shown that this Pn l is given by P, + Znl (A4) where r is the radius of the circle and Pvl is the point, on the surface vertically above the line, that is closest to the shot position, and is given by txC’(znl)»PvX P'i = hQ\znx)- — „i (A5) Stations where Equation A3 did not have a real solu- tion, or where c did not satisfy 0 < c < 1, were assumed to be instances where the above assumptions did not hold. Zeldis et ai: An estimate of biomass of Hoplostethus atlanticus 595 Another situation in which the assumptions clearly did not hold was at the 13 stations where znl > wv For these stations, z x was set equal to wv and Pnl was taken to be vertically below PvV Calculating the net path during hauling From assumptions 1 and 3, the position of the vessel and length of the warp at any time during hauling was calcu- lated by using In solving Equation A13, the solution using the positive square root was ignored because it led to large values of p(t) and large vertical oscillations in the net path. Flow of water through the net Because, by assumption 4 above, the net mouth was al- ways perpendicular to the warp, the flow of water through the net at time t (in m3/s) is given by Fit) = 2V_nw *Uwp, (A14) Pv(t) (*2 -*>£■! +<*-*■>£. 2 «2-fl) where Vnw is the velocity of the net relative to the water, given approximately by and wit) = Pv(t)~ Pnit) = (t2 - t)w1 %-k) (A7) V Pnit+8t)-Pnit) ~5t C(znit)), (A15) To calculate the position of the net during hauling, we made one further assumption: that the velocity of the net is the sum of the water velocity and a vector in the direc- tion of the warp. With this assumption, an iterative pro- cedure was used to calculate P it) at times t{+*t, tx+2*t, etc, for a small time interval *t. The basis of this proce- dure is the ability to calculate Pnit+*t) once Pnit) is known. This is done as follows. The velocity assumption may written approximately as H is a unit vector in the direction of the warp, given by jj _ £,(*)-£„(*) ~uip \Pv(t)-Pn(t)\’ (A16) and factor 2 is the net mouth area in m2. Calculation of correction factors Pn(t + 8t) -P.it) ... Pv(t)-Pn(t) . . .... pU)j~ +C[zn(t)), (A8) St \P .it) -P.it) where pit) is an unknown scalar which varies with t. Re- placing t by t+*t in Equation A7, we get | (A9) and substituting for P it+*t) from Equation A8, Equation A9 may be rewritten as where and |p(f )A + JBl = — 2 * 6t)Wl, A = — (0.05) from that for mature males (0.822); however, the regression fit of the relation- ship (df=55, F=4680, P=<0.001, r2=0.31) shows that CF increased with length for small fish, reached a peak at 34.5 cm, and declined in larger fish (Fig. 3B). Hepatosomatic index varied significantly among female maturity groups (single-factor ANOVA; df=5, F= 59.0, P<0.001) and was significantly greater in both the late vitellogenesis stage (2.281) and migra- tory nucleus stage (2.921). In general, HSI increased Zimmermann: Maturity and fecundity of Atheresthes stomias 605 4.500 -| 4 000 - 3 500 - 3.000 - 2.500 - 2.000 - 1.500 - 1.000 - 0.500 - A o Early perinucleus □ Late perinucleus x Cortical alveoli * Early vitellogenesis a Late vitellogenesis * Migratory nucleus A A O O o o o o Dgo □“ v A** . ° □ Aa a X A* X , A A* V * X * □ □ * A * > ,B0 £□* ° *□ °i X 3n « 8x X 0D A S 0 000 10 20 30 40 50 60 70 80 90 cc Q. CD X 2.5 n B □ Immature * Mature 1.5 - 1 - 0 5 - 0 -I 1 1 1 1 1 1 1 1 1 10 15 20 25 30 35 40 45 50 55 Length (cm) Figure 4 Hepatosomatic index by length and maturity stage for female arrowtooth flounder (A) and for male arrowtooth flounder (B). with increasing female length (Fig. 4A). There was no difference in HSI between mature and immature males (unpaired f-test; df=27, P>0.05), and no trends in the data (Fig. 4B). Gonadosomatic index varied significantly among female maturity groups (single-factor AN OVA; df=5, F=91.9, PcO.OOl). The Tukey test revealed that de- spite more than doubling, the GSI did not change significantly between the early and late perinucleus stages (Table 3). After that, the GSI was significantly greater in each succeeding stage of maturity (Table 3). Figure 5A shows that GSI remained low until fish reached lengths over 45 cm. GSI was significantly greater in mature males than in immature males 606 Fishery Bulletin 95(3), 1997 12.000 n 10.000 - 8.000 - 6.000 - 4.000 2 000 - $ 0 000 o Early perinucleus □ Late perinucleus x Cortical alveoli x Early vitellogenesis a Late vitellogenesis * Migratory nucleus A * A A ti A A X X*„ X„ * D Bx xx i 5 *x * x**« Jx X„ X X*. OO qOOq ° □ QgQ0BDBDBann°DDD90BBBBBKn □§ □ □” 10 20 30 40 50 60 70 80 03 E o CO o n cs c S 1 000 -I 0.900 - 800 - 700 - 600 - 500 - 400 - 300 - 200 - 100 - B 0 000 □ Immature a Mature A A 0 n □ □ D □ □ □ B 9 n 0 □ A □ □ □ ° □ O B 90 10 15 20 25 30 35 40 45 50 55 60 Length (cm) Figure 5 Gonadosomatic index by length and maturity stage for female arrowtooth flounder (A) and for male arrowtooth flounder (B). (unpaired £-test; df=27, P<0.001); the increase began between 30 and 35 cm in length (Fig. 5B). According to histological analysis, L50 for females occurred at 46.9 cm (df=174, F= 897, r2=0.76, Fig. 6A) and at 42.2 cm (df=56, F=69, r2=0.60, Fig. 6B) for males. According to macroscopic staging, L50 for females oc- curred at 49.9 cm (df=172, F=748, ^=0.76, Fig. 6A). Total fecundity In general, the eyed-side ovary lobes were heavier than the blind-side lobes (paired £-test; df=154, t= 2.663, P=0.009). Linear regression analysis (df=153, F= 2365, P<0.001, r2=0.94) described the relation between the weight of ovarian lobes as Zimmermann: Maturity and fecundity of Atheresthes stomias 607 WB = 0.843 (WE) + 1.398, where WB = the weight of the blind-side lobe; and WF = the weight of the eyed-side ovary lobe. All the frequency distributions of the diameter of maturing oocytes were unimodal and had a long tail on the left. A cubic transform (diameter3) normal- ized the distributions. The transformed values of the mean diameter of oocytes did not vary significantly among positions (anterior, medial, posterior) within the same ovarian lobe (two-factor ANOVA, df=2, P>0.05) or between lobes (paired £-test, df=9, P>0.05). Density of oocytes (count of oocytes per gram of ova- rian tissue) also did not vary significantly between positions within ovarian lobes (two-factor ANOVA, df=2, P>0.05) or among ovarian lobes (paired f-test, df=9, P>0.05). Thus samples for total fecundity were combined from the different positions and ovarian lobes. Total fecundity was significantly related to fish length (df=ll, F=125.7, P<0.001, r2=0.92) by the equation 608 Fishery Bulletin 95(3), 1997 ln(F) = 4.020 ln(L)- 3.149, from which was derived F = 0.0429 (L)4020, where F = total fecundity; and L = fork length in centimeters (Fig. 7A). Length-based total fecundity estimates ranged from 246,000 (48 cm) to 2,224,000 (83 cm) oocytes. Total fecundity was also significantly related to so- matic fish weight (df=ll,F=264.7, P<0.001, r2=0.96): F = 350.4(W) - 138,482, where F = total fecundity; and W = somatic weight in grams (Fig. 7B). Weight-based total fecundity estimates ranged from 255,000 (1,122 g) to 2,339,000 (7,070 g) oocytes. Discussion Maturity Histological analyses demonstrated that the assign- ment of macroscopic maturity stages was not always reliable. Although it was confirmed through histol- Zimmermann: Maturity and fecundity of Atheresthes stomias 609 ogy that none of the “immature” females had begun acquiring yolk, some had begun the process of ma- turing oocytes, and several had atresia of large, pre- viously yolked oocytes, indicating that these females had probably spawned during the previous year. All females classified macroscopically as “developing” had maturing oocytes to spawn, most of them in the later stages of vitellogenesis. Few of the fish classi- fied as “spent or resting” were actually resting: most were in vitellogenesis and should have been classi- fied as “developing.” In addition, the author was un- able to assign a single maturity stage to ovaries with a mixed appearance and to assign a stage for ovaries from some females. The several females correctly classified as “spent or resting” deserve some discussion. These fish had oocytes only as advanced as the late perinucleus stage and had degenerating oocytes present, which had previously been yolked. These fish were classified as mature because they had previously contained yolked oocytes, but they were unable to spawn soon, despite the upcoming spawning season, and they did not show signs of recent spawning. Hunter and Macewicz (1985, a and b) have reported histological details on creation and resorption of atretic oocytes and postovulatory follicles in the northern anchovy, Engraulis mordax. Their results have shown that the relatively rapid resorption of yolked oocytes (after 23 days of starvation; Hunter and Macewicz, 1985a) and postovulatory follicles (after 3-4 days; Hunter and Macewicz, 1985b) seems to contradict my asser- tion that atretic vitellogenic oocytes from the previ- ous spawning season were still being resorbed in arrowtooth flounder just prior to the spawning sea- son. A comparison of the reproductive cycle of arrowtooth flounder (which is a determinate-spawn- ing benthic flatfish, dwelling in relatively cold north- ern waters) with that of the northern anchovy (an indeterminate-spawning pelagic roundfish, occupy- ing much warmer southern waters) is not without merit. It is important to note, however, that signifi- cant differences could occur in the rate in which these species cycle between reproductive stages. Hunter and Macewicz (1985b, p. 87) caution that “the dura- tion of postovulatory stages must be newly estimated for each species, and an assumption that the dura- tion of these stages in a new species is similar to the northern anchovy is highly speculative.” Perhaps further sampling of this arrowtooth population closer to or during the spawning season could have pro- vided more information on the further development of these “spent or resting” fish. Macroscopic classification was not successfully applied to the males. Only by histologic work was male maturity confidently assessed. Rickey (1995, p. 130), in her nearly year-round sampling of arrowtooth flounder, also had difficulty with assigning macro- scopic maturity stages to males, stating that males “. . . did not show grossly apparent developmental changes over time ... no spawning males were seen.” In the present study, females were classified as mature if they had oocytes as advanced as the corti- cal alveoli stage (Rickey, 1995), or showed atresia of previously vitellogenic oocytes. Significant differ- ences in GSI and HSI occurred between the late perinucleus and cortical alveoli stages, and fish at the cortical alveoli stage were also longer and heavier. The insignificant but noticeable decline in condition factor in the cortical alveoli and early vitellogenic stages also indicates an emphasis in gonad growth over somatic growth. Rickey (1995), in examining Washington coast arrowtooth flounder collected dur- ing the spawning season, found that all of the fish had either matured beyond the cortical alveoli stage or had not yet matured that far. This finding sup- ports the idea that fish in the cortical alveoli stage prior to the spawning season will mature during the same spawning season. Histology is regarded as the best method available to assess maturity (Hunter et al., 1992; West, 1990), but Hunter et al. (1992) con- cluded that even with the broadest criteria defining maturity, some spent fish are not identifiable as post- spawners when sampling is done after spawning has commenced. Hosie and Barss1 determined that Oregon arrow- tooth flounder males reach L50 at 29 cm and females reach L50 at 44 cm. Rickey (1995) determined that Washington males reach L50 at 28.0 cm and females reach L50 at 36.8 cm. Fargo et al. ( 1981) determined that British Columbia males reach L50 at 31 cm and females reach L50 at 37 cm. All the above studies determined maturity by using macroscopic classifi- cation of arrowtooth flounder gonads. It was thought that macroscopic observations of maturity would result in a lower L50, as all the other arrowtooth flounder maturity studies showed, but instead the L50 based on macroscopic observations was 3 cm higher. This finding is the opposite of that found in a study by Walsh and Bowering (1981) who compared macroscopic and histological staging of Greenland halibut ( Reinhardtius hippoglossoides) ovaries and demonstrated that L50 was 3 cm higher in the maturity ogive derived from histological work. Time of sampling in relation to the spawning sea- son may have been a factor in determining female L50. Hunter et al. (1992) showed that estimates of L50 for female Dover sole (Microstomus pacificus ) taken during the spawning season were higher than estimates of L5Qfor female Dover sole taken just prior to the spawning season, whereas Rickey ( 1995) found 610 Fishery Bulletin 95(3), 1997 the opposite for arrowtooth flounder; her highest L50 values were derived from fish taken prior to spawn- ing and her lowest value was derived from fish taken during the spawning season. The Oregon study oc- curred from September through June (Hosie and Barss1), the Washington study occurred nearly year- round (Rickey, 1995), and the British Columbia study occurred only in June (Fargo et al., 1981). Histological examination revealed that most ma- ture males in this study had only a small portion of their testes filled with spermatozoa; and thus they were not yet ready to spawn. It is likely that, as these large males continued to develop sexually during the season, other smaller males would have become sexu- ally mature, thus lowering the male L50. The male L50 value of 42.2 cm determined in this study should be viewed as a high estimate. Male GSI values started increasing at around 30 cm in length, and CF values began declining at 34.5 cm; both trends indicate a transition from somatic growth to gonad maturation at a much smaller size than that for the L50 reported here. In general, the largest females were the most ma- ture in this study, indicating that they might spawn the earliest. The high values of CF, GSI, and HSI for these largest, most mature females also show that these fish are best able to support the burden of spawning. The noticeable but insignificant drop in CF for females in the cortical alveoli and early vitel- logenesis stages (Fig. 2), at around 50-60 cm in length (Fig. 3A), if real, can be explained by two pos- sibilities. Either these mid-size fish are affected more by vitellogenesis than are the larger fish, or all fe- males suffer losses in CF in the early stages of vitel- logenesis and recover during later maturity stages. Gonadosomatic index was also highest in the largest males; thus they appear to mature earlier in the sea- son than smaller males. The largest, most mature males had decreasing CF values that indicated an impact of maturing testes on body composition. The spawning habits of arrowtooth flounder are not well known. Shuntov (1970) was unable to de- termine accurate spawning times for arrowtooth flounder in the eastern Bering Sea but nonetheless stated that they were close to those of Kamchatka flounder (Atheresthes evermanni), which were found in spawning condition in January and March. Fargo et al. (1981), using macroscopic observations of go- nads collected in June from Hecate Strait, concluded that spawning takes place prior to June, probably in spring months. Rickey (1995) showed that spawning occurred off the Washington coast from September through December, and possibly as late as February. Hosie and Barss1 reported a December-March spawning period for arrowtooth flounder off the Or- egon coast. Pertseva-Ostroumova (1961) reported arrowtooth flounder spawning in the Bering Sea from January through March. The results presented here, that a spawning season begins after September, are supported by all of the studies mentioned above. Total fecundity Both macroscopic and microscopic observations showed that this study was made prior to the spawn- ing season: none of the females were “ripe and run- ning,” had hydrated oocytes, or had postovulatory follicles, and none of the males were ready to spawn. Thus no bias due to loss of oocytes was expected. Total fecundity estimates for arrowtooth flounder had not been previously reported in the literature. The only other member of the genus, Kamchatka flounder, has an estimated fecundity range of 130,000-500,000 oocytes (Pertseva-Ostroumova, 1961), which is much lower than what is reported here for arrowtooth flounder. As with many other flat- fish species, arrowtooth flounder total fecundity in- creases linearly with fish weight and in a curvilin- ear fashion with length (Hempel, 1979). The largest arrowtooth flounder in this study had about 10 times as many oocytes as the smallest fish for which total fecundity was estimated. The unimodal frequency distribution of maturing oocyte diameters is sup- ported by Rickey’s (1995) determination that arrowtooth flounder is a group-synchronous spawner. The significant difference in weight between the eyed-side and blind-side lobes has not been previ- ously reported for arrowtooth flounder but has been reported for sole ( Solea solea, Witthames and Walker, 1995). Nichol2 found that blind-side lobes were sig- nificantly larger than eyed-side lobes in yellowfin sole ( Pleuronectes asper). This finding suggests that in flatfish species both ovarian lobes should be consid- ered when calculating GSI or total fecundity. Because there were no significant differences in mean oocyte diameter or mean oocyte density within or between ovarian lobes, total fecundity samples may be taken from any portion of the ovary. In his review paper, West ( 1990) mentioned that typically there is no dif- ference in oocyte size or diameter frequency distri- bution between ovarian lobes, but differences along the length of the ovary and in cross sections do occur in some species. Hunter et al. (1992) found no differ- ences in oocyte density between ovarian lobes, along the length of a lobe, or by cross section of a lobe in Dover sole. 2 Nichol, D. G. 1995. Resource Assessment and Conservation Engineering Div., Alaska Fish. Sci. Center, 7600 Sand Point Way NE, Seattle, WA 98115. Personal commun. Zimmermann. Maturity and fecundity of Atheresthes stomias 61 1 Although this study provides the first information on total fecundity of arrowtooth flounders, sub- sampling of ovarian tissue for total fecundity, histol- ogy of males, comparison of macroscopic and histo- logical methods, and relation of ancillary body mea- sures such as CF, GSI, and HSI, there is still much that is unknown about this species. The lag of male maturity in comparison to that of females in this study, and the lack of spawning males in Rickey’s (1995) study are parts of an interesting riddle. Pos- sible seasonal migrations proposed by Shuntov ( 1970) and Rickey ( 1995) need to be thoroughly documented, as well as possible spawning migrations. Differences in size at maturity found in this study, compared with those found in studies conducted in southern waters, need to be explored. A study, in which collections are made at preselected depths and in which age and histological data are collected at regular intervals throughout a year, would answer many questions. Acknowledgments I would like to thank D. Somerton for suggesting this project and M. Wilkins for giving me the time to com- plete it. S. Hinckley, S. McDermott, N. Merati, D. Nichol, and M. Rickey provided helpful discussions on sampling design, processing, and analysis. In ad- dition, I would like to thank L. Mooney and F. Morado for teaching me histological processing and for use of their laboratory. D. Benjamin, J. Haaga, D. King, R. Macintosh, D. Smith, and D. Somerton assisted in collecting the samples. Literature cited Allen , M. J., and G. B. Smith. 1988. Atlas and zoogeography of common fishes in the Bering Sea and northeastern Pacific. U.S. Dep. Commer., NOAATech. Rep. NMFS 66, 151 p. Fargo, J., L. A. Lapi, J. E. Richards, and M. Stocker. 1981. Turbot biomass survey of Hecate Strait, June 9-21, 1980. Can. Manuscr. Rep. Fish. Aquat. Sci. 1630, 84 p. Greene, D. H., and J. K. Babbitt. 1990. Control of muscle softening and protease-parasite interactions in arrowtooth flounder, Atheresthes stomias. J. Food Sci. 55(21:579-580. Hempel, G. 1979. Early life history of marine fish. Washington Sea Grant, Univ. Washington Press, 70 p. Hunter, J. R., and B. J. Macewicz. 1985a. Rates of atresia in the ovary of captive and wild northern anchovy, Engraulis mordax. Fish. Bull. 83(2): 119-136. 1985b. Measurement of spawning frequency in multiple spawning fishes. In R. Lasker (ed.l, An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax, p. 79-94. U.S. Dep. Commer., NOAATech. Rep. NMFS 36. Hunter, J. R., B. J. Macewicz, N. C. Lo, and C. A. Kimhrell. 1992. Fecundity, spawning, and maturity of female Dover sole Microstomus pad ficus, with an evaluation of assump- tions and precision. Fish. Bull. 90(11:101-128. Martin, M. H., and D. M. Clausen. 1995. Data report: 1993 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer., NOAATech. Memo. NMFS- AFSC-59, 217 p. Morrison, C. M. 1990. Histology of the Atlantic cod, Gadus morhua : an at- las, part three. Reproductive tract. Can Spec. Publ. Fish. Aquat. Sci. 100, 177 p. Pertseva-Ostroumova, T. A. 1961. The reproduction and development of far eastern flounders. Izdatel’stvo Akad. Nauk. SSSR, 483 p. [Transl. by Fish. Res. Board Can., 1967, Transl. Ser. 856, 1003 p.] Porter, R. W., B. J. Kouri, and G. Kudo. 1993. Inhibition of protease activity in muscle extracts and surimi from Pacific whiting, Merluccius productus , and arrowtooth flounder, Atheresthes stomias. Mar. Fish. Rev. 55( 3 ): 10—15. Reppond, K. D., D. H. Wasson, and J. K. Babbitt. 1993. Properties of gels produced from blends of arrowtooth flounder and Alaska pollock surimi. J. Aquat. Food Prod. Technol. 2( 1 ):83— 98. Rickey, M. H. 1995. Maturity, spawning, and seasonal movement of arrowtooth flounder, Atheresthes stomias, off Wash- ington. Fish. Bull. 93(1):127-138. Shuntov, V. P. 1970. Seasonal distribution of black and arrow-toothed halibuts in the Bering Sea. Tr. Vses. Nauchno-issled. Inst. Morsk. Rybn. Khoz. Okeanogr. 70 (Izv. Tikhookean. Nauchno-issled. Inst. Rybn. Khoz. Okeanogr. 721:391-401. [Transl. in Soviet Fisheries Investigations in the North- eastern Pacific, Part V, p. 397-408, by Israel Prog. Sci. Transl., 1972. Avail. Natl. Tech. Info. Serv., Springfield, VA, as TT71-50127.] Smith, R. L., A. J. Paul, and J. M. Paul. 1990. Seasonal changes in energy and the energy cost of spawning in Gulf of Alaska Pacific cod. J. Fish. Biol. 36:307-316. Wallace, R. A. 1985. Vitellogenesis and oocyte growth in nonmammalian vertebrates. In Developmental biology, Vol. 1, p. 127-177. Plenum Publ. Corporation. Walsh, S. J., and W. R. Bowering. 1981. Histological and visual observations on oogenesis and sexual maturity in Greenland halibut off northern Labrador. NAFO Sci. Counc. Stud. 1:71-75. Wasson, D. 11., K. D. Reppond, J. K. Babbitt, and J. S. French. 1992. Effects of additives on proteolytic and functional prop- erties of arrowtooth flounder surimi. J. Aquat. Food Prod. Technol. 1(3/4):147-165. West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar. Freshwater Res. 41:199-222. Witthames, P. R., and M. G. Walker. 1995. Determinacy of fecundity and oocyte atresia in sole (.Solea solea ) from the Channel, the North Sea and the Irish Sea. Aquat. Living Resour. 8:91-109. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Inc., Engle- wood Cliffs, NJ, 718 p. 612 The reproductive biology and early life stages of Podothecus sachi (Pisces: Agonidae) * Hiroyuki IVSur&ehara Usujiri Fisheries Laboratory, Faculty of Fisheries, Hokkaido University Minamikayabe-cho Usujiri 152, Hokkaido, 041-16, Japan E-mail address: hm@fl hines.hokudai.ac.jp Within the agonids, 20 genera and 50 species are recognized, with most distributed from the bottom of the north Pacific Ocean to the Bering Sea (Nelson, 1984). The agonid body is covered with bony plates and is unusual among teleost fishes. Several taxonomic studies have been conducted (Jordan and Evermann, 1898; Matsubara, 1955; Kanayama, 1991); however, little ecological information on the agonids exists because of their poor com- mercial value and small population density. Even for the sail-fin poacher, Podothecus sachi, the most common of the Japanese agonids, only larvae and juveniles have been reported from the adjacent waters of north- ern Japan (Maeda and Amaoka, 1988). Past reports concerning the reproductive ecology of agonids have suggested that they have in- ternal fertilization and that fe- males produce small clutches of eggs almost daily (lioka and Gunji, 1979; Sugimoto, 1987; Aoyama and Onodera, 1989). Recently, eggs of co- pulating cottids that were thought to undergo internal fertilization were shown to be fertilized by the received spermatozoa only after eggs were deposited (Munehara et al., 1989, 1991, 1994a, in press; Koya et al., 1993), i.e. there is an internal deposition of sperm that do not penetrate the ova or egg until after the latter are spawned and free in the environment. This spawning mode, named the inter- nal gametic association (Munehara et al., 1989), is characterized by the deposition of unfertilized eggs whose paternity has been fixed be- fore spawning (Munehara et al., 1990; 1994b). This spawning mode is so unique that it is not included in other categories of the parity mode of fishes, as determined on an evolutionary basis (Wourms et al., 1988). A comparative osteological and myological study on Scor- paeniformes indicated that the Agonidae family is most closely re- lated to the Cottidae family (Yabe, 1985). Therefore, it is possible that the internal gametic association mode of spawning that occurs in agonid fishes may provide informa- tion concerning the relation of the two families. In this study, I report on the reproductive biology and the early life stages of P. sachi, compar- ing them with those of other agonid fishes. Materials and methods Three adult females of P. sachi were collected with gill nets from off- shore bottom waters (60-80 m depth) at Usujiri, southern Hok- kaido, 7 October 1992. After the live fish had been transported to the laboratory, their ovaries were sur- gically removed, and ripe eggs and ovarian fluid were extracted. Great care was taken to prevent contami- nation by seawater, urine, and blood. The ovarian fluid was iso- lated with a pipette for use in the following test. To determine if egg development began before or after contact with seawater, eggs of each female were placed in separate petri dishes containing either ova- rian fluid or seawater at 6°C. After 20 hours, the number of develop- ing eggs were counted on the basis of occurrence of cleavage and for- mation of the blastodisc, which were regarded as signs of initial fertilization and autoactivation, respectively. Histological observations were carried out on several eggs before exposure to seawater to determine if internal fertilization occurred. Ovaries were examined to decide their developing mode. The ovaries were fixed in Bouin’s fluid. Serial paraffin sections, 5-8 |im, were pre- pared and stained with Delafield’s hematoxylin and eosin. The crite- ria of the maturing oocytes followed the classification of Yamamoto (1956). Eggs remaining from the above observations were used for morpho- logical observation of the early life stages of this fish. The eggs not used for artificial insemination were kept in a 1-L glass dish at a mean water temperature of 5°C. No bubbling stone was placed in the dish, but half of the seawater was replaced once a week. Juveniles were fed nauplii of Artemia salina. Measurement and observation of embryonic development were con- ducted once every 1-3 days. Sam- pling of hatched fish was done at intervals of 1-2 weeks until 93 days after hatching. Specimens were observed under a microscope after fixation in 5% neutral formalde- hyde solution. Terminology of the * Contribution 117 from Usujiri Fisheries Laboratory, Faculty of Fisheries, Hok- kaido University, Hokkaido, Japan. Manuscript accepted 19 March 1997. Fishery Bulletin 95:612-619 (1997). NOTE Munehara: Reproductive biology and early life stages of Rodothecus sachi 613 Figure 1 Retractable genital duct and mature ovary. (A) Genital duct (arrow) and eggs with ovarian fluid extracted from the top of the duct. (B) Ovulated eggs in the ovarian cavity and the genital duct. bony plates followed Gruchy (1969) and Maeda and Amaoka (1988). Resuits General anatomy and histology of the ovary The paired ovary of P. sachi was bilobed anteri- orly but fused together from the middle region; its posterior end reached be- yond the genital duct, which was located at the middle of the abdominal cavity (Fig 1). The anterior part of the genital duct was retractable and could be everted by pressing the fish’s belly. The protruded duct was tapered, about 2 cm length in 27.5-29.1 cm standard-length (SL) specimens (Fig. 1A). Blood vessels in the ovary ran radiately in the tunica of the dorsal side. The ova- rian cavity passed through the center of the ovary and then directly faced the ovarian wall lined with epi- thelia near the genital duct. The genital pore opened just behind the pelvic fins. The cavity and the genital duct contained several hun- dred ripe eggs (Fig. IB). Sections of the ovary contained oocytes in vari- ous developing stages, in- cluding the chromatin-nucleolus, perinucleolus, yolk- vesicle, oil-droplet, yolk-globule, migratory-nucleus, premature, and ripe stages (Fig. 2). In addition, postovulatory follicles in the stage just after ovula- tion or in the stage of regenerating were also found. These observations indicated that female P. sachi produce multiple clutches in a breeding season. Appearance of eggs Eggs were demersal, adhesive, and almost spherical in shape. The mean egg size was 1.73 mm in dia- meter, ranging from 1.70-1.75 mm (n= 30). The yolk was light pink, and many oil droplets and small, whitish, granular material were observed within the yolk. Initiation of egg development Although the eggs from the three females were not artificially inseminated, most eggs placed in seawa- ter had developed to the 2-cell stage after 20 hours (Table 1). In contrast, none of the eggs kept in the ovarian fluid showed any sign of development. When 614 Fishery Bulletin 95(3), 1997 Table 1 Embryonic development of Podothecus sachi eggs from 3 specimens 20 hours after immersion in seawater or ovarian fluid at 6°C. Rearing medium Percent (no.) of 2-cell stage eggs Percent (no.) of undeveloped eggs Percent (no.) of dead eggs Seawater 97.5 (118) 1.7 (2) 0.8 (1) Ovarian fluid 0 (0) 98.5 (133) 1.5 (2) 24 h after egg transfer from ovarian fluid to seawater 96.3 (130) 1.5 (2) 2.2 (3) Seawater 95.5 (106) 3.6 (4) 0.9 (1) Ovarian fluid 0 (0) 98.3 (117) 1.7 (2) 24 h after egg transfer from ovarian fluid to seawater 95.8 (114) 1.7 (2) 2.5 (3) Seawater 97.2 (141) 1.4 (2) 1.4 (2) Ovarian fluid 0 (0) 98.4 (124) 1.6 (2) 24 h after egg transfer from ovarian fluid to seawater 93.7 (118) 4.8 (6) 1.6 (2) such eggs were transferred into seawater, they de- veloped to the 2-cell stage within 24 hours. This find- ing indicates that egg development was initiated only after the eggs came in contact with seawater. It ap- peared that every female used in this study had copu- lated and that spermatozoa had already been trans- ferred into the ovarian cavity. Histological observation of gametes before exposure to seawater The micropyle of P. sachi eggs consisted of a micro- pylar vestibule, a funnellike depression about 100 pm across at the level of the egg surface, and a mi- cropylar canal penetrating the approximately 90-gm NOTE Munehara: Reproductive biology and early life stages of Podothecus sachi 615 c _ Figure 3 Photomicrograph of the eggs internally asso- ciated with spermatozoa. (A) Irregular par- ticles (arrows) on and near the vestibule, and micropylar canal (MC). EE = egg envelope, O = ooplasm. (B) High magnification of micro- pylar cone in A. Spermatozoa (SP) have en- tered into micropylar canal. (C) Chromosomes and metaphase spindle of second meiotic di- vision. Bars in A and B indicate 50 pm and bar in C indicates 10 pm. thick egg envelope (Fig. 3 A). The external opening of the canal was centrally located at the bottom of the vestibule. The canal was slightly tapered, and its inner opening was situated at the center of the micropylar cone. The external opening of the canal was about 5 pm in diameter. Many irregular par- ticles stained with hematoxylin were deposited on and near the vestibule (Fig. 3A). In eggs fixed before exposure to seawater, a num- ber of spermatozoa were found to have entered the micropylar cone (Fig. 3B), but membrane fusion had not yet occurred. Furthermore, in the region of the ooplasm near the micropylar cone, chromosomes and a metaphase spindle of the second meiotic division were detected (Fig. 3C). Embryonic development After developing to the 2-cell stage after 20 hours of immersion in seawater, eggs reached the 32-cell, morula, and blastula stages on the 1st, 5th, and 8th days, respectively (Fig. 4A). The embryo became vis- ible on the 13th day. The early embryo was smaller than the egg size, its length approximately 1/5 of the yolk’s circumference. A pair of optic vesicles and op- tic lenses appeared on the 16th and the 21st day, respectively. Myomeres began forming on the 22nd day (Fig. 4B). On the 29th day, the tail of the embryo began to grow free from the yolk. The heart was pul- sating and the embryo was moving occasionally on the 31st day. On the 35th day, a pair of otoliths was observed and the eyes began blackening. The em- bryo elongated to encircle the yolk completely by the 42nd day (Fig. 4C). A pair of pectoral fins began to extend at this time. Melanophores first appeared on the abdominal membrane on the 57th day. They be- gan forming on the side of the trunk on the 62nd day. On the 76th day, the intestine and the liver were differentiated, and blood vessels appeared along the yolk below the thoracic region of the embryo. Just before hatching, the embryo measured 1.5 times the yolk circumference, and some projections of supra- lateral and infralateral bony plates were observed (Fig. 4D). Hatching began on the 92nd day and ended by the 104th day. Larvae and juveniles Newly hatched larvae were 6.9-7. 1 mm in notochord length (NL) (Fig. 4E). Their bodies were slender, white, and pigmented on the head, trunk, and finfold. Yolks had not been completely absorbed yet, and an oil droplet remained in each yolk’s anterior part. The larvae were weak swimmers and usually lay on the bottom of the tank. Most hatched 101 days after fer- tilization. On the 3rd day of hatching, not only supralateral and infralateral bony plates, but dorsal and ventral bony plates began to form. The larvae occasionally fed on Artemia salina. The urinary blad- ders of larvae were always swollen with transparent liquid. On the 14th day, the yolk was completely ab- sorbed; a 7.4-mm-NL specimen had melanophores on the lateral sides of the abdomen and two pairs of fronto- 616 Fishery Bulletin 95(3), 1 997 Figure 4 Eggs, larvae and juveniles of Podothecus sachi kept at 5°C. (A) 2-cell stage, 20 hours after immersion in seawater. (B) Formation of optic lenses, auditory vesicles, Kupffer’s vesicle, and myomeres, 22 days after immersion in seawater. (C) Embryo almost in full circle, 42 days after immersion in seawater. (D) Just before hatching, 92 days after immersion in seawater. (E) Newly hatched larva 7.0 mm in notochord length (NL), 92 days after immer- sion in seawater. (F) 7.4-mm-NL larva, 14 days after hatching. (G) 7.7-mm-NL larva, 28 days after hatching. (H) Juvenile 8.7 mm in standard length (SL), 42 days after hatching. (I) Juvenile 15.6 mm in SL, 60 days after hatching. Dashed vertical line indicates the position of genital duct’s opening. parietal ridges with one spinule (Fig. 4F). The pecto- ral fins began to enlarge, and the larvae spent more time swimming near the surface than lying on the bottom. A 7.7-mm-NL specimen observed on the 28th day had 14 rays in its pectoral fins, melanophores on the ventral sides of the abdomen, and both jaws protruded slightly (Fig. 4G); four spines on preop- ercular and 2—4 spinules on each frontoparietal ridge were found. Another specimen measuring 8.3 mm NL on the 28th day was found to be in flexion; the larva NOTE Munehara: Reproductive biology and early life stages of Podothecus sachi 617 had 13, 16, and 15 rays in its dorsal, anal, and pecto- ral fins, respectively. On the 42nd day, a specimen measuring 8.7 mm SL had 10 spines and 13 rays on its dorsal fin and 15 and 16 rays on its anal and pec- toral fins, respectively, forming a full complement (Fig. 4H). The pelvic fins were still buds. The open- ing position of the anal and the genital pore began moving from just anterior to the anal fin toward the base of the pelvic fins. Whiskers, a mustache, and pelvic fins first appeared in a 15.6-mm-SL specimen on the 60th day (Fig. 41) and were completely formed in a 24.6-mm-SL specimen on the 93rd day. Juvenile P. sachi almost corresponded to adults in morphologi- cal features, but movement of the anal and genital pore openings were only half finished. The 24.6-mm speci- men swam, using its elongated pectoral fins, but never used its posterior trunk for propulsion owing to hardened bony plates. Discussion Internal gametic association and external fertilization As noted, P. sachi eggs placed in seawater and with- out artificial insemination began to develop and grow to juveniles, whereas eggs maintained in ovarian fluid showed no signs of development. In addition, histological observations of eggs directly obtained from the ovary showed that spermatozoa had entered the micropyle before contact with seawater, indicat- ing that fertilization was not initiated in the ova- rian fluid. This observation demonstrates that the spawning mode of P. sachi is of the internal gametic association type, reported in previous studies of copu- lating cottids (Munehara et al., 1989, 1991, 1994a; Koya et al., 1993). Atlantic wolffish, Anarichas lu- pus (Perciformes, Anarhichadidae), is also known to undergo copulation before egg laying, but the spawn- ing mode of this fish is not comparable to that of P sachi because inseminated eggs of Atlantic wolffish develop internally (Pavlov, 1994). This information seems to support Yabe’s hypothesis (1985), based on comparative osteological and myological observa- tions, that the family Agonidae may be the most closely related family to the Cottidae. Early life history Many larvae and juveniles, 8.3-25.1 mm SL, have been collected by plankton nets (Maeda and Amaoka, 1988). In the present study, the largest specimen took 93 days to grow to 24.6 mm SL after hatching. Thus, P. sachi probably inhabits the pelagic ocean during its first three months. It seems reasonable that whis- kers and mustache, which function as sensory or- gans for detection of benthic prey (Sato, 1977), are completed before juveniles become benthic. Reproductive style of agonid species Information on the reproduction of agonids is avail- able for only 6 of 50 species (Table 2). Many common characteristics are recognized among these species. First, the reproductive behavior of the Agonidae involves copulation. All the agonid species whose re- productive styles are known ( Agonomalus probo- scidalis, Occella iburia, and Bracdyopsis rostratus) have been described as internally fertilizing species on the basis of the development of eggs deposited without male involvement (Iioka and Gunji, 1979; Sugimoto, 1987; Aoyama and Onodera, 1989). These agonids probably exhibit external fertilization with internal insemination, as does P. sachi, because B. rostratus has been determined to be of the internal gametic association type from the same investiga- tions done for P. sachi, and information on internal fertilization of the agonids was proposed prior to the first recognition of the internal gametic association in teleost fishes (Munehara et al., 1989). A second characteristic of the reproductive style is that agonids have a long embryonic period, ranging from 100 days to 1 year. In addition to the Agonidae, such extraordinarily long embryonic periods in te- leost fishes are known for only a few trichodontids and cottid species (Okiyama, 1990; Munehara and Shimazaki, 1991). The third and fourth characteristics are egg depo- sition in concealed sites and lack of parental care. Naturally deposited egg masses of Agonus cata- phractus were collected from the roots of kelp ( Breder and Rosen, 1966). Spawning of captive Agonomalus mozinoi and A. proboscidalis was performed by con- cealing eggs in the exoskeletons of invertebrates or between rocks (Marliave, 1978; Aoyama and Onodera, 1989). Egg masses of B. rostratus were found deposited on the bottom of a tank, but the spawner was kept in a bare tank with no suitable substrate (Sugimoto, 1987). It is still unknown where P. sachi spawns its eggs. However, it is probable that the spawning of this fish involves brood hiding and lack of parental care because this species has a re- tractable genital duct and a long incubating period, similar to other copulating cottids that deposit their eggs into sponges, polychaete tube colonies, and nar- row fissures (Gomelyuk and Markevich, 1986; Munehara, 1991, 1992, 1996). Deposition of egg masses on such spawning substrates limits preda- tion on the eggs. In addition, flagellar movements of 618 Fishery Bulletin 95(3), 1997 Table 2 Comparison of reproductive characteristics in some agonid fishes. Species name Spawning mode Embryonic period and egg diameter Spawning substrate Parental care Spawning system References Agonomalus mozinoi unknown unknown 1 mm barnacle, tube worm2 without remarkable iteroparity of small clutches Marliave, 1978 A. probosciadalis internal fertlization' 110-114 days 2.05-2.30 mm between rocks and sand on bottom without remarkable iteroparity of small clutches Iioka and Gunji, 1979 Aoyama and Onodera, 1989 Aigonus cataphractus unknown 1 year 1.76-2.23 mm on roots of kelp unknown remarkable iteroparity of small clutches Eherenbaum, 1936 in Breder and Rosen, 1966 Bracdyopsis rostaratus internal gametic association' 287-324 days 2.1 mm on bottom2 without remarkable iteroparity of small clutches Sugimoto, 1987; this study Ocella iburia internal fertilization' unknown unknown unknown unknown unknown Sugimoto, 1987 Podothecus sachi ' internal gametic association' 92-104 days 1.70-1.75 mm unknown unknown remarkable iteroparity of small clutches this study 1 The studies had been reported before publication of internal gametic association (Munehara et al., 1989). 2 The findings were observed in aquaria. host invertebrates, for aspiration and filtration, in- cidentally supply oxygen to eggs deposited inside or on the invertebrates (Munehara, 1991). The pro- tracted period of incubation may have promoted egg deposition in any safe cradle for embryos rather than parental care. Multiple clutches are produced by agonid species during a breeding season, as demonstrated by histo- logical observation of the ovary of P. sachi; moreover, observations of A. mozinoi, A. proboscidalis, and B. ostratus indicate that females spawn small clutches almost daily in aquaria (Marliave, 1978; Iioka and Gunji, 1979; Sugimoto, 1987; Aoyama and Onodera, 1989). The fifth characteristic is the remarkable iteroparity of small clutches, which may have evolved i n association with the laying of eggs, both into narrow spaces and without male involvement. In summary it is suggested that five common char- acteristics, i.e copulation, a protracted period of in- cubation, concealment of deposited eggs, lack of pa- rental care, and remarkable iteroparity of the repro- ductive ecology of agonids have been closely corre- lated with each other through evolutionary construc- tion of their distinctive reproductive style. Copula- tion enabling impregnated females to spawn eggs without subsequent involvement of male fish seems to be a principal element of these characteristics. Acknowledgments The authors wish to thank the crew of the Yuki-maru and the staff of Usujiri Fisheries Laboratory, Hokkaido University, for collecting specimens. This study was supported in part by Grant-in-Aid for sci- entific Research (05760145) from the Ministry of Education, Science and Culture, and The Special Grant-in-Aid for Promotion of Education and Science at Hokkaido University. Literature cited Aoyama, S., and H. Onodera. 1989. Breeding habits, larvae, and juveniles of the agonid fish, Agonomalus proboscidalis in captivity. J. Jpn. Assoc. ofZoological garden and Aquarium 31:14—20. [In Japanese.] Breder, C. M., and C. E. Rosen. 1966. Modes of reproduction in fishes. The American Museum of Natural History, New York, NY, 941 p. Gomelyuk, V. E., and A. I. Markevich. 1986. On the strength of the egg case of the sea raven, Hemitripterus villosus (Cottidae). J. Ichthyol. 26:148-150. Gruchy, C. G. 1969. Canadian records of the warty poacher Occa verrucosa, with notes on the standardization of plate ter- minology in Agonidae. J. Fish. Res. Board Can. 26:1467- 1472. NOTE Munehara: Reproductive biology and early life stages of Podothecus sachi 619 Iioka, M., and Y. Gunji. 1979. Breeding of Agonomalus proboscidalis in the aquarium. J. Japan. Assoc, of Zoological Garden and Aquarium 21:21-25. [In Japanese.] Jordan, D. S., and B. W. Evermann. 1898. The fishes of North and Middle America: a descrip- tive catalogue of the species of fishlike vertebrates found in the waters of North America, north of the Isthmus of Panama. Part II. Bull. U.S. Nat. Mus. 47:1241-2183. Kanayama, T. 1991. Taxonomy and phylogeny of the family Agonidae (Pi- sces: Scorpaeniformes). Mem. Fac. Fish. Hokkaido Univ. 38:1-199. Koya, Y., K. Takano, and H. Takahashi. 1993. Ultrastructural observations on sperm penetration in the egg of elkhom sculpin, Alcichthys alcicornis, showing in- ternal gametic association. Zool. Sci. (Tokyo) 10:93-101. Maeda, K., and K. Amaoka. 1988. Taxoo+mic study on larvae and juveniles of agonid fishes in Japan. Mem. Fac. Fish. Hokkaido Univ. 35:47-124. Marliave, J. B. 1978. Spawning and yolk-sac larvae of the agonid fish spe- cies, Agonomalus mozinoi Wilimovsky and Wilson. Syesis 11:285-286. Matsubara, K. 1955. Fish morphology and hierarchy. Part II. Ishizaki- shoten, Tokyo, 1605 p. [In Japanese.] Munehara, H. 1991. Utilization and ecological benefits of a sponge as a spawning bed by the little dragon sculpin Blepsias cirrhosus. Jpn. J. Ichthyol. 38:179-184. 1992. Utilization of polychaete tubes as spawning substrate by the sea raven Hemitripterus villosus (Scorpaeniformes). Environ. Biol. Fish. 33:395-398. 1996. Sperm transfer during copulation in the marine sculpin Hemitripterus villosus (Pisces: Scorpaeniformes) by means of a retractable genital duct and ovarian secre- tion in females. Copeia 1996:452-454. Munehara, H., and K. Shimazaki. 1991. Embryonic development and newly hatched larvae of the little dragon sculpin Blepsias cirrhosus. Jpn. J. Ichthyol. 38:31-34. Munehara, H., Y. Koya, and K. Takano. 1991. The little dragon sculpin Blepsias cirrhosus, another case of internal gametic association and external fertilization. Jpn. J. Ichthyol. 37:391-394. 1994a. Conditions for initiation of fertilization of eggs in the copulating elkhom sculpin. J. Fish Biol. 45:1105-1111. Munehara, H., A. Takenaka, and O. Takenaka. 1994b. Alloparental care in the marine sculpin Alcichthys alcicornis (Pisces: Cottidae): copulating in conjunction with paternal care. J. Ethol. 12:115-120. Munehara, H., H. Okamoto, and K. Shimazaki. 1990. Paternity estimated by isozyme variation in the ma- rine sculpin Alcichthys alcicornis (Pisces: Cottidae) exhib- iting copulation and paternal care. J. Ethol. 8( 1 ):2 1— 24. Munehara, H., K. Takano, and Y. Koya. 1989. Internal gametic association and external fertiliza- tion in the elkhorn sculpin, Alcichthys alcicornis. Copeia 1989:673-678. Munehara, H., Y. Koya, Y. Hayakawa, and K. Takano. In press. Extracellular environments for the initiation of external fertilization and micropylar plug formation in a cottid species, Hemitripterus villosus (Pallas) (Scor- paeniformes) with internal insemination. J. Exp. Mar. Biol. Ecol. Nelson, J. S. 1984. Fishes of the world, 2nd ed. John Wiley and Sons, Inc., New York, NY, 475 p. Okiyama, M. 1990. Contrast in reproductive style between two species of sandfishes (family Trichodontidae). Fish. Bull. 88:543- 549. Pavlov, D. A. 1994. Fertilization in the wolffish, Anarhichas lupus : ex- ternal or internal? J. Ichthyol. 34:140-151. Sato, M. 1977. Histology of barbels of Blepsias cirrhosus draciscus (Cottidae). Jpn. J. Ichthyol. 23:220-224. Sugimoto, T. 1987. Embryonic development, larvae and juveniles of the longsnout poacher Bracdyopsis rostratus. Marine Snow 6:1-2. [In Japanese.] Wourms, J. P., B. D. Grove, and J. Lombaridi. 1988. The maternal-embryonic relationship in viviparous fishes. In B. W. S. Hoar and D. J. Randall (eds.), Fish physiology XI, p.1-134. Academic Press, Inc., London. Yabe, M. 1985. Comparative osteology and myology of the superfam- ily cottoidea (Pisces: Scorpaeniformes), and its phylogenetic classification. Mem. Fac. Fish. Hokkaido Univ. 32:1-130. Yamamoto, K. 1956. Studies on the formation of fish eggs. I, annual cycle in the development of ovarian eggs in the flounder, Liopsetta obscura. J. Fac. Sci. Hokkaido Univ. Ser. VI Zool. 12:362-373. 620 Age and growth of totoaba, Totoaba macdonaldi (Sciaenidae), in the upper Gulf of California Martha J. Roman Rodriguez Instituto del Medio Ambientey el Desarrollo Sustentable del Estado de Sonora (IMADES) Reyes y Aguascalientes esq. Col. San Benito. C.P 83190, Hermosillo, Sonora, Mexico M. Gregory Hammann* Centro de Investigacion Cientlfica y Educacion Superior de Ensenada (CICESE) Dept, of Ecology, Fisheries Ecology Group Km. 107 Carr. Tij.-Ens. Ensenada, Baja California, Mexico E-mail address: ghammann@cicese.mx The totoaba, Totoaba macdonaldi (Gilbert), also known as Mexican giant bass, is found only in the Gulf of California. This species during 1934-45 supported one of the most important sport and commercial fisheries in the Gulf, with total an- nual landings exceeding 2,000 met- ric tons (Rosales-Juarez and Ra- rmrez-Gonzalez* 1 ). At present, it is considered endangered (Flanagan and Hendrickson, 1976; NMFS2 ) as a result of 1) a high mortality of ju- veniles in shrimp trawl nets, 2) past overexploitation, 3) current illegal fishing during its reproductive sea- son (early February to early May), and 4) ecological alterations of its spawning and nursery grounds. Flanagan and Hendrickson (1976) suggested that there was a high probability that this species would become extinct by 2000 AD. In 1975 the Mexican government declared a moratorium on fishing totoaba. This paper reports on the age and growth of totoaba as determined from sectioned-otolith readings and contrasts current population age composition with what was known about the early population. Previ- ous studies have reported ages based on nonvalidated scale read- ings (Nakashima, 1916; Berdegue, 1955; Molina et al.3). Materials and methods Totoaba were sampled in 1986-91 from the northern part of the up- per Gulf of California between 31° and 32° N Lat. and 114° and 115° W Long. (Fig. 1). In 1989-91, juve- niles were collected from shrimp trawl nets. Adults were sampled with gill nets during their repro- ductive season (Feb-Apr) of 1986, 1987, and 1989-91. After determining individual standard (SL) and total length (TL) in millimeters and weight in grams, we extracted otolith (sagittal) pairs from 118 fish and embedded them in epoxy resin. For comparison with other age studies of totoaba, a lin- ear regression was performed for converting total length into stan- dard length. Lowerre-Barbieri et al. (1994) reported that sectioned otoliths were the best structure for ageing weakfish, Cynoscion regalis. To permit data recovery when only severed heads were available, 118 otoliths were weighed (OW; +/— 0.001 g) and measured to determine their relation with SL (Pauly, 1984); only whole individuals were used in the present study. Maximum otolith length (OL; +/- 0.05 mm) was measured from rostrum to postrostrum margins (anterior- posterior), and maximum otolith thickness (OT; +/- 0.05 mm) from the dorsal to ventral margins (dis- tal-proximal plane). A transverse section was made from 101 otoliths with an Isomet low speed saw following a tech- nique described by Beckman et al. (1990), Lowerre-Barbieri et al. (1994), and Secor et al.4 and the otolith ring counts were read to determine age. Each thin section was read three times with trans- mitted light in a bright field by the same person. Following the crite- ria of Beamish and Fournier (1981), we calculated an index of average percent error for the single reader. Three different axes (Fig. 2) were explored to measure the otolith ra- dius (OR) of 94 thin sections. An- nuli were most clearly counted and measured along axis 1; thus otolith radius (OR) was defined as the dis- tance from the center of the core to the otolith outer edge along the ven- * Author to whom correspondence should be sent. 1 Rosales-Juarez, F., and E. Ramirez- Gonzalez. 1987. Estado Actual del Conocimiento Sobre la Totoaba (Cynoscion macdonaldi), Gilbert 1890. Secretaria de Pesca, Mexico, 41 p. ISBN 968-817-086- 0. [In Spanish; available at CICESE library.] 2 NMFS (National Marine Fisheries Ser- vice). 1991. Endangered Species Act Status Review, totoaba ( Cynoscion mac- donaldi). Prot. Spec. Manage. Admin. Rep. SWR-91-01, 9 p. 3 Molina, D., M A. Cisneros-Mata, R. Urias, C. Cervantes y M. A. Marquez. 1988. Prospeccion y evaluacion de la totoaba ( Totoaba macdonaldi) en el Golfo de California. Tech. Report, 18 p. [In Span- ish, available from Cisneros-Mata, CRIP- Guaymas, Calle 20, No. 605 Sur, Guay- mas, Sonora, Mexico c.p. 85400.] 4 Secor, D. H., J. M. Dean, and E. H Laban. 1990. Manual for otolith re- moval and preparation for microstructural examination. Electric Power Research Inst., The Belle W. Baruch Inst, for Ma- rine Biology and Coastal Research, 85 p. Manuscript accepted 27 February 1997. Fishery Bulletin 95:620-628 ( 1997). NOTE Rom^n-Rodriguez and Hammann: Age and growth of Totoaba macdonaldi 621 Figure 1 Location of sampling sites in the northern upper Gulf of California. (▲) = adults; juve- niles were collected within the 40-m depth contour. tral arm of the sulcal groove. The relation between otolith radius and fish length was fitted by using the Gompertz function (Ricker, 1979) and the computer program Fishparm 3.0 (Prager et al., 1989). To de- termine size at ages that were not collected, we backcalculated past ages following Bagenal and Tesch (1978) and Jerald (1983), using the Gompertz relation between OR and SL, not the otolith length to fish length regression. A von Bertalanffy growth model (VBGM) was fit- ted to the observed SL at the midpoint of each age group represented in our sample (n=101) and also to the back-calculated sizes (n= 346) determined from 81 otolith thin sections, by using the computer pro- gram Fishparm 3.0 (Prager et al., 1989). The growth equation was calculated for pooled sexes because ju- veniles can be sexed only by using histological tech- niques; age and length differences between males and females have not been reported in the literature. Three juveniles captured in trawl nets during July and August 1989 were kept alive and transferred to the Centro Ecologico de Sonora Research Aquarium (Hermosillo, Sonora, Mexico); (see Almeida Paz et al. [1990] for more details). One fish died after almost 12 months of captivity; the other two were sacrificed 24 months after capture. Lengths and weights of these fish were taken every month beginning five months after capture. These lengths and the ring counts found in the otoliths of these fish were used to validate annual otolith ring deposition. 622 Fishery Bulletin 95(3), 1997 Table 1 Meristic relations between totoaba standard length (SL) and total length (TL), otolith length (OL), otolith thickness (OT), and otolith weight (OW). For otolith measurements, n = 118. Parameter Equation T 2 Fish total length (TL) SL = 0.91 TL - 26.34 (n=951) 0.99 Otolith length (OL) SL = 16.91 exp(4.7( l-exp(-0.96 x OL))) 0.99 Otolith thickness (OT) SL = 8.53 exp(5.4( l-exp(-0.17 x OT))) 0.98 Otolith weight (OW) SL = 788.8 x OW° 4518 0.99 Results From shrimp trawl nets, 1,125 juvenile totoaba ( 100- 600 mm SL, mean=223, SD=65 mm SL) and 157 adults were collected (600-1,850 mm SL, mean = 1,360, SD=89 mm SL). Figure 3 shows the length- frequency distribution for the specimens collected. Totoaba sagittal otoliths are large and beanlike, as are most sciaenid otoliths (Secor et al.4). Table 1 shows the linear equation describing the relation between SL and TL, the Gompertz relations between SL and otolith length (OL), otolith thickness (OT), and the allometric equation for the SL and otolith weight (OW) relation (Fig. 4). Otolith growth in weight changes in relation to fish growth in length and is an allometric, not isometric relation (6=2.176, Ho: b =3.00, student’s Gtest (0 025 m) =21.41). Transverse sections of the totoaba otolith (Fig. 5) typically showed clear, opaque, and translucent zones and when the otolith is sectioned exactly through the focus, the core appears as a dense opaque zone, next to which the first annulus is found. The distance be- tween the core and the first annulus is variable but is typically greater than increment sizes between the remaining annuli. The relation between otolith ra- dius (OR) along radius 1 and SL (Fig. 6) was fitted to a Gompertz function: SL = 30.92 x exp(3.86(l-exp(0.99 x OR))), [r2=0.98, n=94]. We found specimens representing 15 year classes between 0 and 24 years (Table 2). Of the 101 read- able otolith sections, 66 fish were young-of-the-year (110-377 mm SL). Ten juveniles were of age class 2 NOTE Rom^n-Rodriguez and Hammann: Age and growth of Totoaba macdonaldi 623 200 180 160 >N 140 o c 120 CD 3 100 c r ^ , 1“ C\J CO Standard length (mm) Figure 3 Length-frequency distribution for totoaba specimens collected in the Gulf of California, n = 1,282. Figure 4 Relation between otolith weight and totoaba standard length, n = 118. (378-620 mm SL), and one was of age class 3 (740 mm SL). The remaining specimens were adults between age classes 4 and 24. The index of average percent error was 16.10% for the single reader. Observed lengths and otolith ring counts were used to fit the von Bertalanffy model to obtain the growth curve for the totoaba population in the Gulf of California (Fig. 7); the fit was good (r2=0.98, n = 101). Past ages (n = 346) were backcalculated from the thin-section otolith ra- dius (OR) to fish length relation of 81 fish, and the von Bertalanffy growth model was also de- termined (Fig. 7). Table 2 compares the observed standard lengths with those calculated from both growth models. Figure 8 shows the relation be- tween maximum whole otolith length and age as an exponential function. In the case of the fish that died after a year of captivity ( 11 mo 21 d), its otoliths showed only one ring; the second fish held for two years, had two rings. The otoliths from the third specimen were decalcified and readings, un- fortunately, were not possible. Fish held in captivity were captured in the same trawls as the rest of the juveniles used for age deter- mination in this study. Otoliths of juveniles sac- rificed at the time of capture did not present any rings or marks similar to those detected in otoliths of the individuals kept in captivity. Discussion The relation between otolith dimensions and fish length can be used to obtain data from the totoaba heads commonly found on the beaches of the northern Gulf. This relation is particularly important when one consid- ers the restrictions and the potential impact of sampling an endangered species. Otolith growth and fish growth are proportional re- gardless of how growth rate changes with time; in early stages, both fish and otoliths increase faster than in adult stages after maturity is reached. Barbieri et al. (1994) reported for Atlantic croaker ( Micropogonias undulatus ) that age has an impor- tant effect on the otolith dimension to fish length relation. For totoaba, however, fish length alone de- scribed over 98% of the variability in otolith size. This growth pattern is common for other sciaenid species (Ross, 1988; Murphy and Taylor, 1989). The relation between standard length and otolith radius in sciaenids has been fitted to several growth equations. Maceina et al. (1987) and Blake and Blake (1981) found good fit with a linear model, but Barger (1985) found the best fit with a power function. For totoaba in our study, the best fit was found with a Gompertz model for radius along axis 1. The wide range of radii at the maximum fish length (shown in Fig. 6), supports our assumption that annuli are formed throughout the life of the fish, as is also suggested by the power function fitted to the otolith length and age relationship (shown in Fig. 8). Barbieri et al. (1994) also reported the formation of annuli throughout the life of the Atlantic croaker. Although the otolith length to age relation could be used to estimate fish age, the fit is not as good as that between standard length and 624 Fishery Bulletin 95(3), 1997 Sulcal groove Figure 5 Photograph of transverse thin section of an adult totoaba otolith showing 24 annual rings, sulcal groove, and core. Core 2 mm age. No “Lee’s phenomenon” (Ricker, 1969) was ob- served, and the von Bertalanffy growth model result- ing from back-calculated data was very similar to that derived from observed data (Fig. 7 ; Table 2). Gauldie and Nelson (1990) commented that otolith growth is typically repressed on the ventral plane because this part of the otolith is in direct contact with the skull, which restricts otolith growth; al- though otolith growth ceases in the ventral plane, it continues to grow in the sulcular re- gion. For totoaba, otolith growth seems to con- tinue on the proximal side along the sulcal groove. The ventral arm of the sulcal groove has been reported as the best area for otolith reading in other sciaenids (Beckman et al. , 1990; Lowerre-Barbieri et al., 1994). Major increases in length occur during the first and second years, diminishing once totoaba reach their sixth or seventh year, the age at first maturity. After this stage, the growth curve reaches an asymptote from about the twelfth to fourteenth year. The observed adult mean standard lengths at age in our study are very close to those calculated by VBGM from observed data and also from back-calculated data. In a comparison of age determinations with scales (Nakashima, 1916; Berdegue, 1955; NOTE Rom^n-Rodriguez and Hamm ann: Age and growth of Totoaba macdonaldi 625 Table 2 Observed and predicted mean standard length at year-class midpoints, n = sample size. SD = standard deviation. VBGM = von Bertalanffy growth model. BC = back-calculated model. Year class midpoint (yr) n Observed SL: mean (mm) Observed SD Predicted VBGM (mm) Predicted VBGM BC (mm) 0.5 66 218 66 216 319' 1.5 10 503 75 525 595 2.5 1 740 — 750 797 3.5 1 1,080 — 914 945 4.5 0 — — 1,034 1,053 5.5 0 — — 1,121 1,133 6.5 0 — — 1,184 1,192 7.5 1 1,260 — 1,231 1,234 8.5 0 — — 1,264 1,266 9.5 0 — — 1,289 1,289 10.5 1 1,271 — 1,307 1,306 11.5 0 — — 1,320 1,318 12.5 2 1,363 38 1,329 1,327 13.5 5 1,318 88 1,336 1,334 14.5 4 1,309 102 1,341 1,339 15.5 5 1,340 24 1,345 1,342 16.5 1 1,390 — 1,348 1,345 17.5 1 1,300 — 1,350 1,347 18.5 0 — — 1,351 1,348 19.5 1 1,280 — 1,352 1,349 20.5 0 — — 1,353 1,350 21.5 1 1,490 — 1,354 1,350 22.5 0 — — 1,354 1,351 23.5 0 — — 1,354 1,351 24.5 1 1,453 — 1,354 1,351 Total 101 1 Otolith core to margin measurements along ventral arm of sulcal groove were used for year-class 0, and predicted lengths were derived from the resulting equation. Flanagan, 1973; Molina et al.1), versus otoliths, as used in our study (Table 3), the greatest difference is found with Nakashima (1916). He estimated the maximum age for a fish of 1,980 mm (SL) to be nine years; our study shows that this fish could be older than 24 years. The von Bertalanffy parameters we estimated are very similar to those reported by Berdegue (1955) and Flanagan ( 1973). In the Molina et al.1 study there was a large underestimation of maximum age in comparison with our results, re- flected in the K value. This difference could be due to our use of a greater range of year classes from young-of-the-year to adults, whereas Molina et al.1 used only adult fish. Juveniles or young-of-the-year should be included to fit the von Bertalanffy growth curve because if only adults are used, there is a ten- dency to obtain low K values (Beckman et al., 1990). Scales often result in the underestimation of age (Beamish and McFarlane, 1987) owing to difficulty in reading, especially in the outer rings which are very close. Furthermore, there is a possibility of us- ing regenerated scales in which the first rings that were formed are not included. Authors working with sciaenids have mentioned that reading scales is easier for short-lived species, like some species of Cy noscion (Villamar, 1972; De Vries and Chittenden, 1982). On the basis of growth and otolith marks observed in two juveniles held in captivity, we suggest that ring formation in totoaba from the Gulf of California is annual; annual deposition patterns have been re- 626 Fishery Bulletin 95(3), 1 997 ported for other sciaenids (Beckman et al. 1990, Murphy and Taylor 1991), including species of Cynoscion, a closely related genus (Gonzalez, 1977; Blake and Blake, 1981; Shlossman and Chittenden, 1982; Barbieri et al., 1994; Lowerre-Barbieri et al., 1994). The well-defined seasonality in temperature in the Gulf of California (Alvarez Borrego et al. , 1973; Paden et al., 1991) also suggests that the marks seen in totoaba otoliths are annual because such tempera- ture changes are an important factor in ring deposi- tion (Brothers, 1978; Beckman et al., 1990). Berdegue (1955) considered scale rings to be annual on the basis of the migratory pattern and reproductive pe- riod of totoaba. The recent creation of the Upper Gulf of Califor- nia and Colorado River Delta Biosphere Reserve will enhance conservation efforts for totoaba by protect- ing important spawning and nursery habitat. Fur- thermore, fishing pressure from commercial shrimp trawls and gill nets will be greatly decreased. Barrera-Guevara (1990) reported that 92% of young-of-the-year totoabas were killed in the commercial shrimp fishery. In our study we were not able to sample organ- isms between ages 5 and 11 because they were not available to trawls and gill nets and because we did not sample in areas where prerecruit totoaba concentrate. These areas are difficult to sample because of their depth; the Guaymas basin reaches more than 200 m depth. It has been sug- gested that the summer migration of totoaba is toward deep waters in the cen- tral Gulf of California (Berdegue, 1955; Arvizu and Chavez, 1972; Flanagan and Hendrickson, 1976). These fish, however, are accessible to hook and line fishing, and “catch and release” sport fishing practices should be encouraged. It is clear that the current available habitat for totoaba will not allow signifi- cant population increase. Nevertheless, we found a population age-structure similar to that existing during the early 1890’s (as- suming that fish of ages 3-11 years exist but were unavailable to our sampling as previously described), and we suggest that continued conservation efforts should al- low for the survival of a stable but small population of totoaba in the Gulf of Cali- fornia. Estimates of adult survival pro- posed by Cisneros-Mata et al. (1995) be- fore and after the 1975 moratorium also support evidence of stability in the current population age-structure. Acknowledgments We thank the fishermen of the Gulf of Santa Clara and Puerto Penasco for their collaboration in sampling adult totoaba, especially Heriberto Amaya, Rosario Angulo, and their families. We thank per- sonnel of the Centro Regional de Investi- Years Figure 7 Von Bertalanffy growth model curve fitted to observed ( ) and back- calculated ( — ) data of age (number of rings) and standard length (the year-class midpoint was used for age). Open triangles are observed data. NOTE Rom^n-Rodriguez and Hammann: Age and growth of Totoaba macdonaldi 627 Table 3 Comparison of von Bertalanffy growth parameters for dif- ferent studies of totoaba. Author K *0 Max. age (yr) Present observed data 0.3162 1,355 -0.0499 24 Present back- calculated data 0.3103 1,352 -0.3679 24 Nakashima (1916) — 1,980 — 9 Berdegue (1955) — 1,330 — 15 Flanagan (1973) 0.16 1,467 — 20 Molina et al. 1 0.271 1,373 -2.264 19 1 See Footnote 1 in the text. gacion Pesquera en Guaymas for assistance in the field with sampling juveniles. The Fish and Wildlife Foundation is acknowledged for support of field work during 1989. Sampling was carried out under Fish- eries Research Permit No. 1229, Secretary of Fish- eries, Mexico. We acknowledge the valuable sugges- tions of two anonymous reviewers. Literature cited Almeida-Paz, M., G. Morales-Abril, and M. J. Roman Rodriguez. 1990. Aspectos sobre aclimatacion y crecimiento de juve- niles de totoaba, Totoaba macdonaldi , (Gilbert) (Pisces: Sciaenidae ) en condiciones de cautiverio. Ecologica ( Mexico ) 2( 1 ):7— 12. [In Spanish.] Alvarez-Borrego, S., L. A. Galindo-Bect, and B. P. Flores-Baez. 1973. Hidrologia. In Estudio Quimico sobre la contami- nation por insecticidas en la desembocadura del Rio Colo- rado. Reporte final de la segunda etapa, a la direction de acuacultura de la secretaria de recursos hidraulicos, p. 5- 177. Univ. Auton. Baja California, Ciencias Marinas, tomo I, sec. I. [In Spanish.] Arvizu, J., and H. Chavez. 1972. Sinopsis sobre la biologia de la totoaba, Cynoscion macdonaldi Gilbert 1890. [In Spanish.] FAO Fish. Synop- sis 108. Bagenal, T. B., and F. W. Tesch. 1978. Age and growth. In T.B. Bagenal (ed.), Methods for assessment offish production in freshwater, 3rd ed., p.101- 136. Blackwell Scientific Publication. Oxford, England. Barhieri, L. R., M. E. Chittenden Jr., and C. M. Jones. 1994. Age, growth, and mortality of Atlantic croaker, Micropogonias undulatus, in the Chesapeake Bay region, with a discussion of apparent geographic changes in popu- lation dynamics. Fish. Bull. 92:1-12. Barger, L. E. 1985. Age and growth of Atlantic croakers in the northern Gulf of Mexico, based on otolith sections. Trans. Am. Fish. Soc. 114:847-850. Barrera Guevara, J. C. 1990. The conservation of Totoaba macdonaldi (Gilbert), (Pisces: Sciaenidae), in the Gulf of California, Mexico. J. Fish. Biol. 37(suppl. A):201-202. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determination. Can. J. Fish. Aquat. Sci. 38:982-983. Beamish, R. J., and G. A. McFarlane. 1987. Current trends in age determination metho- dology. In R. C. Summerfield and G. E. Hall (eds.), Age and growth of fish, p.15-42. Iowa State Univ. Press. 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. Berdegue, A. J. 1955. La pesqueria de la totoaba (Cynoscion macdonaldi) en San Felipe, Baja California. Rev. Soc. Mex. Hist. Nat. 16( l-4):45-78. [In Spanish.] Blake, C., and B. F. Blake. 1981. Age determination in six species of fish from a Mexi- can Pacific coastal lagoon. J. Fish Biol. 18(4):471-478. Brothers, E. B. 1978. Exogenous factors and the formation of daily and sub- daily growth increments in fish otoliths. Am. Zool. 18:631 . Cisneros-Mata, M. A., G. Montemayor-Lopez, and M. J. Roman-Rodriguez. 1995. Life history and conservation of Totoaba macdo- naldi. Conserv. Biol. 9(4):806-814. De Vries, D. A., and M. E. Chittenden Jr. 1982. Spawning, age determination, longevity and mortal- ity of silver sea trout, Cynoscion nothus in the Gulf of Mexico. Fish. Bull. 80(3):487-500. Flanagan, C. A. 1973. Study of the feasibility of modeling the totoaba fish- ery of the northern Gulf of California, with preliminary estimation of some critical parameters. Univ. Arizona, Biol. Sci. Rep. 249, 58 p. Flanagan C. A., and J. R. Hendrickson. 1976. Observation on the commercial fishery and reproduc- tive biology of the totoaba Cynocion macdonaldi , in the northern Gulf of California. Fish. Bull. 74(3 ):53 1-544 . Gauldie, R. W., and D. G. A. Nelson. 1990. Otolith growth in fishes. Comp. Biochem. Physiol. 97A(2):119-135. Gonzalez, L. W. 1977. Aspectos tecnicos de preparation de otolitos para estudios de edad de algunas especies del genero Cynoscion (Pisces: Sciaenidae). Lagena nos. 39-40:43^8. [In Spanish.] Jerald, A., Jr. 1983. Age determination. In L. A. Nielsen and D. L. Johnson (eds.), Fisheries techniques, p. 301-324. Am. Fish. Soc., Bethesda, MD. Lowerre-Barbieri, S. K., M. E. Chittenden Jr., and C. M. Jones. 1994. A comparison of a validated otolith method to age weakfish, Cynoscion regalis , with the traditional scale method. Fish. Bull. 92:555-568. Maceina, M. J., D. N. Hata, T. L. Linton and A. M. Landry Jr. 1987. Age and growth analysis of spotted sea trout from Galveston Bay, Texas. Trans. Am. Fish. Soc. 116: 54-59. Murphy M. D., and R. G. Taylor. 1989. Reproduction and growth of black drum Pogonias cromis, in northeast Florida. Northeast Gulf Sci. 10(2): 127-137. 1991. Direct validation of ages determined for adult red 628 Fishery Bulletin 95(3), 1997 drums from otolith sections. Trans. Am. Fish. Soc. 120:267-269. Nakashima, F. 1916. Cy noscion macdonaldi, Gilbert. Copeia (37):85-86. Paden, C. A., M. R. Abbott, and C. D. Winant. 1991. Tidal and atmospheric forcing of the upper ocean in the Gulf of California: 1: Sea surface temperature varia- bility. J. Geoph. Res. 96:18,337-18,359. Pauly, D. 1984. Fish population dynamics in tropical waters: A manual for use with programmable calculators. ICLARM Studies and Reviews ( 8 ): 1—325. Prager, M. H., S. B. Saila, and C. W. Recksiek. 1989. FISHPARM: A microcomputer program for parameter estimation of nonlinear models in fisheries science, sec- ond ed. Old Dominion Univ. Oceanogr. Tech. Rep. 87-10. 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. 1979. Growth and rate models. Chapter 11 in W. S. Hoar, D. J. Randall and J. R. Brett (eds.), Fish physiology, vol. 8, p.301-324. New York Academic Press. Ross, S. W. 1988. Age, growth, and mortality of the banded drum, Lari- mus fasciatus (Sciaenidae) in North Carolina. Northeast Gulf Sci. 10(1): 19-31. Shlossman, P. A., and M. E. Chittenden Jr. 1982. Spawning, age determination, longevity and mortal- ity of the silver sea trout, Cynoscion nebulosus\ red drum, Sciaenops ocellata\ and the black drum, Pogonias cromis in south central Louisiana. La. Dep. Wild. Fish. Contrib. Mar. Res. Lab. Tech. Bull. 31:41-48 . Villamar, A. 1972. Age determination in fishes of the family Sciae- nidae. J. Ichthyol. 13(4):550-561. 629 Mark retention and growth of jet-injected juvenile marine fish John F. Thedinga* Adam Moles Jeffrey T. Fujioka Auke Bay Laboratory, Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 1 1305 Glacier Highway, Juneau, Alaska 99801-8626 *E-mail address: John.Thedinga@noaa.gov The type of mark or tag used for a particular study depends on the objectives of the study. Retention and recognition of a mark are criti- cal to the success and reliability of a study. External tags have been the most common fish tags used (McFarlane et al., 1990), but they may affect survival, behavior, and growth of fish (Andersen and Bagge, 1963; McFarlane and Beamish, 1990). Each tag type has limita- tions and capabilities, but ideally external tags should be easily and rapidly applied to many fish, be inexpensive, easily identified, not easily lost or entangled, and not be stressful enough to alter fish sur- vival, behavior, or growth. Studies that require such characteristics, therefore, are restricted in the type of tag that can be used and thus must rely on internal marks or dyes to identify fish. Identification of internal marks, however, generally requires that fish be killed, thus eliminating any possibility of re- peated measurements of individual fish. Choice of mark is further re- stricted when marking juveniles that are of small size and that ex- hibit rapid growth. Jet injection is a method of ap- plying external marks to fish (Hart and Pitcher, 1969) that is relatively fast and can apply a variety of col- ors for either batch or individual marking. Jet injection does not af- fect survival or growth of juvenile salmonids in the laboratory (Cane, 1981; Herbinger et al., 1990; The- dinga and Johnson, 1995) but may contribute to increased mortality in field situations (Thedinga et al., 1994). Injection by Panjet has been used primarily on freshwater fish species (Hart and Pitcher, 1969) in addition to salmonids (Cane, 1981; Pauley and Troutt, 1988; Laufle et al., 1990). Juvenile flatfish have been marked with needle-injected latex (Riley, 1966) as well as by freeze branding (Dando and Ling, 1980), but not by jet injection. To our knowledge, juvenile sablefish have been marked only with Floy anchor tags (Rutecki and Meyers, 1992). Our objectives were to evalu- ate retention of jet-injected marks and their effect on growth of four marine fish species, as well as their effect on the tissue structure of three marine species held in the laboratory. Methods We tested the retention of jet-in- jected marks and effects of marks on growth of four species of marine fish and histological effects on three species of marine flatfish. We in- jected two substances into juvenile sablefish, Anoplopoma fimbria, and one substance into juvenile yellow- fin sole, Pleuronectes asper, rock sole, Pleuronectes bilineatus, and Pacific halibut, Hippoglossus sten- olepis. Sablefish were captured by hand-jigging in St. John Baptist Bay near Sitka, Alaska, September 1993 (Rutecki and Meyers, 1992). Sole were captured by beach sein- ing in Auke Bay, Alaska, May to July 1994, and halibut were col- lected by trawling in Sitkinak Strait and Ugak Bay near Kodiak Island, Alaska, August 1994. A total of 28 sablefish and 30 flat- fish were injected with a Panjet in 1993-94. Sablefish were marked with alcian blue dye (65 mg/mL aqueous solution) and fluorescent orange acrylic paint (Liquitex, 50% aqueous solution); 10 of each flat- fish species were marked with alcian blue dye. All sablefish were marked on the abdomen between the pelvic fins (Fig. 1). Flatfish, however, were marked with indi- vidual identifying marks on the ventral surface at one to four loca- tions along the lateral margin and on the caudal peduncle (Fig. 2); 12 sablefish and 10 of each flatfish spe- cies were left unmarked as controls. The Panjet was held about 25 mm from the fish’s skin during marking. Sablefish were anesthe- tized with tricaine methanesul- fonate ( MS-222 ) before marking but flatfish were not. After marking, excess dye or paint was rinsed off with water to check mark quality. If a mark was good, the fish was put in a recovery tank; if poor, the fish was remarked. Sablefish and flatfish were held in different environments. After being marked, sablefish were held in 600-L flow-through tanks for 238 days. Because of space restrictions, blue-marked and control fish were held in one tank, but each group was kept in separate compart- ments. Orange-marked fish were held in another tank but died pre- maturely in a laboratory accident Manuscript accepted 7 January 1997. Fishery Bulletin 95:629-633 (1997) 630 Fishery Bulletin 95(3), 1997 Figure 1 Panjet-injected alcian blue dye mark between the pelvic fins of a juvenile sablefish. after 189 days and were frozen prior to analysis. Flat- fish were held for 90 d in six 70-L flow-through tanks on their preferred bottom type (mud substrate for soles [Moles and Norcross, 1995], sand substrate for halibut). For each flatfish species, control and marked fish were held in separate tanks. Sablefish were held indoors and were provided about 8 h of fluorescent light daily, whereas flatfish were held outdoors un- der an awning with 12 h of fluorescent light daily. Sablefish wei e fed ad libitum, and flatfish were fed 10% of their initial body weight per day throughout the study. The substrates in the flatfish tanks were initially frozen three days to kill meiofauna and macrofauna Mark recognition for sablefish was checked abou every 3 weeks, flatfish about every 4 weeks. Blue marks were viewed under fluorescent light, orange marks under fluorescent and ultravio- let (UV) light. Mark retention was rated subjectively as acceptable (retained) or unacceptable (not re- tained) by the same person each time the fish were checked. Fish lengths were recorded at the begin- ning and end of the study: fork length (FL) for sable- fish, total length (TL) for flatfish. Because we were obtaining additional data (histological) from flatfish, we also recorded flatfish weights when we recorded their lengths. Differences in acceptable mark retention were tested with a chi-square test, and differences in fish size and growth rate were tested with a btest. Absolute growth rate in length of flatfish was cal- culated as L = TL2- (7^/90X10), where L = absolute growth in length; TL2 = total length at 90 days; and TLl = total length at day 1 (beginning of the study). Instantaneous growth rate in weight of flatfish was calculated as W = (logeW2 - logeWj)/90, where W = instantaneous rate of increase in weight; W2 = weight at 90 days; and W j = weight at day 1 (beginning of the study). All flatfish were examined for histological changes. Fish were examined for gross pathology at 50x with a dissecting microscope. Gill and liver tissues were NOTE Thedinga et at: Mark retention and growth of jet-injected marine fish 631 Figure 2 Panjet-injected alcian blue dye mark on the ventral lateral margin of a yellowfin sole. excised and fixed in 10% buffered seawater. Tissues were then placed in 70% ethanol for two days, dehy- drated in a graded ethanol series, cleared in xylene, embedded in paraffin, and sectioned at 6 p. Sections were stained with hematoxylin and eosin and exam- ined for lesions and evidence of wound healing. Results For sablefish, mark retention varied with mark color and method of detection. Retention of marks was sig- nificantly higher (P<0.001) for alcian blue dye ( 100%) than for fluorescent orange acrylic paint when or- ange marks were viewed under fluorescent light but was similar (P=0.99) when orange marks were viewed under UV light (Fig. 3). Retention of orange marks viewed under fluorescent light was 100% af- ter 21 days but decreased to less than 20% after 84 days. Retention of orange marks viewed under UV light, however, decreased only to 92% after 84 days and remained at that level throughout the remain- der of the study. Mean length at the end of the study was similar (P=0.24) between blue-marked sablefish (280 mm) Table 1 Mean fork length of jet-injected and control juvenile sable- fish at time of injection and after 33 weeks. Fish were in- jected with alcian blue dye and fluorescent orange acrylic paint. Standard error is in parentheses. Mean fork length (mm) Initial After 33 weeks Blue 212(0.80) 280(1.07) Orange 212(0.96) 270(1.18) Control 218(1.16) 288(1.48) and controls (288 mm) (Table 1). Despite the poten- tial for tank-linked effect due to holding the marked fish in a separate tank, mean length of the orange- marked sablefish (270 mm) was not significantly dif- ferent from that of the controls (288 mm) (P=0.06) (Table 1). Mortality was zero. For all flatfish, mark retention was 100% through- out the study, and growth was similar between marked and control fish (Table 2). The instantaneous rate of increase in length and weight at the end of 632 Fishery Bulletin 95(3), 1997 the study was similar (P>0.10) for marked and con- trol fish, indicating that marking did not affect growth. Again, mortality was zero. Histological examination of flatfish indicated that the marks were nonirritating and nontoxic. Necropsy of the flatfish revealed no evidence of damage to skin or musculature and no alteration in the structure of dyed tissue. All liver hepatocytes were normal, indicating no toxic exposure. There was no evidence of increased macrophage aggregations in the liver or hyperplasia in the gills to suggest cellular responses to dyes. 120 irks (%) OO o o o HP HP ™ US) lLj i_j h n m '"m- — •- - 1 60 3 w ft 40 Blue H Orange (UV) — Orange O o < 20 0 '♦ i i i i ♦ i i t 0 40 80 120 160 200 240 Figure 3 Percentage of acceptable jet-injected marks on juvenile sablefish by color: alcian blue dye and fluorescent orange acrylic paint. Dashed line is orange marks viewed under ultraviolet light (UV), and dotted line is orange marks viewed under fluorescent light. Discussion Marine fish can be jet injected rapidly with many indi- vidual marks, often without anesthesia. For example, in this study we marked several nonanesthetized flat- fish per minute. Jet-injected alcian blue dye produced a highly visible intradermal mark that was retained for at least 8 months by juvenile sablefish, 3 months by juvenile flatfish. Detection of alcian blue dye marks on flatfish is easy and requires minimal handling. Usu- ally marks can be detected without anesthesia and without turning fish over. Increased visibility of jet-injected marks, however, could make fish more conspicuous to predators. Fluorescent orange acrylic paint marks, however, faded rapidly, mak- ing marks less visible to predators but necessi- tating the use of UV light for detection. Mark retention was similar to that reported for other species. Thedinga and Johnson (1995) re- ported 96% retention of alcian blue dye and fluo- rescent orange acrylic paint marks on the caudal fin of juvenile coho salmon, Oncorhynchus kisutch, and sockeye salmon, O. nerka, after nearly 4 months, and Herbinger et al. ( 1990 ) reported 96% retention of alcian blue dye-marked Atlantic salmon, Salmo salar, after 6 months. Few stud- ies have been published that used marked juve- nile sablefish or flatfish, and only one used a dye mark. Kelly (1967) injected Fast Blue 8GXM and hydrated chromium oxide by needle into the heads of juvenile winter flounder, Pleuronectes ameri- canus, and had 100% retention after 4 months. Jet-injected marks did not affect growth or mortality. Unlike Petersen disc and roll tags, which depressed growth rates (Andersen and Table 2 Mean initial total length (mm) and weight (g) and mean absolute growth rate in length and instantaneous growth rate in weight of marked (jet-injected with alcian blue dye) and control juvenile yellowfin sole, rock sole, and halibut 90 d after marking. Stan- dard error is in parentheses. Initial size Growth rate after 90 d Marked Control Marked Control Length Yellowfin sole 67.9(1.04) 74.2 (1.45) 0.198(0.067) 0.196(0.054) Rock sole 59.7 (0.92) 71.6 (1.56) 0.161 (0.068) 0.137 (0.073) Halibut 72.3 (0.81) 71.8 (0.88) 0.228 (0.059) 0.241 (0.050) Weight Yellowfin sole 3.5 (0.49) 5.2 (0.64) 0.823 (0.144) 0.841 (0.120) Rock sole 2.3 (0.32) 5.2 (0.68) 0.610 (0.171) 0.817(0.153) Halibut 4.1 (0.36) 4.1 (0.36) 1.193 (0.109) 1.175(0.111) NOTE Thedinga et al.: Mark retention and growth of jet-injected marine fish 633 Bagge, 1963) and resulted in increased mortality in sablefish (McFarlane and Beamish, 1990), jet-in- jected marks did not affect flatfish and sablefish growth or survival. Marks, however, may not be re- tained as long under natural conditions where growth is faster: sablefish in their natural habitat average 31-33 cm FL in spring (McFarlane and Beamish, 1983) in contrast with 28 cm FL recorded at the end of our laboratory study in spring. Jet-injected marks did not affect fish histology. There was no evidence of lesions in skin or muscula- ture and no alterations in either the cells or the struc- ture of dyed tissues. Changes in liver hepatocytes occur when fish have been exposed to toxicants (Hinton et al., 1992) but test hepatocytes in our preparations were normal. Jet-injection of either alcian blue dye or fluores- cent orange acrylic paint is a good method for mass marking or individually marking juvenile marine fish and meets most criteria for an effective external marker (Kelly, 1967). The marks are effectively re- tained and nontoxic and nonirritating; they do not affect mortality, can be used rapidly, and are inex- pensive, readily visible, and permit numerous dif- ferent mark combinations (Thedinga et al., 1994). Their application requires minimal training and equipment. Most importantly, jet injection does not alter the growth or tissue structure of fish. A limita- tion of jet-injection marking is the nonpermanent nature of the mark (Thedinga and Johnson, 1995). As a moderate-lasting marking method of juvenile marine fish, it is superior to most available external marking methods. Acknowledgments We thank Mike Murphy and Scott Johnson for re- viewing the manuscript and Mark Carls for photo- graphing jet-injected sablefish and yellowfin sole. Literature cited Andersen, K. P., and O. Bagge. 1963. The benefit of plaice transplantation as estimated by tagging experiments. In International Commission for the Northwest Atlantic Fisheries, North Atlantic fish mark- ing symposium, Woods Hole, MA, May 1961, p. 164- 171. Dartmouth, Nova Scotia, Spec. Publ. 4. Cane, A. 1981. Tests of some batch-marking techniques for rainbow trout ( Salmo gairdneri Richardson). Fish. Manage. 12( 1 ): 1 — 8. Dando, P. R., and R. Ling. 1980. Freeze-branding of flatfish: flounder, Platichthys flesus, and plaice, Pleuronectes platessa. J. Mar. Biol. Assoc. U.K. 60:741-748. Hart, P. J. B., and T. J. Pitcher. 1969. Field trials of fish marking using a jet inoculator. J. Fish Biol. 1:383-385. Herbinger, C. M., G. F. Newkirk, and S. T. Lanes. 1990. Individual marking of Atlantic salmon: evaluation of cold branding and jet injection of Alcian Blue in several fin locations. J. Fish Biol. 36:99-101. Hinton, D. E., P. C. Bauman, G. R. Gardner, W. E. Hawkins, J. D. Hendricks, R. A. Murchelano, and M. S. Okihiro. 1992. Histopathologic biomarkers. In R. J. Huggett, R. A. Kimerie, P. M. Mehrle, and H. L. Bergman (eds.), Biomarkers: biochemical, physiological and histological markers of anthropogenic stress, p. 155-209. Lewis Pub- lishers, Boca Raton, FL. Kelly, W. H. 1967. Marking freshwater and a marine fish by injected dyes. Trans. Am. Fish. Soc. 96:163-175. Laufle, J. C., L. Johnson, and C. L. Monk. 1990. Tattoo-ink marking method for batch-identification offish. Am. Fish. Soc. Symp. 7:38-41. McFarlane, G. A., and R. J. Beamish. 1983. Preliminary observations on the juvenile biology of sablefish ( Anoplopoma fimbria) in waters off the west coast of Canada. In Proceedings of the International Sablefish Symposium, March 29-31, 1983, Anchorage, AK, p. 119- 135. Alaska Sea Grant Rep. 83-8. 1990. Effect of an external tag on growth of sablefish ( Anoplopoma fimbria), and consequences to mortality and age at maturity. Can. J. Fish. Aquat. Sci. 47:1551-1557. McFarlane, G. A., R. S. Wydoski, and E. D. Prince. 1990. Historical review of the development of external tags and marks. Am. Fish. Soc. Symp. 7:9-29. Moles, A., and B. L. Norcross. 1995. Sediment preference in juvenile Pacific flat- fish. Neth. J. Sea Res. 34:177-182. Pauley, G. B., and D. A. Troutt. 1988. Comparison of three methods of fluorescent dye ap- plication for marking juvenile steelhead. Trans. Am. Fish. Soc. 117:311-313. Riley, J. D. 1966. Liquid latex marking technique for small fish. J. Cons. Perm. Int. Explor. Mer 30:354-357. Rutecki, T. L., and T. R. Meyers. 1992. Mortality of juvenile sablefish captured by hand-jig- ging and traps. N. Am. J. Fish. Manage. 12:836-837. Thedinga, J. F., and S. W. Johnson. 1995. Retention of jet-injected marks on juvenile coho and sockeye salmon. Trans. Am. Fish. Soc. 124:782-785. Thedinga, J. F., M. L. Murphy, S. W. Johnson, J. M. Lorenz, and K V. Koski. 1994. Determination of salmonid smolt yield with rotary- screw traps in the Situk River, Alaska, to predict effects of glacial flooding. N. Am. J. Fish. Manage. 14:837-851. 635 Erratum Erratum: Fishery Bulletin 95(1 ):1 1-24. Crockford, Susan J. Archeological evidence of large northern bluefin tuna, Thunnus thynnus, in coastal waters of British Columbia and northern Washington Correction: Susan Crockford has brought to our at- tention an error in the last column of data in her original Table 4 (page 18). The correct data for estimated FL (cm) should read as follows: Table 4 Archeological bluefin tuna vertebrae measurements and fork length estimates, by vertebrae number. All specimens. Measure- ments are defined in Figure 2. Vertebra no. Centrum GL (mm) Centrum GB(p) (mm) Estimated FL (cm) Vertebra no. Centrum GL (mm) Centrum GB(p) (mm) Estimated FL (cm) 01 27.0 44.4 198.7 21 38.3 45.1 189.6 02 26.3 48.2 191.5 22 39.2 44.8 188.6 04 22.5 39.6 164.7 22 39.2 46.5 193.5 04 31.6 65.1 224.9 24 39.4 47.5 195.3 05 27.1 58.7 212.4 24 45.1 209.8 06 30.8 58.9 218.2 25 40.8 48.3 194.8 09 25.6 34.4 168.2 26 40.1 47.6 190.7 (09) 48.7 220.8 (28) 30.2 35.9 159.3 (09) 22.3 30.5 153.6 29 33.7 36.3 155.3 (10) 33.4 45.0 206.7 29 45.3 49.5 201.1 11 24.4 34.6 165.5 29 50.3 58.2 221.1 (11) 31.0 37.8 177.5 30 28.1 32.0 138.5 (12) 35.0 200.5 30 36.7 167.1 (12) 36.7 46.5 206.9 30 38.3 44.0 172.3 (12) 34.9 41.9 192.0 30 46.7 199.2 14 33.5 187.0 30 47.3 50.0 201.1 (14) 33.9 40.9 185.4 30 47.5 51.2 201.7 (14) 34.1 42.0 189.3 31 48.6 52.8 201.3 (14) 32.5 42.1 189.7 31 52.7 53.8 215.3 (14) 27.7 32.8 157.8 32 41.7 45.6 179.3 15 35.4 42.5 188.4 32 43.7 185.6 (15) 31.4 42.5 188.4 32 49.7 54.4 204.4 16 35.9 45.6 191.2 33 26.6 27.0 122.4 16 36.1 42.9 191.9 33 43.8 44.8 178.9 (16) 31.3 35.4 175.0 33 44.5 47.2 186.5 (16) 33.8 40.9 183.9 33 44.7 45.6 181.4 (16) 40.2 47.2 206.0 33 47.9 53.9 207.3 (16) 41.2 51.8 209.4 33 52.2 53.7 206.7 17 36.0 42.8 187.3 33 53.7 55.2 211.3 17 37.5 47.3 200.2 34 48.1 202.4 (17) 37.0 43.9 190.4 34 48.2 202.6 (17) 38.2 44.2 191.3 35 41.2 42.0 200.8 18 37.2 46.5 195.4 36 30.5 198.9 (18) 26.8 32.4 153.5 38 11.9 28.2 210.3 19 39.4 47.2 196.0 38 16.5 31.8 243.2 (19) 27.2 32.7 151.9 39 23.0 198.3 20 35.5 44.5 182.4 39 19.5 174.3 (20) 38.0 45.0 191.3 (20) 44.3 54.0 213.4 636 Fishery Bulletin 95(3), 1 997 Superintendent of Documents Publications Order Form *5178 □yes, please send me the following publications: subscriptions to Fishery Bulletin for $34.00 per year (S42.50 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) Charge your order. It’s Easy! (Daytime phone including area code) (Purchase Order No.) Please Choose Method of Payment: I | Check Payable to the Superintendent of Documents ] GPO Deposit Account — j ] VISA or MasterCard Account 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! Fishery Bulletin Guide for Contributors Content Articles published in Fishery Bulletin de- scribe original research in fishery marine science, engineering and economics, and the environmental and ecological sciences, in- cluding modeling. Articles may range from relatively short to extensive; notes are re- ports of 5 to 10 pages without an abstract and describing methods or results not sup- ported by a large body of data. Although all contributions Eire subject to peer review, responsibility for the contents of papers rests upon the authors and not upon the editor or the publisher. It is therefore im- portEmt that the contents of the manuscript are carefully considered by the authors. Submission of an Eirticle is understood to imply that the article is original and is not being considered for publication elsewhere. Manuscripts should be written in English. Authors whose native language is not En- glish are strongly advised to have their manuscripts checked by English-speEiking colleagues prior to submission. Preparation Title page should include authors’ full names and mailing addresses, the corre- sponding author’s telephone, FAX number, and E-mail address, and a list of key words to describe the contents of the manuscript. Abstract should not exceed one double- spaced typed page. It should state the main scope of the research but emphasize its con- clusions Emd relevant findings. Because ab- stracts are circulated by abstracting agencies, it is important that they represent the research clearly and concisely. Text must be typed double-spaced through- out. A brief introduction should portray the broad significance of the paper; the remain- der of the paper should be divided into the following sections: Materials and meth- ods, Results, Discussion (or Conclusions), and Acknowledgments. Headings within each section must be, short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e. there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbrevia- tions, and technical terms should be spelled out the first time they are mentioned. The ☆ U S, Gov. Printing Office 1997 5890 17/400 10 scientific names of species must be written out the first time they are mentioned; sub- sequent mention of scientific names may be abbreviated. Follow the U.S. Government Printing Office Style Manual (1984 ed.) and the CBE Scientific Style and Format (6th ed.) for editorial style, Emd the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada for fish nomenclature. Dates should be written as follows: 11 November 1991. Measurements should be expressed in metric units, e.g., metric tons as (t); if other units of measure- ment are used, please make this fact explicit to the reader. The numeral one (1) should be typed as a one, not as a lower-case el (1). Use of appendices is discouraged. Text footnotes should be numbered with Arabic numerals. Footnote all personal communications, unpublished data, and un- published manuscripts with full address of the communicator or author, or, as in the case of unpublished data, where the data are on file. Authors are advised to avoid ref- erences to nonstandard (gray) literature, such as internal, project, processed, or ad- ministrative reports. Where these refer- ences Eire used, please include whether they are available from NTIS (National Techni- cal Information Service) or from some other public depository. Literature cited comprises published works and those accepted for publication in peer- reviewed literature (in press). Follow the name smd year system for citation format. In the text, cite as follows: Smith Emd Jones (1977) or (Smith smd Jones, 1977). If there is a sequence of citations, list by year. Smith, 1932; Smith and Jones, 1986; Smith and Allen, 1986. Abbreviations of serials should conform to abbreviations given in Serial Sources for the BIOSIS Previews Database. Authors Eire responsible for the accuracy and completeness of all citations. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be ex- plEiined in the table legend. Other inciden- tal comments may be footnoted with italic numerals. Use the asterisk only to indicate probability in statistical data. Because ta- bles Eire typeset, they need only be submit- ted typed and formatted, with double-spaced legends. Zeros should precede all decimal points for values less than one. Figures must be cited in numerical order in the text. The senior author’s last nsime and , the figure number should be written on the back of each one. Illustrations should be submitted as originals and not as photocop- ies. Submit photographs as glossy prints or slides that show good contrast, otherwise we cannot guarantee a good final printed copy. We may need to edit electronically generated graphics; therefore do not submit them as encapsulated postscript (EPS) files. Save them in the software program in which they were created. Label all figures with Helvetica typeface and capitalize the first letter of the first word in axis labels. Italicize species name and variables in equations. Use zeros before all decimal points. Use Times Roman bold typeface to label the parts of a figure. Send only photo- copies of figures to the Scientific Editor; original figures will be requested when the mEmuscript has been accepted for publica- tion. Each figure legend should explain all symbols and abbreviations in the figure and should be double-spaced and placed at the end of the manuscript. Copyright law does not cover government publications; they fall within the public do- main. If an author; reproduces any part of a government publication in his work, refer- ence to source is appreciated. Submission Send printed copies (original and three cop- ies) to the Scientific Editor: Dr. John B. Pearce, Scientific Editor Northeast Fisheries Science Ceiiter Nationsd Marine Fisheries Service 166 Water Street Woods Hole, MA 02643-1097 Once the manuscript has been accepted for publication, you will be Eisked to submit a softwEire copy of your manuscript to the Man- aging Editor. The software copy should be submitted in WordPerfect or Microsoft Word text format (or in stEmdsird ASCII text for- mat if neither progrEim is available) and placed on a 3.6-inch disk that is double- sided, double or high density, and that is compatible with either DOS or Apple Mac- intosh systems. A copy of page proofs will be sent to the au- thor for final approval prior to publication. Copies of published articles and notes are available free of charge to the senior author (50 copies) and to his or her laboratory (50 copies). Additional copies may be purchased in lots of 100 when the author receives page proofs. umber 4 October 1997 cgVTHSO/VM/v U.S. Department of Commerce William M. Daley Secretary National Oceanic and Atmospheric Administration D. James Baker Under Secretary for Oceans and Atmosphere National Marine Fisheries Service Rolland A. Schmitten Assistant Administrator for Fisheries Sr4T ES O* The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, Wash., and additional offices. POSTMASTER send address changes for subscriptions to Fishery Bulletin, Super- intendent of Documents, Attn: Chief, Mail List Branch, Mail Stop SSOM, Washington, DC 20402-9373. Although the contents have not been copyrighted and may be reprinted en- tirely, reference to source is appreci- ated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary in the transaction of the public business required by law of this Department. Use of funds for printing of this periodical has been ap- proved 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: $32.00 domestic and $40.00 foreign. Cost per single issue: $13.00 domestic and $16.25 foreign. See back page for order form. Scientific Editor Dr. John B. Pearce Editorial Assistant Kimberly T. Murray Northeast Fisheries Science Center National Marine Fisheries Service, NOAA 1 66 Water Street Woods Hole, Massachusetts 02543-1097 Editorial Committee Dr. Andrew E. Dizon National Marine Fisheries Service Dr. Harlyn O. Halvorson University of Massachusetts, Dartmouth Dr. Ronald W. Hardy University of Idaho, Hagerman Dr. Richard D. Methot National Marine Fisheries Service Dr. Theodore W. Pietsch University of Washington, Seattle Dr. Joseph E. Powers National Marine Fisheries Service Dr. Harald Rosenthal Universitat Kiel, Germany Dr. Fredric M. Serchuk National Marine Fisheries Service Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C 1 5700 Seattle, Washington 98115-0070 The Fishery Bulletin carries original research reports and technical notes on investiga- tions in fishery science, engineering, and economics. The Bulletin of the United States Fish Commission was begun 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. Begin- ning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin.. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 95 Number 4 October 1997 Fishery Bulletin Contents The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprietary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or proprietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the adver- tised product to be used or purchased because of this NMFS publication. Articles 637 Barbieri, Luiz R., Mark E. Chittenden Jr., and Cynthia M. Jones Yield-per-recruit analysis and management strategies for Atlantic croaker, Micropogonias undulatus , in the Middle Atlantic Bight 646 Bradbury, Carole, John M. Green, and Michael Bruce-Lockhart Daily and seasonal activity patterns of female cunner, Tautogolabrus adspersus (Labridae), in Newfoundland 653 Broadhurst, Matt K., and Steven J. Kennelly The composite square-mesh panel: a modification to codends for reducing unwanted bycatch and increasing catches of prawns throughout the New South Wales oceanic prawn-trawl fishery 665 Buckel, Jeffrey A., and David O. Conover Movements, feeding periods, and daily ration of piscivorous young-of-the-year bluefish, Pomatomus saltatrix, in the Hudson River estuary 680 Craig, Peter C., J. Howard Choat, Lynda M. Axe, and Suesan Saucerman Population biology and harvest of the coral reef surgeonfish Acanthus lineatus in American Samoa 694 DeVries, Douglas A., and Churchill B. Grimes Spatial and temporal variation in age and growth of king mackerel, Scomberomorus cavalla, 1 977-1992 709 Donohoe, Christopher J. Age, growth, distribution, and food habits of recently settled white seabass, Atractoscion nobilis, off San Diego County, California Fishery Bulletin 95(4), 1 997 ii 722 Gaughan, Daniel J., and Ian C. Potter Analysis of diet and feeding strategies within an assemblage of estuarine larval fish and an objective assessment of dietary niche overlap 732 Harris, Patrick J., and John C. McGovern Changes in the life history of red porgy, Pagrus pagrus, from the southeastern United States, 1972-1994 748 Johnson, Scott W., Mark G. Carls, Robert R Stone, Christine C. Brodersen, and Stanley D. Rice Reproductive success of Pacific herring, Clupea pallasi, in Prince William Sound, Alaska, six years after the Exxon Valdez spill 762 Myers, Ransom A., Gordon Mertz, and P. Stacey Fowlow Maximum population growth rates and recovery times for Atlantic cod, Gadus morhua 773 Mixon, Stephen W., and Cynthia M. Jones Age and growth of larval and juvenile Atlantic croaker, Micropogonias undulatus, from the Middle Atlantic Bight and estuarine waters of Virginia 785 Rotunno, Teresa, and Robert K. Cowen. Temporal and spatial spawning patterns of the Atlantic butterfish, Peprilus triacanthus, in the South and Middle Atlantic Bights 800 Smith, Peter J., Peter G. Benson, and S. Margaret McVeagh A comparison of three genetic methods used for stock discrimination of orange roughy, Hoplostethus atlanticus: allozymes, mitochondrial DNA, and random amplified polymorphic DNA 812 Unwin, Martin J. Survival of Chinook salmon, Oncorhynchus tshawytscha, from a spawning tributary of the Rakaia River, New Zealand, in relation to spring and summer mainstem flows 826 Weinrich, Mason, Malcolm Martin, Rachel Griffiths, Jennifer Bove, and Mark Schilling A shift in distribution of humpback whales, Megaptera novaeanghae, in response to prey in the southern Gulf of Maine 837 Zhao, Boxian, and John C. McGovern Temporal variation in sexual maturity and gear-specific sex ratio of the vermilion snapper, Rhomboplites aurorubens, in the South Atlantic Bight Notes 849 Oialoupka, Milani, and George R. Zug A polyphasic growth function for the endangered Kemp's ridley sea turtle, Lepidochelys kempii 857 Grama-Raffucci, Felix A., and Richard S. Appeldoorn Age determination of larval strombid gastropods by means of growth increment counts in statoliths 863 Murphy, Michael D. Bias in Chapman-Robson and least-squares estimators of mortality rates for steady-state populations 869 Ramon, Darlene, and fslorm Bartoo The effects of formalin and freezing on ovaries of albacore, Thunnus alalunga 873 Wallin, Julie E., John M. Ransier, Sondra Fox, and Robert H. McMichael Jr. Short-term retention of coded wire and internal anchor tags in juvenile common snook, Centropomus undecimal is 637 Abstract.-The effect of different fishing mortality (F) and natural mor- tality (M), and age at first capture (tc) on yield-per- recruit of Atlantic croaker, Micropogonias undulatus, in the lower Chesapeake Bay and North Carolina were evaluated with the Beverton-Holt model. Independent of the level of M (0.20-0.35) or F (0.01-2.0) used in simulations, yield-per-recruit values for Chesapeake Bay were consistently higher at tc = 1 and decreased continu- ously with increases in tc (2-5). Al- though maximum yield-per-recruit al- ways occurred at the maximum level ofF(F=2.0), marginal increases in yield beyond F = 0.50-0.75 were negligible. Current F ( FCUR ) is estimated to be be- low the level that produces maximum potential yield-per-recruit ( FMAX ) and at or below the level of F0 x if M > 0.25. Although modeling results indicated yield-per-recruit could be maximized by reducing the current level of tc (tc= 2), the resultant gains were small and did not appear to justify such management measures. Instead, it is suggested that regulatory measures be directed at maintaining the current level of tc in the lower Chesapeake Bay. Simulation results for North Carolina showed a pattern opposite to that shown for Chesapeake Bay, with yield-per-recruit curves increasing consistently with in- creases in tc. Estimates of FCUR for tc = 1 were consistently higher than F0 l as well as Fmax, indicating that during the period 1979-81 Atlantic croaker were being growth-overfished in North Caro- lina. However, differences between Chesapeake Bay and North Carolina seem to reflect temporal rather than spatial differences in Atlantic croaker population dynamics, because data for North Carolina came from a period co- inciding with the occurrence of unusu- ally large Atlantic croaker along the east coast of the United States. Manuscript accepted 11 March 1997. Fishery Bulletin 95:637-645 (1997). Yield-per-recruit analysis and management strategies for Atlantic croaker, Micropogonias undulatus, in the Middle Atlantic Bight* Luiz R. Barbieri** Mark E. Chittenden Jr. Virginia Institute of Marine Science School of Marine Science The College of William and Mary Gloucester Point, Virginia 23062 **Present address: University of Georgia Marine Institute Sapelo Island, Georgia 31327 E-mail address: Barbieris@msn.com Cynthia M. Jones Applied Marine Research Laboratory Old Dominion University Norfolk, Virginia 23529 The Atlantic croaker, Micropogonias undulatus (Linnaeus), is one of the most important commercial and rec- reational fishery resources of the southeastern coast of the United States (Wilk, 1981; Schmied and Burgess, 1987; Mercer* 1 ). Along the Atlantic coast, commercial fisheries for Atlantic croaker are centered in Chesapeake Bay and in North Caro- lina waters (Joseph, 1972; Roths- child et al., 1981; Ross, 1988; Mer- cer1); both inshore and offshore catches are distributed according to the seasonal migratory patterns of Atlantic croaker. From late spring to early fall Atlantic croaker are caught in estuarine areas, primarily by haul-seine, pound-net, and gill- net fisheries (Ross, 1988; Chitten- den, 1991; Barbieri et al., 1994a). From late fall through winter, after adults have moved out of estuaries, they are caught in continental shelf waters by otter-trawl and gill-net fisheries (Wilk, 1981; Ross, 1988; Mercer1). Commercial landings of Atlantic croaker have fluctuated widely over the past 50-60 years (Joseph, 1972; Rothschild et al., 1981; Wilk, 1981). Landings exceeded 20,000 metric tons (t) between 1937 and 1940, peaked at ca. 29,000 t in 1945 and dropped to less than 1,000 1 between 1967 and 1971 (Wilk, 1981; McHugh and Conover, 1986). The most re- cent peak in landings occurred in 1977 and 1978 at just over 13,000 t annually (Mercer1). Recreational catches in the mid-Atlantic and South Atlantic regions during 1979- 93 have also fluctuated, although they do not reflect fluctuations in commercial landings for the same period. Commercial landings from Virginia and North Carolina— the * Contribution 2057 from Virginia Institute of Marine Science, School of Marine Sci- ence, College of William and Mary, Glouce- ster Point, Virginia 23062. 1 Mercer, L. P. 1987. Fishery manage- ment plan for Atlantic croaker ( Micro- pogonias undulatus). North Carolina Dep. Nat. Res. Comm. Dev., Div. Mar. Fish., Spec. Sci. Rep. 48, 90 p. [Available from North Carolina Department of Envi- ronment, Health, and Natural Resources, Div. Marine Fisheries, PO Box 769, Morehead City, NC 28557-0769.] 638 Fishery Bulletin 95(4), 1997 two states with 98% of the Atlantic catch — have de- clined since 1987, whereas recreational catches peaked in 1991 with an estimated 21 million fish (Newlin, 1992; Speir et al., 1994). A lack of accurate catch and effort data from both the commercial and recreational fisheries makes it difficult to evaluate to what extent these long-term fluctuations represent natural changes in population abundance or reflect historic changes in Atlantic croaker exploitation. There has been a growing con- cern, however, that recent low landings may be re- lated to the large numbers of young fish killed as bycatch in the southern shrimp fishery and as part of the scrap catch in pound-net, haul-seine, and trawl fisheries (Speir et al., 1994; Mercer1). In response to these concerns, the 1993 review of the Atlantic States Marine Fisheries Commission Fishery Management Plan for Atlantic croaker (Speir et al., 1994) has rec- ommended the use of bycatch reduction devices and the establishment of a coast- wide minimum size limit that would maximize Atlantic croaker yield-per- recruit. Yield-per-recruit models, widely used in fish popu- lation dynamics studies (Beverton and Holt, 1957; Ricker, 1975; Gulland, 1983), can be a useful tool in defining routine fisheries management measures such as minimum size limits, closed seasons, etc. (Gulland, 1983; Deriso, 1987). However, the only published application of yield-per-recruit models to Atlantic croaker is based on data from the northwest- ern Gulf of Mexico (Chittenden, 1977) and points out that results may or may not apply to other areas. In this paper we use stock assessment data from the Chesapeake Bay (years 1988-91; Barbieri et al., 1994a) and from North Carolina (years 1979-81; Ross, 1988) to evaluate the effect of different fishing (-induced) and natural mortality, and age-at-first- capture schedules on Atlantic croaker yield-per-re- cruit. Implications of this analysis for management of Atlantic croaker are discussed. Methods Yield-per-recruit analysis Yield-per-recruit curves were calculated with the Beverton-Holt yield-per-recruit model (Beverton and Holt, 1957): 3 tj -nK(tc-t0) Y/R = Fe~M(tc~tr)W00 > (1) ^ F + M + nK n= 0 where Y/R = yield-per-recruit in weight (g); F = instantaneous fishing mortality coefficient; M = instantaneous natural mortality coefficient; W ^ = asymptotic weight (von Bertalanffy growth parameter); Un = summation parameter (U0=l, U^-3, U2=3, u3=- D; tc = mean age at first capture; tr = mean age (years) at recruitment to the fish- ing area; tQ = hypothetical age at which fish would have been zero length (von Bertalanffy growth pa- rameter); and K = the Brody growth coefficient (von Berta- lanffy growth parameter). Computations were performed with the computer program B-H3 available in the Basic Fisheries Sci- ence Programs package (Saila et al. , 1988). Parameter values used in simulations are summa- rized in Table 1. Estimates of growth parameters ( K, and fQ) for Chesapeake Bay and North Carolina were obtained from Barbieri et al.( 1994a) and Ross ( 1988), respectively. For both areas, was converted from by using an allometric length-weight rela- tion (6=3.23; Ross, 1988; and 6=3.30; Barbieri et al, 1994a). One of the assumptions of the Beverton-Holt yield-per-recruit model is that growth is isometric — i.e. the coefficient 6 in the length-weight relation is equal to 3 (Beverton and Holt, 1957; Ricker, 1975). We, however, considered that departure from the as- sumption of isometric growth did not affect interpre- tation of our modeling results because the factor of interest in these simulations is the relative differ- ence in yield resulting from varying tc and F at dif- ferent levels of M. The relative error in such differ- ences, when using an incorrect 6, tends to be much less than that in absolute levels (Ricker, 1975). Estimates of tr, the mean age at recruitment to the fishing area, were based on Atlantic croaker life his- tory information (Chao and Musick, 1977; Ross, 1988). Estimates of current tc, the mean age at first capture, was based on Atlantic croaker age composi- tions reported for the pound-net, haul-seine, and gill-net catches in the lower Chesapeake Bay for the period 1988-91 (tc= age 2; Barbieri et al., 1994a) and from age compositions reported for the haul-seine fishery in North Carolina for the period 1979-81 (tc= age 1; Ross, 1988). Because of the uncertainty associated with estimates of M in fish populations (Vetter, 1988), simulations for both areas were con- ducted over a range of M values (0.20-0.35; Table 1). The instantaneous total annual mortality rate, Z, for fully recruited Atlantic croaker in North Caro- lina is 1.3 (Ross, 1988) and ranges from 0.55 to 0.63, with a mean value of 0.59 for the lower Chesapeake Barbieri et al.: Yield-per-recruit analysis for Micropogonias undulatus 639 Table 1 Parameter estimates or range of values used in yield-per- recruit simulations for Atlantic croaker, Micropogonias undulatus, in the lower Chesapeake Bay (period 1988-91) and North Carolina (period 1979-81). See Equation 1 for definitions of parameter variables. Parameter Chesapeake Bay North Carolina K 0.36 0.20 409.9 g 3,814 g *0 -3.26 yr -0.60 yr tr 0 yr 0 yr tc 1-5 yr 1-5 yr F 0.01-2.0 0.01-2.0 M 0.20-0.35 0.20-0.35 Bay (Barbieri et al., 1994a). To estimate current lev- els of fishing mortality (FCUR) for different values of M, we used Z = 0.60 for Chesapeake Bay and Z = 1.3 for North Carolina, as F cur = Z - Mi , (2) where i - 0.20, 0.25, 0.30, and 0.35. The value of F0 x (the level of F for which the mar- ginal increase in yield-per-recruit due to a small in- crease in F is 10% of the marginal yield-per-recruit in a lightly-exploited fishery [Gulland and Boerema, 1973; Anthony2 ]), was estimated for Chesapeake Bay with F = 0.01 and tc = 2 (Barbieri et al., 1994a) and for North Carolina with F = 0.01 and tc = 1 (Ross, 1988). Cohort biomass and harvesting time In general, the maximum possible yield for a given year class occurs at the critical age tCRITIC, the age where biomass of a cohort is maximum in the ab- sence of fishing. For comparison with the Beverton- Holt yield-per-recruit modeling results, we estimated t critic f°r Atlantic croaker following Alverson and Carney (1975) and Deriso (1987) as tCRITlC =t0 + — ln(3if/M+ 1), (3) where t0, K, and M are defined as in Equation 1. Parameter estimates or the range of values used in calculations are listed in Table 1. 2 Anthony, V. 1982. The calculation of F0 p a plea for standard- ization. Northwest Atlantic Fisheries Organization, SCR Doc. 82/VI/64 Ser. No. N557, 15 p. NAFO, PO Box 638, Dartmouth, Nova Scotia, Canada B2Y 3Y9. To evaluate the proportion of the potential growth span (P ) remaining when Atlantic croaker enter the exploited phase of life (Beverton and Holt, 1957), we used the quantity (Beverton, 1963): Pg = (l-lc/LJ, (4) where Lm, the asymptotic length, was obtained from Barbieri et al. (1994a) and Ross (1988) and lc, the average length at first capture, was obtained by con- verting t to length with the von Bertalanffy growth curve reported for Atlantic croaker in Chesapeake Bay (Barbieri et al., 1994a) and North Carolina (Ross, 1988). Both parameters are based on total length (TL) in mm. Results Chesapeake Bay Curves of yield-per-recruit on F (Fig. 1) showed that the yield of Atlantic croaker in Chesapeake Bay could be maximized by decreasing the current level of tc = 2 (265 mm TL) to tc = 1 (245 mm TL). Independent of the level of M or F used in simulations, yield-per- recruit values were consistently higher at t = 1 and decreased continuously with increasing tc. However, increases in yield from t = 2 to t = 1 were generally small and gradually increased with increases in M. For example, at the estimated current levels of fish- ing mortality for Atlantic croaker in the Chesapeake Bay (Fcur), increases in yield between tc = 2 and tc = 1 would be 7.1% at M = 2.0, 12.6% at M = 0.25, 18.4% at M = 0.30, and 24.6% at AT = 0.35. The curves of yield-per-recruit for Atlantic croaker on F for different levels of M and tc showed no clearly defined peaks. Although the magnitude of yield curves was dependent on the level of M used in simu- lations, relative changes in yield as a function of F and tc were very similar, regardless of M (Fig. 1). For all levels of M and t , yield curves increased rapidly in the range of F between 0 and 0.50-0.75, and re- mained relatively flat thereafter. Although yield val- ues increased continuously with F, i.e. maximum yield-per-recruit always occurred at the maximum value of F used in simulations (F=2.0), increases in yield beyond F = 0.50-0.75 were very small. For ex- ample, increases in yield from F = 0.75 to FMAX ranged from 5.3% to 22.7%, depending on the level of M and t used in the model (Table 2). However, this rela- tively small gain in yield corresponds to an increase in F of 166.7%. For the range of M used in our study, estimates of F CUR are below the levels that give maximum poten- tial yield-per-recruit ( FMAX ) and, for M > 0.3, below 640 Fishery Bulletin 95(4), 1997 Table 2 Percent increase in yield-per-recruit of Atlantic croaker, Micropogonias undulatus, from fishing mortality (F) = mortality at the level that gives maximum potential yield-per-recruit ( FMAX ) for mean age-at-first-capture (tc) = 1 mortality (M) = 0.20-0.35 for Chesapeake Bay. 0.75 to fishing -5 and natural M tc Yield-per-recruit (g) % increase M tc Yield-per-recruit (g) % increase F= 0.75 F MAX F= 0.75 F 1 MAX 0.20 1 160.4 168.9 5.3 0.30 1 129.2 142.5 10.3 2 153.9 165.1 7.3 2 112.8 128.9 14.3 3 140.3 154.1 9.8 3 93.4 109.0 16.7 4 123.5 137.8 11.6 4 74.6 88.2 18.2 5 106.3 119.8 12.7 5 58.2 69.4 19.2 0.25 1 143.7 153.5 6.8 0.35 1 116.5 132.4 13.6 2 131.6 145.8 10.8 2 97.0 114.0 17.5 3 114.3 129.5 13.3 3 76.5 91.7 19.9 4 95.8 110.3 15.1 4 58.1 70.7 21.7 5 78.5 91.2 16.2 5 43.1 52.9 22.7 Barbieri et al.: Yield-per-recruit analysis for Micropogonias undulatus 641 F Figure 2 Curves of yield-per-recruit on fishing mortality (F) for Atlantic croaker, Micropogonias undulatus, in the lower Chesapeake Bay (period 1988-91) estimated for mean age-at- first-capture ( t ) = 2 and natural mortality (M) = 0.20-0.35. F0 7 = the level ofF at which the marginal increase in yield-per-recruit due to a small increase in F is 10% of the marginal yield-per-recruit in a lightly exploited fishery; FCUR = the estimated current levels of fishing mortality. the level of F0 l (Fig. 2; Table 3). For M = 0.20, FCUR is higher than F() p indicating that, although it pro- duces slightly higher yield values, current fishing mortality is not at its most economically efficient level. For example, for tc = 1 and tc = 2, over 90% of the yield obtained at FCUR can be achieved by lower- ing fishing mortality to the level of F0 r For M = 0.25, both Fci;r andF0 j equal 0.35, indicating that, although below the maximum potential yield-per-recruit, esti- mated current levels of harvest probably correspond to the most efficient level of F. In contrast, if M ranges from 0.30 to 0.35, F0 1 is higher than FCUR (Table 3), suggesting there would still be room to increase yield efficiently with increases inF. However, at these higher levels of M, increases in F necessary to achieve the yields at F0 , may be unrealistically high (Table 3). Values of tCR1TIC estimated with different values of M were relatively low for Atlantic croaker in Chesa- peake Bay. For M equal to 0.20, 0.25, 0.30, and 0.35, values of tCRITIC were 1.9, 1.4, 1.0, and 0.6 years, re- spectively. These values indicate that, for the range of M considered herein, maximum theoretical cohort biomass in the absence of fishing would be achieved before Atlantic croaker reach age 2 (years). Table 3 Estimated value of current levels of fishing mortality (FCUR) and level of fishing mortality at which the marginal in- crease in yield-per-recruit due to a small increase in F is 10% of the marginal yield-per-recruit in a lightly exploited fishery ( F0 , ) for Atlantic croaker, Micropogonias undulatus, in the Chesapeake Bay region for a range of fishing mor- tality (Ml = 0.20-0.35, and the percent increase or decrease in Fcur necessary to make it equal to F01. M F r CUR F). i % Difference 0.20 0.40 0.27 -48 0.25 0.35 0.35 0 0.30 0.30 0.45 +50 0.35 0.25 0.64 + 156 Estimated values of P g for Atlantic croaker in Chesapeake Bay were also relatively low. For = 3 12 mm, and the current estimated level of lc (265 mm, corresponding to tc= 2), Pg = 0.15, i.e., on the aver- age, only 15% of their potential growth still remains when Atlantic croaker in Chesapeake Bay enter the exploited phase at age 2. For alternative values of t 642 Fishery Bulletin 95(4), 1997 equal to 1, 3, 4 and 5 years, values of are 0.21, 0.10, 0.07, and 0.05, respectively. North Carolina Curves of yield-per-recruit on F for Atlantic croaker in North Carolina (Fig. 3) showed an opposite trend from that shown in Chesapeake Bay. For all levels of ForM used in simulations, yield values continuously increased from t(; = 1 (177 mm TL) to tc = 5 (434 mm TL), indicating that yield could be maximized by in- creasing t . However, the shape of yield-per-recruit curves differed among different levels of M and tc (Fig. 3). For M = 0.20 and 0.25, curves for tc = 1-3 peaked at low to intermediate levels of F ( FMAX = 0.20-0.60) and gradually decreased after that, whereas for t = 4-5 they increased rapidly in the range of F between 0 and 0.35-0.60 and remained relatively flat thereafter. For M = 0.30-0.35, the peaks in yield at low to intermediate levels of F oc- curred only for t = 1-2 and were a lot less pronounced than those at lower levels of M. For the range of M used in our simulations, esti- mates of Fcjjr (Fig. 4) indicated that during the pe- riod 1979-81 the level of fishing mortality for Atlan- tic croaker in North Carolina was well above the lev- els of F0 j and FMAX. At t = 1, estimated losses in potential yield-per-recruit from FMAX to FCUR were equal to 45%, 35%, 25%, and 4% for M = 0.20, 0.25, 0.30, and 0.35, respectively. Estimated losses if fish- ing mortality were kept at the level of FQ 1 would be 44%, 22%, 20%, and 14%, respectively. Estimated values of tCRITIC and P for Atlantic croaker in North Carolina were much higher than those estimated for Chesapeake Bay. For M equal to 0.20, 0.25, 0.30, and 0.35, values of tCRITIC were 7.5, M = 0.20 M = 0.25 roo M = 0 30 M = 0 35 500 f*'0.0 1*0.0 Figure 3 Curves of yield-per-recruit on fishing mortality (F) for Atlantic croaker, Micropogonias undulatus, in North Carolina (period 1979-81) estimated for mean age-at-first-capture (fc) = 1-5 and natural mortality (M) = 0.20-0.35. Barbieri et al.: Yield-per-recruit analysis for Micropogonias undulatus 643 6.7, 6.1, and 5.6 years, respectively. ForL^ = 645 mm and the estimated level of lc for the period 1979-81 (177 mm, corresponding to tc- 1), P ' = 0.72, i.e., on the average, 72% of their potential growth still remained when Atlantic croaker in North Carolina entered the exploited phase at age 1 during the period 1979-81. For alternative values of tc = 2-5 years, values of P g were 0.59, 0.49, 0.39, and 0.33, respectively. Discussion Our modeling results indicate that, for the range of M and F used in simulations, yield-per-recruit of Atlantic croaker in the lower Chesapeake Bay could be maximized by a management strategy that incor- porates early age at first capture (£c=l) and high rates of fishing mortality (F=2.Q). However, the analysis for Chesapeake Bay also showed this is probably not the most efficient management option for this spe- cies. Because of the essentially asymptotic relation between yield-per-recruit and F, harvesting Atlantic croaker at or near their maximum potential yield (i.e. at Fmax ) would require a disproportionate increase in fishing mortality making it an economically inef- ficient management option. In addition, given the multispecies nature of the main fisheries for Atlan- tic croaker in Chesapeake Bay (Austin, 1987; Chit- tenden, 1991), raising current levels of F would greatly increase overall rates of exploitation and probably interfere with management of other spe- cies such as weakfish, Cynoscion regalis, and spot, Leiostomus xanthurus. Decreasing the current level of t for Atlantic croaker in Chesapeake Bay would not be recom- mended for two reasons. First, for the range of M used in simulations, gains in yield-per-recruit from t = 2 to t = 1 were relatively small at FCUR. Second, because of the magnitude of the scrap catch of At- lantic croaker in Chesapeake Bay (Mercer1), it is likely that this species is already entering the ex- ploited phase at age 1 or younger. The current esti- mate of t (tc= 2; Barbieri et al., 1994a) may be an overestimate because it was based on arbitrarily defined commercial market grades instead of over- all catches — including the scrap. Because the mar- ket accepts only fish above a certain size, a reduc- tion in mesh sizes to attempt to increase the proportion of age-1 Atlantic croaker in the catches would probably only increase the number of fish sold as scrap and have little or no effect on commercial market grades. Nevertheless, the analysis showed no indication that fully recruited Atlantic croaker in Chesapeake Bay are being growth-overfished (i.e. that the fish were being caught before they had a chance to grow to their ideal size). Yield-per-recruit modeling results and estimated values of Fcur indicated that, over a likely range of M, current levels of harvest are below the levels at F and, under most scenarios, at or below the levels atF0 r In addition, yield- per-recruit curves showed no signs of decrease at higher lev- els of F, even if M is as low as 0.20. This pattern suggests that stocks of Atlantic croaker in the Chesapeake Bay region show the same great biologi- F Figure 4 Curves of yield-per-recruit on fishing mortality (F) for Atlantic croaker, Micropogonias undulatus , in North Carolina (period 1979-81) estimated for mean age-at-first-capture (tc) = 1 and natural mortality (M) = 0.20-0.35. F0 1 = the level of F at which the marginal increase in yield-per-recruit due to a small increase in F is 10% of the marginal yield- per-recruit in a lightly exploited fishery; FCUR = the estimated current levels of fishing mortality. 644 Fishery Bulletin 95(4), 1997 cal capacity to resist growth overfishing as those stocks in the northwestern Gulf of Mexico (Chit- tenden, 1977). The low values of tCRITIC andPa agree with yield-per-recruit modeling results and indicate that 1) for a reported maximum longevity of 8 years in Chesapeake Bay (Barbieri et al. , 1994a), maximum theoretical biomass is achieved very early in life, before fish reach age 2; and 2) very little potential for a growth span still remains when fish enter the exploited phase at age 2. As a precaution against future problems — especially considering that annual recruitment is reported to be highly variable and strongly density independent — we suggest that regu- latory measures for Atlantic croaker in the lower Chesapeake Bay be directed at maintaining the ap- parent current level of t (age 2; lc= 265 mm TL; Barbieri et al., 1994a). In addition, the magnitude and composition of the scrap catch for the main fish- eries in this area need to be estimated, and their ef- fect on estimates of FCUR and tc need to be assessed more precisely before any definite conclusion on At- lantic croaker yield-per-recruit can be reached. In contrast to what we found for the lower Chesa- peake Bay, results for North Carolina indicated that Atlantic croaker were being severely growth-over- fished. First, independent of the level of F or M used in simulations, yield-per-recruit values were consis- tently higher at higher levels of tc, indicating that age and size limits during the period 1979-81 (f =1, Z =177 mm TL; Ross, 1988) were unrealistically low. Second, estimates of FCUR for Z = 1 were not just con- sistently higher than F0 x but were also well above FMAX. The pattern of declining yield-per-recruit values with increasing F at lower levels of tc agrees well with the high estimates of tCRITIC and Pg and indicates that, con- trary to the pattern shown in Chesapeake Bay, maxi- mum cohort biomass is attained later in life (ages 5-7). However, differences in yield-per-recruit modeling results between Chesapeake Bay and North Caro- lina seem to reflect temporal rather than spatial dif- ferences in Atlantic croaker population dynamics. Parameters used in simulations for North Carolina were obtained from a study (Ross, 1988) conducted during a period (1979-81) that coincides with the occurrence of unusually large Atlantic croaker (350- 520 mm TL; Ross, 1988) along the east coast of the United States (Barbieri et al., 1994a). However, since 1982, Atlantic croaker catches in North Carolina have been dominated by smaller fish. Modal lengths of Atlantic croaker in the long haul-seine fishery dur- ing 1982-92 ranged from 215 to 245 mm TL; in the winter trawl fishery, they ranged from 215 to 240 mm TL. In both fisheries, less than 10% of the fish were older than age 3 (Wilson, 1993). Therefore, yield-per-recruit modeling results presented here for North Carolina should not reflect current conditions, but rather be considered representative of temporal changes in Atlantic croaker population dynamics. The specific value of M used in our simulations had no effect on the levels of F or Z that produce maximum yield-per-recruit values and would not change conclusions for either Chesapeake Bay or North Carolina. However, these conclusions are still critically dependent on how realistic is the range of M used in these simulations. Methods currently used to estimate M have strong limitations and disadvantages (Vetter, 1988), and the method used here is no exception. How- ever, we feel comfortable with the range of M used in this study because it agrees with values of M reported for other sciaenids with similar life spans, e.g. spotted seatrout, Cynoscion nebulosus (Rutherford et al., 1989). Yield-per-recruit analysis is only part of a fishery management strategy (Beverton and Holt, 1957; Gulland, 1983; Deriso, 1987). It must be applied in conjunction with eggs-per-recruit (Prager et al. , 1987) and spawning stock biomass per recruit models (Gabriel et al., 1989; Goodyear, 1993; Schirripa and Goodyear, 1994) to allow managers to examine the effects of different policies on both reproduction (i.e. egg production) and biomass yield. The pattern of early maturation, multiple spawning, long spawn- ing season, and indeterminate fecundity in Atlantic croaker (Barbieri et al., 1994b) suggest that repro- duction would be compromised only at extremely high levels of fishing. However, eggs-per-recruit and spawning stock biomass models must be applied be- fore this issue can be properly evaluated. Acknowledgments We would like to thank Sue Lowerre-Barbieri and two anonymous reviewers for helpful suggestions that improved the manuscript. Financial support for this project was provided by the College of William and Mary, Virginia Institute of Marine Science, and by a Wallop/Breaux Program Grant for Sport Fish Restoration from the U. S. Fish and Wildlife Service through the Virginia Marine Resources Commission, Project No. F-88-R3. Luiz R. Barbieri was partially supported by a scholarship from CNPq, Ministry of Science and Technology, Brazil (process No. 203581/ 86-OC) and by a postdoctoral fellowship from the University of Georgia Marine Institute. Literature cited Alverson, D. L., and M. J. Carney. 1975. A graphic review of the growth and decay of popula- tion cohorts. J. Cons. Cons. Int. Explor. Mer 36:133-143. Barbieri et a\.: Yield-per-recruit analysis for Micropogonias undulatus 645 Austin, H. 1987. Chesapeake Bay fisheries: an overview. In S. K. Majumdar, L. W. Hall, and H. M. Austin (eds.), Contami- nant problems and management of living Chesapeake Bay resources, p. 33—53. The Pennsylvania Academy of Sci- ence, Philadelphia, PA. Barbieri, L. R., M. E. Chittenden Jr., and C. M. Jones. 1994a. Age, growth, and mortality of Atlantic croaker, Micropogonias undulatus, in the Chesapeake Bay region, with a discussion of apparent geographic changes in popu- lation dynamics. Fish. Bull. 92:1-12. Barbieri, L. R., M. E. Chittenden Jr., and S. K. Lowerre-Barbieri. 1994b. Maturity, spawning, and ovarian cycle of Atlantic croaker, Micropogonias undulatus, in the Chesapeake Bay and adjacent coastal waters. Fish. Bull. 92:671-685. Beverton, R. J. H. 1963. Maturation, growth, and mortality of clupeid and engraulid stocks in relation to fishing. Rapp. Reun. P.-V. Cons. Int. Explor. Mer 154:44-67. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. U.K. Min. Agric. Fish. Food., Fish. Invest, (ser. 2) 19:1-533. Chao, L. N., and J. A. Musick. 1977. Life history, feeding habits, and functional morphol- ogy of juvenile sciaenid fishes in the York River estuary, Virginia. Fish. Bull. 75:675-702. Chittenden, M. E., Jr. 1977. Simulations of the effects of fishing on the Atlantic croaker, Micropogon undulatus . Proc. Gulf Caribb. Fish. Inst. 29:68-86. 1991. Operational procedures and sampling in the Chesa- peake Bay pound-net fishery. Fisheries (Bethesda) 16:22-27. Deriso, R. B. 1987. Optimal F0 j criteria and their relationship to maxi- mum sustainable yield. Can. J. Fish. Aquat. Sci. 44(suppl. 21:339-348. Gabriel, W. L., M. P. Sissenwine, and W. J. Overholtz. 1989. Analysis of spawning stock biomass per recruit: an example for Georges Bank haddock. North Am. J. Fish. Manage. 9:383-391. Goodyear, C. P. 1993. Spawning stock biomass per recruit in fisheries man- agement: foundation and current use. In S. J. Smith, J. J. Hunt, and D. Rivard (eds.), Risk evaluation and biologi- cal reference points for fisheries management, p. 67- 81. Can. Spec. Publ. Fish. Aquat. Sci. 120. Gulland, J. A. 1983. Fish stock assessment. John Wiley & Sons, New York, NY, 223 p. Gulland, J. A., and L. K. Boerema. 1973. Scientific advice on catch levels. Fish. Bull. 71:325- 335. Joseph, E. B. 1972. The status of the sciaenid stocks of the middle At- lantic coast. Chesapeake Sci. 13:87-100. McHugh, J. L., and D. Conover. 1986. History and condition of food finfisheries in the Middle Atlantic region compared with other sections of the coast. Fisheries (Bethesda) 11:8-13. Newlin, K. (ed.) 1992. Fishing trends and conditions in the southeast re- gion 1991. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-311, 84 p. Prager, M. H., J. F. O’Brien, and S. B. Saila. 1987. Using lifetime fecundity to compare management strategies: a case history for striped bass. North Am. J. Fish. Manage. 7:403-409. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics of fish populations. Bull. Fish. Res. Board Can. 191, 382 p. Ross, S. W. 1988. Age, growth, and mortality of Atlantic croaker in North Carolina, with comments on population dynam- ics. Trans. Am. Fish. Soc. 117:461-473. Rothschild, B. J., P. W. Jones, and J. S. Wilson. 1981. Trends in Chesapeake Bay fisheries. Trans. 46th N. Am. Wildl. Nat. Resour. Conf. 1981:284-298. Rutherford, E. S., J. T. Tilmant, E. B. Thue, and T. W. Schimidt. 1989. Fishery harvest and population dynamics of spotted seatrout, Cynoscion nebulosus, in Florida Bay and adja- cent waters. Bull. Mar. Sci. 44:108-125. Saila, S. B., C. W. Recksiek, and M. H. Prager. 1988. Basic fisheries science programs. Elsevier, New York, NY, 230 p. Schirripa, M. J., and C. P. Goodyear. 1994. Simulation modeling of conservation standards for spotted seatrout ( Cynoscion nebulosus) in Everglades Na- tional Park. Bull. Mar. Sci. 54:1019-1035. Schmied, R. L., and E. E. Burgess. 1987. Marine recreational fisheries in the southeastern United States: an overview. Mar. Fish. Rev. 49:2-7. Speir, H., H. Austin, and L. Mercer. 1994. 1993 annual review of the Atlantic States Marine Fisheries Commission fishery management plan for Atlan- tic croaker (Micropogonias undulatus ). In 1993 annual review of the Atlantic States Marine Fisheries Commis- sion Interstate Fishery Management Plans, p. 11-16. Spe- cial Report 25 of the Atlantic States Marine Fisheries Com- mission, Washington, D.C. Vetter, E. F. 1988. Estimation of natural mortality in fish stocks: a review. Fish. Bull. 86:25-43. Wilk, S. J. 1981. The fisheries for Atlantic croaker, spot, and weak- fish. In H. Clepper, (ed.), Proceedings of the 6th annual marine recreational fisheries symposium, p. 15-27. Sport Fish. Inst., Washington. Wilson, C. 1993. North Carolina report for Atlantic croaker. In L. L. Kline and H. Speir (eds.), Proceedings of a workshop on spot (Leiostomus xanthurus ) and Atlantic croaker ( Micro- pogonias undulatus), p. 34-54. Atlantic States Marine Fisheries Commission Special Report 25. 646 Abstract. -Eight female cunners, Tautogolabrus adspersus, were tagged externally with ultrasonic transmitters in Newfoundland, and their activity pattern was recorded. They were active diurnally, commencing activity, on av- erage, 55 minutes after sunrise and ceasing activity about 50 minutes af- ter sunset. The diurnal activity period was interrupted by periods of inactiv- ity usually lasting 5-15 minutes. Lev- els of activity varied daily and season- ally; seasonal changes were the most dramatic. On average, female cunners were active for more than 12.5 h/day in June-July and for only 3 h/day in October-November. Decrease in activ- ity reflected decreasing day length; as photoperiod became shorter, cunners spent a much larger portion of the day- light period inactive (22.9% in June- July compared with 71.8% in October- November). Decrease in cunner activ- ity in the fall occurred while water tem- perature was as high as that in June and July and is speculated to be con- trolled by an endogenous rhythm. Manuscript accepted 7 May 1997. Fishery Bulletin 95:646-652 (1997). Daily and seasonal activity patterns of female cunner, Tautogolabrus adspersus (Labridae), in Newfoundland Carole Bradbury John M. Green* Department of Biology Memorial University St. John's, Newfoundland, Canada A1 B 3X9 E-mail address (for John Green): JMGREEN@morgan.ucs.mun.ca Michael Bruce-Lockhart Faculty of Engineering Memorial University St. John's, Newfoundland, Canada A 1 B 3X9 Cunner, Tautogolabrus adspersus, is the most northerly distributed labrid fish in the western North Atlantic, reaching the northern ex- tent of its range in waters off New- foundland. A member of a large, es- sentially tropical, family, it is well known for its annual state of pro- longed physiological torpor, which in Newfoundland may last for more than 6 months (Green and Farwell, 1971). The ability of cunners to un- dergo a long period of torpor is ap- parently one of the factors that have enabled this species to flourish in a low temperature environment (Cur- ran, 1992). In waters off Newfound- land, cunners are abundant and in- habit sites where summer maxi- mum water temperature is less than 11°C and the winter minimum is below -1°C. Newfoundland cun- ners enter and remain in torpor when seawater temperature is be- low about 5°C (Green and Farwell, 1971). Throughout their range, Chesa- peake Bay to the Strait of Belle Isle, cunners are associated with inshore habitats that provide shelter dur- ing nocturnal quiescence (a charac- teristic of labrids) as well as during overwintering torpor (Pottle and Green, 1979a). Rather than migrat- ing to deeper, warmer water as the temperature declines in the fall, Newfoundland cunners take shelter under boulders and rocks at their summer feeding and reproductive sites. There they remain until the seawater temperature approaches 5°C the following year, usually in early June (Green and Farwell, 1971). In Conception Bay, New- foundland, territorial males estab- lish territories within a week of emerging from winter torpor and maintain them until just prior to reentering winter torpor, usually in late November or early December (Pottle and Green, 1979a). Spawn- ing commences in mid to late July, depending upon water temperature, and lasts for 2-3 weeks. All spawn- ing activity occurs within male ter- ritories (Pottle and Green, 1979b; Martel and Green, 1987). We recently reported on a home- range size for adult female cunners in Conception Bay, Newfoundland, that was based on telemetry data * Author to whom correspondence should be sent. Bradbury et a I.: Daily and seasonal activity patterns of female Tautogolabrus adspersus 647 from fish tagged with ultrasonic transmitters (Bradbury et al., 1995). Female cunners occupy small home ranges (300-2,353 m2) and exhibit seasonal variation in the size of their home range. The larg- est home ranges occur during June and July, a time when cunners are replacing energy stores depleted during winter torpor (Bradbury et al., 1995). This is also the time of year with the longest photoperiod and hence with the maximum potential foraging pe- riod for a diurnal species. In this study, we report on the daily activity pat- terns of the female cunners in Conception Bay. Cun- ners at the northern extent of their range are of in- terest because their prolonged winter torpor may reduce annual foraging time. We expected that as the photoperiod decreased, female cunners would be active for more of the diurnal period, both to maxi- mize growth and increase energy stores in prepara- tion for six months of winter torpor. Methods Tracking A fixed hydrophone array tracking system ( Bradbury et al., 1995) was used to monitor the activity of eight female cunners tagged with ultrasonic transmitters in Broad Cove, Conception Bay. The system provided positional information (fixes) on a tagged fish once every 15 seconds. The change in the quality of the transmitter signal when a tagged cunner sought shel- ter by going under a boulder or into a crevice be- tween two rocks enabled us to monitor activity-inac- tivity patterns accurately in much the same way as Chapman et al. (1975) had done with Norway lobster (Nephrops norvegicus). For a description of the track- ing system, tagging procedure, and study site see Bradbury et al. (1995). With our tagging procedure, a tag holder and dummy tag were initially attached to fish in the field. After a week (minimum period), a fish with a tag holder and dummy tag was recaught by a diver, and the dummy transmitter was replaced with a functional one. The latter procedure involved handling the fish for 1-2 min, from capture to release. Tagged cunners were from 194 to 250 mm in total length (Table 1). At this size female cunners in Con- ception Bay are sexually mature (Pottle and Green, 1979a). Fish were tracked from June until Novem- ber, i.e. during most of the period between the end and start of winter torpor. Lightning damage to the tracking system limited the amount of tracking that could be done in August. Individual fish were tracked for 4 to 32 days during which they all remained in the area encompassed by the hydrophone array. For Table 1 Total length and tracking dates for female cunner tagged with ultrasonic transmitters in Broad Cove, Conception Bay, Newfoundland. Fish identification Total length (mm) Tracking dates Track duration (days) A 194 17 Jun-30 Jun 14 5 Jul-6 Jul 2 B 215 18 Jun-4 Jul 17 C 250 28 Jun-21 Jul 24 D 235 12 Jul-21 Jul 10 E 195 15 Aug- 18 Aug 4 F 225 30 Aug-22 Sep 16 G 245 23 Sep-20 Oct 22 H 240 21 Oct -24 Nov 32 the duration of the tracking period, information on water temperature, sea state, tidal phase, and cloud cover was available on a daily basis, as described by Bradbury et al. (1995). Activity-inactivity A tagged cunner was determined to be active or in- active based on information from the tracking sys- tem. If a strong transmitter signal was received, and positional information was obtained, the subject was considered active. If, on the other hand, signals were weak and no positional fix could be determined for more than 3 min ( 12 possible fixes), the fish was con- sidered inactive. With scuba or snorkel equipment, divers documented over a period of >20 h that cun- ners had retreated into cracks and crevices or un- derneath objects when transmitter reception was poor. These observations also showed that female cunners do not “rest” on the substrate in open sites. During the night, cunners seek shelter and un- dergo a period of nocturnal quiescence during which positional fixes cannot be obtained. The first and last positional fixes of the day therefore marked the be- ginning and end of diurnal activity. Onset of diurnal activity was expressed as the number of minutes before or after sunrise, whereas cessation of diurnal activity was expressed as the number of minutes before or after sunset. The duration of diurnal activity for an individual was defined as the total elapsed time between the onset and cessation of its daily activity. Cunners also entered shelter (became inactive) at various times throughout the day; sites where cun- ners were inactive are termed day-rest sites. The duration of each inactive period (the time between 648 Fishery Bulletin 95(4), 1997 the initiation and the end of a poor transmitter sig- nal) was recorded. On the basis of these data, the portion of the diurnal activity period spent inactive was calculated and expressed as a percentage. In describing the activity-inactivity patterns of female cunners, five parameters were used: 1) onset of activity, 2) cessation of activity, 3) duration of di- urnal activity, 4) length of inactivity bouts, and 5) percent of diurnal activity period spent inactive. Analysis of data Although a fish was handled for only 1-2 min when a transmitter was inserted into its tag holder and although field observations did not detect any changes in the behavior of fish following this proce- dure, nonparametric paired t-tests were used to ex- amine whether female cunners showed similar ac- tivity on the first complete day of tracking (day 2) compared with the following day of tracking (day 3). All five activity parameters were tested. A nonparametric analysis (Wilcoxon matched-pairs signed-ranks test) was used to compare inter- individual differences in activity between the two pairs of subjects tracked during the same periods (Table 1). Comparisons were made between fish A and fish B for a total of 11 days (i.e. June 18, 19, 20, 21, 22, 24, 25, 27, 28, 29, and 30) and between fish C and fish D for a total of 9 days (i.e. July 13-21 inclu- sive) for all five activity parameters. A least-squares multiple regression analysis was used to determine whether date, time of day (i.e. morning vs. afternoon), or environmental variables (water temperature, cloud cover, and sea state) had a significant effect on activity. All activity param- eters were tested. Because the activity data were normally distributed, no transformations were carried out. Although there were 141 tracking days, some days could not be used for certain activity parameters. For example, there were only 72 tracking days during which both the time of onset and cessation of activity were known for any given fish, both of which are required to calculate the duration of diurnal activity. The tidal cycle was divided into four phases: low- tide, flood-tide, high-tide, and ebb-tide as described by Bradbury et al. (1995). For fish A, B, C, and D, mean activity parameters (i.e. percentage of time spent inactive and length of inactivity bouts) were determined for each tidal phase for the duration of the tracking period. An analysis of variance with three factors was used to test for intra- and inter- individual differences in activity during the tidal phases. We included only the fish by tide interaction term in our analyses because we did not expect any temporal variation in tide (tide x date) or activity Table 2 Mean number of minutes before sunrise and after sunset when tracked female cunners began and ceased their diur- nal activity. Because the data for onset of activity for fish E consisted of a single point, no standard deviation is given. Onset Cessation Fish (minutes before sunrise) (minutes after sunset) identi- fication Mean SD n Mean SD n A 57.6 10.02 9 57.6 26.70 9 B 40.2 23.95 12 63.6 18.10 12 C 48.7 10.93 18 74.3 19.64 22 D 31.7 18.48 6 43.0 43.59 4 E 45.0 — 1 50.3 34.46 4 F 52.0 21.28 3 36.5 41.16 4 G 49.4 32.69 14 54.1 20.33 17 H 72.4 52.54 28 18.9 39.12 27 All 54.7 36.63 91 49.1 36.15 99 (fish x date), given the relatively short time (i.e. 11 and 9 days) over which observations were made. Paired comparison £-tests were used to examine whether the one female cunner tracked during both the prespawning and spawning period had the same activity patterns during both periods. All activity parameters were tested. Statistical analyses were performed with Minitab (Minitab, Inc., 1992) or SPSSX (SPSS Inc., 1990) sta- tistical software packages. Results There were no significant differences (P>0.05, non- parametric paired £-test) in activity parameters be- tween the first complete day of tracking (day 2) and the following day. All tagged fish were active during the day, inac- tive at night. Activity commenced, on average, 55 minutes (SD=36.6) before sunrise and ceased 49 min- utes (SD=36.2) after sunset; however, there was considerable daily variation among individuals (Table 2). Throughout the day, activity was inter- rupted by periods of inactivity, usually lasting 5-15 minutes. Among those fish tracked on the same day, there were no significant differences between sub- jects for any of the activity parameters (Table 3). When water temperature was below 5°C, cunners were inactive. On 23 and 24 June, for example, strong northwesterly winds forced cold water into the study area causing the water temperature to drop from 6°C to 3°C and the cunners to be inactive for two days. On the morning of 26 June the water temperature Bradbury et al.: Daily and seasonal activity patterns of female Tautogolabrus adspersus 649 Table 3 Results of the Wilcoxon matched-pairs signed-ranks test (t-value) used to compare interindividual differences in activity between pairs of cunners tracked during the same periods (n represents the number of days during which comparisons were made). Critical values of t are derived from Rohlf and Sokal (1969). There were no significant differences between fish for any of the activity parameters tested. Comparisons between fish A and fish B Comparison between fish C and fish D Behavioral parameter n t-value Critical t-value (significance level) n t-value Critical <-value (significance level) Percent of time inactive n 26 13 (0.0415) 14 (0.0508) 9 11 8 (0.0488) 9 (0.0645) Length of inactivity bout ii 26 13 (0.0415) 14 (0.0508) 9 14 8 (0.0488) 9(0.0645) Onset of activity 9 15 8 (0.0488) 9 (0.0645) 7 6 3 (0.0391) 4 (0.0547) Cessation of activity 8 10 5 (0.0391) 6 (0.0547) 7 5 3 (0.0391) 4(0.0547) Duration of diurnal activity 8 13 5 (0.0391) 6 (0.0547) 7 7 3 (0.0391) 4 (0.0547) Table 4 Summary of multiple regression analysis on the effects of time of day (prior to 1200 h vs. after 1200 h), date, and environmental variables on activity of female cunner. A minimum of 62 days and a maximum of 89 days were incorporated in the regression analysis for the last three behavioral parameters. Percentage of variation accounted for by each variable is given. * = significant at 0.05 level; ** = significant at 0.01 level; and *** = significant at 0.001 level. Water Time of temperature Combined Behavioral parameter n day Date >5°C Sea state Cloud cover variables Percent of time inactive 205 0.6 58.7*** 0.2 0.3 0.1 59.9*** Length of inactivity bout 229 0.1 12 7*** 2.8** 1.7* 0.2 17.5*** Onset of activity 75 NA 6.6* 1.9 0.0 0.1 8.6 Cessation of activity 83 NA 22.3*** 0.2 0.0 0.5 23.0*** Duration of diurnal activity 62 NA 92.3*** 0.9** 0.1 0.1 93.4*** again dropped to 3°C, resulting in cunners being in- active for the remainder of the day. Temperatures above 5°C had small but significant effects on length of inactivity bouts and duration of diurnal activity (Table 4). Length of bouts of inactivity tended to de- crease with increasing water temperature, whereas the trend was reversed for the duration of diurnal activity. Water temperatures above 5°C had no sig- nificant effect on the onset or cessation of activity or on the percentage of time spent inactive. Sea state had a significant (P<0.05) effect on length of cunner inactivity bouts (Table 4), with bouts of inactivity tending to be longer on days with high surface waves. Sea state did not have a significant effect on other activity parameters. Cloud cover had no significant effect on any of the activity param- eters. There was a trend, however, for females to re- main inactive for longer periods (i.e. percentage of time spent inactive increased as well as length of inactivity bouts) as cloud cover increased. There was also a tendency for the duration of diurnal activity to decrease with increasing cloud cover. Neither per- centage of diurnal activity period spent inactive or length of inactivity bouts differed between morning and afternoon (Table 4). There was no significant difference in fish behav- ior owing to tides (Table 5), indicating that both fish A and fish B responded similarly to the tidal cycle. Furthermore, there was no significant difference in activity (i.e. percentage of time spent inactive) be- tween the various tidal phases for fish A or fish B (Table 5). Finally, there were no significant differ- 650 Fishery Bulletin 95(4), 1997 Table 5 Results of analysis of variance performed on the percent- age of time spent inactive by fish A and fish B during the four phases of the tidal cycle, df = degrees of freedom; SS = sum of squares; MS = mean of squares. Probability Source df SS MS F-value value P Date 10 5,492.5 549.3 1.74 0.089 Fish 1 33.1 33.1 0.10 0.747 Tide 3 1,320.8 440.3 1.39 0.252 Fish x tide 3 771.5 257.2 0.81 0.490 Error 70 22,117.3 316.0 Total 87 29,735.3 ences between the tidal phases for the length of in- activity bouts of fish Aand fish B (F3 70=1.57, P=0.204; F1 ?=0.4, P= 0.756), percent of time spent inactive by fish C and fish B (P356=0.50, P= 0.685; F13=0.79, P=0.505) and length of inactivity bouts of fish C and fish B (P356=0.82, P=0.4986; Fx 3=1.94, P=0.134). Time of1 year, i.e. seasonal factors, accounted for 58.7% of the variation in the percentage of the diur- nal activity period spent inactive, i.e. the amount of time spent in shelter between the commencement and cessation of daily activity (Table 4). Cunners spent an increasing proportion of the diurnal period inactive from June through November. This factor accounted for 12.7% of the variation in length of inactivity bouts (which increased over time as well) and for 6.6% and 22.3% of the variance in onset and cessation of activ- ity, respectively (Table 4). The trend was for activity to begin later in the morning and to end earlier in the evening (in relation to sunrise and sunset) as the season progressed. Time of year accounted for 92.3% of the variation in duration of diurnal activity. As the sea- sons progressed and the photoperiod became progres- sively shorter, there was a corresponding decrease in the duration of diurnal activity (Fig. 1). Thus the duration of cunner diurnal activity, per- centage of time inactive, and length of inactivity bouts were all closely related to day length. By com- bining data from all subjects, the average elapsed time between onset and cessation of activity (i.e. duration of diurnal activity) was 16.5 h (n= 38, SE=0.10) during June-July, 13.5 h (n= 8, SE=0.19) during August-September, and 11.0 h (n=27, SE=0.25) during October-November. Cunner spent 22.9% (n=35, SE=15.5) of the diurnal period inactive between June and July, 47.0% (n=13, SE 24.3) of the diurnal period inactive in August-September, and 71.8% (n= 30, SE=13.2 ) of the diurnal period inactive during October-November; length of inactivity bouts, on the other hand, increased from an average of 6.4 min (n=64, SE=0.75) in June— July to 9.3 min (n= 15, SE 1.56) in August-September, to 12.7 min (tz=41, SE=1.14) in October-November. These data were used to calculate the number of hours a female cunner spent out of its shelter per day. On average, dur- ing June-July, female cunners were active for 12.5 h/day, during August- September 7 h/day, and during Octo- ber-November only 3 h/day. The re- mainder of the year was spent in win- ter torpor. There were no significant differ- ences (P>0.05, paired comparison t- test) in any of the activity parameters for the single female cunner tracked during both the prespawning and spawning periods. Discussion 20 15 io » Figure I Relation between length of daily photoperiod and duration of diurnal activity (US) for eight female cunners tracked in Conception Bay, Newfoundland, be- tween June 17 and November 24. Maximum daily seawater temperature at the study site is also shown. Pottle (1979) reported that in New- foundland territorial male cunners undergo periods of daylight quies- cence under cover. Whoriskey (1983) also observed cunners in Massachu- setts underneath boulders at those times during the day when they were Bradbury et al. : Daily and seasonal activity patterns of female Tautogolabrus adspersus 651 not foraging. The female cunners we studied exhib- ited similar behavior, seeking shelter beneath rocks or in crevices during the day. Some workers have assumed that temperate wrasses use cover to avoid predation ( Olla et al., 1979; Hobson et al., 1981), although threat of predation has not been well documented as a factor. Whoriskey (1983) interpreted the diurnal use of shelter by cun- ners in Massachusetts as predator avoidance. Al- though he may have been correct, our field observa- tions in Newfoundland do not support this hypoth- esis. Predation on adult cunners in Conception Bay is very low as judged by over 400 hours of diving ob- servations during which no predation, or attempted predation, on adults was observed. Females may seek shelter to avoid conspecifics with courting and chasing behaviors, especially ter- ritorial males. This explanation, however, is inad- equate in elucidating why males exhibit the same behavior or why the behavior occurs so frequently outside the spawning period. A reduction in energy expenditure may be associ- ated with such behavior because cunners probably require less energy to maintain a position in a shel- ter than in the water column, even when water move- ment from currents and waves is minimal, and be- cause the length of inactivity bouts increased with increased water turbulence. However, such an ex- planation is weak unless it can be shown that con- tinued foraging would result in a net loss of energy. In many diurnal fishes, including cunners, the onset and cessation of daily activity coincides closely with the rising and setting of the sun (e.g. Hobson, 1972, 1973; Hawkins et al., 1974; Olla et al., 1974, 1975; Clark and Green, 1990). As expected, females exhibited a marked seasonal decrease in the dura- tion of their diurnal activity (from 16.5 h in June- July to 11.0 h in October-November) as day length decreased. Light intensity at sunrise and sunset was affected seasonally by the surrounding topography at the study site (e.g. in the fall the sun “set” behind a range of hills rather than at sea level), and this topography may have accounted for some of the sea- sonal change in the onset and cessation of diurnal activity. However, the considerable variation in the onset and cessation of daily activity among cunners suggests that this variation is not simply a response to a threshold light intensity. Contrary to the expectation that female cunners would maximize foraging opportunities prior to en- tering winter torpor, they spent a larger percentage of the diurnal period inactive as the length of the photoperiod decreased. Why cunners should signifi- cantly reduce their foraging activity, at a time when food is still available and they could acquire more energy for somatic growth and winter torpor, is not clear. Although water temperature is decreasing dur- ing this period, our analyses show that above ~5°C, temperature has little direct effect in determining the ratio of activity to inactivity. Mean daily water temperatures during June-July and October-No- vember were approximately the same (8.2°C and 9.2°C, respectively )(Fig. 1), yet there were large dif- ferences in the amount of time cunners spent in shel- ter. During June and July, females were outside their shelter for about 12.5 hours of the day, whereas from October to November cunners were active, on average, only 3 hours of the 11-h sunrise to sunset period. Fall or winter decreases in the activity or feeding behavior (or both) of fishes in the absence of changes in water temperature are common, although the mechanisms underlying these decreases are not un- derstood. Smith et al. (1993) for example found that in Atlantic salmon ( Salmo salar), seasonal reductions in swimming activity and feeding were more closely related to day length and changes in day length than to other environmental variables, including water temperature. This also seems to be true for cunners. Presumably their temporal pattern of activity is adaptive and important to their success at northern latitudes. Cunners survive six months or more of torpor that can begin at a time not predictable by exogenous cues in the marine environment. At our study site, the date at which winter torpor com- mences (i.e. when seawater temperature remains below ~5°C) can vary year to year by at least four weeks. Perhaps an endogenous mechanism sensitive to changes in day length, or to some other environ- mental cue, regulates the physiological processes associated with successful winter torpor. Although such a mechanism may exist in cunners, the identi- fication of endogenous rhythms in fishes is difficult (Boujard and Leatherland, 1992). Our findings concerning seasonal changes in the activity patterns of cunners have implications for estimating the size of their populations. For example, population estimates based on visual surveys by divers should take into account that, depending on when the survey is conducted, a significant and vari- able proportion of the population will be out of sight, under cover. Significant errors in estimates of popu- lation size are likely, and errors will not be consis- tent for different times of the year. This caution may apply to other species with similar behavior patterns. Acknowledgments We thank Robert Dunbrack for many helpful discus- sions and his review of the manuscript. John Gibson 652 Fishery Bulletin 95(4), 1997 also provided helpful advice. Steve Carr and Malcolm Grant wrote computer programs to help analyze te- lemetry data. Graham Skanes, Bill Warren, Brian McCleran, and David Schneider provided statistical advice. We appreciated the good spirits and field as- sistance provided by Duane Barker, Lorri Mitchell, Natalie Miller, and Wayne Chiasson. Bruce Cocker and Ed Thistle brought the tracking system back to life af- ter it had been damaged by lightning. The comments of three anonymous reviewers improved the paper. This work was supported by an NSERC operating grant to JMG and funding from the School of Gradu- ate Studies, Memorial University, to CB. Literature cited Boujard, T., and J. F. Leatherland. 1992. Circadian rhythms and feeding time in fishes. Environ. Biol. Fishes 35:109-131. Bradbury, C., J. M. Green, and M. Bruce-Lockhart. 1995. Home ranges of female cunner, Tautogolabrus adspersus (Labridae), as determined by ultrasonic telemetry. Can. J. Zool. 73:1268-1279. Chapman, C. J., A. D. F. Johnstone, and A. L. Rice. 1975. The behaviour and ecology of the Norway lobster, Nephrops norvegicus ( L. ). Proc. 9th Europ. Mar. Biol. Symp., p 59-74. Aberdeen Univ. Press, Scotland. Clark, D. J., and J. M. Green. 1990. Activity and movement patterns of juvenile Atlantic cod, Gadus morhua, in Conception Bay, Newfoundland, as deter- mined by sonic telemetry. Can. J. Zool. 68:1434—1442. Curran, M. C. 1992. The behavioral physiology of labroid fishes. Doctoral diss., Massachusetts Institute of Technology /Woods Hole Oceanographic Institution Joint Program in Oceanogra- phy, Woods Hole, MA, 124 p. Green, J. M., and M. Farwell. 1971. Winter habits of the cunner, Tautogolabrus adspersus (Walbaum 1792), in Newfoundland. Can. J. Zool. 49:1497- 1499. Hawkins, A. D., D. N. Maclennan, G. G. Urquhart, and C. Robb. 1974. Tracking cod Gadus morhua L. in a Scottish sea loch. J. Fish Biol. 6:225-236. Hobson, E. J. 1972. Activity of Hawaiian reef fishes during the evening and morning transitions between daylight and dark- ness. Fish. Bull. 70:715-740. 1973. Diel feeding migrations in tropical reef fishes. Helgol. Wiss. Meeresunters. 24:361-370. Hobson, E. S., W. N. McFarland, and J. R. Chess. 1981. Crepuscular and nocturnal activities of California nearshore fishes, with consideration of their scotopic vi- sual pigments and the photic environment. Fish. Bull. 79:1-30. Martel, G., and J. M. Green. 1987. Differential spawning success among territorial male cunners, Tautogolabrus adspersus (Labridae). Copeia 3:643-648. Minitab Inc. 1992. Minitab reference manual: release 9. Minitab Inc., State College, PA. Olla, B. L., A. J. Bejda, and A. D. Martin. 1974. Daily activity, movements, feeding, and seasonal oc- currence in the tautog, Tautoga onitis. Fish. Bull. 72: 27-35. 1975. Activity, movements, and feeding behaviour of the cunner, Tautogolabrus adspersus, and comparison of food habits with the young tautog, Tautoga onitis, of Long Is- land, New York. Fish. Bull. 73:895-900. 1979. Seasonal dispersal and habitat selection of the cun- ner, Tautogolabrus adspersus, and young tautog, Tautoga onitis, in Fire Island Inlet, Long Island, New York. Fish. Bull. 77:255-261. Pottle, R. A. 1979. A field study of territorial and reproductive behaviour of the cunner, Tauogolabrus adspersus, in Conception Bay, Newfoundland. Master’s thesis, Department of Biology, Memorial Univ. Newfoundland, St. John’s, Newfoundland, 104 p. Pottle, R. A., and J. M. Green. 1979a. Field observations on the reproductive behaviour of the cunner, Tautogolabrus adspersus (Walbaum), in Newfoundland. Can. J. Zool. 57:247-256. 1979b. Territorial behaviour of the north temperate iabrid, Tautogolabrus adspersus. Can. J. Zool. 57:2337-2347. Rohlf, F. J., and R. R. Sokal. 1969. Statistical tables. W. H. Freeman and Company, San Francisco, CA, 253 p. Smith, I. P., N. B. Metcalfe, F. A. Huntingford, and S. Kadri. 1993. Daily and seasonal patterns in the feeding behaviour of Atlantic salmon ( Salmo salar L.) in a sea cage. Aqua- culture 117:165-178. SPSS Inc. 1990. SPSS for VAX/VMS: operations guide. SPSS Inc., 258 p. Whoriskey, F. G. 1983. Intertidal feeding and refuging by cunners, Tauto- golabrus adspersus (Labridae). Fish. Bull. 81:426- 428. 653 Abstract .—The effectiveness of a new bycatch reduction device (BRD) was tested across a wide geographical range to determine its use in the NSW oceanic prawn-trawl fishery. Using four commercial trawlers, each from a dif- ferent port located in the fishery, we compared the catches and bycatches from conventional trawls with those from trawls containing composite pan- els of netting (60 mm and 40 mm) hung on the bar and inserted into the top anterior section of the codend (termed the composite-panel codend). This panel was designed so that the 40-mm mesh 1) would allow some small fish to escape and 2) would distribute the load anterior and lateral to the 60-mm mesh (which was located in an area where waterflow was thought to be greatest), allowing the 60-mm mesh to remain open and thus facilitate the removal of larger fish. Simultaneous comparisons against a control codend showed that the composite-panel codend signifi- cantly reduced the weights of discarded bycatch at all four locations (means re- duced by 23.5% to 41%) and the num- bers of juveniles of commercially impor- tant species, such as whiting, Sillago sp. (by up to 70%). At three of the loca- tions the composite-panel significantly increased the catches of the prawn Penaeus plebejus (5.5% to 14%) and, although not statistically significant, showed a similar trend at the fourth location (mean increase of 4%). As a result of this study, the composite-panel codend has been adopted and voluntar- ily used by fishermen throughout the New South Wales oceanic prawn-trawl fishery. Manuscript accepted 4 April 1997. Fishery Bulletin 95:653-664 ( 1997). The composite square-mesh panel: a modification to codends for reducing unwanted bycatch and increasing catches of prawns throughout the New South Wales oceanic prawn-trawl fishery Matt K. Broadhurst Steven J. Kennedy N.S.W Fisheries Research Institute RO. Box 21 Cronulla, New South Wales 2230, Australia E-mail address: broadhum@fisheries.nsw.gov.au In New South Wales (NSW), Aus- tralia, oceanic prawn-trawling in- volves over 300 vessels operating from 11 major ports along 1,000 km of coastline and is valued at ap- proximately A$17 million per an- num. Vessels operating in this fish- ery primarily target the eastern king prawn, Penaeus plebejus, although a significant portion of the total value in the fishery is derived from the sale of legally retained bycatch ( termed “by-product” ) — comprising several species offish, crustaceans, and cephalopods. As in the major- ity of the world’s prawn-trawl fish- eries, however, significant numbers of nontarget organisms are also cap- tured and discarded in this fishery (for reviews see Saila, 1983; Andrew and Pepperell, 1992; Alverson et ah, 1994; Kennedy, 1995). In NSW, this discarded bycatch includes indi- viduals of byproduct species that are smaller than the minimum com- mercial size and a large assemblage of noncommercial species (see Kennedy, 1995). Unwanted bycatch has been re- duced in several of the world’s prawn- trawl fisheries by means of modifica- tions to codends that contain bycatch reduction devices (BRD’s) (e.g. Wa- tson et al., 1986; Matsuoka and Kan, 1991; Isaksen et ah, 1992; Rulifson et ah, 1992; Renaud et ah, 1993; Christian and Harrington1 ). In gen- eral, these modifications have in- volved either 1) some form of rigid structure that functions by me- chanically separating larger un- wanted individuals or 2) a strategi- cally placed escape “window” made of netting that works by exploiting behavioral differences between prawns and smaller finfish. Al- though many of these modifications have proven effective in reducing bycatch from prawn trawls, some- times they have not been favored by commercial fishermen (see Kendall, 1990; Renaud et ah, 1993) because of their size (in relation to the codend), their often complex design (e.g. Mounsey et ah, 1995), and, in some cases, their failure to main- tain prawn catches at the same lev- els as conventional trawls (e.g. Rulifson et ah, 1992; Robins- Troeger et ah, 1995; Christian and Harrington1). One modification that has been successfully tested and adopted in 1 Christian, P., and D. Harrington. 1987. Loggerhead turtle, finfish and shrimp retention studies on four excluder devices (TEDs). In Proceedings of the nongame and endangered wildlife sympo- sium; 8-10 September 1987, Georgia, p. 114-127. Dep. Nat. Resources, Social Circle, GA. 654 Fishery Bulletin 95(4), 1997 several trawl fisheries in the North Atlantic involves inserting large panels of square-mesh in codends (Robertson and Stewart, 1988; Carr, 1989; Briggs, 1992; Isaksen and Valdermarsen2 ; Suuronen3 ). These studies have shown that square-mesh panels often reduce the bycatch of juvenile roundfish while retaining a large proportion of the targeted catch. In previous experiments (Broadhurst and Kennelly, 1994, 1995, 1996; Broadhurst et al., 1996), we have shown that relatively small panels of square-mesh, inserted into the top anterior sections of penaeid prawn-trawl codends, allowed large numbers of small fish to escape without any losses of prawns. In these experiments, the majority of fish were thought to have been herded together in the narrow anterior section of the codend, immediately in front of the catch (see also Wardle, 1983). This concentration of fish was thought to upset the balance of the school and to initiate a response in the fish to escape by swimming towards the sides and top of the net and out through the open square-meshes. In addition, we showed that codend circumference and differences in hydrodynamic pressure had significant effects on the rates of movement of these fish through the square-mesh panel. The reaction of prawns to these stimuli was considered to be fairly limited, given their inability to maintain an escape response to trawls (see Lochhead, 1961; Main and Sangster, 1985). In a recent experiment ( Broadhurst and Kennelly, 1996) in one location in NSW, we tested a new de- sign of codend, comprising composite panels of square-shaped mesh (referred to as the composite- panel codend), designed for and located in the codend, to take advantage of the theory discussed above. The results showed that this design was effective in re- ducing up to 40% of the total unwanted bycatch and 2 Isaksen, B., and J. W. Valdemarsen. 1986. Selectivity experi- ments with square mesh codends in bottom trawl. Int. Coun. Explor. Sea council meeting 1986/B: 28, 18 p. 3 Suuronen, P. 1990. Preliminary trials with a square mesh codend in herring trawls. Int. Coun. Explor. Sea, council meet- ing 1990/B: 28, 14 p. up to 70% of the numbers of juveniles of commer- cially important species with no significant reduc- tion in the catches of prawns and other commercially important species. Although not validated statistically, there was also some evidence to suggest that the trawls with the composite square-mesh panel retained, on average, slightly more prawns than a conventional trawl (means increased by up to 3%). This latter re- sult, in particular, led numerous local fishermen to in- stall the composite-panel voluntarily in their trawls and use it as part of normal commercial operations. To assess the performance of this design through- out the full geographic range of this fishery (encom- passing the range of fishing conditions and catches) and to promote its voluntary acceptance, our specific goals in the present study were to investigate the effectiveness of the composite-panel under normal commercial operations at four major ports along the NSW coast in 1) reducing unwanted bycatch, 2) main- taining catches of commercially important byproduct, and 3) increasing catches of prawns. Materials and methods This study was performed between December 1995 and February 1996 with four commercial vessels (see Table 1 for details) on prawn-trawl grounds offshore from four ports (Port Stephens, Southwest Rocks, Yamba, and Ballina) in New South Wales, Australia (Fig. 1). Each vessel was rigged with three Florida flyers (mesh size=42 mm) in a standard triple gear configuration (see Kennelly et al., 1993 for details), towed at 2.5 knots. Each of the identical outside nets on each vessel were rigged with zippers (no. 10 ny- lon open-ended auto-lock plastic slides) to facilitate removal and attachment of the codends. Because each of the middle nets were not rigged in exactly the same way as the outside nets, their catches were excluded from any analysis. The codends used in the study measured 58 meshes long (2.3 m) and were constructed from 40-mm mesh Table 1 Summary of vessels, trawl headline lengths, and depths trawled for each of the four ports. Trawl headline Port Vessel and (length in m) length for each net (m) Depth trawled (m) Port Stephens Fairwind (16) 16.45 75-88 Southwest Rocks Shelley-Anne (13.7) 10.97 47-53 Yamba L- Margo (15.93) 12.8 20-49 Ballina New Avalon (18.5) 14.63 29-55 Broadhurst and Kennedy: Composite square-mesh panel in codends for reducing bycatch in an Australian prawn-trawl fishery 655 Figure 1 Map of New South Wales showing locations of the four ports that were sampled (Ballina, Yamba, Southwest Rocks, and Port Stephens). netting and 48-ply twine (Fig. 2). They comprised two sections: the anterior section was 100 meshes in circumference, 33 meshes in length, and attached to a zipper; the posterior section was 150 meshes in cir- cumference and 25 meshes in length. Two designs of codend were compared. The control codend was made entirely of diamond-shaped meshes (Fig. 2 A). The second codend (termed the composite-panel codend) was similar to the control but had composite panels made of 60-mm and 40-mm netting cut on the bar and inserted into the top of the anterior section (Fig. 2B — see also Broadhurst and Kennedy, 1996). The composite-panel codend was designed so that the load was distributed anteriorly and laterally to the panel of 60-mm square-mesh, allowing this 60-mm panel to remain open. We predicted that 1) large numbers of fish would escape through this panel, located at the point where waterflow was thought to be great- est and that 2) in addition to reducing load on the 60-mm panel, the 40-mm square-mesh would also facilitate the escape of smaller fish. The two codends were compared with each other in independent, paired trials, with the two outside nets of each vessel at each location. The codends were used in normal commercial tows of 90-min duration and alternated after each shot (to eliminate biases between different trawls and sides of the vessels). Because some significant effects of a delay in haul- back (the period between slowing the vessel and en- gaging the winch to haul the trawl) were detected in a previous experiment (Broadhurst et ah, 1996), all tows were performed with no delay in haulback. The location of each tow was randomly selected from the available prawn-trawl locations that were possible under the fishing conditions. During a period of four nights at locations offshore from each of the four ports, we completed a total of 16 replicate tows (i.e. four separate paired comparisons of 16 replicate tows each throughout the fishery). After each tow, the two codends were emptied onto a partitioned tray. Prawns and all commercially im- portant species larger than the minimum legal size (retained commercials) were separated. The remain- ing bycatch (termed “discarded by-catch”) was then sorted. This included individuals of commercially important species that were smaller than the mini- mum legal size (“discarded commercials”). Data col- lected from each tow were as follows: the total weight of king prawns and a subsample (50 prawns from each codend) of their sizes (to the nearest 1-mm cara- pace length); the weight of the discarded bycatch; the weights, numbers, and sizes (to the nearest 0.5 cm) of retained and discarded commercial species; the weights and the numbers of the most commonly oc- curring noncommercial species; and the total num- bers of discarded commercial species. Several spe- cies (commercial and noncommercial) were caught in sufficient numbers to enable meaningful compari- sons (see Table 2). Data at each port for all replicates that had suffi- cient numbers of each variable (defined as >2 indi- viduals in at least 8 replicates) were analyzed with one-tailed, paired £-tests (i.e. four separate analyses). Because a previous experiment had shown that trawls with the composite-panel have the potential to retain more prawns than conventional trawls (Broadhurst and Kennedy, 1996), we tested the hy- pothesis that the composite-panel codend caught more prawns but less total bycatch than the control codend. Where analyses provided similar results for weights and numbers of taxa, only data about num- bers were included in the figures to conserve space. Size frequencies of prawns, as well as discarded stout 656 Fishery Bulletin 95(4), 1 997 whiting, red spot whiting, and retained red mullet (where there were sufficient numbers) were plotted for each port and compared with two-sample Kolmogorov-Smirnov tests (P=0.05). Results Compared with the control codend, the composite- panel codend significantly reduced the weights of discarded bycatch (means reduced from 23.5% to 41%) at all four ports and significantly increased the catches of prawns at Port Stephens, Yamba, and Ballina (means increased by 14%, 5.5%, and 6%, re- spectively) (Fig. 3, A and B; Table 3). Although not to a significant degree (4%), the composite-panel codend used at Southwest Rocks also retained, on average, more prawns than the control codend (Fig. 3A). There were no significant reductions detected in the num- bers and weights of commercial species retained by the composite-panel codend at any of the four ports (Fig. 3; Table 3). The mean numbers and weights of discarded red spot whiting and stout whiting were reduced by the composite-panel codend at all locations where there were sufficient numbers to enable meaningful analy- sis (means reduced by up to 73%) (Fig. 3, F-G; Table 3). At Port Stephens, the composite-panel codend sig- nificantly reduced the numbers and weights of dis- carded john dory (by 50% and 57%, respectively) and blackeyes (by 45%) (Fig. 3, H and M; Table 3). There was a significant reduction in the numbers and weights of flutefish at Southwest Rocks (by 37% and 34%, respectively) and in the numbers and weights of red bigeye at Yamba (by 38.5% and 44%, respec- tively) and Ballina (by 35%) (Fig. 3, L and K; Table 3). There was also a significant reduction in the numbers and weights of leatheijacket (by 17% and 31%, respec- tively) and gurnard (by 41.5%) with the composite-panel codend at Ballina (Fig. 3, J and N; Table 3). Broadhurst and Kennedy: Composite square-mesh panel in codends for reducing bycatch in an Australian prawn-trawl fishery 657 Figure 3 Differences in mean catches (+ SE) between the control and composite-panel codends at each of the four ports: the weights of (A) prawns and (B) discarded bycatch; and the num- bers of (C) retained octopus, (D) retained red mullet, (E) retained red spot whiting, (F) dis- carded red spot whiting, (G) discarded stout whiting, (H) discarded john dory, (I) discarded eastern blue spot flathead, (J) discarded leather jacket, (K) discarded red bigeye, (L) discarded flute fish, (M) discarded blackeyes, and (N) discarded gurnard. ** = significant (P<0.01); * = significant (P<0.05); PS = Port Stephens ; SWR = Southwest Rocks; Y = Yamba; and B = Ballina. Two-sample Kolmogorov-Smimov tests, comparing the size-frequency distributions for prawns, discarded red spot whiting, and retained red mullet measured from each sample at each site showed no differences in the relative size compositions of fish retained by the two codends (Figs. 4, 5, and 6C). There were no signifi- 658 Fishery Bulletin 95(4), 1 997 Figure 3 (continued} Table 2 List of species caught in sufficient quantities to permit analyses. Scientific name Common name Scientific name Common name Penaeus plebejus Octopus spp. Sepia spp. Sepioteuthis australia Ibacus sp. Pecten fumatus Upeneichthys lineates Sillago flindersi Sillago robusta Zeus faber eastern king prawn octopus cuttlefish southern calamary smooth bug scallop red mullet red spot whiting stout whiting john dory Platycephalus caeruleopunctatus Platycephalus richardsoni Centroberyx affinis Paramonacantus filicauda Priacanthus macracanthus Macrorhamphosus scolopax Apogonops anomalus Lepidotrigla argus eastern blue spot flathead tiger flathead redfish threadfin leatheijacket7 big redeye7 flute fish7 blackeye7 gurnard7 1 Denotes noncommercial species Broadhurst and Kennelly: Composite square-mesh panel in codends for reducing bycatch in an Australian prawn-trawl fishery 659 Table 3 Summaries of one-tailed paired t-tests comparing the composite-panel and control codends. pt-v = paired i -value; n = number of replicates; all weights are in kilograms, disc = discarded; ret = retained; s. calamari = southern calamari; s. bug = smooth bug; rsw = red spot whiting ; sw = stout whiting; ebs = eastern bluespot flathead; and comm. sp. = commercial species. Significant P-values are in bold; insufficient data are marked by a dash. Port Stephens Southwest Rocks Yamba Ballina pt-v P n pt-v P n pt-v P n pt-v P n Wt of prawns 2.139 0.024 16 1.366 0.090 16 2.104 0.026 16 1.963 0.034 16 Wt of disc bycatch 4.467 0.0002 16 2.930 0.0001 16 5.518 0.0001 16 8.254 0.0001 16 No. of ret octopus — — — -0.913 0.812 16 -1.959 0.964 15 0.904 0.190 16 Wt of ret octopus — — — -0.868 0.800 16 -0.298 0.615 15 -0.341 0.631 16 No. of disc octopus — — — 0.000 ® 10 — — — — — — Wt of disc octopus — — — -0.171 0.566 10 — — — — — — No. of ret cuttlefish — — — -0.324 0.624 13 — — — — — — Wt of ret cuttlefish — — — -0.434 0.664 13 — — — — — — No. of disc cuttlefish 0.631 0.272 10 0.500 0.312 16 — — — — — — Wt of disc cuttlefish 0.165 0.436 10 0.995 0.167 16 — — — — — — No. of ret s. calamari — — — 1.011 0.166 13 — — — — — — Wt of ret s. calamari — — — 1.532 0.075 13 — — — — — — No. of disc s. calamari — — — 0.703 0.248 12 — — — — — — Wt of disc s. calamari — — — 0.887 0.197 12 — — — — — — No. of ret s. bug 0.452 0.329 13 — — — — — — — — — Wt of ret s. bug 1.214 0.124 13 — — — — — — — — — No. of disc s. bug — — — -0.254 0.597 8 -0.541 0.701 15 — — — Wt of disc s. bug — — — 0.344 0.371 8 -0.593 0.718 15 — — — No. of disc scollop — — -0.377 0.644 16 -1.109 0.542 9 — — — Wt of disc scollop — — — 0.501 0.312 16 0.348 0.368 9 — — — No. of ret red mullet — — — — — — — — — -1.012 0.833 12 Wt of ret red mullet — — — — — — — — — -0.345 0.632 12 No. of ret rsw — — — 0.893 0.194 14 — — — — — — Wt of ret rsw — — — 1.270 0.113 14 — — — — — — No. of disc rsw — — — 4.911 0.0001 16 3.593 0.004 8 3.704 0.001 15 Wt of disc rsw — — — 4.574 0.0002 16 2.554 0.019 8 3.979 0.0007 15 No. of disc sw — — — — — — 2.776 0.011 10 2.958 0.005 16 Wt of disc sw — — — — — — 2.566 0.015 10 3.077 0.004 16 No. of disc john dory 2.611 0.012 12 — — — — — — — — — Wt of disc john dory 3.174 0.004 12 — — — — — — — — — No. of disc ebs — — — — — — -1.139 0.862 14 -1.037 0.842 16 Wt of disc ebs — — — — — — -0.919 0.812 14 -0.971 0.826 16 No. of ret tiger flathead -0.349 0.634 13 — — — — — — — — — Wt of ret tiger flathead -0.602 0.721 13 — — — — — — — — — No. of disc tiger flathead 1.282 0.111 14 — — — — — — — — — Wt of disc tiger flathead 1.71 0.055 14 — — — — — — — — — No. of disc redfish 0.947 0.179 15 — — — — — — — — — Wt of disc redfish -0.131 0.551 15 — — — — — — — — — No. of leatherjacket — — — — — — — — — 2.404 0.014 16 Wt of leather jacket — — — — — — — — — 2.15 0.024 16 No. of red bigeye — — — — — — 3.344 0.002 16 2.528 0.012 15 Wt of red bigeye — — — — — — 4.122 0.0004 16 2.548 0.012 15 No. of flutefish — — — 1.841 0.045 13 — — — — — — Wt of flutefish — — — 1.851 0.044 13 — — — — — — No. of blackeyes 4.364 0.0003 16 — — — — — — — — — Wt of blackeyes 5.459 0.0001 16 — — — — — — — — — No. of gurnard — — — — — — — — — 2.392 0.018 15 Wt of gurnard — — — — — — — — — 2.034 0.033 15 No. of disc comm sp 1.168 0.1306 16 -2.282 0.981 16 0.436 0.334 16 -0.674 0.744 16 cant differences detected in the size-compositions of stout whiting at Yamba (Fig. 6A); however, at Ballina, the control codend caught proportionally more small stout whiting than the composite-panel codend (Fig. 6B). Discussion This study has shown the effectiveness of square mesh panels in allowing nontarget organisms to es- 660 Fishery Bulletin 95(4), 1997 cape trawls (see also Briggs, 1992; Fonteyne and M’Rabet, 1992; Broadhurst and Kennedy, 1994, 1995, 1996; Broadhurst et al., 1996) while maintaining catches of commercially important species. By con- ducting independent experiments on different ves- sels across four ports over a range of fishing condi- tions and catches, we have also provided informa- tion on the relative performance of the composite- panel throughout the full operational range of the NSW oceanic prawn-trawl fishery and have docu- mented, for the first time, a significant increase in the catch of targeted prawns with this design. The composite-panel codend was most effective in excluding large quantities of those discarded species that are relatively fusiform and of a size small enough to pass through the square-meshes. Species such as Broadhurst and Kennelly: Composite square-mesh panel in codends for reducing bycatch in an Australian prawn-trawl fishery 661 HI Control, n = 937 > o c 0 13 cr 0) c o o 0 Q_ 25 n B 20- 15- 10- 5- 0 | Control, n = 261 □ Composite-panel, n= 166 _j£l .hlWllM Figure 5 Size-frequency distributions of discarded red spot whiting caught with control and com- posite-panel codends from (A) Southwest rocks, (B) Yamba, and (C) Ballina. blackeyes, flute fish, red bigeye, and, in particular, stout and red spot whiting, were all significantly re- duced by the composite-panel, which contributed to- wards a reduction in the mean weight of discarded bycatch at all locations from 23.5% to 41% (Fig. 3). Assuming minimal differences between the various vessels and their gear, the relative availability of these fusiform species throughout waters off New South Wales may explain the variations in the mean reductions of total discarded bycatch at each of the ports and across the fishery. For example, there were no red spot or stout whiting captured at Port Stephens (Fig. 3, E-G), and there was only a 23.5% reduction in total discarded bycatch by the compos- ite-panel at that location (Fig. 3B). In contrast, the discarded bycatch at Yamba and Ballina included large numbers of whiting and red bigeye (up to 500 fish and 1,000 fish, respectively, from each tow in the control net) (Fig. 3, E-F and K) and correspond- ingly large percentage reductions in total discarded bycatch (41% and 39.5%, respectively) (Fig. 3B). The above reductions in total discarded bycatch with the composite-panel provide a possible expla- nation for the significant increase in catches of 662 Fishery Bulletin 95(4), 1 997 [1 Control, n = 510 Length (cm) Figure 6 Size-frequency distributions of discarded stout whiting caught with control and com- posite-panel codends from (A) Yamba, (B) Ballina, and (C) size-frequency distributions of retained red mullet from Ballina. prawns at Port Stephens, Yamba, and Ballina (by 14%, 5.5%, and 6%, respectively) and for the nonsig- nificant increase of 4% at Southwest Rocks (Fig. 3 A). By reducing the amount of total discarded bycatch and therefore the weight and drag in the codend, the trawl with the composite-panel may have achieved greater spreads between the otter boards (i.e. an in- creased swept area) than did the control, thereby covering more of the seabed and capturing more prawns. These prawns were probably the same sizes as those that we sampled, because Kolmogorov- Smirnov tests failed to detect any significant differ- ences in prawn sizes between the codends for any of the ports (Fig. 4). In support of the theory discussed above, there was also an increase (although not statistically signifi- cant) in the mean numbers of retained octopus at Southwest Rocks and Yamba (by 11% and 14%, re- spectively), retained red mullet (by 17%) at Ballina, and discarded eastern blue spot flathead at Yamba and Ballina (by 19.5% and 14.5%, respectively) with the composite-panel (Fig. 3, C-D, and I; Table 1). Broadhurst and Kennelly: Composite square-mesh panel in codends for reducing bycatch in an Australian prawn-trawl fishery 663 Given the physical profile of these individuals and their large size, it is unlikely that once captured by the trawl, they would have been able to fit through the small square-meshes of the composite-panel. In a previous study (Broadhurst and Kennelly, 1996), we showed that large quantities of small individuals of long spined flathead, Platycephalus longispinis, escaped through the square-meshes in the compos- ite-panel (62% reduction compared with a conven- tional codend). Because the tiger and eastern blue spot flathead captured in the present study are physi- cally similar to this species, it may be possible to fa- cilitate their escape simply by increasing the size of mesh in the panel (assuming they display similar responses to stimuli from the trawl). Such a modifi- cation, however, would likely result in less retention of smaller individuals of commercially important species such as red spot and stout whiting (see Figs. 5 and 6, A-B), cuttlefish, and southern calamari. In addition, the composite-panel has been designed so that the load is distributed across the many bars of the 40-mm square-shaped mesh. Any major increase in this mesh size would result in the distribution of load across fewer bars, possibly altering the geometry of the codend and its overall performance. In the present study, we have shown that the com- posite-panel codend consistently increased catches of prawns over a range of operational conditions while removing large quantities of unwanted bycatch throughout the entire geographic range of the NSW oceanic prawn-trawl fishery. In another study in Australia, Robins-Troeger et al. (1995) tested a large and comparatively complex BRB (termed the “AusTED”) off northern Australia and, despite re- ports of significant losses of prawns, concluded that “the AusTEB system has the potential to be devel- oped to suit trawling conditions encountered in dif- ferent Australian prawn fisheries.” It is unlikely, however, that any design of a BRB would be accepted and endorsed by fishermen if it did not consistently maintain catches of the target species throughout the range of the fishery — as is shown to be the case in the present paper for the composite-panel codend (see also Kendall, 1990; Renaud et al., 1992). In terms of promoting a large-scale voluntary adop- tion of BRB’s, like the composite-panel described in the present paper, it is useful to provide industry not only with evidence of catch rates similar to those obtained with conventional gear but also with evi- dence of additional benefits, such as a potential for increasing duration of tows, improving quality of catches (due to less damage from bycatch in the codend), increasing savings in labor and fuel, reduc- ing sorting times, and reducing conflicts with other user groups (e.g. recreational and commercial fish- ermen targeting stocks of bycatch species). The real- ization of these incentives, along with the results from the present study, have resulted in many commercial fishermen using the composite-panel throughout the entire NSW oceanic prawn -trawl fishery. Acknowledgments This work was funded by the Australian Fishing In- dustry Research and Bevelopment Corporation (Grant No. 93/180). The authors are grateful to Gerry O’Boherty for his valuable expertise and assistance in the field, Tommy Richardson, Barry Williams, Buck Anderson, and Sparrow Castle for the use of their respective vessels, and to Chris Paterson for providing technical support. Literature cited 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. Paper 339, 233 p. Andrew, N. L., and J. G. PepperelL 1992. The by-catch of shrimp trawl fisheries. Oceanogr. Mar. Biol. Annu. Rev. 30:527-565. Briggs, R. P. 1992. An assessment of nets with a square mesh panel as a whiting conservation tool in the Irish Sea Nephrops fishery. Fish. Res. 13:133-152. Broadhurst, M. K., and S. J. Kennelly. 1994. Reducing the by-catch of juvenile fish (mulloway Argyrosomus hololepidotus) using square-mesh panels in codends in the Hawkesbury River prawn-trawl fishery, Australia. Fish. Res. 19:321-331. 1995. A trouser-trawl experiment to assess codends that exclude juvenile mulloway (Argyrosomus hololepidotus) in the Hawkesbury River prawn-trawl fishery. Mar. Fresh- water Res. 46:953-58. 1996. Effects of the circumference of codends and a new design of square-mesh panel in reducing unwanted by-catch in the New South Wales oceanic prawn-trawl fishery, Australia. Fish. Res. 27:203-214. Broadhurst, M. K., S. J. Kennelly, and G. O’Doherty. 1996. Effect of square-mesh panels in codends and of haulback delay on by-catch reduction in the oceanic prawn-trawl fish- ery of New South Wales, Australia. Fish. Bull. 94:412-422. Carr, H. A., ed. 1989. Proceedings of the square mesh workshop, held at the world symposium on fishing vessel and fishing gear design; November 1988, Department of Fisheries and Oceans, St. John’s, Newfoundland, 133 p. Massachusetts Div. Mar. Fish., 100 Cambridge St., Boston, MA. Fonteyne, R., and R. M’Rabet. 1992. Selectivity experiments on sole with diamond and square mesh codends in the Belgian coastal beam trawl fishery. Fish. Res. 13:221-233. Isaksen, B., J. W. Valdemarsen, R. B. Larsen, and L. Karlsen. 1992. Reduction of fish by-catch in shrimp trawl using a rigid separator grid in the aft belly. Fish. Res. 13:335-352. 664 Fishery Bulletin 95(4), 1997 Kendall, D. 1990. Shrimp retention characteristics of the Morrison soft TED: a selective webbing exclusion panel inserted in a shrimp trawl net. Fish. Res. 9:13-21. Kennelly, S. J. 1995. The issue of bycatch in Australia’s demersal trawl fisheries. Rev. Fish Biol. Fish. 5:213-234. Kennelly, S. J., K. J. Graham, S. S. Montgomery, N. L. Andrew, and P. A. Brett. 1993. Variance and cost-benefit analyses to determine op- timal duration of tows and levels of replication for sam- pling relative abundances of species using demersal trawling. Fish. Res. 16:51-67. Lochhead, J. H. 1961. Locomotion. In T. H. Waterman ( ed. ), The physiol- ogy of the Crustacea, p. 313-356. Academic Press, New York, NY. Main, J., and G. I. Sangster. 1985. The behavior of the Norway lobster Nephrops novegicus (L.) during trawling. Scot. Fish. Res. Rep. 34, 8 p. Matsuoka, T., and T. T. Kan. 1991. Passive exclusion of finfish by trawl efficiency device (TED) in prawn trawling in Gulf of Papua, Papua New Guinea. Nippon Suisan Gakkaishi 57(7):1321-1329. Mounsey, R. P., G. A. Baulch, and R. C. Buckworth. 1995. Development of a trawl efficiency device (TED) for Australian prawn fisheries. I: The AusTED design. Fish. Res. 22:99-105. Renaud, M., G. Gitschlag, E. Kilma, A. Shan, D. Koi, and J. Nance. 1993. Loss of shrimp by turtle excluder devices (TEDs) in coastal waters of the United Sates, North Carolina to Texas: March 1988-August 1990. Fish. Bull. 91:129-137. Robertson, J. H. B., and P. A. M. Stewart. 1988. A comparison of size selection of haddock and whit- ing by square and diamond mesh codends. J. Cons. Cons. Int. Explor. Mer. 44:148-161. Robins-Troeger, J. B., R. C. Buckworth, and M. C. L. Dredge. 1995. Development of a trawl efficiency device (TED) for Australian prawn fisheries. II: Field evaluations of the AusTED. Fish. Res. 22:107-117. Rulifson, R. A., J. D. Murray, and J. J. Bahen. 1992. Finfish catch reduction in South Atlantic shrimp trawls using three designs of by-catch reduction devices. Fisheries (17)1:9-19. Saila, S. B. 1983. Importance and assessment of discards in commer- cial fisheries. FAO Fish. Circ. 765, 62 p. Wardle, C. S. 1983. Fish reactions to towed fishing gears. In A. Mac- Donald and I. G. Priede (eds.), Experimental biology at sea, p. 167-195. Academic Press, New York, NY. Watson, J. W., J. F. Mitchell, and A. K. Shah. 1986. Trawling efficiency device: a new concept for selec- tive shrimp trawling gear. Mar. Fish. Rev. 48(l):l-9. 665 AbStr3Ct.— Young-of-the-year(YOY) bluefish, Pomatomus saltatrix, were collected during the summers of 1992 and 1993 in the Hudson River estuary with beach seine, surface trawl, and gill nets. The temporal and spatial patterns of catch-per-unit-of-effort (CPUE) and gut-fullness values were used to infer bluefish movement and feeding periods, respectively. Estimates of daily ration were made from gut-fullness values and previously published estimates of gas- tric evacuation rate. Nearshore beach- seine CPUE was highest during day collections and lowest at night. Offshore gill-net CPUE was highest during cre- puscular or night periods and lowest during day sets. Hence, YOY bluefish appear to occupy nearshore environ- ments during the day and move away from shore at night. Gut-fullness val- ues for bluefish captured with beach seines were highest at diurnal and cre- puscular periods and declined at night; however, there were indications of night feeding on some dates. The magnitude and pattern of daily ration estimates of YOY bluefish in the Hudson River estuary were similar to values mea- sured in previous studies with other methods. Interannual differences in the magnitude of daily ration were ob- served and may be a result of day-to- day variation in feeding or differences in available prey type and size. Clu- peids, striped bass, and bay anchovy were important prey in 1992, whereas striped bass, bay anchovy, and Atlan- tic silversides were the dominant prey of YOY bluefish in 1993. Improved un- derstanding of the spatial and tempo- ral patterns of bluefish feeding, as well as fine-scale temporal resolution of es- timates of bluefish consumption rates, will aid in assessing the impact ofYOY bluefish predation on fish populations within the Hudson River estuary. Manuscript accepted 28 May 1997. Fishery Bulletin 95:665-679 (1997). Movements, feeding periods, and daily ration of piscivorous young-of-the-year bluefish, Pomatomus saltatrix, in the Hudson River estuary* Jeffrey A. Buckel** David O. Conover Marine Sciences Research Center State University of New York Stony Brook, New York 1 1 794-5000 **Present address: James J. Howard Marine Sciences Laboratory National Marine Fisheries Service, NOAA 74 Magruder Road Highlands, New Jersey 07732 E-mail address: jbuckel@sh.nmfs.gov The movements of fishes are con- trolled by both biotic and abiotic phenomena. In estuaries, fish may move in relation to the availability of prey and to reduce the risk of pre- dation, as well as in response to fluc- tuations in light, tide, salinity, tem- perature, and dissolved oxygen lev- els (Miller and Dunn, 1980). Fish distribution is often controlled by the interacting effects of these fac- tors (Miller and Dunn, 1980; Gibson et al., 1996). An understanding of the factors that govern the move- ments and distributions of predators and prey is prerequisite to quantify- ing predator-prey interactions. The bluefish, Pomatomus salt- atrix, is a marine piscivorous preda- tor of circumglobal distribution. On the U.S. east coast, spawning occurs offshore over the continental shelf, but young-of-the-year (YOY) mi- grate abruptly into estuaries at ~60 mm fork length (Kendall and Wal- ford, 1979; Nyman and Conover, 1988; McBride and Conover, 1991; Juanes and Conover, 1995). Blue- fish spawned in the South Atlantic Bight in the spring (spring-spawned) are advected northward in waters associated with the Gulf Stream (Hare and Cowen, 1996) and move into New York-New Jersey estuaries in June (Nyman and Conover, 1988; McBride and Conover, 1991). A sec- ond wave of recruits made up of sum- mer-spawned fish (spawned in the Middle Atlantic Bight) appear in nearshore waters in mid- to late- summer. The habitat shift from oce- anic waters to estuarine areas co- incides with a shift from feeding that is zooplanktivorous to one that is pi- scivorous (Marks and Conover, 1993). This study is part of a larger project designed to estimate the predatory impact that YOY bluefish have on their piscine prey popula- tions in the Hudson River estuary. Young-of-the-year bluefish are known to prey on larval and juve- nile fishes in marine embayments and estuaries along the U.S. east coast (Grant, 1962; Friedland et al., 1988; Juanes et al., 1993; Creaser and Perkins, 1994; Hartman and Brandt, 1995a; Juanes and Cono- ver, 1995). In the Hudson River es- tuary, YOY bluefish prey include the young of several resource species such as striped bass, Morone sax- * Contribution 1067 of the Marine Sciences Research Center, State University of New York, Stony Brook. 666 Fishery Bulletin 95(4), 1997 Locations where beach seines and gill nets were used at Croton Point, New York, in the lower Hudson River estuary. atilis, and American shad, Alosa sapidissima (Juanes et al., 1993; 1994). Mortality caused by YOY bluefish predation may be intense given the relatively high consumption rates of this species (Juanes and Conover, 1994; Buckel et al., 1995). In order to assess the effect of YOY bluefish predation, an under- standing of the location and timing of blue- fish prey interactions, as well as accurate and fine-scale temporal measurements of bluefish consumption rates, are needed. Consumption rates of fish are measured with direct methods (laboratory- or field- based) and indirect methods. The field-based method requires measurements of gut full- ness over a diel cycle coupled with estimates of gastric evacuation rate (Elliott and Persson, 1978; Eggers, 1979). The indirect method most widely used is a bioenergetic approach that requires knowledge of the predator’s growth trajectory, physiological parameters, and environmental data (Kit- chell et al, 1977). Because all of these meth- ods have their drawbacks and their use is controversial (Hewett et al., 1991; Boisclair and Leggett, 1991), we used a combination of different techniques in order to compare methods and cross-validate results. Juanes and Conover ( 1994) and Buckel et al. (1995) measured YOY bluefish consump- tion rates in the laboratory. Steinberg (1994) estimated daily ration of Hudson River YOY bluefish with a bioenergetics modeling ap- proach. However, the only two field estimates of blue- fish consumption rates that exist were made in Great South Bay, NY (Juanes and Conover, 1994), an envi- ronment that differs from the Hudson River estuary. Here we report on the results of diel field collec- tions of YOY bluefish during the summers of 1992 and 1993 in the Hudson River estuary. These collec- tions allowed us to determine temporal and spatial (e.g. inshore vs. offshore) patterns of YOY bluefish movements and feeding. Gut-fullness values were coupled with previously determined estimates of gastric evacuation rates (Buckel and Conover, 1996) to estimate YOY bluefish daily ration. Methods Die! coHections — beach seine Spring- and summer-spawned YOY bluefish (cohorts easily identified by size) and their prey were collected from the lower Hudson River estuary in 1992 and 1993 on the north shore of Croton Point (410ll'N, 73°54'W; Fig. 1). Ten diel beach-seine collections were made: five in 1992 (16-17 July, 28-29 July, 13-14 August, 26-27 August, and 19-20 September) and five in 1993 (7-8 July, 20-21 July, 4-5 August, 18— 19 August, and 11-12 September). Collections were made every three hours beginning at 1200 h and ending at 1200 h the following day. Sampling began 30 min before and ended 30 min after each time point (e.g. sampling for the 1200 time point began at 1130 and ended at 1230). A 60 x 3 m beach seine (13-mm- mesh wings, 6-mm-mesh bag) and a 30 x 2 m beach seine (6-mm-mesh wings, 3-mm-mesh bag) were used for nearshore sampling. The 60-m seine was set by boat. A minimum of two 60-m seine hauls were made during each one-hour sampling period. Additional seine hauls were sometimes made to increase blue- fish sample size. Bluefish and samples of prey were immediately preserved in 10% formalin. Catch-per- unit-of-effort (CPUE) of bluefish was calculated as the number of fish caught per seine haul in the first two 60-m seine hauls. Potential prey were counted Buckel and Conover: Movements, feeding, and daily ration of Pomatomus saltatrix 667 from one haul of each of the 60- and 30-m seines in 1993. Temperature (thermometer), salinity (refrac- tometer), and dissolved oxygen (modified Winkler’s method) were measured at each time point. Periodic functions were used to obtain quantitative values of tide and light levels for statistical analyses. Tides at the study site are semidiurnal and the tidal ampli- tude is ~1.5 m. The state of tide (T) for any given time point was calculated from the following equation: T = cos(((27t/ 12.42) x time of day) - 0 ), where 0 = the time of high tide in radians. A value for illumination (watts/cm2) was calculated for each time point on each specific date (Kuo-Nan, 1980). The effects of light, tide, salinity, dissolved oxy- gen, and temperature on CPUE of spring-spawned bluefish were examined with a forward step-wise multiple regression. Data from 1992 and 1993 were analyzed separately and logp (x+1) transformed to remove heterogeneity of variances. Diel collections — gill net and surface trawl A 90 x 2 m (4-cm-stretch, 2-cm-square) monofilament surface gill net was used to collect spring-spawned YOY bluefish from offshore shoal areas. One end of the gill net was anchored approximately 30 m east of the northernmost tip of Croton Point at a depth of -3 m (low tide) and set parallel to the north shore of Croton Point (see Fig. 1). The nearest beach-seine site was -200 m away in the cove just south of the gill-net set location. Gill-net collections were made concurrently with diel beach-seine collections on 20- 21 July (approximate mid-set times were 1500, 2100, 0600), 4-5 August (2100, 0130, 0600), and 18-19 August (1500, 2100, 0130, 0600) in 1993. Soak times always lasted two hours. After net retrieval, blue- fish were removed and immediately frozen on dry ice. Fish from gill-net collections were not used in the calculation of daily ration. Relative abundances of bluefish were calculated as the number of fish caught per hour of soak time (CPUE). A surface trawl collection (8.2-m head-rope, 6.7-m foot-rope, 0.9-m-opening height, 2.5-cm-mesh net, 0.6-cm-mesh codend) towed between two boats was made on 15-16 July 1993. It was conducted 1) to supplement night beach-seine collections, 2) to de- termine YOY bluefish movement patterns, and 3) to estimate daily ration. Collections were made every three hours beginning at 1230 and ending at 1230 the following day. Two to three ten-minute tows were made at each time point. Tow speed was approxi- mately 4 knots. Bluefish and prey were immediately preserved in 10% formalin. Relative abundances of bluefish were calculated as the number of fish caught per trawl (CPUE). Diel collections — "movement collection" A combination gill-net and beach-seine collection was made on 11-12 August 1993 with the sole purpose of examining bluefish movement at crepuscular peri- ods. Fish from these collections were not used in the calculation of daily ration. Collections with beach seines were performed at 1200, 2400, and one hour before and after sunrise (0520 and 0720) and sunset (1900 and 2100). The temporal resolution of this sam- pling scheme with respect to sunrise and sunset was higher than that of evenly spaced intervals of beach seining described above (every three hours). Gill-net collections lasted two hours and were made through- out the diel cycle. Feeding period Values of gut fullness were used to examine the feed- ing periods of beach-seine-collected YOY bluefish. Gut-fullness values (F) were calculated as F = G/W, where G - prey wet weight; and W - bluefish wet weight (total weight minus prey wet weight; see “Diet analysis” below). Arc-sin square root and logt, (jc+1) transformations of individual gut-fullness values did not remove heteroscedasticity; therefore, the effect of time on gut- fullness values from beach-seine data (excluding 11- 12 Aug and 11-12 Sep 1993) was examined with a nonparametric Kruskal-Wallis ANOVA. If treatment effects were significant, a nonparametric multiple comparison test for unequal sample sizes (Zar, 1984) was used to compare means. Daily ration estimates Values of gut fullness for spring-spawned bluefish from beach-seine (five dates in 1992 and four dates in 1993) and surface trawl (15-16 July 1993) collec- tions were used in estimating daily ration. Daily ra- tion was also estimated for summer-spawned blue- fish captured during beach-seine collections (19-20 Sept. 1992). The Elliott and Persson ( 1978) food con- sumption model was used to estimate bluefish daily ration: 668 Fishery Bulletin 95(4), 1 997 CA,= {.Fh-Fh-e~^yRjt 1-e R,t where CAt F. and F. l2 R. t food consumption between sampling periods at time t2 and 1 7; the geometric mean gut-fullness values (back-transformed from loge (x+1)) at these time points (time points with n < 3 fish were not used in daily ration cal- culation); the exponential gastric evacuation rate; and t2-tv Daily ration was calculated by summing estimates ofCAr The method of Boisclair and Marchand (1993) and Trudel and Boisclair ( 1993) was used to estimate 95% confidence intervals for daily ration estimates. There were four steps in the analysis. First, an estimate of exponential evacuation rate (Rg) was made from the average water temperature during a given sampling period. Estimates of Re were calculated from the equation Re = 0.015e(0103T), where T = water temperature (°C ) from Buckel and Conover (1996). Periods of declining gut fullness can be used as a validation of laboratory-based gastric evacuation rates (see Parrish and Margraf, 1990) and were used for seven out of ten diel beach-seine collections with the same data analysis techniques as those described in Buckel and Conover (1996). The mean field-derived estimate of Rg for these dates was 0.24 1/h, and the labo- ratory-derived estimate was 0.20 1/h (± 0.038 SE). Second, a normal distribution of 1,000 pseudo val- ues of R were calculated as e Re*=Re + (SERexRN), where R = K = Re ~ SE RN = the pseudo value of evacuation rate; the estimated mean evacuation rate; the standard error of Re (Buckel and Conover 1996); and a random number (different for each calculation) from a normal distribution with a mean of 0 and standard devia- tion of 1. Third, a normal distribution of 1,000 pseudo values of gut fullness (F) were calculated for each time point as Ft*=Ft + (SEFt x RN), where Ff * = the pseudo value of gut fullness; Ft = the mean loge(F+l) transformed gut fullness; SEpt = the standard error of Ff; and RN = a random number (different for each calculation) from a normal distribution with a mean of 0 and standard devia- tion of 1. Fourth, Monte-Carlo simulations were used to esti- mate CAt from the above equations by randomly choos- ing values of Re*, Ft*, and Ft* from the distributions of 1,000 pseudo values (values of Ft* and Ft* were back-transformed before calculation of CAt). Simulated values of CAt were generated for each of the eight time intervals (nine sampling points; less if a time point had n< 3 fish) and summed to estimate a daily ra- tion. This calculation was repeated 1,000 times. The 2.5 and 97.5 percentiles of these daily ration esti- mates were taken as the 95% confidence intervals. Diet analysis Diets of bluefish captured with beach seines, surface trawls, and gill nets were quantified. In the labora- tory, bluefish were measured for total length (TL, ± 1.0 mm), weighed (± 0.01 g), and their stomachs were extracted. Stomach contents of bluefish were identi- fied to the lowest possible taxon, enumerated, blot- ted dry, weighed (± 0.01 g), and measured (TL, ± 1.0 mm; eye diameter, ±0.1 mm; caudal peduncle depth, ±0.1 mm). Regressions relating prey eye-diameter and caudal peduncle depth to TL were used to esti- mate prey TL from prey pieces (see Scharf et al., 1997). A reference collection of Hudson River fish species (whole fish, scales, and bones) was used to aid in identification of digested prey. Two indices were computed to describe diets (see Hyslop, 1980). The indices were 1 ) number of stomachs in which a taxon was found, expressed as a percentage of the total number of stomachs containing food (%F=percent frequency of occurrence), and 2) weight of taxon, ex- pressed as a percentage of the total weight of food items (%W=percent weight). Results Die! collections — beach seine A total of 1,204 spring-spawned and 64 summer- spawned bluefish were collected during diel beach- seine collections. There were five successful diel col- Buckel and Conover: Movements, feeding, and daily ration of Pomatomus saltatrix 669 lections in 1992 (571 spring- and 64 summer- spawned) and four in 1993 (633 spring-spawned fish). The sample size of bluefish from the 11-12 Septem- ber 1993 beach-seine collection was too small (n= 23) for all analyses except diet. In the forward stepwise multiple-regression analy- sis, illumination of the surface waters was the only factor that explained a significant amount of the variation in spring-spawned bluefish CPUE for both 1992 (P= 0.006) and 1993 (P<0.001). The influence of illumination on CPUE of spring-spawned bluefish was positive in both 1992 and 1993 (Fig. 2, A and B): more bluefish were captured by day than by night but daytime CPUE was more variable. The CPUE pattern was also seen with summer-spawned blue- fish (Fig. 2A). The number of prey captured in the 60- and 30-m seine hauls at each time point ranged from 1 to 1,910 in 1993. On three out of the four dates examined, the relation between numbers of prey and bluefish (from identical seine hauls) was positive; however, none of these correlations were significant. Diel collections — gill net and surface trawl A total of 154 bluefish were captured in gill-net sets on three diel collections in 1993. Mean CPUE was highest during sunset and midnight collections and lowest during afternoon and sunrise sets (Fig. 20. A total of 94 bluefish were captured during surface trawl collections on 15-16 July 1993. Bluefish sur- face trawl CPUE was highest during the day (1500) and lowest at sunset ( 2 lOOKFig. 2D). Diel collections — "movement collection" Gill nets and beach seines captured 29 and 47 blue- fish on 11-12 August 1993, respectively. Gill-net CPUE was low during midday, increased through the evening to a peak at midnight (Fig. 2E), and then declined to zero by morning. Beach-seine CPUE was high during the day and low at night: the drop and increase in CPUE corresponded with sunset and sun- rise, respectively. Feeding period Time of collection had a highly significant (Kruskal- Wallis ANOVA, P<0.001) effect on the gut-fullness values of spring-spawned bluefish in 1992 and 1993 (Figs. 3, A-F, and 4, A-E). Mean gut-fullness pat- terns from seine-collected bluefish in 1992 increased throughout the afternoon, peaked in late afternoon or evening, decreased throughout the night, and in- creased during the morning hours (Fig. 3A). LU 3 CL O Q) C <1) C/3 JZ o CD LU 3 CL O LB 3 CL O 03 ( D LU 3 CL 0 1 LU 3 O. o 0) c 0) g> 2. o “0 c m Time (h) Figure 2 Catch-per-unit-of-effort (CPUE) of bluefish, Pomatomus saltatrix, versus time of capture dur- ing 1992 and 1993 diel collections. (A) Mean 1992 beach-seine CPUE (circles; ± SE) of spring- spawned bluefish averaged over all dates of col- lection ( 16-17 July, 28-29 July, 13-14 August, 26- 27 August, and 19-20 September) and summer- spawned bluefish CPUE (squares) on 19-20 Septem- ber. (B) Mean 1993 beach-seine CPUE (± SE) of spring-spawned bluefish averaged over all dates of collection ( 7-8 July, 20-21 July, 4—5 August, and 18- 19 August). (C) Mean 1993 gill-net CPUE averaged over all dates of collection (20-21 July, 4—5 August, and 18— 19 August). (D) Surface trawl CPUE on 15- 16 July 1993. (E) CPUE of spring-spawned bluefish during the beach-seine (circles, solid line) and gill- net (squares, broken line) “movement collection” on 11-12 August 1993. The time periods from sunset to sunrise are indicated by dark horizontal bars. 670 Fishery Bulletin 95(4), 1997 Time (h) Figure 3 Gut-fullness values (mean ± SE) of spring-spawned bluefish, Pomatomus saltatrix, versus time of capture during 1992 diel beach-seine collections. (A) Mean gut fullness (± SE) averaged across all dates of collection: (B) 16-17 July, (C) 28-29 July, 0.05). Daily ration estimates Daily ration estimates for YOY spring-spawned bluefish dur- ing 1992 beach-seine collec- tions were highest on our first sampling date 16-17 July at 22.2% body weight/d (95% con- fidence interval (Cl) 13.3-32.3) and dropped to 7.3% body weight/d (1.7-13.6) by 19-20 September (Fig. 5A). Although there was a decline in daily ra- tion, these values had overlap- ping CFs and were therefore not statistically different. In 1993, daily ration values from beach-seine-captured spring-spawned bluefish were highest on 20-21 July at 14.7% body weight/d (8.5- 21.6) and lowest on 7-8 July at 10.1% body weight/d (6.7-14.0) (Fig. 5B). The diel collection made with surface trawls on 15-16 July 1993 yielded a daily ration estimate of 8.6% body weight/d (4.7-12.8). Buckel and Conover: Movements, feeding, and daily ration of Pomatomus saltatrix 671 Time (h) Time (h) Figure 4 Gut-fullness values (mean ± SE) of spring-spawned bluefish, Pomatomus saltatrix, versus time of capture during 1993 diel beach-seine collections. (A) Mean gut full- ness (± SE) averaged across all dates of collection: (B) 7-8 July, (C) 20-21 July, (D) 4-5 August, and (E) 18-19 August 1993. The time periods from sunset to sunrise are indicated by dark horizontal bars. Numbers above each gut-fullness estimate represent bluefish sample size. Upward and downward facing arrows represent the time of high and low tide, respectively. Estimates of daily ration in 1993 were not statistically dif- ferent from each other. Daily ration of summer-spawned fish on 19-20 September 1992 was 5.7% body weight/d (2. 6-9. 4). Diet analysis Diets of YOY bluefish collected with beach seines during 1992 and 1993 were dominated by fish. Fish prey represented 97- 100% of bluefish diet by weight (Table 1 and 2). In 1992, the diet ofYOY blue- fish was dominated by clupeids, moronids, and bay anchovy Anchoa mitchilli (Table 1). Clu- peids (American shad, blueback herring [. Alosa aestivalis ], ale- wife [ Alosa pseudoharengus ], and Alosa spp. — clupeids that could not be identified to spe- cies) were the dominant prey of 1992 bluefish collected by beach seine and were found in 30% of stomachs containing prey and represented 27% of bluefish prey weight (Table 1). Striped bass, white perch, Morone amer- icana, and Morone spp. (mor- onids that could not be identified to species) were found in 19% of bluefish stomachs and made up 27% of their diet by weight. Bay anchovy were an important component of the diet on 19-20 Sept, representing 47% of the diet by weight (41 %F). Atlantic silversides, Menidia menidia, and Atlantic tomcod, Micro- gadus tomcod , were found in bluefish diets during August and September. Invertebrates (zoeae, copepods, and sand shrimp) were a small compo- nent of bluefish diet (Table 1). In 1993, the diet ofYOY bluefish was dominated by striped bass, bay anchovy, and Atlantic silversides (Table 2). Striped bass was the dominant prey ofYOY bluefish in 1993 beach-seine collections, i.e. in 22% of bluefish stomachs and accounting for 37% of their diet by weight. Bay anchovy was also a major prey of bluefish in 1993 (11%F, 22% Wj, particularly during the July collections. The Atlantic silverside became an important prey in August (Table 2). As in 1992, invertebrates were a small component of bluefish diet. Striped bass ( 11 %F, 35% W) and Atlantic silversides (24%F, 21%W) were dominant prey items ofYOY bluefish captured with the gill net in 1993 (Table 3). 672 Fishery Bulletin 95(4), 1997 Clupeids and bay anchovy were also important prey items of YOY blue fish captured in the gill net. Diets of YOY bluefish captured in the surface trawl on 15- 16 July 1993 were dominated by bay anchovy (56 %F, 52 %W) and striped bass (7 %F, 20%W) (Table 3). Discussion Diei movements We found large differences in the CPUE of bluefish with the diel cycle in beach-seine, gill-net, and sur- face trawl collections. There are several mechanisms 2 C o Q_ E D V) C o O # /V, /\ JO <3- A A A) A) a a .p w ^ ■O' ^ <» O ^ O Date V A Beach seine (spring-spawned) H Beach seine (summer-spawned) ® Surface trawl (spring-spawned) Figure 5 Estimates of consumption rates versus date for spring-spawned (triangles) and summer- spawned (squares) bluefish, Pomatomus saltatrix, captured by beach seines in 1992 (A) and for spring-spawned bluefish cap- tured by beach seine (triangles) and surface trawling (circles) in 1993 (B). Bars repre- sent 95% confidence intervals. that could account for these patterns. Rountree and Able (1993) distinguished two types of diel sampling bias: 1) direct avoidance of the gear or 2) a change in fish behavior. They further divided the second bias into diel movement between habitats (into or away from the gear sampling area) and diel changes in local activity (e.g. foraging). Catch-per-unit-of-effort of YOY bluefish (both spring- and summer-spawned cohorts) was higher during day beach-seine collections than during night collections in both 1992 and 1993 (Fig. 2, A-B). Fish would more likely detect and avoid beach-seine gear during the day than at night. Additionally, we used a boat to set the seine, which helped to standardize set time so that there were probably limited avoid- ance biases between diurnal and nocturnal collec- tions due to “operator” efficiency. We therefore rule out direct avoidance of the gear (bias one) and ac- cept a diel behavioral change (bias two) as an expla- nation for low night CPUE. The surface trawl CPUE in 1993 was also highest during daylight hours (Fig. 2D). Because the pattern of CPUE in 1993 was not that expected if fish were visually avoiding the gear (bias one), we propose that a behavioral change that increases the susceptibil- ity of bluefish to the surface trawl gear during the day is most likely responsible for the pattern. Bluefish CPUE with the gill net was highest at sunset and night sets in 1993 (Fig. 2C). The pattern of gill-net CPUE was the opposite of what we saw with the 1993 beach-seine and surface trawl CPUE data. During the gill-net and beach-seine “movement collection” on 11-12 August 1993, beach-seine catches were higher an hour before sunset than an hour af- ter (Fig. 2E). The opposite pattern was seen at sun- rise. On this date, gill-net catches were low during the day and increased to a midnight peak before de- clining to zero after sunrise (Fig. 2E). We attempted to determine the direction of bluefish movement from the orientation of individual bluefish in the gill net; however, data were inconclusive. According to beach-seine, surface trawl, and gill- net collections, bluefish occupy nearshore and sur- face waters during the day and then move offshore and below surface waters at night. Although we can- not rule out avoidance of gill-net gear (bias one) as a possible explanation of low day gill-net CPUE’s, con- comitant declines in beach-seine CPUE of bluefish in nearshore areas suggest that increased gill-net catches are at least partly a result of bluefish mov- ing offshore. Support for our findings comes from field collections in other estuaries. Pristas and Trent (1977) found significantly higher catches of adult bluefish at night with monofilament and multifila- ment gill nets in shallow-water (0.7-1. 1 m), mid- Buckel and Conover: Movements, feeding, and daily ration of Pomatomus saltatrix 673 TabSe 1 Stomach contents of spring- and summer-spawned bluefish, Pomatomus saltatrix, captured during diel beach-seine collections in the Hudson River estuary in 1992. %F = frequency of occurence, %W = percent wet weight. Prey type Spring-spawned Summer- spawned Spring- spawned 16-17 July 28-29 July 13- 14 Aug. 26-27 Aug. 19-20 Sept. 19-20 Sept. Total Species Common name %F %W %F %W %F %W %F %W %F %W %F %W %F %W Anchoa mitchilli bay anchovy 13.5 11.0 5.1 1.9 3.0 0.7 10.3 1.1 41.4 47.1 19.2 18.5 15.0 18.4 Morone saxatilis striped bass 17.5 24.4 5.1 3.9 15.2 23.7 11.8 39.0 4.6 9.5 11.8 19.6 Morone americana white perch 2.4 1.8 6.8 5.6 6.1 10.6 1.5 1.7 1.2 2.6 3.4 4.8 Morone spp. 5.6 2.3 3.4 1.5 5.1 4.3 2.9 1.9 1.2 1.0 3.9 2.2 Alosa sapidissima American shad 6.4 15.6 15.3 24.1 5.1 7.3 7.4 6.6 9.2 9.7 2.1 9.3 8.0 10.5 Alosa aestivalis blueback herring 3.2 4.9 15.3 12.7 1.0 1.4 3.2 2.1 Alosa pseudoharengus alewife 3.4 7.1 1.0 3.0 0.7 1.5 Alosa spp. 21.4 22.9 25.4 26.1 18.2 17.4 17.7 8.7 10.3 5.4 14.9 26.4 18.5 13.0 Menidia menidia Atlantic silverside 2.3 2.3 7.1 8.6 13.2 13.8 2.3 5.2 2.1 7.7 4.8 6.9 Microgadus tomcod Atlantic tomcod 0.8 0.4 1.0 2.2 2.9 13.4 2.3 2.5 1.4 4.0 Other fish7 1.0 0.1 1.5 3.0 0.4 0.5 Unidentified fish remains 38.9 12.9 52.5 16.9 49.5 20.5 41.2 10.8 39.1 14.2 55.3 33.6 43.5 15.3 Total fish 98.5 99.8 99.8 99.5 97.2 95.5 98.8 Crangon spp. sand shrimp 1.5 0.2 4.6 1.0 8.5 4.0 1.1 0.2 Zoeae and copepods 0.8 0.5 1.7 <0.1 2.0 <0.1 2.9 0.3 1.2 1.3 3.6 0.7 Other2 3.4 0.2 9.1 0.2 1.5 <0.1 8.0 0.5 14.9 0.8 4.9 0.3 Total stomachs analyzed 179 83 125 85 99 64 571 Number containing prey 126 59 99 68 87 47 439 Mean bluefish size (g) (SE) 4.48 11.30 25.04 41.16 43.71 10.29 (0.18) (0.69) ( 1.27) (2.78) (2.01) (1.29) 1 “Other fish” include Atlantic menhaden, Brevoortia tyrannus, and bluefish, Pomatomus saltatrix. 2 “Other” includes vegetation, gravel, sand, and rope fibers. water (2. 2-2. 6 m), and deep-water (5. 2-5. 6 m) zones of a Florida estuary. Using a subtidal weir in a polyhaline marsh creek in New Jersey, Rountree and Able (1993) captured a significantly higher number of YOY bluefish during day sampling than during night sampling. They concluded that this CPUE pat- tern was a result of diurnal foraging or increased activity (or both). Juanes and Conover (1994) made two diel beach-seine collections in Great South Bay, NY. Although they made no comparison between night and day abundance, their mean diurnal catch was two to three times higher than their mean noc- turnal catch. These field studies confirm that blue- fish activity patterns are influenced by light and dark cycles and that this pattern exists in diverse envi- ronments beyond the Hudson River estuary. Factors that may be responsible for changes in diel activity or movements of fishes include foraging (Sciarrotta and Nelson, 1977; Wurtsbaugh and Li, 1985), reduction in predation risk (Clark and Levy, 1988; see Hobson, 1991), spawning (Conover and Kynard, 1984), and thermoregulation (Caulton, 1978; Rountree and Able 1993; Neverman and Wurtsbaugh, 1994). These factors may be interdependent. For ex- ample, Neverman and Wurtsbaugh ( 1994) found that Bear Lake (Utah-Idaho) YOY sculpin were able to digest their gut contents in a short period (overnight) by moving into warm surface waters at night. By digesting their food overnight, these fish were able to feed the following day. Clark and Levy (1988) showed that the vertical migration of juvenile sock- eye salmon in an Alaskan lake during the day could be explained as a tradeoff between foraging and pre- dation risk. For juvenile estuarine fishes, Miller and Dunn (1980) considered foraging as the primary cause of diel movements. If bluefish movements are directly related to for- aging, we might expect a strong correlation between the abundance of bluefish and their prey. Bluefish may congregate where prey density is high, or prey 674 Fishery Bulletin 95(4), 1997 Table 2 Stomach contents of spring-spawned bluefish, Pomatomus saltatrix, captured during diel beach-seine collections in the Hudson River estuary in 1993. %F = frequency of occurence, %W = percent wet weight. Prey type Date 7-8 July 20-21 July 4-5 Aug 11-12 Aug' 18-19 Aug 11-12 Sept' Total Species Common name %F %W %F %W %F 1 %W %F %W %F %W %F %W %F %W Anchoa mitchilli bay anchovy 27.9 13.5 30.2 27.3 15.1 8.9 15.4 5.9 12.2 7.8 12.5 3.2 21.7 9.9 Morone saxatilis striped bass 32.0 67.8 17.4 41.2 16.3 37.0 7.7 13.8 13.0 28.9 12.5 23.9 20.7 32.2 Alosa sapidissima American shad 2.3 8.5 1.0 1.4 0.6 2.4 Alosa spp. 2.3 5.8 5.8 5.0 2.6 0.5 4.1 4.0 2.5 3.3 Menidia menidia Atlantic silverside 1.2 1.2 1.2 2.5 16.3 23.1 41.0 69.7 22.5 35.9 12.5 21.0 11.5 32.8 Other fish2 3.5 1.7 4.1 9.0 12.5 13.2 3.4 4.3 Unidentified fish remains 40.1 14.8 48.8 21.8 40.7 10.8 25.6 7.5 45.0 12.0 75.0 37.1 42.1 13.2 Total Fish 97.3 98.6 95.0 97.4 99.0 98.4 98.0 Crangon and sand and Palaemonetes spp. grass shrimp 14.0 4.6 10.3 2.6 3.1 0.3 12.5 1.6 4.1 1.6 Zoeae and copepods 6.4 2.4 1.2 0.2 2.0 0.4 Other3 5.2 0.3 11.0 1.2 7.0 0.4 10.2 0.3 7.0 0.4 Total stomachs analyzed 248 137 119 47 129 23 703 Number containing prey 172 86 86 39 98 8 489 Mean bluefish size (g) (SE) 4.96 8.96 19.84 23.29 33.41 75.67 (0.10) (0.51) (0.94) (1.41) (1.59) (8.79) 1 Bluefish that were captured on 11-12 Aug and 11-12 September were not used to calculate daily ration. 2 “Other fish” includes killifish, Fund ulus spp., American eel, Anguilla rostrata, white perch, Morone americana, Morone spp., Atlantic menhaden, and unidentifiable sciaenids. 3 “Other” includes vegetation, gravel, sand, and rope fibers. may leave an area of high bluefish densities. How- ever, we found no evidence of a correlation between prey and predator abundance; although positive re- lations between bluefish and prey abundance were found on three out of four dates in 1993, these corre- lations were nonsignificant. Sea-surface illumination was the only environmen- tal factor describing a significant amount of the varia- tion in nearshore CPUE of YOY bluefish. Young-of- the-year fish of several shallow-water marine fishes move inshore at night (Keats, 1990; Burrows et al., 1994). We, however, observed an opposite pattern for bluefish in our study. Given that bluefish are visual predators, diurnal schooling and foraging in the nearshore zone is a possible explanation for relatively high and variable daytime beach-seine CPUE. Olla and Studholme (1972) found that adult bluefish in the laboratory had higher activity (swimming speed) and a larger schooling group size during the day than at night. The difference in diel activity was also seen in YOY bluefish and shown to be endogenous (Olla and Studholme, 1978). Olla and Marchioni ( 1968) found that photomechanical changes in the retina of YOY blue- fish were also internally controlled and thus lessened the time required for light and dark adaptation. Such diurnal rhythms would offer a selective advantage for a predator dependent on vision for prey capture. Feeding period During 1992 and 1993, gut-fullness values from blue- fish caught in beach seines were highest during day, evening, and morning collections. However, there were dates in 1993 when bluefish gut-fullness levels indicated nocturnal feeding; these dates occurred with a recent full moon (7-8 July, 4-5 August) and a new moon (20-21 July). Therefore, moonlight is not entirely responsible for the nocturnal feeding seen in 1993. In laboratory tanks, YOY bluefish are ca- pable of feeding in complete darkness (Juanes and Conover, 1994). Tide may also influence the timing of feeding; however, the timing of low and high tide had no consistent influence on peaks in gut-fullness levels (Fig. 3-4). Buckel and Conover. Movements, feeding, and daily ration of Pomatomus saltatrix 675 Table 3 Stomach contents of spring-spawned bluefish captured during surface trawl and gill-net collections in the Hudson River estuary in 1993. %F = frequency of occurence, %W = percent wet weight. Date Prey type Surface trawl Gill net 15-16 July 20 July-4 Aug 11-12 Aug 18-19 Aug Species Common name %F %W %F %W %F %W %F %W Anchoa mitchilli bay anchovy 56.3 51.7 3.0 5.9 6.3 10.1 2.6 0.6 Morone saxatilis striped bass 7.0 20.4 18.2 47.6 25.0 33.9 18.4 33.3 Morone spp. 6.1 5.1 Alosa sapidissima American shad 3.0 5.1 12.5 19.8 Alosa aestivalis blueback herring 2.6 3.5 Alosa pseudoharengus alewife 6.3 8.7 Alosa spp. 1.4 2.5 15.2 8.5 5.3 2.8 Menidia menidia Atlantic silverside 6.1 2.9 18.8 9.0 31.6 39.1 Other fish7 1.4 10.2 3.0 8.7 6.3 15.2 2.6 1.2 Unidentified fish remains 43.7 13.9 51.5 15.3 25.0 1.4 44.7 19.2 Total fish 98.7 99.2 97.9 99.7 Crangon spp. sand shrimp 3.0 0.8 12.5 2.1 2.6 0.3 Zoeae and copepods 4.2 0.5 Total stomachs analyzed 94 83 29 71 Number containing prey 71 33 16 38 Mean bluefish size (g) (SE) 6.83 (0.65) 50.93 (1.81) 55.25 (3.68) 52.95 (2.44) 1 “Other fish” are bluefish, Atlantic menhaden, unidentified sciaenid, and American eel. Declining gut-fullness values probably represent periods when fish do not feed. In 1992, these periods occurred mostly after sunset during nocturnal hours for both spring- and summer-spawned bluefish (Fig. 3, B-F). In 1993, declining gut-fullness values were more variable and followed sunset or sunrise peaks in gut fullness, or else not at all (Fig. 4, B-E). Blue- fish that Juanes and Conover (1994) captured dur- ing diel sampling showed peaks in gut-fullness val- ues during crepuscular periods and a subsequent decline in gut-fullness values and a higher percent- age of empty guts at night. Many freshwater, estuarine, and marine fish spe- cies show periodicity in their daily feeding (Helfman, 1979; Miller and Dunn, 1980; Reis and Dean, 1981; Popova and Sierra, 1985; Wurtsbaugh and Li, 1985; Nico, 1990; Jansen and Mackay, 1992). This period- icity is exhibited in fish that feed either diurnally, nocturnally, or during crepuscular periods. Young- of-the-year bluefish appear to be mainly diurnal and crepuscular foragers but are also able to feed at night. Daily ration estimates Daily ration estimates from this study are consis- tent with prior laboratory and field studies in which YOY bluefish were shown to have relatively high consumption rates for a temperate fish (Juanes and Conover, 1994; Buckel et al., 1995). Field estimates of bluefish consumption rates in the Hudson River estuary in early 1992 approached 25% body wt/d. In 1992, daily rations declined as fish grew. Largest values for consumption-rate rations occurred in mid- July (22.2%) and dropped to a low in mid-September (7.3%) (Fig. 5A). The pattern and magnitude of field estimates of consumption rate were similar to val- ues of consumption rate from laboratory-mesocosm experiments made at the same time on similar-size fish (Buckel et al., 1995). In 1993, however, the early (10.1 %) and mid-July (8.6%) estimates of daily ra- tion from 24-hour collections made with beach seines ( 7-8 July) and surface trawls (15-16 July) were lower than the mid-July beach-seine estimate in 1992 ( lb- 17 July). The last three daily ration estimates in 1993 676 Fishery Bulletin 95(4), 1 997 were similar to those estimated for similar dates in 1992 (Fig. 5, A and B). A comparison of these field consumption rate estimates with estimates from bioenergetic models (Steinberg, 1994; Hartman and Brandt, 1995b) is dealt with elsewhere (Steinberg and Conover1 ). There are several possible explanations for the relatively low estimates of daily ration in early July 1993. First, these are point estimates of feeding rate that may not reflect average daily feeding over longer periods (see Smagula and Adleman, 1982; Trudel and Boisclair, 1993). Although Trudel and Boisclair ( 1993) found low day-to-day variation (7.0-16.3%) of daily ration for minnows in a field study, Smagula and Adelman (1982) found large day-to-day variation (30— 40%) in daily ration estimates of piscivorous large- mouth bass in the laboratory. Alternatively, other factors known to affect fish consumption rates include temperature, fish size, prey availability, prey biomass, prey type and size composition, and risk of predation. Both tempera- tures and bluefish sizes were similar in 1992 and 1993 (Tables 1-4). Prey abundance was not recorded during diel collections in 1992 and thus cannot be compared, but there were differences in the types and sizes of prey consumed by bluefish in these years (Table 4). The much larger clupeid species were a dominant part of bluefish diet in 1992 but were not a dominant prey in 1993. This finding reflects riverwide relative abundance estimates from a sepa- 1 Steinberg, N. D., and D, O. Conover. 1997. Young-of-the-year bluefish ( Pomatomus saltatrix) consumption in the Hudson River estuary: a bioenergetic modeling approach. Marine Sci- ences Research Center, State University of New York, Stony Brook, NY 11794-5000. Manuscr. in prep. rate beach-seine study (Buckel, 1997). In July 1992, clupeids, striped bass, and bay anchovy represented 90% of the available forage fish. Striped bass alone represented 63% of the available forage fish at this time in 1993. Moreover, striped bass and bay anchovy were larger for a given bluefish size in 1992 than in 1993. A combination of the size and type of prey avail- able, along with the possible behavior differences between the prey (e.g. clupeids are more pelagic and less refuge oriented), may have caused the large dif- ferences in daily ration in July. We have no informa- tion on relative abundances of predators of juvenile bluefish in July 1992 and 1993 and therefore cannot discount predation risk as a potential mechanism explaining differences in daily ration in July. Diet analysis Diets of YOY bluefish in 1992 and 1993 were domi- nated by fish prey, confirming past studies in the Hudson River estuary that have shown YOY blue- fish to be largely piscivorous (Texas Instruments, 1976; Juanes et al., 1993). Diets ofYOY bluefish were dominated by important anadromous resource spe- cies: clupeids in 1992 and striped bass in 1993. Interannual variation in diet was also observed by Friedland et al. (1988) in their study ofYOY blue- fish in a New Jersey marine embayment. Because YOY bluefish are believed to be opportunistic preda- tors (Friedland et al., 1988; Juanes et al., 1993), the diet differences we observed appear to be a result of the availability of different prey types in 1992 and 1993 (see riverwide abundances above). However, our diet data may be biased because spatial coverage was limited to the Croton Point region of the Hudson River. Table 4 Percentages of fish with empty stomachs, prey types, mean prey size, and mean values of temperature and gut fullness for diel collections in 1992 and 1993 (beach-seine and the 15-16 July 1993 surface trawl collection (ST)). Prey types are C = clupeids; SB = striped bass; BA = bay anchovy; and AS = Atlantic silversides. Prey types are listed in their order of importance in bluefish diet on each date. Mean prey sizes follow the order of prey type. Year Date % with empty stomach Prey type Mean prey size (mm) Mean temp (°C) Mean gut fullness (g of prey/g of bluefish) (%) 1992 16-17 June 29.6 C, SB, BA 42.9, 32.9, 27.3 25.6 4.31 28-29 July 28.9 C 45.7 25.3 3.40 13-14 Aug 20.8 C, SB, AS 50.9, 56.1, 68.3 25.0 3.26 26-27 Aug 20.0 SB, C, AS 68, 52, 45.2 26.7 2.27 19-20 Sept 12.1 BA, C 43.2, 60.2 22.4 3.62 1993 7-8 July 30.6 SB, BA 26.3, 16.5 26.9 1.52 15-16 July (ST) 24.5 BA 20.4 25.6 1.69 20-21 July 37.2 SB, BA 39.4, 24.5 26.1 2.13 4—5 Aug 27.7 SB, AS, C, BA 52.5, 47.0, 57.3, 35.7 25.9 2.08 18-19 Aug 24.0 SB, AS, BA 68.3, 61.5, 40.4 24.8 2.10 Buckel and Conover: Movements, feeding, and daily ration of Pomatomus saltatrix 677 Bluefish collected by gill net and surface trawls had diets that were similar to those of bluefish cap- tured with beach seines (Table 1-3). This finding suggests that YOY bluefish feed nearshore and then move offshore or that prey types in offshore shoal feeding areas do not differ from prey nearshore (or both). However, bluefish captured by surface trawls in 1993 had a larger amount of bay anchovy in their diet than did bluefish captured with beach seines during the previous week. This finding may reflect increased availability of bay anchovy in offshore sur- face waters than in nearshore environments. Implications This study provides an improved understanding of the temporal and spatial patterns of bluefish feed- ing ecology in the Hudson River estuary. Knowledge of the temporal and spatial scales at which preda- tors forage is required for a variety of predator-prey studies. Densities of predator and prey at scales rel- evant to predator foraging should be used for calcu- lations of prey-type selectivity (O’Brien and Vinyard, 1974), in encounter rate models (see Brandt and Mason, 1994), in functional and numerical response calculations (Peterman and Gatto, 1978), and in cal- culations of a predator’s growth or impact (Brandt and Kirsch, 1993; Petersen, 1994). Empirical data on the diel changes in spatial over- lap of fish and their prey and the timing of foraging activity is often lacking from spatially explicit mod- els of fish feeding and growth. For example, Brandt and Kirsch ( 1993) used estimates of prey density from offshore nighttime collections. If the sampling design for estimating the densities ofYOY bluefish and their prey in the Hudson River were constrained to only offshore or night (or both), the peaks in feeding ac- tivity that occur during day and crepuscular periods in the nearshore would be missed. Clearly, a detailed understanding of the spatiotemporal movement and feeding patterns of predators and prey are necessary to produce realistic models of feeding and growth (Mason and Patrick, 1993). Acknowledgments This work would not have been possible without the help of many volunteers, including T. Anderson, J. Brown, B. Childers, B. Connelly, A. Divadeenam, A. Ehtisham, L. LeBlanc, D. Lewis, C. Lobue, A. Matthews, N. Murray, R. Pantol, A. Parrella, B. Puccio, K. Reynolds, F. Scharf, J. Schell, K. Stansfield, G. Stone, and especially F. Edwards, D. Gardella, T. Hurst, and N. Steinberg. We thank them, as well as R. Cerrato, M. Fogarty, and E. Schultz, for statistical and programming advice, and M. Wiggins, for suggestions with surface trawl methods. K. McKown and B. Young provided advice and data which aided in site location. We also acknowledge personnel from Westchester County, NY Parks and Recreation, for access to Croton Point Park and per- sonnel from Washington Irving Yacht Club for the use of their boat storage and ramp facilities. Critical reviews of the manuscript were provided by R. Cerrato, R. Cowen, M. Fogarty, G. Lopez, and N. Steinberg. This research was funded by a research grant from the Hudson River Foundation for Science and Environmental Research Inc. and by the Na- tional Oceanic and Atmospheric Administration award number NA90AA-D-SG078 to the Research Foundation of SUNY for the New York Sea Grant Institute. Literature cited Boisclair, D., and W. C. Leggett. 1991. If computers could swim or fish could be programmed: a response to comments by Hewett et al. (1991). Can. J. Fish. Aquat. Sci. 48:1337-1344. Boisclair, D., and F. Marchand. 1993. The guts to estimate fish daily ration. Can. J. Fish. Aquat. Sci. 50:1969-1975. Brandt, S. B., and J. Kirsch. 1993. Spatially explicit models of striped bass growth in the mid-Chesapeake Bay. Trans. Am. Fish. Soc. 122:845-869. Brandt, S. B., and D. M. Mason. 1994. Landscape approaches for assessing spatial patterns in fish foraging and growth. In D. J. Stouder, K. L. Fresh, and R. J. Feller (eds. ), Theory and application in fish feeding ecol- ogy, p. 211-238. Univ. South Carolina Press, Columbia, SC. Buckel, J. A. 1997. Impact of bluefish (Pomatomus saltatrix) predation on estuarine and continental shelf fishes. Ph.D. diss., State Univ. New York, Stony Brook, NY, 199 p. Buckel, J. A., and D. O. Conover. 1996. Gastric evacuation rates of piscivorous young-of-the- year bluefish. Trans. Am. Fish. Soc. 125:591-599. Buckel, J. A., N. D. Steinberg, and D. O. Conover. 1995. Effects of temperature, salinity, and fish size on growth and consumption of juvenile bluefish. J. Fish Biol. 47:696-706. Burrows, M. T., R. N. Gibson, L. Robb, and C. A. Comely. 1994. Temporal patterns of movement in juvenile flatfishes and their predators: underwater television obser- vations. J. Exp. Mar. Biol. Ecol 177:251-268. Caulton, M. S. 1978. The importance of habitat temperatures for growth in the tropical cichlid Tilapia rendalli Boulenger. J. Fish Biol. 13:99-112. Clark, C. W., and D. A. Levy. 1988. Diel vertical migrations by juvenile sockeye salmon and the antipredation window. Am. Nat. 131:271-290. Conover, D. O., and B. E. Kynard. 1984. Field and laboratory observations of spawning peri- odicity and behavior of a northern population of the Atlan- 678 Fishery Bulletin 95(4), 1 997 tic silverside, Menidia menidia (Pisces: Atherinidae). Environ. Biol. Fishes 11:161—171. Creaser, E. P., and H. C. Perkins. 1994. The distribution, food, and age of juvenile bluefish, Pomatomus saltatrix, in Maine. Fish. Bull. 92:494-508. Eggers, D. 1979. Comments on some recent methods for estimating food consumption by fish. J. Fish. Res. Board Can. 36:1018-1019. Elliott, J. M., and L. Persson. 1978. The estimation of daily rates of food consumption for fish. J. Anim. Ecol. 47:977-991. Friedland, K. D., G. C. Garman, A. J. Bejda, A. L. Studholme, and B. Olla. 1988. Interannual variation in diet and condition in juve- nile bluefish during estuarine residency. Trans. Am. Fish. Soc. 117:474-479. Gibson, R. N., L. Robb, M. T. Burrows, and A. D. Ansell. 1996. Tidal, diel and longer term changes in the distribu- tion of fishes on a Scottish sandy beach. Mar. Ecol. Prog. Ser. 130:1-17. Grant, G. C. 1962. Predation of bluefish on young Atlantic menhaden in Indian River, Delaware. Chesapeake Sci. 3:45-47. Hare, J. A., and R. K. Cowen. 1996. Transport mechanisms of larval and pelagic juvenile bluefish (Pomatomus saltatrix ) from South Atlantic Bight spawning grounds to Middle Atlantic Bight nursery habitats. Limnol. Oceanogr. 41:1264-1280. Hartman, K. J., and S. B. Brandt. 1995a. Trophic resource partitioning, diets, and growth of sympatric estuarine predators. Trans. Am. Fish. Soc. 124:520-537. 1995b. Comparative energetics and the development of bio- energetics models for sympatric estuarine piscivores. Can. J. Fish. Aquat. Sci. 52:1647-1666. Helfman, G. S. 1979. Twilight activities of yellow perch, Perea flavescens. J. Fish. Res. Board Can. 36:173-179. Hewett, S.W., C. E. Kraft, and B. L. Johnson. 1991. Consumption, growth, and allometry: a comment on Boisclair and Leggett ( 1989 a,b,c,d). Can. J. Fish. Aquat. Sci. 48:1334-1337. Hobson, E. S. 1991. Trophic relationships of fishes specialized to feed on zooplankters above coral reefs. In P. F. Sale ( ed. ), The ecology of fishes on coral reefs, p. 69-95. Academic Press, Inc., San Diego, CA. Hyslop, E. J. 1980. Stomach contents analysis — a review of methods and their applications. J. Fish Biol. 17:411-429. Jansen, W. A., and W. C. Mackay. 1992. Foraging in yellow perch, Perea flavescens'. biologi- cal and physical factors affecting diel periodicity in feed- ing, consumption, and movement. Environ. Biol. Fishes 34:287-303. Juanes, F., J. A. Buckel, and D. O. Conover. 1994. Accelerating the onset of piscivory: intersection of predator and prey phenologies. J. Fish Biol. 45(A):41-54. Juanes, F., and D. O. Conover. 1994. Rapid growth, high feeding rates, and early piscivory in young-of-the-year bluefish (Pomatomus saltatrix). Can. J. Fish. Aquat. Sci. 51:1752-1761. 1995. Size-structured piscivory: advection and the linkage between predator and prey recruitment in young-of-the- year bluefish. Mar. Ecol. Prog. Ser. 128:287-304. Juanes, F., R. E. Marks, K. A. McKown, and D. O. Conover. 1993. Predation by age-0 bluefish on age-0 anadromous fishes in the Hudson River Estuary. Trans. Am. Fish. Soc. 122:348-356. Keats, D. W. 1990. A nocturnal inshore movement of juvenile cod Ga- dus morhua L. in eastern Newfoundland. J. Exp. Mar. Biol. Ecol 139:167-173. Kendall, A. W., Jr., and L. A. Walford. 1979. Sources and distribution of bluefish, Pomatomus saltatrix, larvae and juveniles off the east coast of the United States. Fish. Bull. 77:213-227. Kitchell, J. F., D. J. Stewart, and D. Weininger. 1977. Application of a bioenergetics model to yellow perch ( Perea flavescens) and walleye (Stizostedion vitreum vitreum). J. Fish. Res. Board Can. 34:1922-1935. Kuo-Nan, L. 1980. An introduction to atmospheric radiation. Academic Press, Inc., New York, NY, 392 p. Marks, R. E., and D. O. Conover. 1993. Ontogenetic shift in the diet of young-of-year blue- fish Pomatomus saltatrix during the oceanic phase of the early life history. Fish. Bull. 91:97-106. Mason, D. M., and E. V. Patrick. 1993. A model for the space-time dependence of feeding for pelagic fish populations. Trans. Am. Fish. Soc. 122: 884-901. McBride, R. S., and D. O. Conover. 1991. Recruitment of young-of-the-year bluefish Poma- tomus saltatrix to the New York Bight: Variation in abun- dance and growth of spring- and summer-spawned cohorts. Mar. Ecol. Prog. Ser. 78:205-216. Miller, J. M., and M. L. Dunn. 1980. Feeding strategies and patterns of movement in ju- venile estuarine fishes. In V. S. Kennedy (ed.), Estuarine perspectives, p. 437-448. Academic Press, New York, NY. Neverman, D., and W. A. Wurtsbaugh. 1994. The thermoregulatory function of diel vertical migra- tion for a juvenile fish, Cottus extensus. Oecologia 98: 247-256. Nico, L. G. 1990. Feeding chronology of juvenile piranhas, Pygocentrus notatus, in the Venezuelan llanos. Environ. Biol. Fishes 29:51-57. Nyman, R. M., and D. O. Conover. 1988. The relation between spawning season and the re- cruitment of young-of-the-year bluefish, Pomatomus saltatrix, to New York. Fish. Bull. 86:237-250. O’Brien, W. J., and G. L. Vinyard. 1974. Comment on the use of Ivlev’s electivity index with planktivorous fish. J. Fish. Res. Board Can. 31:1427- 1429. Olla, B. L., and W. W. Marchioni. 1968. Rhythmic movements of cones in the retina of blue- fish, Pomatomus saltatrix, held in constant darkness. Biol. Bull. 135:530-536. Olla, B. L., and A. L. Studholme. 1972. Daily and seasonal rhythms of activity in the blue- fish (Pomatomus saltatrix). In H. E. Winn and B. L. Olla (ed.), Behavior of marine animals, p. 303-326. Plenum Publishing, New York, NY. 1978. Comparative aspects of the activity rhythms of tau- tog, Tautoga onitis, bluefish, Pomatomus saltatrix, and Atlantic mackerel. Scomber scombrus, as related to their Buckel and Conover: Movements, feeding, and daily ration of Pomatomus saltatrix 679 life habits. In J. E. Thorpe (ed.), Rhythmic activity of fishes, p. 131-151. Academic Press, New York, NY. Parrish, D. L., and F. J. Margraf. 1990. Gastric evacuation rates of white perch, Morone americana, determined from laboratory and field data. Environ. Biol. Fishes 29:155-158. Peterman, R. M., and M. Gatto. 1978. Estimation of functional responses of predators on juvenile salmon. J. Fish. Res. Board Can. 35:797-808. Petersen, J. H. 1994. Importance of spatial patterns in estimating preda- tion on juvenile salmonids in the Columbia River. Trans. Am. Fish. Soc. 123:924—930. Popova, O. A., and L. M. Sierra. 1985. Feeding of Black Jack, Caranx ruber, in the Gulf of Batabano on the Cuban Shelf. II: Digestion rate and daily rations. J. Ichthyol. 25:105-118. Pristas, P. J., and L. Trent. 1976. Comparisons of catches of fishes in gill nets in rela- tion to webbing material, time of day, and water depth in St. Andrew Bay, Florida. Fish. Bull. 75:103-108. Reis, R. R., and J. M. Dean. 1981. Temporal variation in the utilization of an intertidal creek by the bay anchovy ( Anchoa mitchilli). Estuaries 4:16-23. Rountree, R. A., and K. W. Able. 1993. Diel variation in decapod crustacean and fish assem- blages in New Jersey polyhaline marsh creeks. Estuarine Coastal Shelf Sci. 37:181-201. Scharf, F. S., J. A. Buckel, F. Juanes, and D. O. Conover. 1997. Estimating piscine prey sizes from partial remains: testing for shifts in foraging mode by juvenile bluefish. Environ. Biol. Fishes 49:377-388. Sciarrotta, T. C., and D. R. Nelson. 1977. Diel behavior of the blue shark, Prionace glauca, near Santa Catalina Island, California. Fish. Bull. 75:519-528. 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. Steinberg, N. S. 1994. Young-of-the-year bluefish ( Pomatomus saltatrix) consumption in the Hudson River estuary: a bioenergetics modeling approach. M.S. thesis, State Univ. New York at Stony Brook, New York, NY, p. 100. Texas Instruments Inc. 1976. Predation by bluefish in the lower Hudson River. Consolidated Edison Company, New York, NY, 26 p. Trudel, M., and D. Boisclair. 1993. An in situ evaluation of the day-to-day variation in the quantity of food consumed by fish. Can. J. Fish. Aquat. Sci. 50:2157-2165. Wurtsbaugh, W., and H. Li. 1985. Diel migrations of a zooplanktivorous fish ( Menidia beryllina) in relation to the distribution of its prey in a large eutrophic lake. Limnol. Oceanogr. 30:565-576. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, NJ, 718 p. 680 Abstract .“The herbivorous blue- banded surgeonfish, Acanthurus line- atus, was a major species harvested on coral reefs in American Samoa, ac- counting for 39% by weight of artisanal catches in 1994. Spawning occurred year-round but peaked during the aus- tral spring and summer (October-Feb- ruary). A dense pulse of recruits (0.4— 0.6 recruits/m2) settled onto the outer reef flat in March-April. Apparent sur- vival was low during the first year but increased thereafter (80%/year). The fish were strongly site-attached on a daily basis, but an estimated 60% of the adults switched territories at least once during a 3-year period, thereby negat- ing attempts to estimate mortality through attrition rates of marked indi- viduals. Estimates of fish condition changed through the year, generally paralleling seasonal changes in a suite of environmental factors. The fish grew rapidly, attaining 70-80% of their to- tal growth during their first year, fol- lowed by slow growth and long life (up to 18 years), characteristics that con- founded standard growth analyses by producing age-specific growth para- meters. Growth was best described by a two-phase von Bertalanffy growth curve for ages 0-3 (if=l.l) and ages 4— 18 (if=0. 12, Lj= 22. 1 cm), with the sepa- ration based on the age at which 50% of the population reached maturity. In- dicators of fishing pressure over a 9- year period were equivocal but did not point to significant overfishing. Manuscript accepted 4 April 1997. Fishery Bulletin 95:680-693 (1997). Population biology and harvest of the coral reef surgeonfish Acanthurus lineatus in American Samoa Peter C. Craig Department of Marine and Wildlife Resources PO Box 3730, Pago Pago, American Samoa 96799 Present address: PO Box 532 Klamath, California 95548 E-mail address: Peter Craig J. Howard Choat Lynda M. Axe Department of Marine Biology James Cook University, Townsville Queensland, 4811, Australia Suesan Saucerman Department of Marine and Wildlife Resources PO Box 3730, Pago Pago, American Samoa 96799 Coral reef fishes are harvested for food throughout the South Pacific islands (Wright, 1993; Balzell et al.1 ). In American Samoa, well over 100 species are caught in artisanal and subsistence fisheries, but bio- logical information about these spe- cies and their responses to exploi- tation is sparse. Moreover, these fish are often treated as taxonomic groupings rather than as individual species, and the information that is available often pertains to geo- graphic areas distant from and dis- similar to isolated oceanic islands such as American Samoa. The purposes of this study were to examine life history characteris- tics and harvest of one of the most abundant species caught in Samoa, the bluebanded surgeonfish, Acan- thurus lineatus. This species is one component of a multispecies subsis- tence fishery that has declined in total catch in recent years for un- clear reasons (Craig et al., 1993; Ponwith2 ; Saucerman3 ). Thus, we also examined whether overfishing accounted for declines in catches. The herbivorous A. lineatus is broadly distributed throughout the Indo-Pacific and Indian Ocean re- gions and has been the subject of several studies on behavioral ecol- ogy (Robertson et al., 1979; Robert- son and Polunin, 1981; Robertson, 1983 and 1985; Choat and Bellwood, 1985; Polunin and Klumpp, 1989; Choat, 1991; Craig, 1996). On the Great Barrier Reef of Australia, this species is long-lived; some fish live 1 Dalzell, P., T. Adams, and N. Polunin. 1995. Coastal fisheries of the South Pa- cific. Workshop on management of South Pacific Inshore Fisheries, New Caledonia, 26 June-7 July 1995. Joint Forum Fish- eries Agency — South Pacific Commission, Biol. Paper 30, 151 p. 2 Ponwith, B. 1991. The shoreline fishery of American Samoa: a 12-year com- parison. Dep. Mar. and Wildl. Res., American Samoa, Biol. Rep. Ser. 23, 51 p. 3 Saucerman, S. 1995. The inshore fish- ery of American Samoa, 1991 to 1994. Dep. Mar. and Wildl. Res., American Sa- moa, Biol. Rep. Ser. 77, 45 p. Craig et al.: Population biology and harvest of Acanthurus lineatus 681 as long as 44 years (Choat and Axe, 1996). In Sa- moa, A. lineatus occurs in high densities on coral reefs (0.4 fish/m2; Craig, 1996). It maintains feeding terri- tories in shallow waters during the daytime but spends nights in deeper-water crevices where it is harvested by spear fishermen. The fisheries American Samoa has several small-scale fisheries for nearshore and offshore fishes and invertebrates (Craig et al., 1993). In 1994, the first year when all components of these fisheries were monitored, A. lineatus ranked second only to skipjack tuna (. Kate vonus pelamis) among all species harvested, accounting for 10% of the total catch of 295 metric tons (t) (DMWR4 ). Acanthurus lineatus, a small fish averaging 18 cm fork length (FL) and 170 g, was caught in two inter- related coral reef fisheries: artisanal and subsistence harvests. Multispecies landings in these two fisher- ies were 76 and 86 t, respectively, in 1994 (Saucer- man3). The artisanal fishery consisted of 56 night- time spear divers, among whom 15 fished regularly (about 15 days per month) by free diving and scuba diving. At 28 t, A. lineatus accounted for 39% by weight of artisanal catches (Fig. 1). The subsistence fishery was more diverse: fish were captured by gill nets, throw nets, rod-and-reels, handlines, and by spear fishing; invertebrates were captured by hand picking and spearing. Many species were taken; A. lineatus 4 DMWR (Dep. Marine and Wildlife Resources), PO Box 3730, Pago Pago, American Samoa 96799. Unpubl. data. Figure 1 Catch composition (by weight) of fish families in the 1994 artisanal fishery in American Samoa. Redrawn from Saucerman (see Footnote 3 in the text). accounted for only 1-3% of subsistence catches. In both artisanal and subsistence fisheries, use of destructive fishing practices (dynamite, poison) was infrequent. Materials and methods Tutuila Island in American Samoa (14°S, 171°W) is a steep volcanic island ( 142 km2) with 55 km of fring- ing coral reef. It has two seasons, a wet summer (Oct- May) and a slightly drier and cooler season (Jun- Sep) characterized by 2.5°C cooler nearshore water temperatures and increased SE trade winds. Nearshore water temperature (taken seaward of the reef flat at 0.3 m depth) ranged from 27°C to 31°C (n= 295 daily measurements). Additional details about physical variables are presented below as they relate to changes in fish condition. Rainfall and wind data were obtained from NOAA.5 Data were collected from 1) field studies conducted by snorkeling in shallow waters (1-4 m) on the coral reefs fronting the villages of Afao, Leone, and Matu’u (Fig. 2) and from 2) market samples of the artisanal fishery. Reef flats at the study sites were narrow (100-250 m), dropped abruptly to a depth of 3-6 m, and descended gradually thereafter to 20 m. The outer reef flat inhabited by A. lineatus consisted of consolidated limestone, encrusting coralline algae, 5 NOAA (National Oceanic and Atmospheric Administration). 1994. Local climatological data, annual summary with com- parative data, Pago Pago, American Samoa. National Climatic Data Center, Asheville, North Carolina, ISSN 0198-4357, 8 p. Tutuila Island, showing locations of study sites at the villages of Afao, Leone, and Matu’u. The main study reef at Afao (see en- largement at right) shows the spawning site of A. lineatus in the outer reef channel (x), the nighttime rest region (R) for A. lineatus using the study area, and the reef flat edge (dashed line). 682 Fishery Bulletin 95(4), 1997 and only 3.5% live coral cover (Craig, 1996). Spawn- ing and nocturnal rest sites at Afao are shown in Figure 2 for those A. lineatus that maintained day- time feeding territories in the study area. Length, weight, sex, and maturity Length frequencies of fish in the artisanal catch were measured at 54 occasional intervals from 1987 to 1995. During 1991-94, fish were purchased at local markets to determine fork length (to the nearest mm), weight (to 0.1 g), sex, and maturity. Pooled monthly samples for these years consisted of 18-33 mature females, 18-49 mature males, and 9-116 immature fish, for a total of 1,139 fish. Maturity was assessed by visual inspection of gonads and by gonadosomatic index (GSI = 102 x gonad weight/ whole body weight). To determine maturity-at-size, immature fish whose sex could not be determined (13% of total sample) were assigned in equal propor- tions to numbers of identified males and females because the sex ratio of identified males and females was equal. Limited samples of rotenone-treated fish were collected at several nearshore sites in August 1990 to compare with sizes of fish taken in the artisanal fishery. Spawning Seasonal spawning patterns were determined from GSI trends and by conducting monthly visual sur- veys in the outer reef channel at the Afao site, 1993- 94. Confirmation of spawning was determined by an upward rush of fish with the production of visible milt clouds. Settlement of young onto reef At each of the three study sites, newly settled fish in five 2 x 20 m permanent transects along the outer edge of the reef flat were censused monthly, approximately one week after the new moon. A repeated-measures multivariate analysis-of-variance (MANOVA with Pillai’s trace test statistic) was used to test for signifi- cant differences in settlement time among the three sites and to accommodate autocorrelation of counts among observed times (Tabachnick and Fidell, 1989). Condition factor fCFJ We used two measures of fish condition: 1) as 105 x body weight/length3, and 2) as the weight of the paired postabdominal fat bodies found in surgeon- fishes (Fishelson et al., 1985). To examine seasonal changes, monthly mean CF values and fat body weights were compared with trends in five environ- mental factors that might affect fish growth: 1) nearshore water temperature, 2) available feeding hours, 3) calm surf conditions, 4) rainfall, and 5) daylength. Available feeding hours were calculated as the number of feeding hours per month during the fish’s daily peak feeding period (1000-1800; Craig, 1996), minus losses of feeding time during the cooler season due to earlier sunsets and increased occurrence of spring low tides that prevented access to the fish’s intertidal feeding territories. An index of calm surf conditions was calculated as the inverse of wind speed, because seasonally strong winds in- crease turbulence in the surf zone, thereby decreas- ing feeding opportunities for A. lineatus (Craig, 1996). Rainfall was used as a possible indicator of the amount of nutrient input into coastal waters that might, in turn, enhance growth of the algae that the fish eat. Similarly, daylength might affect algal pro- duction. Monthly means of these five factors were cal- culated for the years 1991-94 and presented as per- centages of the maximum monthly value that occurred during this period, which were water temperature (29.6°C), available feeding period (225 h), wind speed (27 km/h), rainfall (72 cm), and daylength (13.0 h). Growth Growth data were fitted to the von Bertalanffy growth function (VBGF): lt = L„[ 1 - where l( = length at age t; L ^ = asymptotic length; K = growth coefficient; and /0 = time when length would theoretically be zero. Two independent estimates of fish growth were made. In the first method, sagittal otoliths were used to estimate ages of 94 fish selected to span the widest size range possible (5.3-22.9 cm FL from pooled lo- cations) with the methods described by Choat and Axe (1996), who aged the same species (by including tetracycline verification) from the Great Barrier Reef. Estimates of and K were derived from a Ford- Walford regression of the age-length relation (Pauly, 1983). Estimates of t0 were made with Pauly’s (1979) empirical equation: log(-/0) = -0.392 - 0.275 log - 1.038 log K. Additionally, one of each otolith pair was weighed to 10-4 g for comparison with fish age. Craig et al.: Population biology and harvest of Acanthurus lineatus 683 In the second method, individual growth rates were calculated for a subset of the naturally marked fish described in the field mortality study (see next sec- tion). The 57 fish selected were those for whom a time series of 4-20 size estimates was available for each fish. Lengths were estimated visually under- water; a comparison of visual estimates with actual sizes of the same fish when caught by spear indi- cated that the average error was 8.1 ± 1.5% (mean and SE throughout text, n=19, 6-20 cm FL). The fish were initially selected from three general size classes according to their size at settlement onto the reef (2.5-5 cm) and approximate state of matu- rity based on dissection data (juveniles 6-14 cm, adults 15-23 cm). Sample sizes were 11 newly settled fish (monitored 1.7 ± 0.3 mo), 28 juveniles (5.2 ± 0.6 mo), and 18 adults (14.1 ± 1.0 mo). These fish were grouped into eight size classes of 2.5-cm intervals on the basis of their initial sizes. By using the mean growth rate of each size class, we calculated the time needed to grow to the next size class. These growth increments were plotted sequentially, forming a single growth curve for the population. A Gulland- Holt (1959) plot of growth increments of individual fish produced estimates of and K. Mortality Total mortality (Z), which equals natural mortality (M) plus mortality caused by fishing (F), was esti- mated by monitoring the gradual loss of 145 marked fish for three years at the Afao site and by analyzing the length and age composition of fish taken in the artisanal fishery. Field mortality Earlier work had shown that A. lineatus was highly site-attached (Craig, 1996); thus we initially assumed that a fish had died if it failed to re-occupy its territory or nearby area. Individual adults (n=45) and juveniles (n=50) were recognized by distinctive line patterns behind the eye and on the cheek. Sexual dimorphism was not apparent, thus males and females were not distinguished in the field. Newly settled fish {n= 50) were identified by a com- bination of their specific location, size, color phase,6 and line pattern when discernible. Because newly settled fish were selected on the basis of identifiabil- ity rather than first appearance in the study area, the time elapsed since settlement was not known. 6 Color phases of newly settled A. lineatus are not described in the literature. The “light” phase is that of adult coloration; the “dark” phase is light-to-dark gray (overlying a faint adult color pattern) and the caudal fin is orange (differentiated from dark newly settled Ctenochaetus striatus which have orange only on caudal fin tips). However, surveys were conducted frequently; there- fore most newly settled fish had probably arrived within the previous week or two. On average, about 35 fish were monitored at any one time; new individuals were added when others either outgrew their size class or were not relocated after three successive surveys. Small fish were in- spected at least twice each week, larger fish about once per week. All three size groups were intermixed on the outer reef flat. To calculate mortality, all fish within a size group were aligned to a common starting date. For each size class, mortality at any given time equalled 1 - (no. fish alive + no. fish outgrowing size class)/(ini- tial no. fish in that size class). This approach 1) un- derestimated mortality for newly settled fish if there had been high mortality during the first days of settlement before observations began, or 2) overesti- mated mortality if observed fish emigrated from the study area. Total mortality (Z) was calculated as the slope of the descending limb of the “catch curve” (a plot of the natural logarithm of fish remaining each year versus relative age). Annual mortality was es- timated as 1 - e~z (Ricker, 1975). Total mortality Total mortality for harvested fish was estimated in several ways: 1) length-converted catch curves (Pauly, 1983); 2) the relation between Z and mean length of fish in the catch: Z = K(Lm - Lc)/(Lc - L'), where Lc = the average length of fish greater than length L'; and L' = the size at which fish are assumed to be fully recruited to the fishery (Beverton and Holt, 1957); and 3) Hoenig’s ( 1983) empirical relation between Z and a population’s longevity: ln(Z) = 1.46-1.01 ln(*m(W), where t = maximum age. Natural mortality Natural mortality (M) was esti- mated with two empirical equations: log M = 0.007 - 0.279 log + 0.654 log if + 0.463 log T (Pauly, 1980), (1) where T = the average monthly water temperature in the study area (28.6°C), and In (M/K) = 0.30 ln(Tj - 0.22 (Longhurst and Pauly, 1987). (2) 684 Fishery Bulletin 95(4), 1 997 Biological characteristics of harvest Length-based estimates of maturity and age were calculated for artisanal catches. Additionally, in 1994-95 we measured the catches of 19 groups of 1-4 fishermen (rc=43) who had fished together, to de- termine their average catch per unit of effort (CPUE: kg/h per person). The principal method was spear fish- ing by free diving; data are presented for that gear only. Results Length, weight, sex, and maturity A complete size range of newly settled, juvenile and adult A. lineatus was present in rotenone-treated samples from shallow nearshore waters <2 m deep (Fig. 3), but large fish were underrepresented because some avoided capture. Night spear fishermen har- vested the larger fish, generally 15-21 cm FL. Males and females taken in the fishery were of similar length (t= 0.26, df=993, P=0.8) and the sex ratio was nearly equal (1 male: 1.1 females, n= 995). Length-weight relations for the sexes did not differ sig- nificantly ( ANOVA, F=0.07, P=0.79); thus all fish were pooled, including smaller unsexed fish: log weight (g) = -1.60 + 3.03 log length (FL in cm)(r2=0.99, n=l,047). The relation between FL and standard length in cm was SL=0.86(FL) - 0.38 (^=0.99, n=94). Mature fish of both sexes generally had well-de- veloped gonads (6.2 ± 0.2 g, n= 529) or gonads that appeared to be partly or wholly spawned out (1.2 ± 25 Fork length (cm) Figure 3 Sizes of A. lineatus in the artisanal fishery, all years com- bined ( 1987-95), and those collected in rotenone-treated shal- low waters. 0.1 g, n=108). Immature fish had little gonad devel- opment (0.2 ± 0.01 g, n= 502). Fish reached sexual maturity at 15-21 cm FL (Fig. 4), with males matur- ing at a slightly smaller size than females. Half of both sexes were mature at about 18 cm, i.e. at ap- proximately 4 years of age. Spawning The gonadosomatic index (GSI) was highest during October-February (Fig. 5) and was strongly corre- lated with daylength and feeding hours (Table 1). Spawning also occurred throughout the year. Dur- ing all months, groups of 50-200 fish were observed spawning at dawn in the outer portion of the outer reef channel at Afao (see Fig. 2). Additional details are provided elsewhere (Craig, in press). Settlement of young onto reef Newly settled fish (n= 575) exhibited both light (79%) and dark (21%) color phases.6 Settlement peaked in March-April with densities of 0.4-0. 6 recruits/m2 on Craig et al.: Population biology and harvest of Acanthurus lineatus 685 the outer reef flat (Fig. 6). The settlement pulse oc- curred one month earlier at Matu’u as indicated by the significant site-time interaction detected by the repeated-measures MANOVA (P=0.03, F=4.67, nu- merator df=22, denominator df=6). In previous years, similar large pulses occurred in earlier months (Nov- Mar; senior author, pers. obs.). Seasonal changes in fish condition Fish condition factor (CF) peaked in summer and declined rapidly thereafter (Fig. 7). For mature fish, a decline in CF after the spawning season was ex- pected, but a similar decline was evident among im- mature fish. Postabdominal fat bodies also declined Jun Aug Oct Dec Feb Apr Figure 5 Monthly gonadosomatic index (GSI) for mature males and females, and im- mature A. lineatus, 1991-94 combined. Also shown are months when spawn- ing was directly observed at Afao (stars). Table 1 Correlation coefficients between monthly averages for physical and biological variables: gonadosomatic index (GSI), condition factor (CF), and paired fat bodies of A. lineatus. P = probability value. Water temperature Daylength Calm surf index Rainfall Feeding hours GSI: mature fish 0.034 0.906 0.418 0.36 0.817 P>0.1 P<0.001 P>0.1 P>0.1 P<0.01 GSI: immature fish 0.347 0.129 0.252 0.335 0.311 P>0.1 P>0.1 P>0.1 P>0.1 P>0.1 CF: mature fish 0.495 0.703 0.65 0.667 0.511 P>0.1 P<0.02 P<0.05 P<0.02 P<0.1 CF: immature fish 0.846 0.546 0.925 0.706 0.343 P<0.001 P<0.1 P<0.001 P<0.02 P>0.1 Fat bodies: mature fish 0.77 0.228 0.702 0.491 0.036 P<0.01 P>0.1 P<0.02 P>0.1 P>0.1 686 Fishery Bulletin 95(4), 1 997 Figure 6 Monthly counts of newly settled A. lineatus on the outer reef flat at three sites on Tutuila Island, 1994-95. steadily from early summer through the cooler sea- son (Fig. 8). These losses resulted in CF values that were about 10% below peak summer levels and in fat bodies that were about two thirds below peak levels. Seasonal changes in five environmental factors paralleled changes in fish condition (Fig. 8). During the cooler season, nearshore water temperatures dropped below maximum summer values (-9%), as did available feeding hours (-14%), calm water con- ditions (-44%), rainfall (-76%), and daylength (-13%). Most physical factors were significantly autocorrelated (7 out of 10 comparisons), indicating that there is a distinctive seasonal signal in the nearshore environment, despite Samoa’s open-ocean location near the equator. Monthly CF values for im- mature and mature fish were significantly correlated with 3 of the 5 physical factors (Table 1). It seems clear that the fish were responding to a seasonal change, but causative factors are not known. Growth Otolith-based ages and size determinations of natu- rally marked fish provided two estimates of A. lineatus growth. Otolith weight was highly correlated with the number of annular bands in an otolith (Fig. 9). The fish grew rapidly, attaining 70-80% of their total growth by the end of their first year, and they were long-lived — up to 18 years (Fig. 10). Field estimates of fish growth confirmed the rapid early growth but underestimated later growth in comparison with the otolith-based de- terminations. Similar values of # and were derived from both the otolith-based age-length relation (Ford- Walford regression: #=0.7 , Lj= 20.3 cm, r2=0.93) and from size increments of marked fish (Gulland-Holt plot: #=0.8, Lm= 21.0 cm, ^=0.58). However, Lx and # were highly dependent on the age range of fishes examined (Fig. 11). Iterative Ford- Walford regressions of the smoothed mean size at age showed that # dropped progressively from 0.8 for the whole sample (ages 0-12) to 0.1 when juve- nile fish ages 0-3 were excluded. Asymptotic length (Lm) increased in a similarly systematic manner from about 20 cm to 22 cm. Therefore, separate VBGF growth curves were generated for the juvenile phase (ages 0-3, #=1.1, tQ =-0.2, [L^=18.3 cm]) and adult phase (ages 4-12, #=0.12, L=22. 1 cm, t0=-15.6), a separation based on the age at which 50% of the popu- lation was mature (age 4, Fig. 4). The two-phase curve captured the precocious growth and attained a larger, more realistic that approached the maximum size of fish taken in the fishery (see Fig. 3). Using the relation that longevity is approximated by 3/# (Pauly, 1983) and the adult #- value derived for ages 4-12, we predicted that the maximum age would be 25 years, which compared favorably with the observed maximum age of 18 years. Mortality Mortality indices differed among the three size classes of naturally marked fish at the Afao site (Fig. 12). Only 2% of the newly settled fish and 34% of the juveniles appeared to survive and grow into the next Craig et al.: Population biology and harvest of Acanthurus lineatus 687 Jan Mar May Jul Sep Nov Figure 7 Monthly changes in “condition factor” of juvenile and adult A. lineatus (top), and weights of postabdominal fat bodies (bottom). larger size class. At these rates, only 1% (2% x 34%) of the population would survive their first year on the reef. Thereafter numbers of marked adults de- clined rapidly at a loss of 48%/yr (Z=0.65). At these rates, the life span of combined life history stages would only be about 4-5 years. However, longevity, as revealed by otolith analysis, indicated that field mortality of marked fish was greatly overestimated. Mean ages of artisanal catches were 4-6 years (see below) and some fish lived up to 18 years. It is there- fore likely that some marked fish emigrated from the study area rather than died (see “Discussion” section). To obtain a more realistic estimate of adult mor- tality, the annual loss of fish in each age class of the 1994-95 fishery was examined by length-converted catch curves calculated in two ways: 1) after conver- sion with the VBGF parameters for adult fish (Pauly, 1983), and 2) after graphical conversion of lengths to ages based on the weighted length-age relation derived by otolith analysis. The latter was included because of the variability of the VBGF parameters shown in Figure 11. For fish that were assumed to be fully recruited to the fishery, total mortality was low in both cases (Z=0.24 and 0.23; Fig. 13), equat- 688 Fishery Bulletin 95(4), 1 997 100 80 60 40 20 £ 2 CO S' Figure 8 Relative seasonal changes in five environmental variables in the study area (see text for definitions): water temperature, feeding hours, daylength, rainfall, and calm surf. Note different y-axes. ing to an annual loss of about 20%. Simi- lar values were obtained with Hoenig’s equation (Z=0.23) and Beverton and Holts’ relation between Z and mean length of fish in the catch (Z=0.19, K= 0.12, 1^=22.! cm, Lc=19.6, L'=18 cm). Natural mortality (M) was estimated to be 0.2 and 0.45 with the empirical equations of Longhurst and Pauly ( 1987) and Pauly (1980), respectively, and with the adult values of K and Lm. The lower value indicated that M is equivalent to Z; the latter value was spurious given that it exceeded Z. Biological characteristics of harvest The flattened growth curve exhibited by older A. lineatus limited analyses based on length-converted ages, but the conver- sion did indicate that most fish taken in the fishery were relatively young (Fig. 14). During the 9-year period 1987-95, annual catches varied moderately in mean age (3. 6-6. 2 years), mean length (17.5-19.3 cm FL), proportions of immature fish taken (29-60%), and total mortal- ity (0.16-0.31 XTable 2). No trends in these variables were apparent, with one exception: the maximum size of fish decreased. Maximum sizes of fish in 1987-88, however, seem unrealistically high (Fig. 14), and in any case, mean sizes in later years (1994-95) were significantly larger than in earlier years (1987- 88)(<=9.7, df=5088, PcO.001). Although the more detailed 1994-95 market data spanned a relatively short period (Fig. 15), available data indicated no decrease in fish size (r=0.28, df=17, P>0.1) or CPUE over time (r=0.41, df=17, P=0.085), and no relation between fish size and CPUE at vari- ous island-wide fishing sites, i.e. sites with low CPUE did not have smaller fish (r=0.28, df=17 P>0.1). Figure 10 Age-length relations for A. lineatus. The solid line indi- cates a two-phase von Bertalanffy curve for ages 0-3 and 4-18; the dashed line indicates the growth of naturally marked fish based on visual estimates of fish size over time. Craig et al.: Population biology and harvest of Acanthurus lineatus 689 0 8 06 0 4 0.2 - j \ K f"" y" x i 1 ♦ ♦ 22 21 t" 20 0-12 1-12 2-12 3-12 4-12 5-12 6-12 Selected age range (yr) Figure 1 1 Changes in K and L M based on iterative Ford-Walford re- gressions that progressively excluded age classes of younger fish. Age-length data used in this calculation were derived from a smoothed line of mean size at age. Discussion The life history traits exhibited by A. lineatus are com- mon among coral reef fishes (e.g. Sale, 1991): it is a territorial fish that spawns year-round but primarily during the austral summer, its pelagic young settle onto the reef in a dense pulse and suffer high mortality, sur- vivors are sedentary but occasionally relocate to new sites, and the fish grow rapidly, have relatively low mortality rates after their first year and may live for many years. Of particular interest in this study was the opportunity to examine fishing pressure on a coral reef species. In this instance, the possiblity of emigra- tion rates of a “highly site-attached species” and the rapid initial growth pattern provide a useful context for examining the fisheries data. Additionally, because Figure II 2 Survival index (percentage of fish returning to their terri- tories) for newly settled, juvenile and adult A. lineatus at the Afao site. another data set is available for this species, we were able to compare locality-specific demographic traits. Emigration Acanthurus lineatus is strongly site-attached (adult fish have a 99.9%/day return rate to the same site: Craig, 1996), but it occasionally switches territories. After 3 years of monitoring, 6 of the 45 marked adults remained on site (Fig. 12), whereas 23 adults would have been present with an annual mortality rate of 20% as determined by catch curve. The difference (17/45) indicates that 38% of the adults that disap- peared had probably emigrated to other sites. Craig Table 2 Size, age, maturity, and mortality of A. lineatus in the artisanal fishery in American Samoa, 1987-95. n = sample size. 1987 1988 1990 1994 1995 Mean FL (cm) 18.2 17.6 18.9 19.3 18.0 SE 0.04 0.1 1.0 0.5 0.7 Maximum FL (cm) 27.7 28.9 22.4 23.0 23.0 Mean length-converted age 4.6 3.6 5.7 6.2 4.1 Percent immature 46 60 37 29 53 Total mortality (Z1) 0.22 0.23 0.19 0.23 0.25 (Z2) 0.25 0.16 0.31 0.23 0.23 n 2,499 1,126 329 720 745 Zj = Z derived from a catch curve based on length-converted ages with “adult” VBGF parameters. Z2 = Z derived from a catch curve based on graphical conversion of length to age with the age-length relation. 690 Fishery Bulletin 95(4), 1 997 Age (A, B) Figure 13 Catch curves for A. lineatus in the combined 1994- 95 fishery, based on length-converted ages deter- mined (A) graphically from the weighted age-length relationship, and (B) using VBGF parameters for adult fish (Pauly, 1983). Graph axes and Z values: (A) Ln (AO x estimated age, Z=0.23, r2=0.98; (B) Ln ( N/dt ) x relative age, Z=0.24, r2=0.95 (where N is the number of fish in a given length class and dt is the time taken to grow through that length class). ( 1996) also monitored the same marked fish described in the present study and reported that an additional 22% of the adults changed territories to nearby loca- tions and were relocated ( for the purpose of calculat- ing mortality, these fish were not of course consid- ered deaths). Altogether then, about 60% of the adults changed territories at some point during the 3-year period of observation. This analysis is not thought to be complicated by the loss of fish due to fishing mortality (F), because the 20% mortality rate incor- porated F. Further, Afao was a lightly fished area (senior author, unpubl. data). Emigration probably also accounted for the loss of many juveniles and newly settled recruits shown in Figure 12. Although the annual input of recruits to the reef was high, the observed “survival” rate of these fish during their first year ( 1%) could not maintain the standing stock of adult fish. To illustrate, the adult density of 40 fish/100m2 (Craig, 1996 ) would lose 8 fish/ 100m2 per year at an annual loss of 20%. To replace those fish with newly settled fish (with an annual in- put rate of 100 recruits/ 100m2 per year), a survival rate of 8% would be required during their first year. Growth pattern The rapid growth of young A. lineatus was so pro- nounced that initial VBGF analyses produced age- Length-converted age (yr) 10 12 14 16 18 20 22 24 26 28 Fork length (cm) Figure 14 Length-converted age frequency of the 1994-95 fish- ery (top) and a comparison of length frequencies of A. lineatus in the 1987-88 and 1994-95 fisheries (bottom). dependent estimates of Lto and K. The fish attained most of their adult size during their first year, even though the species was long-lived. There is increas- ing evidence that this growth pattern is common among coral reef fishes (Choat and Axe, 1996; Hart and Russ, 1996; Newman et al., 1996; Williams et al.7 ). Standard applications of growth models may be inappropriate for populations exhibiting these growth characteristics. Use of a two-phase von Bertalanffy growth curve (e.g. Soriano et al., 1992; Ross et al., 1995) is a possible solution, although care must be taken to establish consistent procedures for separating the two phases of the curve. 7 Williams, D., S. Newman, M. Cappo, and P. Doherty. 1995. Recent advances in the ageing of coral reef fishes. Work- shop on management of South Pacific Inshore Fisheries, New Caledonia, 26 June-7 July 1995. Joint Forum Fisheries Agency — South Pacific Comm., Biol. Paper 74, 5 p. Craig et al.: Population biology and harvest of Acanthurus lineatus 691 21 pd 19 § 17 s 15 r = 0.28 P>0.1 © o O O o O © o o 7/9 1/9 7/9 1/9 Date (mo/yr) a. . CJ 1 S 21 o u3 19 § V 1-7 s 17 15 r = 0.41 P = 0.09 o o o o o o O o ® © $ 0 © o o o © 79 1/9 7/9 1/« Date (mo/yr) r = 0.28 P>0.1 o $ L O O < > o o ° © o o © o © © CPUE (kg/h) Figure 1 5 Size and CPUE of A. lineatus in the artisanal fish- ery from varied sites around Tutuila Island, 1994-95. Comparison of Samoan and Great Barrier Reef (GBR) populations There are noteworthy differences between A. lineatus populations off Samoa and those on the GBR. Fish grow larger and live up to twice as long on the Great Barrier Reef than those off Samoa (Fig. 16); they are also more sparsely distributed and have lower re- cruitment rates of newly settled young (Choat and Bellwood, 1985; Choat and Axe, 1996; Craig, 1996; Choat, unpubl. data). Reasons for these differences are speculative and may be complicated by the ex- istence of a fishery in Samoa (see section below). Perhaps A. lineatus in Samoa is comparatively short- lived and maintains its abundance by a high annual input of newly settled young. Why there should be a greater abundance of newly settled fish in Samoa, an oceanic island, than in the extensive reef network of the GBR is unclear. Fishing pressure For at least the past 18 years, the catch composition of fish taken by night spear divers has not changed greatly, particularly with respect to the prominent catch of surgeonfishes (Fig. 17). Although Wass8 did not identify the species composition of the 1977-80 subsistence catch, local residents report that A. lineatus has always been a plentiful and popular food fish. Indicators of current levels of fishing pressure were ambiguous. Some evidence indicated that overall fishing pressure was low: 1) survival rates of fish age 1 year and older were high (80% per year), 2) estimates of total mortality and the mean size of fish in the fishery changed little over a 9-year period, 3) there was no relation between fish size and CPUE, and 4) an estimate of natural mortality was similar to that of total mortality. However, some of these points are not overly persuasive. First, estimates of natural mortality were derived from empirical equa- tions that embody considerable variability (Gulland, 1984). Second, trends based on fish size are of uncer- tain value as indicators of fishing pressure due to fish behavior. At night, when A. lineatus is harvested, there is an apparent spatial separation of small and large fish. Fish less than about 14 cm are not often encountered in the areas fished (senior author, pers. obs.), perhaps because they remain in shallower ar- eas or hide within smaller crevices during the night. Thus the larger sizes of fish taken by the spear fish- ermen represent those that were available to them, i.e. there was little opportunity for size selection. Consequently, the mean size of fish harvested could remain relatively stable under increasing levels of fish- ing pressure until there were no more fish left to catch. Indications that fishing pressure was affecting the population included 1) decreases in maximum size of fish over a 9-year period, 2) the absence of very old fish in the Samoan population compared with the GBR population, as might be expected in a fished population, and 3) a possible decrease in CPUE. These points, too, are less than compelling. First, the Wass, R. 1981. The shoreline fishery of American Samoa, past and present. In J. Munro (ed. ), Marine and coastal pro- cesses in the Pacific: ecological aspects of coastal zone manage- ment: proceedings of the UNESCO seminar at Motupore Island Research Center, 1980, p. 51-63. United Nations Education, Scientific and Cultural Organization, Paris. 692 Fishery Bulletin 95(4), 1 997 decrease in maximum size was counterbalanced by a significant increase in the mean size of the catch during the same period (Fig. 14). Second, the com- parison with GBR may not be valid, because natural longevities of the two populations may differ. The two populations are nearly 5,000 km apart and they dwell in different water temperature regimes; thus some geographic variation is likely. Third, the pos- sible CPUE decline was not statistically significant and was also compromised by occasions when fish- ermen targeted species other than A. lineatus. One 0 10 20 30 40 50 Age (yr) Figure 16 Growth rates of A. lineatus from American Samoa and the Great Barrier Reef (latter data from Choat and Axe, 1996). group of fishermen we interviewed had been told before diving that the fish buyer already had enough A. lineatus, thus some low CPUE values may indi- cate a saturated market rather than a depleted stock. Given the indefinite nature of these indicators, it remains unclear whether fishing pressure is having a significant effect on the demographics of A. lineatus, but the composite picture does not indicate substan- tial overfishing. We acknowledge, however, that lo- calized overfishing is a distinct possibility. Villagers often complained that the night spear fishermen had depleted fish stocks along their village coastline. Al- though the artisanal fishermen generally rotated areas fished to maximize their CPUE, the villagers living near the area fished might be left with a di- minished resource for a period of time. An additional concern is that an increase in fish- ing efficiency was in progress in 1995, with a con- version from free diving to scuba diving. Catch-per- unit-of-effort for scuba diving was more than twice that of free diving (3.8 vs. 10.5 kg/h for all species combined ). A further concern is that the human popu- lation of American Samoa, like that on many other South Pacific islands, is increasing rapidly; thus the demand on nearshore resources seems likely to in- crease (Craig9 ). The lack of overt signs of fishing stress for A. lineatus in the artisanal fishery is at odds with ob- served declines in the subsistence fishery on the same coral reefs. Multispecies subsistence catches dropped from 265 and 311 t in 1979 and 1991 (Ponwith2; Wass8), to 48 t 1995 (Saucerman3). Catches of A. lineatus, a minor component in this fishery, dropped from 8 t in 1991 to 1 1 in 1995. Although some of this decline may be attributed to reduced fishing effort, CPUE for most gear types de- clined as well (Saucerman3). Causes of reduced subsistence catches are not clear but may include a variety of factors such as fishing for selected species, a reduced reliance on subsistence fishing, and habitat degradation (Craig et al.10). Coral reefs in American Samoa have been severely damaged in the past 15 years by three hurricanes, an Acanthaster starfish invasion, temperature rises that resulted in mass coral bleaching, and sedimentation from land. Whether these en- 9 Craig, P. 1995. Are tropical nearshore fisheries manageable in view of projected population increases? Workshop on management of South Pacific Inshore Fish., New Caledonia, 26 June-7 July 1995. Joint Forum Fisheries Agency — South Pacific Comm., Biol. Paper 1, 6 p. 10 Craig, P, A. Green, and S. Saucerman. 1995. Coral reef troubles in American Samoa. South Pacific Comm., New Caledonia, Fish. Newsletter 72:33-34. FSgure 1 7 Comparison of catch composition (by weight) of fish families taken by subsistence night divers in 1977-80 (see Footnote 8 in the text) and artisanal night divers in 1994 (see Footnote 3 in the text). Craig et al.: Population biology and harvest of Acanthurus lineatus 693 vironmental disturbances have affected fish catches is not known, but it seems possible that these changes may have contributed to the current abundance of A. lineatus by creating expansive areas of denuded habi- tat suitable for the growth of turf algae that A. lineatus eats (Craig, 1996). As previously mentioned, live coral covered only 3.5% of the outer reef flat zone inhabited by this species, i.e. over 90% of the habitat seemed ideal for turf algae and A. lineatus. As the reefs recover, we speculate that A. lineatus may decrease in abundance and become less dominant in artisanal catches. Acknowledgments Support for this study was provided by the Federal Aid in Fish Restoration Act and an ARC major grant to J. H. C. We thank Julian Caley, Paul Dalzell, Ali- son Green, Gary Russ, and Tanielu Su’a for review comments. Literature cited Beverton, R., and S. Holt. 1957. On the dynamics of exploited fish populations. Fish Invest. Minist. Agric. Fish. Food, England, ser. II, 533 p. Choat, J. H. 1991. The biology of herbivorous fishes on coral reefs. In P. Sale (ed.), The Ecology of fishes on coral reefs, p. 120- 155. Academic Press, San Diego, CA, 574 p. Choat, J. H., and L. Axe. 1996. Growth and longevity in acanthurid fishes; an analysis of otolith increments. Mar. Ecol. Prog. Ser. 134:15-36. Choat, J. H., and D. Bellwood. 1985. Interactions amongst herbivorous fishes on a coral reef: influence of spatial variation. Mar. Biol. (Berl.) 89:221-234. Craig, P. 1996. Intertidal territoriality and time-budget of the surgeonfish, Acanthurus lineatus, in American Samoa. Environ. Biol. Fish. 46:27-36. In press. Temporal spawning patterns of several surgeon- fishes and wrasses in American Samoa. Pac. Sci. 52. Craig, P., B. Ponwith, F. Aitaoto, and D. Hamm. 1993. The commercial, subsistence, and recreational fish- eries of American Samoa. Mar. Fish. Rev. 55:109-116. Fishelson, L., W. Montgomery, and A. Myrberg. 1985. A new fat body associated with the gonads of surgeon fishes (Acanthuridae: Teleostei). Mar. Biol. 86:109-112. Gulland, J. 1984. Best estimates. Fishbyte 2:3-5. Gulland, J., and S. Holt. 1959. Estimation of growth parameters for data at unequal time intervals. J. Cons. Cons. Int. Explor. Mer 25:47-49. Hart, A., and G. Russ. 1996. Response of herbivorous fishes to crown-of- thorns starfish Acanthaster planci outbreaks. Ill: Age, growth, mortality and maturity indices of Acanthurus nigro- fuscus. Mar. Ecol. Prog. Ser. 136:25-35. Hoenig, J. 1983. Empirical use of longevity data to estimate mortal- ity rates. Fish. Bull. 82:898-903. Longhurst, A., and D. Pauly. 1987. Ecology of tropical oceans. Academic Press, Inc., New York, NY, 407 p. Newman, S., D. Williams, and G. Russ. 1996. Variability in the population structure of Lutjanus adetii (Castelnau, 1873) and L. quinquelineatus (Bloch, 1790) among reefs in the central Great Barrier Reef, Aus- tralia. Fish. Bull. 94:313-329. Pauly, D. 1979. Theory and management of tropical multispecies stocks: a review, with emphasis on the Southeast Asian demersal fisheries. ICLARM Studies and Review 1. Int. Centre for Living Aquatic Resources Management, Manila, Philippines, 35 p. 1980. On the interrelationships between natural mortality, growth parameters and mean environmental temperature in 175 fish stocks. J. Cons. Cons. Int. Explor. Mer 39:175-192. 1983. Some simple methods for the assessment of tropical fish stocks. FAO Fish. Tech. Paper 234:1-52. Polunin, N., and D. Klumpp. 1989. Ecological correlates of foraging periodicity in her- bivorous reef fishes of the Coral Sea. J. Exp. Mar. Biol. Ecol. 126:1-20. Ricker, W. 1975. Computation and interpretation of biological statistics of fish populations. Bull. Fish. Res. Board Can. 191, 382 p. Robertson, D. R. 1983. On the spawning behavior and spawning cycles of eight surgeonfishes (Acanthuridae) from the Indo- Pacific. Environ. Biol. Fishes 9:193-223. 1985. Sexual size dimorphism in surgeonfishes. Proc. Fifth Int. Coral ReefSymp. Tahiti 5:403-408. Robertson, D. R., N. Polunin, and K. Leighton. 1979. The behavioral ecology of three Indian Ocean surgeonfishes (Acanthurus lineatus, A. leucosternon and Zebrasoma scopas ): their feeding strategies, and social and mating systems. Environ. Biol. Fishes 4:125-170. Robertson, D. R., and N. Polunin. 1981. Coexistence: symbiotic sharing of feeding territories and algal food by some coral reef fishes from the Western Indian Ocean. Mar. Biol. (Berl.) 62:185-195. Ross, J., T. Stevens, and D. Vaughan. 1995. Age, growth, mortality, and reproductive biology of red drums in North Carolina waters. Trans. Am. Fish. Soc. 124:37-54. Sale, P. (ed.) 1991. The ecology of fishes on coral reefs. Academic Press, San Diego, CA, 754 p. Soriano, M., J. Moreau, J. Hoenig, and D. Pauly. 1992. New functions for the analysis of two-phase growth of juvenile and adult fishes, with application to Nile perch. Trans. Am. Fish. Soc. 121:486-493. Tabachnick, B., and L. Fidell. 1989. Using multivariate statistics. Harper and Row, New York, NY, 746 p. Wright, A. 1993. Shallow water reef-associated finfish. Chapter 6 in A. Wright and L. Hill (eds.), Nearshore marine resources of the South Pacific, p. 203-264. Inst. Pacific Studies, Fiji; Forum Fish. Agency, Honiara, Solomon Islands; and Int. Centre Ocean Development, Canada. 694 Abstract.— A total of 12,180 king mackerel, Scomberomorus cavalla , col- lected from 1986 to 1992 from North Carolina to Yucatan, Mexico, and 2,033 collected in 1977 and 1978 from North Carolina to Texas were aged with whole or sectioned sagittal otoliths. Data were analyzed by region — Atlantic Ocean, eastern Gulf of Mexico, and western Gulf — reflecting the currently recog- nized stocks. Maximum sizes of females aged were 152, 158, and 147 cm FL in the Atlantic, eastern Gulf, and western Gulf, whereas the largest males were 121, 127, and 117 cm FL in those same regions. Maximum ages from the 1986- 92 fish were 26, 21, and 24 yr for fe- males and 24, 22, and 23 yr for males in the Atlantic, eastern Gulf, and west- ern Gulf, respectively. Females grew faster and larger than males at every age in each region. A very consistent pattern of greatest growth in the east- ern Gulf, intermediate in the western Gulf, and least in the Atlantic was present each year during 1986-92, most noticeably among females. Dur- ing 1977-78, Atlantic females also had distinctly lower growth than Gulf fish. These consistent regional differences support the current hypothesis that there are three stocks as suggested by previous analyses of other types of data. Within a region and sex, growth was lower in 1977-78 than in 1986-92 in both the Atlantic and eastern Gulf, but higher for western Gulf females. Manuscript accepted 11 April 1997. Fishery Bulletin 95:694-708 (1997). Spatial and temporal variation in age and growth of king mackerel, Scomberomorus cavalla, 1 977-1 992 Douglas A. DeVries ChurchiSS B. Grimes Panama City Laboratory, Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 3500 Delwood Beach Road Panama City, Florida 32408 E-mail address (for DeVries): devries@bio.fsu.edu King mackerel, Scomberomorus cavalla, are economically valuable and highly sought after by U.S. rec- reational and commercial fisher- men from North Carolina to Texas (Manooch, 1979). They also support a substantial commercial fishery in Mexico (Gulf of Mexico and South Atlantic Fishery Management Councils1 ) Some populations have been overfished and since 1983 the species has been managed by a joint fishery management plan of the Gulf of Mexico and South Atlantic Fishery Management Councils.2 The species is managed as two stocks, an Atlantic migratory group and a Gulf migratory group, al- though the Councils recognize that there are actually two groups in the Gulf — an east and a west (Grimes et al., 1987; Johnson et al., 1994; Gulf of Mexico and South Atlantic Fishery Management Councils3). However, the paucity of data from the large Mexican fishery, which has a major impact on the western Gulf stock, precludes managing the two Gulf groups separately. Because tag return data (Sutter et al., 1991) collected during 1975-78 indicated considerable seasonal movement between the Gulf of Mexico and At- lantic Ocean, the boundary between the Gulf and Atlantic stocks was defined as the Volusia-Flagler County line off northeast Florida during November-March and the Monroe-Collier County line off southwest Florida during April- October. The Gulf stock has been heavily overfished throughout much of its management history, unlike the Atlantic stock, which has never been considered overfished (Mackerel Stock Assessment Panel4 ). 1 Gulf of Mexico and South Atlantic Fishery Management Councils. 1992. Amend- ment 6 to the fishery management plan for coastal migratory pelagics in the Gulf of Mexico and South Atlantic. Gulf of Mexico Fishery Management Council, The Com- mons at Rivergate, 3018 U.S. Highway 301 North, Suite 1000, Tampa, FL 33619-2266; and South Atlantic Fishery Management Council, Southpark Building, Suite 306, 1 Southpark Circle, Charleston, SC 29407- 4699, 35 p. 2 Gulf of Mexico and South Atlantic Fishery Management Councils. 1982. Fishery management plan, final environmental impact statement, regulatory impact re- view, final regulations for coastal migra- tory pelagic resources (mackerels) in the Gulf of Mexico and South Atlantic region. Gulf of Mexico Fishery Management Coun- cil, Tampa, FL; and South Atlantic Fish- ery Management Council, Charleston, SC, var. pagination. 3 Gulf of Mexico and South Atlantic Fishery Management Councils. 1990. Amend- ment Number 5 to the fishery management plan for the coastal migratory pelagic re- sources (mackerels), 33 p. Gulf of Mexico Fishery Management Council, Tampa, FL; and South Atlantic Fishery Management Council, Charleston, SC, 33 p. 4 Mackerel Stock Assessment Panel. 1994. 1994 report of the mackerel stock assess- ment panel. Miami Laboratory, Natl. Mar. Fish. Serv., NOAA, 75 Virginia Beach Dr., Miami, FL 33149-1003. Contrib. Rep. MIA-93/94-42. DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 695 Several studies have examined spatial, temporal, and gear-related variation in life history and fishery parameters of king mackerel. The parameters have included mean back-calculated sizes, (Beaumariage, 1973; Manooch et al., 1987), von Bertalanffy growth rates (Johnson et al., 1983; Manooch et al., 1987), and size, age, and sex composition of the catch (Beaumariage, 1973; Johnson et al., 1983; Trent et al., 1983; Trent et al., 1987). The usefulness of this information for current stock assessments is limited for several reasons. Much of the previous work was based on data collected 15 to 25 years ago when ex- ploitation was much lower, the species unmanaged, and population size, at least in the eastern Gulf of Mexico, higher. In addition, all age estimates were based on examination of whole otoliths, which re- sults in considerable under-ageing of older, larger fish (Collins et al., 1989). In addition, some studies were geographically limited (Beaumariage, 1973; Trent et al., 1987), and because stock boundaries were un- known when the data were collected, none of the data were partitioned according to stock boundaries. The primary objective of this study was to exam- ine variation in age and growth in relation to space, time, and sex of king mackerel collected during 1977- 78 and 1986-92. Methods Most king mackerel used in this study were collected during 1986-92 as part of a continuing cooperative program between the states from North Carolina to Texas and the National Marine Fisheries Service that was designed to provide age and length-frequency data needed to conduct annual stock assessments. Samples from 1977 and 1978 were collected by Johnson et al. ( 1983) for their age and growth study. All fish were measured to the nearest centimeter fork length (FL) and are reported in our study in those units. Three regions, which reflect stock boundaries ac- cording to current hypotheses (Grimes et al., 1987; Johnson et al., 1994; Gulf of Mexico and South At- lantic Fishery Management Councils3), were sampled during 1986-92. The regions were 1) Atlantic: North Carolina to about Miami, FL; 2) eastern Gulf: Florida Keys through Mississippi, and, during April-Octo- ber, Louisiana; and 3) western Gulf: Mexico, Texas, and, during November-March, Louisiana. All Loui- siana samples were collected during April-October; therefore they were classified as eastern Gulf. We did not adjust the Atlantic-eastern Gulf boundary seasonally, as the current fishery management plan does, because only 378 of the 5,490 Atlantic fish aged were collected in the area of mixing off eastern Florida during November-March. These 378 fish were used in the analyses. For the 1986-92 samples, taken from North Caro- lina to Yucatan, Mexico, we used stratified sampling (Ketchen, 1950), attempting to collect sagittal otoliths from 20 fish from each unique year, region, sex, and 10-cm size-interval combination. That quota was of- ten exceeded for abundant size intervals and not reached for rarer size intervals. The fish from Johnson et al.’s (1983) study were collected from recreational hook-and-line catches from North Carolina to Texas. Johnson et al. also used stratified sampling, with each sex and 10-cm size-interval combination comprising a stratum. For the analysis, regional classifications were the same as those used for the 1986-92 samples. In most cases, heads were shipped to our labora- tory where otoliths were removed and stored dry. The majority (>90%) of otoliths collected in the United States were taken from recreational hook-and-line catches, and the remainder from various commer- cial fisheries. All Mexican samples were collected from commercial fisheries. For the 1986-92 samples, otoliths from males <80 cm and females <90 cm were read whole. The whole otoliths were placed in a black-bottomed dish con- taining glycerin and examined with a dissecting mi- croscope at 12-25x with reflected light. For larger fish (males >80 cm and females >90 cm), three trans- verse sections about 0.7 mm thick were made about the focus with a Beuhler Isomet low-speed saw. Sec- tions were mounted on glass slides with FLO-TEXX, a clear polymer mounting medium. Sections were examined under transmitted, polarized light at 50 or 125x with a compound microscope. Annuli of whole otoliths were identified according to the criteria of Johnson et al. (1983), and sections according to the criteria of Waltz.5 The dorsal half of the section was usually read because it was clearer than the ven- tral. Otoliths collected during 1986-88 were read independently by two readers, and if there was dis- agreement, a second reading was made. If the sec- ond reading disagreed with the first, the otolith was excluded from analysis. After 1988, otoliths were read by the senior author alone. Ageing methods for the 1977 and 1978 collections were basically the same, except that we sectioned males >75 cm and females >80 cm FL and used whole ages from Johnson et al. (1983) for fish below these sizes. 5 Waltz, W. 1986. Data report on preliminary attempts to as- sess and monitor size, age, and reproductive status of king mack- erel in the south Atlantic Bight. South Carolina Wildl. Mar. Res. Dep. MARMAP rep. for contract 6-35147. 696 Fishery Bulletin 95(4), 1 997 Ages, to the nearest whole year, were assigned solely on the basis of number of visible annuli for fish collected from mid-July through December. Fish collected 1 January through mid-July had one year added to their age if the marginal increment was estimated to be >80% of the previous annual incre- ment. Ages from samples collected by Johnson et al. (1983) were adjusted similarly with their marginal increment data. This adjustment was necessary be- cause an annulus typically forms during the spring (Beaumariage, 1973; Johnson et al., 1983) but is of- ten difficult to distinguish until later in the summer. Von Bertalanffy growth equations were fitted to quarterly observed lengths-at-age by using Mar- quardt’s nonlinear regression procedure (SAS Insti- tute, Inc., 1988). Annual ages were converted to quar- terly ages by adding 0.25 to the age if the fish was collected during April-June, 0.50 if collected during July— September, and 0.75 if collected during Octo- ber-December. Quarterly ages were used to minimize the variance about the sizes-at-age because obsex’ved annual sizes-at-age, especially for young ( 1-2 yr old ) fish that are growing faster than older fish, can vary considerably depending on month of capture. For the 1986-92 data, we tested for differences in von Bertalanffy equations between sexes within re- gions and among regions within a sex, i.e. we com- pared fitted growth curves, using an F-statistic de- rived from the multivariate Hotelling’s T2 ( Bernard, 1981; Vaughan and Helser, 1990). Estimates of the parameters Lx, K, and tQ are often correlated, mak- ing univariate statistical tests inappropriate for com- paring differences between like parameters from two groups of fish (Bernard, 1981). To analyze the 1977- 78 growth data, we simply examined plots of the von Bertalanffy curves and their 95% confidence limits. We did not use Hotelling’s T2 to test the 1986-92 data for interannual differences, the 1977-78 data for any growth differences, or to compare the 1977-78 and 1986-92 data, primarily because size and age distri- butions of the samples varied considerably among regions (and to some extent between sexes) and sec- ondarily because the sample size was sometimes quite small. Von Bertalanffy parameter estimates, which are used as data for the Hotelling’s T2 test, would certainly be influenced by sample size and age distributions; if the two groups being tested had dis- similar distributions, then a significant difference might not be biologically meaningful. Bernard (1981) noted that one of the assumptions for Hotelling’s T2 is that the two sets of estimates being compared have a common variance structure. However, citing Ito and Schull (1964), “if the vari- ance-covariance matrices are unequal, the probabil- ity of a Type I error and correspondingly the power of the T2 deviate from tabulated values with the same degrees of freedom. However, when both Nx and N2 are equal, different variance-covariance matrices do not effect the error level or the power of the test.” To run each test with equal sample sizes so we could avoid the problems just mentioned, we randomly sampled from the larger group a number of observa- tions equal to the sample size of the smaller group, then used parameter estimates derived from that sample in the test. However, all growth curves shown in the figures in the present study were based on the full number of available observations. Results We aged 14,213 king mackerel — 12,180 collected during 1986-92, 2,033 from 1977-78. The numbers of females and males aged from 1986-92 were 3,407 and 2,083 from the Atlantic, 2,753 and 1,285 from the eastern Gulf, and 1,662 and 990 from the west- ern Gulf. From the 1977-78 collections, the numbers of females and males aged were 323 and 128 from the Atlantic, 1,011 and 343 from the eastern Gulf, and 188 and 40 from the western Gulf (Table 1). The geographical distribution of the 1986-92 samples varied annually, and although fish were collected off every coastal state from Virginia to Texas and in Veracruz, Campeche, and in Yucatan, Mexico, the greatest proportion were collected in North Carolina in the Atlantic region, northwest Florida in the east- ern Gulf, and south Texas in the western Gulf (Table 1). Most fish collected in 1977-78 came from North Carolina, northwest Florida, and Louisiana (Table 1). Size and age distributions Size distributions of aged fish were similar among regions during 1986-92, although females tended to predominate at larger sizes (Fig. 1). In contrast, in 1977-78, size distributions differed markedly, among both regions and sexes (Fig. 1). Males do not grow as large as females, and this difference was reflected in their narrower size distributions (Fig. 1; Table 2). Annual size distributions of aged fish, 1986-92, showed similar ranges each year but some variation in modal sizes (Table 2). The maximum sizes of fe- males aged from 1986-92 were 152 (age 18 [yr]), 158 (age 18), and 147 (age 11) cm for the Atlantic, east- ern Gulf, and western Gulf; sizes of males ranged to 121 (age 20), 127 (age 16), and 117 (age 13) cm for those same regions. Maximum sizes of 1977-78 samples were slightly smaller than those from 1986- 92 in 5 out of 6 region and sex combinations, most likely because the older data had much smaller DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 697 Table 1 Geographical distribution of aged king mackerel for 1977-78, 1986-92, and each year from 1986 to 1992. N.E. Florida = Nassau- Flagler County. E. Florida = Volusia-Palm Beach County. S.E. Florida = Broward-Dade County. S. Florida = Monroe County. S.W. Florida = Sarasota-Collier County. W. Florida = Citrus-Manatee County. N.W. Florida = Escambia-Levy County. N. Texas = Jefferson-Calhoun County. S. Texas = Aransas-Cameron County. Number aged State Females Males or Region area 77-78 86-92 86 87 88 89 90 91 92 77-78 86-92 86 87 88 89 90 91 92 Atlantic Virginia 20 16 1 3 3 1 1 1 Ocean N. Carolina 234 1,982 64 134 68 313 454 402 547 71 1,239 59 101 37 255 274 230 S. Carolina 88 568 55 99 113 70 55 78 98 56 255 31 25 50 38 38 31 42 Georgia — 292 24 5 45 98 63 13 44 — 144 5 9 15 41 35 2 37 N.E. Florida — 52 21 31 — 6 5 1 E. Florida — 459 30 56 22 5 21 171 154 — 379 63 77 74 6 14 12 133 S.E. Florida — 34 6 15 10 3 — — - — 57 14 20 23 — — — — Eastern S. Florida 6 215 5 1 2 29 36 75 67 12 132 6 2 3 27 31 24 39 Gulf S.W. Florida 5 W. Florida — 110 2 — — — — — 108 — 14 2 2 — — — — 10 N.W. Florida 532 1,209 51 94 96 227 128 321 292 283 643 7 49 22 125 79 215 146 Alabama — 303 50 172 54 7 9 3 8 — 122 22 60 29 — 9 1 1 Mississippi 7 290 31 55 51 6 69 47 31 — 104 7 8 31 4 26 12 16 Louisiana 466 628 23 61 56 46 39 195 208 48 265 22 14 12 12 2 108 95 Western Louisiana 147 1 1 2 Gulf N. Texas 41 281 — 45 90 54 14 47 31 38 181 4 27 56 30 14 23 27 S. Texas — 1,026 48 158 130 184 134 206 166 — 553 44 103 39 79 74 102 112 Veracruz — 225 — — 6 47 75 50 47 — 183 — — 9 44 31 53 46 Campeche — 21 — — — 13 8 — - — 16 — — — 7 9 — — Yucatan — 106 — — 62 — 1 37 6 — 57 — — 19 — — 24 14 samples sizes and the sampling was more limited geographically and temporally. During 1986-92, the overall age distributions of samples were quite similar among regions and be- tween sexes within regions, but during 1977-78 they varied noticeably (Fig. 2). Maximum ages of king mackerel from 1986-92 in the Atlantic, eastern Gulf, and western Gulf were 26 ( 137 cm), 21 (127-150 cm), and 24 ( 144 cm) for females and 24 ( 117 cm), 22 ( 110 cm), and 23 (101 cm) for males. Maximum ages from 1977-78 samples from the same respective regions were 20, 19, and 18 for females and 18, 19, and 19 for males. Fish older than age 20 were very rare in the 1986-92 samples — only 22 of 7,822 females (0.15%) and 13 of 4,358 males (0.18%). Growth Growth was significantly different between sexes (P<0.01 in 1986-92) in each region during 1986-92 and 1977-78, and females grew faster and larger than males at every age (Fig. 3; Table 3). Although we did not test the 1977-78 data with Hotelling’s T2, it is obvious that the confidence limits do not over- lap (Fig. 3). During 1986-92, the predicted sizes at age of females were at least 20 cm larger than males by age 13, 9, and 11 in the Atlantic, eastern Gulf, and western Gulf, respectively. Age-at-size was highly variable in all regions for both sexes, especially after fish reached 70 cm FL (Tables 4 and 5). For example, Atlantic females 100.1 to 110.0 cm FL ranged from age 4 to 20, whereas males from that same region and size ranged from age 6 to 22. The pooled 1986-92 data showed that growth was highest in the eastern Gulf, intermediate in the west- ern Gulf, and lowest in the Atlantic for both sexes, and the differences, which were greatest among females, were statistically significant (P<0.01) (Fig. 3; Table 3). Asymptotic length (LJ was the parameter most often (7 of 9 instances) responsible for the significant differ- ences between growth curves (Table 3), although twice it was t0. Estimates ofL^ were 126.7, 134.1, and 137.8 cm for Atlantic, western Gulf, and eastern Gulf females, and 96.4, 102.8, and 102.6 cm for males (Table 6). Above age 7 years, the predicted size at age of eastern Gulf 698 Fishery Bulletin 95(4), 1997 > o c 0) n cr (U 1 986-1 992 Atlantic Ocean 90 100 110 120 130 140 150 160 Eastern Gulf F ema I e s n- 2753 Males rt-1 285 40 50 60 90 100 110 120 130 140 150 160 1977-1978 Atlantic Ocean u M 1 1 F ema 1 e s n- 323 Males n - 128 40 50 60 70 80 90 100 110 120 130 140 150 160 Eastern Gulf F ema I e s 40 50 60 70 80 90 100 110 120 130 140 150 160 We stern Gulf 40 50 60 70 60 90 100 tIO 120 130 140 150 160 Fork length (cm) Figure 1 Length-frequency distribution by sex and region of king mackerel included in the analysis and collected during 1977-78 and 1986-92. females averaged 12.2 cm (SD=0.4) larger than Atlan- tic females, whereas eastern Gulf males averaged 6.9 cm (SB=0.4) larger than Atlantic males. The pattern of highest growth in the eastern Gulf, intermediate growth in the western Gulf, and low- est growth in the Atlantic seen in the pooled 1986- 92 data was very consistent and present each year during that period. These consistent regional differ- ences were especially noticeable among females (Fig. 4). Among males, the eastern Gulf growth curve was clearly higher than that for the Atlantic each year, whereas the growth curve for the western Gulf was intermediate in younger fish but converged with the eastern curve at about age 12-14 (Fig. 5). In 1977 and 1978, as during 1986-92, growth of females was lowest for Atlantic fish; however, unlike DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 699 Table 2 Annual length frequency distributions of aged king mackerel, 1986-1992, by region and sex. See Figure 1 for overall 1977-78 and 1986-92 size distributions. Number aged Size Females Males interval Region FL (cm) 86 87 88 89 90 91 92 86 87 88 89 90 91 92 Atlantic 20-39.9 — 1 — — 1 — — 3 — — — Ocean 40-59.9 23 72 7 14 14 16 17 36 50 12 5 27 20 13 60-79.9 94 83 86 129 135 193 248 102 86 91 122 115 117 239 80-99.9 57 83 91 205 230 289 315 35 88 87 195 186 107 217 100-119.9 21 58 53 127 150 116 196 3 9 9 16 34 31 26 120-139.9 5 42 21 30 57 50 67 — — — — — — 1 140-159.9 - 2 — — 7 — 3 — — — — — — — Total 200 340 258 505 594 664 846 177 233 199 341 362 275 496 Eastern 20-39.9 Gulf 40-59.9 30 26 9 93 20 58 49 7 14 1 45 22 83 23 60-79.9 50 135 94 79 99 242 267 20 69 37 58 60 128 170 80-99.9 34 60 58 65 55 124 206 31 46 53 43 53 129 96 100-119.9 30 73 55 39 66 153 133 8 11 5 22 11 20 17 120-139.9 11 72 35 35 33 56 58 — — 1 — 1 — 1 140-159.9 7 17 8 4 6 8 1 — — — — — — — Total 162 383 259 315 279 641 714 66 140 97 168 147 360 307 Western 20-39.9 6 4 7 2 Gulf 40-59.9 — 3 11 12 14 8 29 — 3 8 9 11 6 35 60-79.9 14 54 76 68 107 171 83 20 47 43 62 58 130 103 80-99.9 19 88 101 124 70 108 104 27 69 65 64 53 56 57 100-119.9 9 42 64 51 32 42 33 1 11 7 18 6 8 4 120-139.9 6 15 36 32 10 7 2 — — — — — — — 140-159.9 — 1 — 5 1 — — — — — — — — — Total 48 203 288 298 234 340 251 48 130 123 160 128 202 199 during 1986-92, western Gulf females grew faster and larger than eastern Gulf females according to the growth curves (Fig. 3). Among males, growth also appeared to be lowest in the Atlantic, although the differences were slight (Fig. 3 ). There was interannual variation in growth within a region and sex during 1986-92; however, it prob- ably reflected sample differences as much as any actual differences (Tables 1 and 2); therefore we did not test these growth curves statistically. Plots of the von Bertalanffy curves (Fig. 6) sug- gested that growth was slightly less in 1977-78 than in 1986-92 in the Atlantic and eastern Gulf for both sexes, whereas western Gulf females grew faster in 1977-78 than in 1986-92. The average differences (and standard errors) in predicted size at age between 1986-92 and 1977-78 for all ages above age 7 were 1) eastern Gulf females: +2.0 ±0.1 cm; 2) eastern Gulf males: +2.9 ±0.1 cm; 3) Atlantic females: +5.8 ± 0.1 cm; 4) Atlantic males: +1.7 ± 0.1 cm; and 5) west- ern Gulf females: -9.6 ± 0.6 cm. Discussion Our findings were based on data collected as part of a long-term, stratified, nonrandom sampling program (following the suggestions of Ketchen [1950]) de- signed to provide age-length keys for annual stock assessments. Most fish sampled were caught by hook and line and gill nets, both of which are size-selec- tive gears. Goodyear ( 1995), using computer simula- tions, demonstrated that samples equally stratified by length and those from size-selective fisheries of- ten yield biased estimates of mean size-at-age; for this reason he recommended that only simple ran- dom sampling be used to generate models of fish growth. He found that most often mean lengths-at- age were overestimated by 5-15% for all but the youngest age classes, which were sometimes under- estimated. Goodyear explained that at the youngest ages, the smaller individuals of an age class are of- ten sampled disproportionately to their true abun- dance, whereas for older ages, the same happens for 700 Fishery Bulletin 95(4), 1997 the larger (faster growing) fish in a given age class. Given Goodyear’s (1995) findings and our nonran- dom sampling design, it may be that our growth models overestimated length-at-age to some degree for all but the youngest age classes, but probably not as much as Goodyear found in his study. Although we had sampling quotas for each year, region, sex, and 10-cm size interval combination, for many dif- ferent reasons we invariably exceeded those quotas for all but the rarest size classes, often greatly for the most common length intervals; Table 2 and Fig- ure 1 provide clear evidence of this. Because of this oversampling, our actual design fell somewhere be- tween simple random sampling and length-stratified sampling, and thus should have reduced the bias to some extent. Given this rather small potential bias and the fact that our sample sizes and spatial and temporal coverages greatly exceeded all previous 1 986-1 992 1977-1978 700 600 500 400 30 0 200 1 00 0 700 600 >. O 500 CD O' 400 CD 300 200 1 00 0 400 300 200 1 00 Age (yr) Figure 2 Age distributions by sex and region of king mackerel included in the analysis and collected during 1977-78 and 1986-92. DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 701 king mackerel studies (all of which sampled size-se- lective fisheries and only one (Beaumariage, 1973) of which clearly used simple random sampling), we feel our growth estimates are the best available. Most important, there is no obvious reason to suspect that the bias would be greatly different among regions or among years; thus our conclusions about the tempo- ral and regional differences in growth should be valid. Our finding of similar maximum longevity for both sexes differs from all previous studies (except Age (yr) Figure 3 Von Bertalanffy growth curves and 95% confidence limits by region and sex for king mackerel collected during 1977-78 and 1986- 92. Growth curves were calculated by using individual quarterly observed sizes-at-age. The upper three curves in each panel represent females; the lower three represent males. Tabie 3 Results of Hotelling’s T2 tests comparing 1986-92 von Bertalanffy growth curves for king mackerel. The larger group in each comparison was randomly subsampled so that it’s sample size equaled that of the smaller group. Underlined Fs in right three columns indicate parameters that did not significantly affect growth differences. Values in bold = parameter which most affected growth differences. NS = not significant, n = sample size for each group in the comparison. AOF = Atlantic Ocean females; EGF = eastern Gulf females; WGF = western Gulf females; AOM = Atlantic Ocean males; EGM = eastern Gulf males; WGM = western gulf males. Groups compared Calculated F7 Denom. df2 Critical value of F needed for 95% Roy-Bose simultaneous confidence limits to bracket zero n K *0 AOF-EGF 363. 82 2,753 5,502 28.4 23 17.7 AOF-WGF 51.2 1,662 3,320 3.2 01 L2 EGF-WGF 37.6 1,662 3,320 05 23 8.2 AOM-EGM 46.3 1,285 2,566 16.5 06 01 AOM-WGM 10.1 990 1,976 9.8 7.9 5.7 EGM-WGM 10.7 990 1,976 OO 03 4.0 AOF-AOM 369.6 2,083 4,162 211.8 53.2 11.4 EGF-EGM 259.5 1,285 2,566 133.3 13.1 Ol WGF-WGM 103.1 990 1,976 46.2 5.0 oo 1 Critical F = 3.12 (a = 0.05, 2-tailed test) for all tests. 2 Numerator df = 3 for all comparisons (3 parameters). 702 Fishery Bulletin 95(4), 1997 Beaumariage, 1973) that reported that females lived longer than males. The 26-year-old female and 24- year-old male we found are the oldest king mackerel reported. Collins et al. (1989), in the only other study that used sectioned otoliths, found a 21-year-old fe- male as well as 16-year-old males. The oldest fish reported in all other studies that used whole otoliths, was age 14 for females and age 12 for males (Beaumariage, 1973; Johnson et al., 1983; Manooch et al., 1987; Sturm and Salter; 1990). Our findings support those of Collins et al. (1989), i.e. that sec- tioned sagittae provide higher age estimates than whole otoliths for king mackerel, especially for fish >85 cm FL. Among females ages 8-12 from the Johnson et al. (1983) study (the 1977-78 collections in our study), our age estimates based on sagittal Table 4 Total age distributions by 10-cm length class, by region, for all female king mackerel collected during 1986-92. Age (yr) uize (cm) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 24 n Atlantic Ocean 35 25.0 50.0 25.0 4 45 9.8 84.3 5.9 51 55 65.1 34.9 126 65 26.6 69.3 3.8 0.3 290 75 5.0 41.0 40.7 11.2 1.8 0.4 735 85 0.7 8.7 22.9 27.0 20.8 11.5 5.7 1.4 0.8 0.3 0.1 0.1 722 95 1.6 7.4 8.8 13.7 18.4 18.0 9.0 7.4 6.3 4.7 2.5 1.4 0.6 0.2 511 105 1.6 5.1 11.0 15.0 10.5 11.3 10.5 10.5 8.6 5.1 3.2 2.9 1.6 1.9 0.5 0.5 0.3 373 115 1.2 2.7 5.9 9.2 11.2 12.1 10.9 12.7 8.9 5.6 6.2 5.3 3.0 1.8 1.8 0.9 0.6 338 125 0.9 2.3 1.8 6.0 7.8 10.6 9.6 11.5 9.2 8.3 7.8 8.7 4.1 5.5 3.2 1.4 1.4 218 135 2.9 2.9 2.9 2.9 8.8 8.8 8.8 8.8 11.8 8.8 17.6 2.9 5.9 2.9; 34 145 42.9 14.3 14.3 14.3 14.3 7 155 25.0 50.0 25.0 4 Eastern Gulf 35 100.0 4 45 12.5 87.5 72 55 91.6 8.4 249 65 34.1 58.8 6.6 0.4 454 75 3.1 62.7 25.2 8.4 0.6 512 85 0.3 16.7 43.4 24.4 7.7 3.9 2.6 0.3 0.3 0.3 311 95 1.8 13.7 31.7 23.2 16.9 4.6 1.8 3.5 1.4 0.7 0.4 0.4 284 105 0.7 11.3 22.5 24.5 13.9 8.9 9.9 5.0 0.3 0.7 1.0 0.3 302 115 1.7 8.4 16.0 22.3 13.0 16.0 11.3 4.6 1.3 1.7 0.8 1.3 1.3 0.4 238 125 2.2 8.6 8.1 11.8 15.1 17.2 6.5 10.8 6.5 5.9 2.7 0.5 0.5 2.2 0.5 0.5 0.5 186 135 0.9 2.8 13.2 7.5 3.8 14.2 11.3 12.3 9.4 4.7 6.6 9.4 1.9 1.9 106 145 2.3 2.3 9.1 2.3 11.4 22.7 9.1 11.4 4.5 9.1 6.8 2.3 6.8 44 155 25.0 50.0 25.0 4 Western Gulf 35 100.0 10 45 91.7 8.3 12 55 2.5 70.0 27.5 80 65 35.1 51.0 9.9 4.0 151 75 1.6 41.4 41.6 12.9 2.3 0.2 442 85 15.2 33.0 29.6 15.7 4.7 0.8 0.8 0.3 382 95 0.5 11.2 27.1 28.5 16.8 9.3 2.8 1.4 0.9 0.9 0.5 214 105 0.6 12.2 15.4 21.8 19.2 10.9 7.1 6.4 3.8 0.6 0.6 0.6 0.6 156 115 1.7 10.3 8.5 12.0 10.3 12.8 13.7 11.1 9.4 4.3 0.9 0.9 1.7 0.9 1.7 117 125 2.9 2.9 4.3 4.3 12.9 7.1 12.9 12.9 10.0 8.6 8.6 7.1 2.9 1.4 1.4 70 135 4.2 4.2 16.7 16.7 4.2 12.5 12.5 8.3 16.7 4.2 24 145 16.7 16.7 33.3 16.7 16.7 6 1 This size class also contained one fish (2.9%) at age 26. DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 703 sections exceeded their original estimates based on whole otoliths 67-100% of the time. The slightly higher maximum ages in the Atlantic than in the eastern or western Gulf during 1986-92 may reflect lower fishing mortality rates for the At- lantic than for the Gulf where age structure was trun- cated by fishing pressure4; alternatively, this find- ing may be a sampling artifact. A much higher pro- portion of Atlantic samples were collected at fishing tournaments, which target larger and older fish, and Atlantic sample sizes exceeded eastern and western Gulf sample sizes by 36% and 107%; therefore the chances of obtaining an older fish were greater. That females grew faster and attained larger maxi- mum sizes than males agrees with previous studies (Beaumariage, 1973; Johnson et al., 1983; Manooch et al., 1987; Collins et al., 1989; Sturm and Salter 1990). The large variation in ages within size intervals that we found was also noted by Johnson et al. (1983). Significant differences in growth among the three regions for both sexes during 1986-92 and the per- sistence of that pattern in each of the seven years support the hypothesis that there are three stocks as suggested by allozyme, mark-recapture, catch and fishing effort, and juvenile birth-date distribution data (Grimes et al., 1987; Johnson et al., 1994; Gulf of Mexico and South Atlantic Fishery Management Councils3). That similar differences between Atlan- tic and eastern Gulf growth were present in 1977- 78 is further evidence that these growth differences Table 5 Total age distributions by 10-cm length class, by region, for all male king mackerel collected during 1986-92. Age (yr) olZG (cm) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 n Atlantic Ocean 35 100.0 4 45 100.0 58 55 61.6 37.6 0.8 125 65 13.7 71.1 14.8 0.4 263 75 0.8 19.3 43.0 21.8 10.0 2.7 1.1 0.6 0.3 0.2 0.2 632 85 0.4 3.5 6.3 16.8 20.7 12.4 10.9 7.6 7.6 5.4 5.0 0.9 0.9 1.1 0.2 0.2 0.2 0.2 542 95 0.3 0.9 2.3 5.8 6.4 7.3 13.1 11.3 11.0 13.1 8.1 7.8 5.8 2.9 2.6 0.3 0.6 0.3 344 105 3.1 3.1 4.1 2.0 7.1 5.1 7.1 11.2 10.2 11.2 10.2 9.2 6.1 4.1 3.1 2.0 1.0 98 115 6.2 6.2 12.5 12.5 18.8 12.5 6.2 12.5 6.2' 16 125 100.0 1 Eastern Gulf 35 45 8.8 91.2 57 55 74.8 25.2 155 65 13.9 73.0 13.1 267 75 0.7 27.8 43.7 19.0 6.0 1.8 0.7 0.4 284 85 0.4 8.4 15.1 24.4 26.1 9.2 5.0 8.0 2.1 0.4 0.8 238 95 1.5 4.5 4.0 11.9 14.9 9.5 18.4 12.4 6.0 8.5 5.0 1.5 1.0 0.5 0.5 201 105 1.6 3.1 4.7 7.8 7.8 17.2 9.4 17.2 4.7 10.9 4.7 3.1 1.6 3.1 1.6 1.6 64 115 4.8 9.5 4.8 14.3 4.8 9.5 4.8 9.5 19.0 4.8 9.5 4.8 21 125 33.3 33.3 33.3 3 Western Gulf 35 9.1 90.9 11 45 85.7 14.3 7 55 70.3 28.4 1.4 74 65 9.3 56.5 29.0 5.2 193 75 2.1 22.5 33.9 24.3 11.4 4.6 0.7 0.4 280 85 1.3 3.9 11.3 22.9 23.4 16.9 10.8 4.8 3.5 0.4 0.9 231 95 2.7 7.5 14.4 15.1 17.1 13.0 10.3 6.2 7.5 2.7 2.1 0.7 0.7 146 105 2.5 7.5 15.0 5.0 5.0 2.5 15.0 12.5 7.5 12.5 5.0 5.0 2.5 2.5 40 115 11.1 11.1 11.1 22.2 11.1 11.1 22.2 9 1 This size class also contained one fish (6.2%) at age 24. 704 Fishery Bulletin 95(4), 1997 are consistent features of king mackerel populations. Assuming that these differences persisted from 1977 to 1992, during which time exploitation rates varied considerably (Mackerel Stock Assessment Panel4), we suggest that these differences are not just tempo- rary density-dependent responses to varying popu- lation sizes or exploitation rates. Our findings of regional (stock) growth differences are also consistent with those of Gold et al. (in press), who compared mtDNA haplotypes and found weak genetic differences between Atlantic and Gulf king mackerel. Although our results are not indicative of genetic discontinuity, our data demonstrate that the three groups of fish experience sufficiently different environmental and fishery conditions to produce identifiable and consistent differences in growth. Contrary to our finding of regional differences in growth within sexes, Beaumariage (1973) reported that growth rates did not differ for either sex between the Gulf and Atlantic coasts of Florida. His results may reflect that many of his Atlantic fish were col- lected off southeast Florida during winter and thus may have been Gulf-group fish. In addition, the use of whole otoliths for ageing undoubtedly introduced error in length-at-age estimates, possibly obscuring regional differences. Johnson et al. (1983) reported that female king mackerel from Louisiana grew faster than females from other areas of the Gulf and from the Atlantic. However, their predicted sizes-at-age for Louisiana females ages 4-8, the ages with adequate sample sizes (n = 16-78) that could be accurately aged with whole otoliths, were no more than 3.1 cm different from our eastern Gulf fish. For fish older than age 8, their estimates were increasingly larger than ours, most likely because the use of whole otoliths resulted in underageing these larger fish. The growth differences between 1977-78 and 1986-92, i.e. lower growth during the former period seen in both sexes in the Atlantic and eastern Gulf (Fig. 6), could be a density-dependent response. Popu- lations were much larger in the late 1970’s and early 1980’s than during 1986-92 (Mackerel Stock Assess- ment Panel4). The key point to remember is that within sexes, the growth differences among regions clearly present in 1986-92 apparently existed as far back as 1977-78. Acknowledgments We would like to thank the Institute Nacional de la Pesca for the cooperation of their personnel at the fishery laboratories in Yucalpeten, Campeche, Alvarado, and Tampico for obtaining samples and data from the Mexican fisheries. K. Burns and her Table 6 Von Bertalanffy parameters and 95% asymptotic confidence intervals for male and female king mackerel by region for fish col- lected during 1986-92 and 1977-78, calculated using quarterly observed sizes-at-age. Collection years Para- meter Females Males n Estimate Asymptotic 95% confidence interval n Estimate Asymptotic 95% confidence interval 1986-92 Atlantic 3,407 126.7 125.0 to 128.5 2,083 96.4 95.7 to 97.1 E. Gulf L„ 2,796 137.8 135.8 to 139.8 1,330 102.6 101.1 to 104.1 W. Gulf Lx 1,662 134.1 130.6 to 137.7 995 102.8 100.5 to 105.2 Atlantic K 3,407 0.145 0.137 to 0.154 2,083 0.262 0.248 to 0.276 E. Gulf K 2,796 0.172 0.163 to 0.181 1,330 0.247 0.227 to 0.267 W. Gulf K 1,662 0.150 0.136 to 0.164 995 0.203 0.180 to 0.226 Atlantic to 3,407 -3.15 -3.41 to -2.90 2,083 -1.98 -2.19 to -1.78 E. Gulf ta 2,796 -1.83 -1.98 to -1.67 1,330 -1.84 -2.09 to -1.59 W. Gulf to 1,662 -2.69 -3.02 to -2.37 995 -2.74 -3.16 to -2.32 1977-78 Atlantic 323 122.7 115.5 to 129.9 128 95.9 92.3 to 99.6 E. Gulf 1,011 137.1 133.4 to 140.8 343 99.0 96.6 to 101.3 W. Gulf 188 151.5 138.2 to 164.8 40 116.0 93.1 to 138.9 Atlantic K 323 0.124 0.096 to 0.151 128 0.211 0.159 to 0.262 E. Gulf K 1,011 0.160 0.145 to 0.175 343 0.269 0.229 to 0.309 W. Gulf K 188 0.127 0.080 to 0.175 40 0.094 0.026 to 0.163 Atlantic t0 323 -4.54 -5.59 to -3.49 128 -3.14 -4.26 to -2.02 E. Gulf tQ 1,011 -2.12 -2.39 to -1.85 343 -1.63 -2.04 to -1.22 W. Gulf to 188 -2.78 -4.52 to -1.03 40 -6.78 -11.1 to -2.45 DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 705 colleagues at the Mote Marine Laboratory were es- pecially helpful both in field sampling in Mexico and in ensuring the shipment of samples and data to us. Special thanks are also due to the North Carolina Division of Marine Fisheries and, in particular, to L. Mercer, L. Nobles, and R. Gregory, for providing us 706 Fishery Bulletin 95(4), 1 997 with large numbers of processed otoliths. Doug Vaughan, NMFS, Beaufort Laboratory, provided pro- grams and valuable statistical advice concerning comparisons of growth curves. Two anonymous re- viewers provided valuable suggestions that helped improve the manuscript. DeVries and Grimes: Spatial and temporal variation in age and growth of Scomberomorus cavalla 707 Literature cited Beaumariage, D. S. 1973. Age, growth and reproduction of king mackerel Scomber- omorus cavalla, in Florida. Fla. Mar. Res. Publ. 1, 45 p. Bernard, D. R. 1981. Multivariate analysis as a means of comparing growth in fish. Can. J. Fish. Aquat. Sci. 38:233-236. Collins, M. R., D. J. Schmidt, C. W. Waltz, and J. L. Pickney. 1989. Age and growth of king mackerel, Scomberomorus cavalla, from the Atlantic coast of the United States. Fish. Bull. 87:49-61. Gold, J. R., A. Y. Kristmundsdottir, and L. R. Richardson. In press. Mitochondrial DNA variation in king mackerel ( Scomberomorus cavalla ) from the Gulf of Mexico and west- ern Atlantic Ocean. Marine Biology. Goodyear, C. P. 1995. Mean size at age: an evaluation of sampling strate- gies with simulated red grouper data. Trans. Amer. Fish . Soc. 124:746-755. Grimes, C. B., A. G. Johnson, and W. A. Fable Jr. 1987. Delineation of king mackerel, (Scomberomorus cav- alla), stocks along the U.S. east coast and in the Gulf of Mexico (ABSTRACT). U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFC- 199: 186-187. Ito, K., and W. J. Schull. 1964. On the robustness of the T20 test in multivariate analysis of variance when variance-covariance matrices are not equal. Biometrika 51:71-82. Johnson, A. G., W. A. Fable Jr., C. B. Grimes, L. Trent and J. V. Perez. 1994. Evidence for distinct stocks of king mackerel, Scomberomorus cavalla , in the Gulf of Mexico. Fish. Bull. 92:91-101. Johnson, A. G., W. A. Fable Jr., M. L. Williams, and L. E. Barger. 1983. Age, growth and mortality of king mackerel, Scomberomorus cavalla, from the southeastern United States. Fish. Bull. 81(1):97-106. Ketchen, K. S. 1950. Stratified subsampling for determining age- distributions. Trans. Am. Fish. Soc. 79:205-212. 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 E. L. Nakamura and H. R. Bullis Jr. (eds.), Proceedings of the mackerel colloquium, p. 33-41. Gulf States Mar. Fish. Comm., Publ. 4 Manooch, C. S., Ill, S. P. Naughton, C. B. Grimes, and L. Trent. 1987. Age and growth of king mackerel, Scomberomorus cavalla, from the U.S. Gulf of Mexico. Mar. Fish. Rev. 49(2):102-108. SAS Institute, Inc. 1988. SAS/STAT user’s guide, release 6.03 edition. SAS Institute, Inc., Cary, NC, 1028 p. Sturm, M. G. de L., and P. Salter. 1990. Age, growth, and reproduction of the king mackerel, Scomberomorus cavalla fCuvier), in Trinidad waters. Fish. Bull. 88:361-370. Sutter, F. C., Ill, R. O. Williams, and M. F. Godcharles. 1991. Movement patterns and stock affinities of king mack- erel in the southeastern United States. Fish. Bull. 89:315-324. 708 Fishery Bulletin 95(4), 1997 Trent, L., W. A. Fable Jr., S. J. Russell, G. W. Bane, and B. J. Palko. 1987. Variations in size and sex ratio of king mackerel, Scomberomorus cavalla, off Louisiana, 1977-85. Mar. Fish. Rev. 49(2):91-97. 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, Scomberomorus cavalla, in the south- eastern United States. Fish. Bull. 81:709-721. Vaughan, D. S., and T. E. Helser. 1990. Status of the red drum stock of the Atlantic coast: stock assessment report for 1989. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFC-263. 709 Abstract .—White seabass, Atract- oscion nobilis, is a valuable recreational and commercial species, but much of its early life history is undescribed. The coast and two bays of San Diego County were sampled each month for two years with a depth-stratified sampling design to determine habitat, food habits, age, and growth rate of recently settled fish. Age was estimated from otolith incre- ments and validated with fish of known age, reared in the laboratory. A few re- cently settled fish were caught in the bays at depths <1 m, but most inhab- ited shallow water (4-8 m) along the coast from May to October. This depth distribution coincides with that of the mysid Metamysidopsis elongata. Fish abundance in this zone was low, how- ever, reaching a maximum of 24/ha in July. The smallest white seabass col- lected were about 7 mm SL and 26 d old, but previous studies indicate that smaller and presumably younger fish were probably extruded through the trawl. According to combined results, most larvae settled 2-3 weeks after being spawned. Juveniles remained at a depth of 4-8 m for 2-3 months, fed primarily on abundant mysids, and as- sociated with drifting macrophytes (r=0.52, P=0.015, n=21). Growth dur- ing this period was 1.3 mm/d, similar to that observed in the laboratory. At about 100 mm SL (-100 d old), juve- niles appeared to move out of the area. The shallow waters just beyond the breaking waves may be preferred by young white seabass because abundant food and warm water promote rapid growth and drifting macrophytes pro- vide a refuge from predators. Manuscipt accepted 30 May 1997. Fishery Bulletin 95:709-721 (1997). Age, growth, distribution, and food habits of recently settled white seabass, Atractoscion nobilis, off San Diego County, California Christopher J. Donohoe Department of Biology San Diego State University San Diego, California 92 1 82 Present address: Department of Fisheries and Wildlife Nash Hall 1 04, Oregon State University Corvallis, Oregon 97331-3803 E-mail address: donohoec@mother.com White seabass, Atractoscion nobilis (family Sciaenidae), is a highly de- sired recreational and commercial species found in waters off the coasts of southern and Baja Califor- nia as well as in the Gulf of Califor- nia. Adults inhabit the nearshore zone over rocky bottoms and in kelp beds and can attain a weight of 38 kg (Young, 1973). Population size has not been estimated, but since the 1920’s, commercial and recre- ational landings off California have continued to decline and the range of the species has contracted (Collins, 1981; Methot, 1983). Management efforts to stabilize and restore the population have been largely unsuc- cessful. Reductions in the catch and distribution of white seabass have been attributed largely to overfish- ing (Thomas, 1968; Vokovich and Reed, 1983; MacCall, 1986), but the importance of other mechanisms, such as increased natural mortal- ity of fish at early life history stages, has not been evaluated. Despite the historic value of this species, much of its early life his- tory was unknown until recently. Moser et al. (1983) described the development of the early life stages and historic distribution of larvae in the California Cooperative Oce- anic Fisheries Investigations (CalCOFI) sampling area off south- ern and Baja California. In several laboratory studies, growth, sur- vival, energetics, and feeding be- havior of larvae have been exam- ined (Kim, 1987; Dutton, 1989; Orhun, 1989), as well as the devel- opment of sensory systems and predator-avoidance behavior (Mar- gulies, 1989). Less is known about the early ju- venile stage because few early ju- veniles have been caught until re- cently. Early studies suggested that juveniles inhabit either the surf zone or kelp canopy along the open coast, or bays and estuaries (Tho- mas, 1968; Feder et al., 1974; Max- well1 ). Allen and Franklin (1988, 1992) have since demonstrated that late larvae and early juveniles in- habit shallow water along the open coast of southern California and Channel Islands and semiprotected embayments in the vicinity of Long Beach Harbor. However the nurs- ery area for white seabass has not been clearly defined and the rela- tive importance of the open coast and bays as nurseries has not been 1 Maxwell, W. D. 1977. Progress report of research on white seabass, Cynoscion nobilis. Calif. Dep. Fish Game, Mar. Resour. Admin. Rep. 77-14, 14 p. [Avail- able from Calif. Dep. Fish Game, 330 Golden Shore, Suite 50, Long Beach, CA 90802.] 710 Fishery Bulletin 95(4), 1 997 evaluated. In addition, food habits, age, and growth of these fish in the wild have not been examined. The specific goals of this study were to determine 1) the depth distribution of early juvenile white seabass along the open coast and in bays of San Di- ego County, 2) size-specific food habits of white seabass, and 3) age and rate of growth. Materials and methods Sampling design Most white seabass were obtained from a survey originally designed to sample settled California hali- but, Paralichthys californicus (Kramer, 1990). Two bays (Mission Bay and Agua Hedionda Lagoon) and the open coast of San Diego County were sampled monthly from September 1986 to September 1988 with a depth-stratified sampling design. The coast was sampled at four primary sites with a 1.6 m x 0.35 m beam trawl (Fig. 1). At each site, four benthic tows were made in each of three bottom depth inter- vals (strata): 4-8, 9-11, and 12-14 m. A few tows were made in water as shallow as 3 m on days when the sea was calm. Tows were made parallel to shore at about 0.6 m/s for 10 min. The exact depth of tows within each stratum was chosen at random. Sam- pling depth was maintained along the chosen 1-m depth contour with the aid of a fathometer. An odometer at- tached to the trawl recorded tow distance, which ranged from 250 to 450 m. An additional four sites were sampled from April to October 1988 by biologists at San Diego State University (SDSU) with identical gear, but only at the 4—8 and 9-11 m depth strata. A similar sampling design was used to sample the two bays. Mission Bay and Agua Hedionda Lagoon were subdivided into five and three blocks respec- tively to sample the various habitats adequately within each bay (Fig. 1). Each block was further sub- divided into three depth strata: 0-1, 1-2, and 2-4 m. Within each stratum and block, three benthic tows were made at random locations with a 1.0 m x 0.35 m beam trawl equipped with an odometer. In the two deeper strata, the trawl was towed by a 5-m skiff for 5 min, covering a distance of 100-250 m. In the 0-1 m stratum, the trawl was towed by hand for a mea- sured distance of 20-50 m. The 0-1 m depth stra- Map of San Diego County showing location of the eight coastal sites and sampling areas (blocks) within Mission Bay and Agua Hedionda Lagoon. Four coastal sites were sampled by Kramer (1990) and four were sampled by San Diego State University (SDSU). The sampling design used along the coast is also shown. Donohoe: Age, growth, distribution, and food habits of Atractoscion nobilis 71 1 turn was also sampled with a 1 m x 6 m beach seine. Three hauls were made in each block over a mea- sured distance of 15-50 m with the width of the seine fixed at 4 m. In addition, three tows were made each month in blocks 2-5 of the 2-4 m stratum in Mission Bay with the 1.6-m beam trawl (Fig. 1). The mesh size of all three nets was 3 mm. All hauls were made during the day. Monthly sampling in Agua Hedionda Lagoon was not initiated until March 1987. Tow distance, water temperature, and the pres- ence and type of drift macrophytes in the net were recorded at the end of every tow. Beginning in April 1988, the weight of drift macrophytes in each tow was also recorded at the four coastal sites sampled by SDSU biologists. White seabass were either fro- zen or preserved in 80% ethanol and later measured to 0.1 mm standard length (SL) in the laboratory. Lengths of alcohol-preserved fish were adjusted by 3.6% to compensate for shrinkage. Average shrink- age was estimated by measuring a subsample of white seabass before and several months after pres- ervation in ethanol. Sagittae and stomach contents were removed and stored in 80% ethanol. Fish were then dried at 60°C for two days and weighed. White seabass were also obtained opportunistically (i.e. sporadically) from the coastal habitat with a 7.6-m headrope otter trawl and a 15.2-m beach seine (6-mm mesh). These fish were used only in the food habits and growth portions of this study. Distribution and abundance Abundance was calculated as the number of fish caught divided by the product of tow distance (from odometers) and net width. Mean abundance of fish along the coast was calculated for each site by depth stratum (n= 4 tows). Monthly differences in abun- dance among the three depth strata were compared by using the Kruskal-Wallis test, with a=3 depths, and n= 4 (1987) or n= 8 (1988) sites (Sokal and Rohlf, 1981). Monthly mean abundance in bays was calcu- lated for each bay by block and depth stratum. Block means were averaged to produce a mean for each bay and the two bays were averaged to yield a grand mean for each depth stratum. Because estimates of monthly mean abundance did not differ among the two gear types (paired £-test; mean difference=0.50 fish/ha, £=0.27, df=8, P=0.79), trawl and seine samples within the 0-1 m depth stratum were pooled to produce an improved estimate of abundance. The relation between abundance of white seabass and drift macrophytes was estimated by testing for a correlation between abundance of white seabass and drift macrophytes in each tow and for a correla- tion between mean abundances at each site (n= 4 tows). Abundance of macrophytes (g/m2) was log- transformed prior to analysis. Biomass of macro- phytes was recorded only from April to October 1988 at the four secondary sites along the coast that were sampled by SDSU biologists. Food habits The stomach contents of 142 white seabass collected in bays and along the coast with all gear types were examined. For each fish, prey items were identified, counted, sorted into one of ten major prey categories, dried at 60°C for 1-2 d, and weighed to either 1 pg (for samples <25 mg) or to 0.1 mg (for samples >25 mg). White seabass were grouped into six length classes: 6- 10, 10-18, 18-25, 25-35, 35-55, and 55-150 mm SL. Class intervals were chosen so that each interval con- tained similar numbers of fish. Mean prey weight and frequency of occurrence of each prey category were cal- culated for the six length classes. Six individuals with empty stomachs were excluded from calculations of fre- quency of occurrence and mean weight of prey. Age and growth The ageing method was validated by using labora- tory-reared fish of known age. Eggs obtained from captive broodstock were placed in 7-m3 flow-through tanks and reared at 17-20°C on a diet of marine ro- tifers, brine shrimp, euphausids, and chopped mack- erel. White seabass were sacrificed at irregular in- tervals between 13 and 76 d after hatching and stored in 80% ethanol. Sagittae were mounted in Eukitt mounting media and ground in the sagittal plane with 15-pm grit sandpaper and polished with 0.3-pm grit lapping film. Increments were counted on the right sagitta from the central primordium to the mid- ventral margin. Each sagitta was read in one ses- sion by one observer, with neither age nor length of the fish known to the reader. The rate of increment deposition and age at first increment formation were estimated by linear regression. A subsample of 50 wild larval and juvenile white seabass was aged with the technique described above. Individuals were selected at random from several length classes to represent equally the size range of fish collected. The subsample included fish caught in bays, on the coast, and in both years. Growth rates were estimated by fitting a Gompertz function (Lt = L oeGl1 ~e4,)) j-0 length-at-age and weight-at-age data. The ages of the remaining individuals were estimated from the resulting age-length relation. The date each fish was spawned was calculated by subtracting the age of the fish and an additional two days (incuba- tion time at ~16°C; Orhun, 1989) from date of cap- 712 Fishery Bulletin 95(4), 1997 ture. The error associated with the estimated spawn dates is therefore the same as that asso- ciated with the age-length relation. Growth of wild white seabass was compared with growth of three groups reared in the labo- ratory. Eggs spawned on 8 May, 24 June, and 25 September 1989 were reared as described above. Mean length-at-age was estimated from random subsamples (n=16-76 fish) taken at ir- regular intervals during rearing. Linear growth models were fitted to the length-at-age data for both reared and wild fish to facilitate statisti- cal comparisons of growth rates with ANCOVA. Although growth of white seabass from hatch- ing to 150 mm SL was nonlinear, growth over a smaller size range of 6-104 mm SL was de- scribed equally well by linear models (r2>0.94). Results Distribution and abundance The overall abundance of white seabass in the bays were caught in 1,250 tows along the coast and 10 and along the coast of San Diego County was low. white seabass were caught in 2,527 tows in the bays. Most tows caught no white seabass. Lengths of fish ranged from 6.2 to 149 mm SL (x= 30.2 mm SL), but 80% were smaller than 40 mm SL (Fig. 2). About 40 fish (33%) were smaller than 15 mm SL, the approximate length at metamor- phosis (Moser et al., 1983). White seabass were caught primarily in shallow water in both coastal and bay habitats. Along the coast, nearly all white seabass (97%) were collected in the shal- lowest (4-8 m) depth stratum, with the highest density at 6 m (Fig. 3). Only three fish were taken in deeper water and these were among the largest caught, ranging from 92 to 149 mm SL. In the bays, all ten fish were caught in the shallowest (0— 1 m) stratum. Settled (demersal) white seabass were present in shallow strata only during spring and summer, although one fish, the 149-mm-SL juvenile, was caught in January 1988. Along the coast, white seabass were caught from June to August 1987, and from May to October 1988, when sampling ended (Fig. 4). In both years, abundance was highest in July, with mean densities of 15/ha (1987) and 24/ha (1988). White seabass were found in 54 of 210 tows (26%) made in the 4-8 m no. tows = 0 0 2 18 68 56 46 21 42 49 79 55 71 25 Depth (m) Figure 3 Distribution of white seabass and the mysid Metamysidopsis elongata along the coast in relation to depth. White seabass abundance is the mean (± 1 SE) during months fish were captured ( Jun-Aug 1987, May- Oct 1988). Number of tows at each depth during this period is indicated at the top of the graph. Mysid data are redrawn from Clutter (1967) and represent total number of M. elongata caught from 1960 to 1962 at a site ~3 km south of the Torrey Pines 2 site (Fig. 1). Sampling effort for mysids was similar for all depths. During regular sampling, a total of 112 white seabass Donohoe: Age, growth, distribution, and food habits of Atractoscion nobilis 713 Figure 4 Monthly mean abundance of white seabass (± 1 SE) in the 4-8 m stra- tum along the coast ( 1987 n=A sites; 1988 n=8 sites) and in the 0-1 m stratum in the bays 0.2 g dry weight (~40 mm SL) fed on mysids of similar mean weight (Fig. 6). Mysids also dominated the diet of the 10 fish caught in the bays. Prey of secondary importance varied with white seabass length class. Larvae (6-10 mm SL) fed on copepods, fish of intermediate length ( 10-55 mm SL) fed on gammarid amphipods, and larger juveniles 714 Fishery Bulletin 95(4), 1997 (35-150 mm SL) preyed on shrimp and fishes. Most fish in the stomachs were well digested and difficult to identify, but the sagittae closely resembled those of white croaker, Genyonemus lineatus, and queen- fish, Seriphus politus. One case of cannibalism was observed; a 7-mm larva was eaten by a 35-mm juve- nile. Most shrimp were well digested, but at least two individuals were identified as belonging to the genus Crangon. Other items found in the stomachs included nematodes, bits of algae and surf grass, portions of crustaceans, and sand. The stomachs of six white seabass (5%) were empty or contained only nonfood items such as sand. Age and growth Otolith increments formed daily in sagittae of labo- ratory-reared white seabass (Fig. 7). The slope of the regression of observed number of increments on age was 0.96 and did not differ from unity (r2=0.96, n= 25, P<0.01, 95% confidence limits on slope: 0.89 and 1.04 increments/d). The first increment formed 3-4 d after hatching, a period that corresponds to yolk absorption and onset of feeding (Kim, 1987; Orhun, 1989). Age was estimated for 50 wild white seabass rang- ing from 6.2 to 104 mm SL. The ten smallest fish that were aged ranged from 6.2 to 9.2 mm SL and were estimated to be 26-32 d old (Fig. 8). The age of a fish that was 15 mm SL, the length at metamor- phosis, was estimated to be about 40 d. The largest juvenile aged (104 mm SL) was estimated to be 108 d old. The range of estimated ages suggests that white seabass remain in the nursery for 2-3 months after settlement. It should be noted that the oldest validated age was 76 d. A 149-mm-SL juvenile was not aged because it was probably much older than the oldest validated age. Growth of these fish was rapid in terms of length and weight. The parameters of the Gompertz model relating age and length were estimated as L0= 0.202, G=6.64, and g=0. 0273, where L0 is length at time t0, G is the instantaneous growth rate at time tQ, and g is the rate of decrease of G (Fig. 8A). This equates to a maximum growth rate of 1.57 mm/d at 70 d. Weight also increased rapidly. The parameters relating weight and age were W0=2.72 x 10~7, G=17.58, and g=0.0256, where W0 is weight at time £0 (Fig. 8B). Wild fish between 6 and 104 mm SL grew at rates similar to those of laboratory-reared fish. Wild fish grew at a linear rate of 1.31 mm/d, compared with laboratory rates of 1.15, 1.32, and 1.04 mm/d for groups spawned in May, June, and September 1989 (Fig. 9). Linear growth models were used to facili- tate statistical analysis by ANCOVA. Linear models fitted the data well, with coefficients of determina- tion (r2) of 0.94-0.99 for the four groups. The rate of growth (slopes) did not differ among the four groups (ANCOVA, F=1.40, P=0.25). However, the June labo- ratory group was significantly larger at a given age than the wild fish (ANCOVA, F=5.05, P < 0.01; Tukey pairwise comparison), but the remaining groups did not differ in their length-at-age. The distribution of spawning dates, based on counts of otolith increments, indicated that spawn- ing occurred from March to July in 1987, and from Donohoe: Age, growth, distribution, and food habits of Atractoscion nobilis 715 Table 1 Mean dry weight in micrograms (|ig) and as a percentage of total weight (in parentheses) and frequency of occurrence of prey items in stomachs of white seabass caught along the coast. Six fish with empty stomachs were excluded from the analysis. n = sample size. Length class and mean length (mm SL) 6-10 10-18 18-25 25-35 35-55 55-150 Prey category (7.9) (13.7) (21.8) (29.1) (44.8) (79.2) Mean Dry Weight in p g and (%) Mysids 46 (75) 154 (91) 467 (99) 1,417 (99) 2,329 (78) 8,270 (74) Copepods 10(16) 2 (1) — — — — Gammarid amphipods — 7 (4) — 5 (<1) 11 (<1) — Fish — — 4(1) — 97 (3) 768 (7) Shrimp — — — — 186(6) 211 (2) Nematodes — — — — 6(<1) 15 (d) Macrophytes — — <1 — 55 (2) 647 (6) Crustacean parts 5 (8) 2 (1) — 5 (<1) — 459 (4) Sand - — — 1 (<1) 185 (6) — Unidentified — 5 (3) — — 99 (3) 814 (7) Total 61 170 471 1,428 2,968 11,184 Frequency of occurrence (%) Mysids 78 91 100 96 100 100 Copepods 11 5 — — — — Gammarid amphipods — 5 — 11 5 — Fish — — 5 — 25 25 Shrimp — — — — 5 5 Nematodes — — — — 5 10 Macrophytes — — 5 — 25 50 Crustacean parts 11 5 — 4 5 15 Sand — — — 4 5 — Unidentified — 5 — — 15 10 n 18 22 20 27 20 20 March to the beginning of September in 1988 (Fig. 10). In both years, most of the young white seabass collected were spawned in June. The distribution of spawning dates derived from estimated ages agreed closely with those based upon direct ageing. Discussion Spawning season Larval and juvenile white seabass collected off San Diego County were spawned from March to September, with the greatest num- ber spawned in June in both years (Fig. 10). This seasonal pattern of spawning, inferred from counts of otolith increments, agrees with previous estimates based on larval abundance and adult spawning condition. Moser et al. (1983) observed that eggs and larvae were most abundant in CalCOFI White seabass dry weight (g) Figure 6 Relation between white seabass dry weight and mean dry weight of mysids (total weight of mysids plus total number of mysids) in the stomachs of 111 white seabass. 716 Fishery Bulletin 95(4), 1 997 plankton samples in July and that 95% of fish were captured between May and August. Adults begin to mature in early March (Clark, 1930) and spawn off south- ern California from April to August. Peak spawning activity is in May and June (Skogsberg, 1939). Size and age at settlement The smallest larvae caught during the present study were 6-7 mm SL, which would seem to indicate that white seabass begin to settle at about this size. However, white seabass as small as 4.2 mm SL were collected along the open coast north of San Diego County in 1988 and 1989 with a trawl containing 2-mm mesh in the codend (Allen and Franklin, 1992); therefore the smallest settled larvae were probably not retained by our net. More than 20% of the individu- als collected in that study were <5 mm SL and 50% were <7 mm SL. This size distribution suggests that many larvae caught off San Diego County had settled at lengths of 4-5 mm SL. Larvae up to 7.2 mm SL have been collected from the water column (Moser et al., 1983), indicating that some in- dividuals settle at >5 mm SL. Given that many white seabass settled at 4-5 mm SL, the average age at settlement must be less than one month. The 10 smallest fish caught and aged in this study ranged from 6.2 to 9.2 mm SL and were 26-32 d old (Fig. 8). Although smaller (4-5 mm SL) fish were not aged, they were probably much younger. In the laboratory, white seabass reared at 15°C hatch at a length of 2.8 mm SL after 2 d and grow to 4 mm SL in 10 d and to 5 mm SL in 15-19 d (Moser et al., 1983; Orhun, 1989). At this rate of growth, a 4-5 mm SL settled fish would have spent only 12-21 d in the pelagic habitat. Allen and Franklin (1992) hypothesized that most white seabass larvae that settle along the coast of southern California are spawned off Baja California and advected northward. However, a short pelagic phase of 2-3 weeks suggests that many of these lar- vae are spawned within the Southern California Bight (SCB). The direction of larval transport is dif- ficult to predict because the behavior and position of white seabass larvae in the water column is un- known. During spring and early summer, poleward- flowing undercurrents over the continental slope and equatorward-flowing surface currents over the con- tinental shelf (Hickey, 1993) could transport larvae along the coast in either direction. However, mean seasonal current velocities in the SCB region in spring and early summer are generally less than 20 cm/s, although short-term velocities can be higher (Hickey, 1993). At 20 crn/s, larvae could be trans- ported a maximum of 200-360 km in 12-21 d. It therefore seems unlikely that the 4-5 mm SL white seabass caught in the northern and middle SCB by Allen and Franklin (1992) were spawned off Mexico. These larvae, which represented a large proportion of the total catch, were almost certainly spawned off Cali- fornia. Of course older larvae collected in the middle SCB or young larvae caught off San Diego County could have been spawned off either California or Mexico. Nursery location The depth distributions of settled white seabass on the coast and within bays suggests that these fish prefer shallow water beyond the surf zone. Along the open coast, nearly all fish were caught at depths of 4-8 m, a region which begins just beyond the break- ing waves. In the bays, all 10 fish were caught just beyond the shore break at a depth of 0-1 m. Previ- ous observations in other regions are consistent with this conclusion. Settled white seabass have been col- lected beyond the breaking waves along semi- protected and exposed shores in and around Long Beach Harbor (Allen and Franklin, 1988). White seabass were not collected along protected shores, but depths <1.5 m were not sampled. On the open coast north of San Diego County, settled white seabass were also more abundant along the 5-m Donohoe: Age, growth, distribution, and food habits of Atractoscion nobilis 717 Age (d) Figure 8 Gompertz growth curves relating (A) length and age, and (B) weight and age of recently settled white seabass. Ages were determined from counts of increments in sagittae. depth contour than along the 10-m con- tour (Allen and Franklin, 1992). Al- though this distribution supports the conclusion that white seabass prefer shallow water, 30-40% of the settled fish were caught at the 10-m contour (Allen and Franklin, 1992). Most of the fish collected at 10 m were taken along one section of coastline between Ven- tura and Point Dume; thus the prefer- ence for shallow water may be modified by local conditions. Greater densities of settled fish along the open coast than in the bays suggests that the primary nursery for white seabass is the open coast. However, this difference in densities may reflect lower capture efficiencies of the 1.0-m trawl and beach seine in comparison with the 1.6-m trawl used on the coast. Although net efficiencies could not be estimated for white seabass because of low abun- dance, Kramer ( 1990) found no difference in efficiency of these same nets for Cali- fornia halibut <40 mm SL. Net efficien- cies probably did not differ greatly for small white seabass either, and thus the low catch of settled fish in the two bays was due to low abundance. In ichthyofau- nal surveys of other southern California bays over the last several decades, only a few early juvenile white seabass have been caught (Dixon and Eckmayer, 1975; Klingbeil et al., 1975; Horn and Allen, 1981). Although depths of 0-1 m may not have been sampled intensively, data from these surveys support the view that bays as a whole are not important nurseries for white seabass. The small size of southern California bays must also limit their importance as nursery areas for white seabass. As an example, Kramer (1990) estimated there were only 92 ha of habitat available between 0 and 1 m in Mission Bay and only 10 ha in Agua Hedionda Lagoon, compared with roughly 2,500 ha of habitat available between 5 and 8 m along the coast of San Diego County. Most of the remaining bays on the southern California coast are also small and many are periodically closed off from the sea by shift- ing sandbars (Zedler, 1982). Nursery features The narrow depth distribution of settled white sea- bass along the coast suggests that one or more fea- tures of this zone enhance survival of young fish. Sur- vival in shallow nurseries may be higher because of faster growth resulting from abundant food or warmer water, or lower predation rates (Bergman et al., 1988; Karakiri et al., 1989). Two features of the white seabass nursery that may promote rapid growth of juveniles are the abundant mysids and warmer water. Mysids, the principal prey of all sizes of white seabass collected, appear to be much more abundant within the nursery than at adjacent depths. Clutter (1967) sampled mysids at depths of 2-14 m during 1960-62 at a site 3 km south of the Torrey Pines 2 site (Fig. 1). Although the center of the depth distribution varied among months by 1-2 m, the mysid Metamysidopsis elongata was most nu- merous in the middle of the white seabass nursery 718 Fishery Bulletin 95(4). 1997 100 80 E E, £ 60 cn c a> ■O •S 40 c CO (75 20 0 20 40 60 80 100 Age (d) Figure 9 Estimated linear growth rates of wild white seabass (n=50) and three groups of juveniles spawned on 8 May, 24 June, and 25 September 1989 and reared at 17-20°C. Symbols represent the mean length (± 1 SD) of a subsample of 16-76 reared fish. For clarity, symbols are not shown for wild fish. Slopes, adjusted least squares (ALS) means, and coefficients of determination (r2) are shown. (Fig. 3). Other species of mysids were much less abundant. Density of M. elongata within the nursery can be quite high. Clutter’s data indicate that the mean density of mysids at 6 m was over 4,000/m3, whereas Roberts et al. (1982) estimated that the mean density of mysids at 6 m near the San Onofre site was over 100/m3. Mysids are also about an order of magnitude more abundant during spring and summer, when white seabass are in the nursery (Clutter, 1967). Mysids are not only abundant within the nursery; their broad size distribu- tion makes them suitable prey for both recently settled larvae and much larger juveniles. Mature M. elongata brood and release relatively large young that re- main in shallow water (Clutter, 1967). This reproductive strategy results in a population of mysids in the nursery that ranges over 100-fold in individual weight (Fig. 6). Although larger fish eat larger mysids, juveniles >40 mm SL can apparently feed on the largest mysids available (Fig. 6). At about this size, a transition from mysids to larger prey such as fish and shrimp also begins. The diets of other closely related sciaenids show a similar shift. Small (10-40 mm SL) sand seatrout, Cynoscion arenarius, small (15-30 mm SL) spotted seatrout, C. nebulosus, and juvenile (50-129 mm SL) weakfish, C. regalis, all feed extensively on mysids and at larger sizes shift to eating fish (Stickney et al., 1975; Sheridan, 1979; McMichael and Peters, 1989). Large juvenile, sub- adult, and adult white seabass feed principally on fish (Quast, 1968; Thomas, 1968). A second feature of the shallow nursery that may enhance growth and survival of settled white seabass is warm water. Temperatures in the 4-8 m stratum during the summer ranged from 16 to 20°C, an aver- age of 1.2 and 1.8°C warmer than the two deeper strata. The effect of a 1-2°C increase on growth has not been calculated for juveniles, but Orhun (1989) observed that a temperature increase from 15 to 17°C resulted in an increase in growth rate (dry weight gain) from 13.8%/d to 16.7%/d in 4-21 d old larvae. Temperature may have less influence on growth of juveniles, but any increase in growth rate will accu- mulate over the 2-3 months that juveniles are in the shallow nursery. Houde ( 1987) has demonstated that a small increase in growth rate acting over a moder- ate time interval can reduce stage duration and theo- retically result in substantial increases in survival and cohort size. Correlations between abundances of drift macro- phytes and white seabass suggest that macrophytes are also an important feature of the nursery. Al- though the correlation among abundances in single tows was weak (r=0.29), the sensitivity of this analy- sis was poor because the catch of white seabass was low; a maximum of only two white seabass was caught in a single tow at these stations. The correla- tion among mean abundances in the four tows at each site — the mathematical equivalent of making longer tows — was stronger (r=0.52) and does suggest that white seabass are more common near drift macro- phytes. Allen and Franklin (1992) also noted that white seabass were rarely caught unless drift algae was present. It is possible that white seabass are sim- ply more vulnerable to the trawl when drift macro- phytes are present, but numerous studies with drop nets and purse seines have demonstrated that many juvenile fishes associate with drift macrophytes (Kulczycki et al., 1981; Robertson and Lenanton, 1984; Kingsford and Choat, 1986). Small white seabass may associate with drift macrophytes because they harbor suitable prey or serve as a refuge from predation. In addition, the nursery area may be preferred by white seabass because the risk of predation could be lower than at adjacent depths. Unfortunately this hypothesis is difficult to evaluate. Surveys of the nearshore areas of southern California show that Donohoe: Age, growth, distribution, and food habits of Atractoscion nobilis 719 Mar Apr May Jun Jul Aug Sep Week Figure 10 Seasonal pattern of spawning by white seabass in 1987 and 1988 as calculated from date of capture, age of settled fish, and an estimated two-day incubation period (Orhun, 1989). The ages of 50 fish were estimated directly from otoliths, the others from the age-length rela- tion shown in Figure 8. predators of small benthic fishes are present both within the white seabass nursery and in deeper water (Love et al., 1986). Some of these predators, such as the California lizardfish ( Synodus lucioceps), are less abundant within the white seabass nurs- ery than at 12-18 m, whereas other spe- cies, such as California halibut ( Para - lichthys calif ornicus), are more abundant within the nursery than in deeper water (Ford, 1965; Love et al., 1986; Allen, 1990). However predation risk will depend not only on the total number of vertebrate and in- vertebrate predators at a particular depth, but also on the size and ontogenetic distri- bution of predators as well as species-spe- cific probabilities of encounter, detection, and capture (Bailey and Houde, 1989; Fuiman and Margurran, 1994). A detailed study is required to determine if the risk of predation to settled white seabass is lower in the nursery than in deeper waters. Conclusion The shallow water along the open coast just beyond the breaking waves appears to be the primary nursery for white seabass. Sur- vival of young white seabass is probably influenced by the abundance of mysids and drifting macrophytes as well as by water temperature in the nursery during spring and summer. However, it is not known if survival in the nursery is an important determinant of year-class success for white seabass. Year-class success in most marine fishes is generally believed to be set during the larval stage. However, poor correlations between larval abundance and subsequent recruitment for some species indicate that survival of older fish, per- haps early juveniles, may be equally important (Sissenwine, 1984; Bradford, 1992). Further studies are needed to evaluate the importance of survival of early life history stages in determining the distribu- tion and abundance of white seabass populations off southern California. Acknowledgments I am deeply indebted to Sharon Kramer for provid- ing specimens and data from her survey of the shal- low water flatfishes of San Diego County. I also thank John Butler who provided laboratory space and equipment; P. Dutton, D. Griffith, M. Maloney, and M. Shane who assisted with field collections; S. Johnson, D. Mayer, R. Orhun and others who reared white seabass; K. Miller-McClune who assisted with data analysis; John Hunter, Richard Ford, and Stuart Hurlbert who reviewed and greatly improved an early version of the manuscript; and three anonymous re- viewers for helpful comments on the final manu- script. This study was supported by the Ocean Re- source Enhancement and Hatchery Program (OREHP) through grants to Richard Ford of San Di- ego State University and Donald B. Kent of Sea World Research Institute. OREHP is administered by the California Department of Fish and Game. This study was conducted in partial fulfillment of the requirements of a Masters degree at San Diego State University. Literature cited Allen, L. G., and M. P. Franklin. 1988. Distribution and abundance of young-of-the-year white seabass, Atractoscion nobilis, in the vicinity of Long 720 Fishery Bulletin 95(4), 1997 Beach Harbor, California in 1984—1987. Calif. Fish Game 74:245-248. 1992. Abundance, distribution, and settlement of young- of-the-year white seabass Atractoscion nobilis in the South- ern California Bight. Fish. Bull. 90:633-641. Allen, M. J. 1990. The biological environment of the California halibut, Paralichthys californicus. In C. W. Haugen (ed.), The California halibut, Paralichthys californicus, resource and fisheries, p. 7-29. Calif. Dep. Fish Game, Fish Bull. 174. Bailey, K. M., and E. D. Houde. 1989. Predation on eggs and larvae of marine fishes and the recruitment problem. Adv. Mar. Biol. 25:1-83. Bergman, M. J. N., H. W. van der Veer, and J. J. Zijlstra. 1988. Plaice nurseries: effects on recruitment. J. Fish Biol. 33A:201-218. Bradford, M. J. 1992. Precision of recruitment predictions from early life stages of marine fishes. Fish. Bull. 90:439-453. Clark, F. N. 1930. Size at first maturity of the white seabass. Calif. Fish Game 16:319-323. Clutter, R. I. 1967. Zonation of nearshore mysids. Ecology 48:200-208. Collins, R. 1981. Pacific coast croaker resources. Mar. Recr. Fish. 6:41-49. Dixon, R. L., and W. J. Eckmayer. 1975. A checklist of elasmobranchs and teleosts in the outer harbor of Anaheim Bay. In E. D. Lane and C. W. Hill (eds.), The marine resources of Anaheim Bay, p. 175-183. Calif. Dep. Fish Game, Fish Bull. 165. Dutton, P. H. 1989. The feeding ecology and growth of white seabass lar- vae ( Atractoscion nobilis). M.S. thesis, San Diego State Univ., San Diego, CA, 146 p. Feder, H. M., C. H. Turner, and C. Limbaugh. 1974. Observations on fishes associated with kelp beds in southern California. Calif. Dep. Fish Game, Fish Bull. 160, 144 p. Ford, R. E. 1965. Distribution, population dynamics and behavior of a bothid flatfish, Citharichthys stigmaeus. Ph.D. diss., Univ. California, San Diego, CA, 243 p. Fuiman, L. A., and A. E. Margurran. 1994. Development of predator defences in fishes. Rev. Fish Biol. Fish. 4:145-183. Hickey, B. M. 1993. Physical oceanography. In M. D. Dailey, D. J. Reish, and J. W. Anderson (eds.). Ecology of the Southern Cali- fornia Bight, p. 19-70. Univ. California Press, Berkeley, CA. Horn, Ml. H., and L. G. Allen. 1981. Ecology of fishes in upper Newport Bay, California: seasonal dynamics and community structure. Calif. Dep. Fish Game, Mar. Resour. Tech. Rep. 45, 102 p. Houde, E. D. 1987. Fish early life history dynamics and recruitment variability. Am. Fish. Soc. Symp. 2:17-29. Karakiri, M., R. Berghahn, and H. Von Westernhagen. 1989. Growth differences in 0-group plaice Pleuronectes platessa as revealed by otolith microstructure analysis. Mar. Ecol. Prog. Ser. 55:15-22. Kim, B. G. 1987. Effects of stocking density and food concentration on survival and growth of larval white seabass (Atractoscion nobilis). M.S. thesis, San Diego State Univ., San Diego, CA, 110 p. Kingsford, M. J., and J. H. Choat. 1986. Influence of surface slicks on the distribution and onshore movements of small fish. Mar. Biol. 91:161-171. Klingbeil, R. A., R. D. Sandell, and A. W. Wells. 1975. An annotated checklist of the elasmobranchs and teleosts of Anaheim Bay. In E. D. Lane and C. W. Hill (eds.), The marine resources of Anaheim Bay, p. 79- 90. Calif. Dep. Fish Game, Fish Bull. 165. Kramer, S. H. 1990. Distribution and abundance of juvenile California halibut, Paralichthys californicus, in shallow waters of San Diego County. In C. W. Haugen (ed.). The California hali- but, Paralichthys californicus, resource and fisheries, p. 99-126. Calif. Dep. Fish Game, Fish Bull. 174. 1991. Growth, mortality, and movements of juvenile Cali- fornia halibut Paralichthys californicus in shallow coastal and bay habitats of San Diego County, California. Fish. Bull. 89:195-207. Kulczycki, G. R., R. W. Virnstein, and W. G. Nelson. 1981. The relationship between fish abundance and algal biomass in a seagrass-drift algae community. Estuarine Coastal Shelf Sci. 12:341-347. Love, M. S., J. S. Stephens Jr., P. A. Morris, M. M. Singer, M. Sandhu, and T. C. Sciarrotta. 1986. Inshore soft substrata fishes in the Southern Cali- fornia Bight: an overview. Calif. Coop. Oceanic Fish. In- vest. Rep. 27:84-106. Mact'aH, A. D. 1986. Changes in the biomass of the California Current ecosystem. In K. Sherman and L. M. Alexander (eds.), Am. Assoc. Adv. Sci. Selected Symposium 99: variability and management of large marine ecosystems, p. 33- 54. Westview Press, Boulder, CO. Margulies, D. 1989. Size-specific vulnerability to predation and sensory system development of white seabass, Atractoscion nobilis, larvae. Fish. Bull. 87:537-552. McMichael, R. H., Jr., and K. M. Peters. 1989. Early life history of spotted seatrout, Cy noscion nebulosus (Pisces: Sciaenidae), in Tampa Bay, Florida. Estuaries 12:98-110. Methot, R. 1983. Management of California’s nearshore fishes. Mar. Recr. Fish. 8:161-172. Moser, H. G., D. A. Ambrose, M. S. Busby, J. L. Butler, E. M. Sandknop, B. Y. Sumida, and E. G. Stevens. 1983. Description of early stages of white seabass, Atractoscion nobilis, with notes on distribution. Calif. Coop. Oceanic Fish. Invest. Rep. 24:182-193. Orhun, M. R. 1989. Early life history of white seabass (Atractoscion nobilis). M.S. thesis, San Diego State Univ., San Diego, CA, 162 p. Quast, J. C. 1968. Observations on the food of kelp-bed fishes. In W. J. North and C. L. Hubbs (eds.). Utilization of kelp-bed resources in southern California, p. 109-142. Calif. Dep. Fish Game, Fish Bull. 139. Roberts, D., E. BeMartini, C. Engel, and K. Plummer. 1982. A preliminary evaluation of prey selection by juve- nile-small adult California halibut (Paralichthys califor- nicus) in nearshore coastal waters off southern Cali- fornia. In G. M. Cailliet and C. A. Simenstad (eds.), Gut- shop ’81, fish food habits studies: proceedings of the third Donohoe: Age, growth, distribution, and food habits of Atractoscion nobilis 721 Pacific workshop, p. 214-223. Washington Sea Grant, Seattle, WA. Robertson, A. I., and R. C. J. Lenanton. 1984. Fish community structure and food chain dynamics in the surf-zone of sandy beaches: the role of detached macrophyte detritus. J. Exp. Mar. Biol. Ecol. 84:265-283. Sheridan, P. F. 1979. Trophic resource utilization by three species of sciaenid fishes in a northwest Florida estuary. Northeast Gulf Sci. 3:1-14. Sissenwine, M. P. 1984. Why do fish populations vary? In R. M. May, (ed.), Exploitation of marine communities, p. 59-94. Springer- Verlag, Berlin. Skogsberg, T. 1939. The fishes of the family Sciaenidae (croakers) of California. Calif. Dep. Fish Game, Fish Bull. 54, 62 p. Sokal, R. R., and F. J. Rohlf. 1981. Biometry, 2nd ed. W. H. Freeman and Company, New York, NY, 859 p. Stickney, R. R., G. L. Taylor, and D. B. White. 1975. Food habits of five species of young southeastern United States estuarine Sciaenidae. Chesapeake Sci. 16:104-114. Thomas, J. C. 1968. Management of the white seabass( Cynoscion nobilis) in California waters. Calif. Dep. Fish Game, Fish Bull. 142, 34 p. Vokovich, M., and R. J. Reed. 1983. White seabass, Atractoscion nobilis , in California- Mexican waters: status of the fishery. Calif. Coop. Oce- anic Fish. Invest. Rep. 24:79-83. Young, P. H. 1973. The status of the white seabass resource and its management. Calif. Dep. Fish Game, Mar. Resour. Tech. Rep. 15, 10 p. Zedler, J. B. 1982. Ecology of southern California coastal salt marshes: a community profile. U.S. Fish and Wildlife Service, Washington, D.C.,110 p. : 722 Analysis of diet and feeding strategies within an assemblage of estuarine larval fish and an objective assessment of dietary niche overlap Daniel J. Gaughan San C. Potter School of Biological and Environmental Sciences Murdoch University Murdoch, Western Australia 6 1 50, Australia Present address for senior author: WA Marine Research Laboratories Bernard Bowen Fisheries Research Institute Fisheries Department of Western Australia PO Box 20, North Beach, WA 6020, Australia E-mail address (for D, J. Gaughan): dgaughan@fish.wa.gov.au Abstract .—Fish larvae and zoo- plankton were sampled during seven consecutive months from four regions of Wilson Inlet, an estuary in south- western Australia. Mouth size, prey size, and dietary composition of larvae of the gobiids Afurcagobius suppositus, Pseudogobius olorum, and Favonigobius lateralis, the blenniid Parablennius tasmanianus, and the syngnathid Uro- campus carinirostris were determined. Dietary niche overlap (DNO) was calcu- lated for co-occurring species pairs, both with and without incorporating a measure of relative prey (zooplankton) abundance. Significance of DNO was assessed 1) objectively, with boot- strapping of the dietary data and 2) subjectively, by assigning significance to values >0.6. The diet of A. suppositus was dominated by harpacticoids, poly- chaete larvae, and the calanoid Gladio- ferens imparipes, whereas diets of the other species were dominated by cope- pod nauplii and postnaupliar stages of the cyclopoid Oithona simplex, the pro- portions of the latter increasing with growth of the larvae. Small numbers of large and small prey items were found in the stomachs of A. suppositus (mean=2.5), which had the largest mouth, whereas large numbers (mean= 28.7) of small prey and no large items were found in the stomachs of P. tasmanianus, which had the second largest mouth. Between these ex- tremes, P olorum, U. carinirostris, and F. lateralis each ate mostly small and intermediate-size prey, supplemented by a few large prey. The data did not support the hypothesis that an increase in the difference in gape size between species would decrease the prevalence of significant DNO. The lack of a con- sistent relation between mouth size and DNO among the five species is evidence that interspecific dietary differences reflect differences in feeding behavior. With bootstrapping, the prevalence of significant (P<0.05) DNO between spe- cies pairs was 32.6% when prey data were included in the analyses and 46.5% when prey data were not in- cluded. By subjectively assigning sig- nificance to DNO values >0.6, we ob- tained substantially less conservative estimates that indicated the prevalence of significant DNO was >53%. Manuscript accepted 31 March 1997. Fishery Bulletin 95:722-731 (1997). Starvation has been considered a major cause of mortality in larval fish (e.g. Hunter, 1984), although evidence from the field has been dif- ficult to obtain (Heath, 1992). If starvation occurs within an assem- blage of larval fish, competition for food is expected to contribute to that starvation. However, indications of any such competition may go unde- tected if the diet of only a single spe- cies is examined or if the influence of other planktivores (Fortier and Harris, 1989) is not considered. In the case of larval fish, relatively few studies have examined in detail the diets of several co-occurring species (e.g. Last, 1980; Laroche, 1982; Govoni et al., 1983; Watson and Davis, 1989). Furthermore, no study has objectively assessed the significance of dietary niche over- lap (DNO), where DNO refers to the amount of sharing of food resources among larval fish. Assessments of whether dietary overlap within or between species (including larval fish) is significant have been based on indices that range from 0 for no overlap to 1 for complete overlap, with values greater than 0.5, 0.6, or 0.7 considered to be significant (e.g. Harmelin-Vivien et al., 1989; Cervel- lini et al., 1993; Vega-Cendejas et al., 1994; Hartman and Brandt, 1995). However, because these cutoff points are arbitrary, they are not necessarily biologically significant, as in the case with fish larvae where individuals of co-occurring species may be confronted with large con- centrations of the same prey type and thus any similarities in diet may be due to chance encounters. Although dietary compositions of larval fish have been considered in the context of the abundance of the zooplankton prey of those larvae (e.g. Dagg et al., 1984; Jenkins, 1987; Hirst and DeVries, 1994; Welker et al., 1994), no study of di- etary overlap between larval fish has incorporated data on their zooplank- ton prey. This analysis is necessary in order to assess whether there is a likelihood of competition for food. Differences in mouth structure of fish may lead to differences in feed- ing success on particular prey types (Lavin and McPhail, 1986). Co-oc- curring larvae of different species with similar-size mouths may therefore exhibit a higher rate of significant DNO than those with mouths of different size. Because fish larvae usually swallow prey Gaughan and Potter: Analysis of diet and feeding strategies within an assemblage of estuarine larval fish 723 whole, mouth size limits prey size; thus prey width is typically the limiting dimension for ingestion (e.g. Hunter, 1984, Heath, 1992). It is therefore impor- tant to examine mouth size and prey width when exploring the trophic relations of larval fish. The first aim of this study was to examine the re- lation between mouth width, prey width, and dietary composition of the larvae of five teleosts in an estu- ary. The dietary data were then used to examine the extent of DNO between these species with a tech- nique that takes into account relative prey concen- trations. With this procedure we were able to test the hypothesis that divergence in gape size between species should be accompanied by a decrease in the prevalence of significant DNO between these species. Bootstrapping was used to assess whether species- pair DNO values were significant. The results of us- ing this robust approach were compared with those obtained when relative prey concentrations were not included in the calculation of DNO and when a sub- jective level of >0.6 was considered to be significant for the DNO values. Materials and methods Sampling methods This study was carried out in Wilson Inlet (35°00'S, 117°24’E), an estuary in southwestern Australia that comprises a 48 km2 basin with two main tributaries, the Denmark and Hay rivers. Although samples were collected monthly between July 1988 and June 1989, the data used in this paper are restricted to those obtained between October 1988 and April 1989 when fish larvae were most abundant. Ichthyoplankton and zooplankton were sampled from open waters of the upper, middle, and lower basin, and the central channel of the lower saline reaches of the Denmark River, located 10.7, 8.3, 2.0, and 7.3 km, respectively, from the estuary mouth. The water depth in each region was between 2 and 3 m. Sampling was initiated soon after sunset to reduce the likelihood of larvae avoiding the plankton nets. Fish larvae were collected with a pair of 500-pm- mesh conical nets, each with a mouth diameter of 0.6 m and a length of 2 m. The nets were attached to either side of a powerboat and towed for 10 min just below the surface of the water at a speed of 1.5 m/s. During each ichthyoplankton tow, three to five zoop- lankton samples were taken from the surface with a conical, 53-pm-mesh net with a mouth diameter of 0.35 m. The volumes of water filtered during each ichthyoplankton and zooplankton tow were measured with flowmeters. The zooplankton tows were 7-10 s in duration. The flowmeter in the zooplankton net was closely observed during each tow. A tow was im- mediately terminated if the propeller speed suddenly decreased — a sign that the net was clogging. Samples were fixed in a 5% solution of formalin, which was replaced with 70% ethanol on the following day. The detailed results of the zooplankton sampling are given in Gaughan and Potter (1995). Laboratory procedures and data analyses Zooplankton were identified and counted under a dissecting microscope from subsamples of the repli- cate samples. Counts were standardized to numbers/ m3; thus mean concentrations of taxa at each region within each month were able to be calculated. Rela- tive proportions of those zooplankton taxa that con- tributed to larval diets at any time during the study were calculated for each sample. These represented relative resource availability (i?.). The gobiids Pseudogobius olorum, Afurcagobius suppositus, and Favonigobius lateralis , the blenniid Parablennius tasmanianus, and the syngnathid Urocampus carinirostris were chosen for the present study because their larvae are abundant in Wilson Inlet from late spring to early autumn (Neira and Potter, 1992), collectively contributing 70.8% of the total open-water assemblage of larval fish in this estuary between September 1987 and April 1989. All larvae of each species in a sample were removed and counted. Body length (BL) of each larva (i.e. the distance from the snout to tip of notochord in preflexion and flexion larvae and from the snout to the posterior end of the hypural plate in postflexion larvae ILeis and Trnski, 1989]) was measured to the nearest 0.1 mm. Since a focus of this study was the comparison of mouth width with prey width, the di- ets of individual size classes of larvae were deter- mined. However, because the analyses of DNO were undertaken for co-occurring species within individual samples, the dietary data for all size classes of each species were pooled (see below). The smaller larvae of P. olorum, P tasmanianus, F. lateralis, and A. suppositus were each grouped into 1.0-mm length classes. Because P. olorum and P. tasmanianus >5 mm BL and F. lateralis and A. sup- positus >6 mm BL were rarely caught, larvae of these four species longer than these respective lengths were each grouped into single length classes. Because the length range of U. carinirostris was relatively wide, the larvae of this species were grouped into length classes with intervals of 3 or 4 mm, depending on the numbers caught. Items in the gut were identified and counted. Maxi- mum widths of intact dietary items, which typically 724 Fishery Bulletin 95(4), 1997 represent the limiting dimension for ingestion, were measured to 0.01 mm with an ocular micrometer in a compound microscope. Mouth width of larvae at the widest point of the upper jaw was similarly measured on a subsample of at least 50 larvae of each species. Widths of prey and mouth widths of larvae of each spe- cies were then plotted against body length of larvae. Proportional utilization (p.) of each prey type was calculated for length classes within each species with data pooled across regions and months. Proportional utilization was also calculated across length classes for the larvae of each species within a sample. Guts that were empty or contained only unidentifiable material were not included in these calculations. Relative feeding prevalence of all larvae within samples were correlated against the corresponding estimates of zooplankton abundance to determine if there was any evidence that zooplankton abundance was limiting feeding success. The average DNO ex- hibited between all species within samples was like- wise compared with zooplankton abundance. Calculation of dietary niche overlap Interspecific DNO was calculated for species-pairs within individual samples, i.e. for each site within a given month. Because this part of our study focused on examining niche relations between species, pooled diets for each species within a sample were consid- ered to represent the average diet of each species. Furthermore, comparisons were limited to those samples in which >10 larvae of each of two or more species contained food. By pooling data across size classes we were able to compare more DNO data. Although the use of average diets would reduce the robustness of a parametric test of significance, we used nonparametric techniques for assessing the sig- nificance of DNO. Sufficient numbers of larvae were obtained for analy- sis in 13 of the 28 ichthyoplankton samples (7 months x 4 regions) to allow 43 pairwise comparisons to be made between the diets of co-occurring species. Besides the DNO values that were calculated and that incorporated prey abundance data, the results of this technique were also evaluated against calcu- lations of DNO that did not incorporate such data. Because prey abundance data are incorporated into prey utilization data prior to calculating DNO (see below), the same formula was used to calculate DNO both with and without consideration of prey concen- trations. DNO was measured with the symmetric niche overlap coefficient (Pianka, 1973) = (X PijPik ) / Pi 5X)’ where ptj and pik = the proportional utilization of prey type i by species j and k, respectively. Using the pt data directly, we were able to provide a basis for calculating DNO without prey abundance data. To incorporate prey abundance data, the geo- metric mean (g;) of p; and electivity (e;) was used (Winemiller and Pianka, 1990), instead of p; as in the original formula. The geometric mean gives a better indication of ecological similarity by reducing those biases within both p; and e; that can result from the presence of very abundant or very rare prey types (Winemiller and Pianka, 1990). Electivity is the value that has been weighted by resource availabil- ity (R}) as e; = p/Rr These values were calculated within the DNO algorithm and are not presented. Note that g: can be used with other overlap indices because it is calculated prior to the calculation of the overlap value. Bootstrapping of the resource matrix of a pair of species was used to obtain a null distribution of 1,000 pseudo-DNO values against which the significance of observed DNO could be assessed (Winemiller and Pianka, 1990). These calculations were performed for species pairs at sites within months. In each one of the 1,000 runs, the algorithm randomly reassigned theg; values for each prey type (e.g. resource states i...p) within each larval species j and k, but among the resource types used by both j and k (e.g. amongst Siy-Snj and Sik-Spk^- A Si value for one of the species pair may thus be reassigned to a resource state which was used only by the other species. The null hypothesis ( H0 ) for each test was that the dietary compositions of the larvae of the two species were not the same. The null hypothesis was rejected if more than 95% of the 1,000 pseudo-DNO values were less that the observed DNO. Such cases indi- cated that the observed value was larger than would be randomly expected at P<0.05. The prevalence of significant DNO calculated with gt and bootstrapping was then compared with that obtained with pt, i.e. when R: was not taken into ac- count. These results were also compared with those obtained when the significance of DNO was arbi- trarily set at values >0.6. Results Zooplankton Zooplankton were very abundant in Wilson Inlet be- tween October 1988 and April 1989; there was a to- tal mean monthly concentration of 342,746 organ- isms/m3 (range = 48,641-2,951,209/m3). Concentra- tions exceeded 100,000/m3 in 26 of the 28 zooplank- Gaughan and Potter: Analysis of diet and feeding strategies within an assemblage of estuarine larval fish 725 ton samples. Mean monthly concentrations of zoop- lankton in Wilson Inlet were similar from October 1988 to April 1989; only January had a significantly (P<0.05) higher concentration (Gaughan and Potter, 1995). All copepods, irrespective of stage, contributed 74.7% of the total mean concentration of zooplank- ton (Table 1). The cyclopoid Oithona simplex , the calanoids Gladioferens imparipes and Acartia sim- plex, and several species of harpacticoids were the only copepods that were common in Wilson Inlet (Gaughan and Potter, 1995). Considering just the copepods, adults of A. simplex and G. imparipes, which represented the largest prey types consumed by fish larvae in Wilson Inlet, contributed only 2.9% of the total mean concentration of this taxonomic group. The smaller species and developmental stages of copepods were thus approximately 33 times more abundant than the adults of A. simplex and G. imparipes collectively. The mean concentrations and relative contributions of other zooplankton taxa that were eaten by larval fish during this study are shown in Table 1. Numbers of prey items consumed and dietary composition of fish larvae The total number of larvae of each species examined during this study are shown in Table 2, and the num- bers of larvae in size classes of each species which contained food are shown in Figure 1. The mean number of prey items found in each larva was less than five for A. suppositus and F. lateralis. Table 1 Mean monthly concentrations and relative contributions of zooplankton in Wilson Inlet between October 1988 and April 1989. Mean Relative concentration contribution Taxa no. organisms/m3) (%) Copepod nauplii 164,827 48.1 Calanoid copepodites 33,755 9.8 Oithona simplex 44,629 13.0 Acartia simplex 6,771 2.0 Gladioferens imparipes 719 0.2 Harpacticoids 5,523 1.6 Polychaete larvae 34,997 10.2 Bivalve larvae 18,435 5.4 Synchaeta cf. baltica 9,787 2.9 Other taxa 23,303 6.8 Total 342,746 between five and ten for P. olorum and 17. carini- rostris and 28.7 for P tasmanianus (Table 2). Like- wise, the maximum numbers of prey consumed were much lower for the first four species than for P tasmanianus (Table 2). The number of prey ingested by larvae increased with body size only in the case of U. carinirostris. Various developmental stages of copepods domi- nated the diets of larval fish in Wilson Inlet; the ro- tifer Synchaeta cf. baltica and the larvae of bivalves and polychaetes were also occasionally important (Fig. 1). Only the postnaupliar stages of copepods in the diets were identified to species. Each of the com- mon types of copepod contributed to the diets of fish larvae. During the growth of P olorum, F. lateralis, U. carinirostris , and P tasmanianus , the contribution of copepod nauplii declined while that of postnaupliar stages increased (Fig. 1). Oithona simplex was par- ticularly important to P. olorum and U. carinirostris, representing over 30% of the diet of the three larg- est size classes of the former species and over 40% of the diet of all size classes of the latter species. By contrast, despite the increased contribution by O. simplex to larger size classes ofP tasmanianus, cope- pod nauplii dominated the diet of all size classes, contributing > 40% to each (Fig. 1). The diet of F. lateralis <4.0 mm BL consisted mainly of copepod nauplii and to a lesser extent of bivalve larvae and O. simplex. The main prey types of F. lateralis from 6. 0-7. 9 mm BL were harpacticoids (40.9%), calanoid copepodites (20.6%), and phytoplankton (15.6%). Polychaete larvae, copepod nauplii, and harp- acticoids each contributed over 25.0% of the diets of the smallest size class of A. suppositus, whereas harpacticoids alone contributed 60.0% to larvae >6.0 mm BL (Fig. 1). Harpacticoids also contributed 15.3 and 27.0% of the diet in the 4. 0-4. 9 and 5. 0-5. 9 mm length classes, respectively. Gladioferens imparipes Table 2 Mean and maximum numbers of prey per larva for five teleost species in Wilson inlet, n = the total number of lar- vae of each species examined. Species n Mean prey/larva Maximum prey/larva Pseudogobius olorum 1,946 7.7 30 Favonigobius lateralis 451 4.0 15 Afurcagobius suppositus 485 2.5 19 Parablennius tasmanianus 469 28.7 103 Urocampus carinirostris 434 9.7 23 726 Fishery Bulletin 95(4), 1997 was abundant only in the diet of the 4. 0-4. 9 and 5.0- 5.9 mm length classes (Fig. 1). Frey width and larval mouth width Urocampus carinirostris had the smallest mouth, P tasmanianus and A. suppositus the widest mouths (Fig. 2, A, D, E). The shapes of the mouths were most similar in the case of P. olorum and F. lateralis (Fig. 2, B and C). Although mouth width of U. carinirostris increased linearly with body length (Fig. 2A), such an increase in the other four species was best de- scribed by a polynomial function (Fig. 2, B-E: Table 3). Mouth width of U. carinirostris increased slowly from 0.19 mm at 8 mm BL to 0.28 mm at 19 mm BL (Fig. 2A). The rate at which mouth width increased with body length was greater for the other four species (Fig. 2, B-E). In A. supposi- tus, mouth width increased from 0.33 mm at 3.5 mm BL to 0.68 mm at 6.0 mm BL (Fig. 2E). The smallest larvae of P olorum, F. lateralis, and P tasmanianus had nar- rower mouths (<0.15 mm) than both A. suppositus and U. carinirostris, owing to their smaller size upon arrival in the plankton (Fig. 2, A-E). However, mouth widths of the first three of these species exceeded the maximum recorded for U. carinirostris (0.28 mm) by the time each had attained 5 mm BL and approached 0.60 mm in larger larvae. The slope describing the relation be- tween prey width and larval length for each species was less than 0.03. The ex- tent to which prey width increased with length was thus very small for each spe- cies. Minimum prey width for each spe- cies was about 0.04 mm (Fig. 2, A-E). The large numbers of prey of each species with widths of 0.04-0.08 mm were predomi- nantly attributable to copepod nauplii. Prey widths of between 0.04 and 0. 18 mm predominated in U. carinirostris, P. olorum, and F. lateralis (Fig. 2, A— C). Parablennius tasmanianus ate mainly prey 0.04-0. 13 mm wide, but with a maxi- mum width of only 0.18 mm (Fig. 2D). Afurcagobius suppositus consumed the largest prey items, i.e. postnaupliar stages of G. imparipes and A. simplex, with widths of 0.12-0.30 mm and 0.10- 0.16 mm respectively. As with the other species, A. suppositus also ate smaller items (0.04-0.10 mm) (Fig. 2E). As U. carinirostris, P. olorum, F. lateralis, and P. tasmanianus grew, they continued to eat many prey <0.10 mm wide, even though the smaller larvae of each were capable of eating prey >0.10 mm (Fig. 2, A-D). Prey width was about 0.10 mm narrower than mouth width for most of the length range of U. carini- rostris (Fig. 2 A). The widths of the larger Pseudogobius olorum 371 395 145 183 2. 0-2. 9 3. 0-3. 9 4 0-4.9 5.0-7 9 Favonigobius lateralis ^ 77 18 10 ^ 100r (3 C O C o Q, O B Others PH Synchaeta cf. baltica H Bivalve larvae B Polychaete larvae I"] Harpacticoids fl Calanoid copepodites ■ Acartia simplex IH Gladioferens imparipes ED Oithona simplex EH Copepod nauplii Parablennius tasmanianus 20 163 90 25 2.0-2 9 3.0-3 9 6 0-7 9 2.0-2 9 3. 0-3. 9 4 0-4 9 5 0-7 9 Afurcagobius suppositus 21 129 38 15 3.0-3.9 4.0-4 9 5 0-5.9 6 0-6.9 Urocampus carinirostris 21 137 85 25 60-9.9 13.0-15.9 10.0-12.9 16.0 9.9 Length class (mm) Figure 1 Proportional utilization (pt, %) for length classes of larvae of five species of fish caught at four sites in Wilson Inlet between October 1988 and April 1989. The numbers of larvae that contained food in each size class are shown above the bars. Gaughan and Potter: Analysis of diet and feeding strategies within an assemblage of estuarine larval fish 727 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 T A Urocampus carinirostris i ._j . i , i ._i . i , — i — . — i — ■ i . — i — . — i — i — i — i_j 7.0 9.0 11.0 13.0 15.0 17.0 19.0 0 mouth width ♦ prey width Figure 2 Prey width and larval mouth width for larvae of five species of fish caught in Wilson Inlet between October 1988 and April 1989. Lateral and ventral views of the head of a representative larva of each species are also given; the upper jaw is indicated in black, the lower jaw is indicated with stippling (scale bar equals 1.0 mm). The examples were each taken from a 5.0-mm-BL larva, except the example of Urocampus carinirostris, which was taken from a 10.5-mm-BL larva. 728 Fishery Bulletin 95(4), 1997 Table 3 The relations between mouth width (MW) and body length (BL) for the larvae of five teleost species in Wilson Inlet. Species n Regression function r2 Urocampus carinirostris 69 MW = 0.139 + 0.006(£L) 0.453 Afurcagobius suppositus 50 MW = -15.666 + 14.256(BL) - 4.687(BL2) + 0.671(BL3) - 0.035(£L4) 0.849 Pseudogobius olorum 98 MW = -0.729 + 0.839(£L) - 0.273 (BL2) + 0.038 (BL3) - 0.002 (BL4) 0.921 Favonigobius lateralis 50 MW = 0.061 + 0.029(BL) + 0.005(£L2) 0.963 Parablennius tasmanianus 66 MW = -0.517 + 0.41KBL) - 0.073(£L2) + 0.005(BL3) 0.889 Table 4 The number of cases of significant dietary niche overlap (DNO) for larvae of five teleost species in Wilson Inlet between October 1988 and April 1989. Co-occurring species were compared only when >10 individuals of each contained food in their guts. The number of cases for which pairwise comparisons could be made is shown as n. The number of significant cases are presented for DNO calculations I) that used zooplankton concentrations and II) that did not use zooplankton concentrations. Within these categories, the number of significant cases were determined a) at P<0.05 from a null distribution derived from bootstrapping and b) with an arbitrary cutoff level for significance at an overlap value >0.6. DNO was measured with a modification of Pianka’s (1973) symmetric niche overlap coefficient. Pseudogobius Afurcagobius Favonigobius Parablennius olorum suppositus lateralis tasmanianus n (I) (II) n (I) (II) n (I) (II) n (I) (II) (a, b) (a, b) (a, b) (a, b) (a, b)(a, b) (a, b) (a, b) Afurcagobius suppositus 5 (0, 2) (0, 0) — Favonigobius lateralis 4(1, 2) (1, 2) 1 (0, 0) (0, 0) — Parablennius tasmanianus 8(1, 4) (3, 6) 1 (0, 0) (0, 0) 4(2, 4) (3, 4) — Urocampus carinirostris 8 (7, 7) (8, 8) 2(1, 1) (0, 1) 4(1, 2X2, 3) 6(1, 4) (2, 4) Totals n (I) (II) (a, b) (a, b) 43 (14, 23) (20, 28) Percent of total (32.6, 53.5) (46.5, 65.1) prey items consumed by P olorum <4.0 mm BL were similar to mouth width, as was also the case with F. lateralis of 2. 0-2. 5 mm BL and A. suppositus of 4.0- 4.5 mm BL. For each of these three gobiid species, the maximum prey width of larvae >5 mm BL was far less than mouth width. This difference exceeded 0.20 mm in the larger gobiid larvae. Likewise, maxi- mum prey widths for larval P. tasmanianus ap- proached mouth width in smaller larvae but were much less for larger larvae (Fig. 2D). Dietary niche overlap Dietary niche overlap between P. olorum and U. carinirostris ranged from 0.543 to 0.983 and was sig- nificant on seven of the eight occasions in which these species co-occurred (Table 4). Although DNO ranged from 0.764 to 0.980 on the four occasions that P tasmanianus and F. lateralis co-occurred, overlap was significant only twice (Table 4). There were a few other cases of significant DNO amongst the larval fish assemblage; DNO was particularly low between A.suppositus and the other species, being significant only with U. carinirostris on one occasion. Of the 43 pairwise comparisons that could be made between the diets of co-occurring larvae of the five species, there were 14 cases (32.6%) of significant DNO (Table 4). This increased to 20 (46.5%) if zoo- plankton data were not included in the calculations of DNO’s. If DNO >0.6 had been considered signifi- cant, the number of significant cases increased from 32.6% to 53.5% for calculations which included zoo- plankton data and from 46.5% to 65.1% for those which did not include these data. The magnitudes of the differences for the preva- lence of significant DNO found by using bootstrap- Gaughan and Potter: Analysis of diet and feeding strategies within an assemblage of estuarine larval fish 729 ping (18.6% and 20.9%) were greater than those ob- tained by accounting for zooplankton abundance data (11.6% and 13.9%, Table 4). Relation between zooplankton abundance and both feeding prevalence and mean DNO Zooplankton abundance at sites within months was not significantly related to either feeding prevalence (P>0.05, r=0.346, n= 26) or mean DNO (P>0.1, r=0.146, n=15) of fish larvae within the correspond- ing samples. Discussion Numbers of prey consumed and prey size Afurcagobius suppositus, because it ingested larger prey types, e.g. G. imparipes, consumed the least number of prey. The other species ate larger num- bers of small and intermediate-size prey, e.g. cope- pod nauplii and O. simplex. The significant increase in numbers of prey with length for U. carinirostris only was probably attributable to the fact that the magnitude of the range of length of individuals ex- amined for this species was 12 mm, whereas that of the other species was less than 6 mm. Despite marked increases in mouth width during the growth of each species except U. carinirostris, average prey width of the five species increased only slightly with growth. Although smaller larvae con- sumed prey almost as wide as their mouths, larger larvae typically ate prey far smaller than their mouth size. Furthermore, the smaller larvae of each spe- cies consumed some prey items almost as wide as those eaten by larvae in the larger size classes. These data indicate that mouth width was not limiting the ingestion of larger prey types among larvae in the larger size classes. The dominance of relatively small prey in the di- ets of larval fish in Wilson Inlet reflects the domi- nance of these types of zooplankton in the environ- ment. During the study period, copepod nauplii, O. simplex, calanoid copepodites and harpacticoids were 33 times more abundant than the adults of G. imparipes and A. simplex collectively, the only com- mon large prey. Thus, as has previously been found for other larval fish (e.g. Ware and Lambert, 1985; Kellermann, 1990), prey availability strongly influ- enced the sizes of prey consumed. From an early age and size, Afurcagobius sup- positus ate larger prey than the other four species. Since this species hatches at a more advanced stage and with better developed fins than the other four species (Neira et al., in press), they were probably su- perior swimmers and thus more efficient at captur- ing larger prey. Greater mobility may have also re- sulted in A. suppositus searching a larger volume of water (Hunter, 1984), which would increase the rate at which the larger and less abundant zooplankton were encountered. The possession of a larger mouth ap- parently allows A. suppositus to take advantage of larger prey in presumably more frequent encounters. Mouth size and DNO Although A. suppositus had the largest mouth and consumed the largest and most diverse prey, the rela- tive differences in mouth size of the other four spe- cies were not accompanied by corresponding differ- ences in the size and composition of prey. A lack of a predictive relation between mouth size and diet has previously been recorded for fish larvae from another estuary (Laroche, 1982) and more recently for lar- vae of freshwater fish in an experimental situation (Bremigan and Stein, 1994). In Wilson Inlet, this situ- ation was further highlighted by the lack of a rela- tion between the prevalence of significant DNO and the mouth structure of the five species. Thus, the prevalence of significant DNO was not particularly high between P. olorum and F. lateralis (Table 4), the species with the most similar mouth structure, whereas P. olorum and U. carinirostris had very simi- lar diets, as indicated by the high prevalence of sig- nificant DNO (Table 4), but had different-size mouths. Conversely, the diet of A. suppositus over- lapped significantly only with that of U. carinirostris, the species with the smallest mouth. Parablennius tasmanianus and F. lateralis, the only other species- pair to exhibit more than one case of significant DNO, also had dissimilar mouths. Finally, A. suppositus and P tasmanianus had the largest mouths but the most divergent diets. Along with the general lack of a relation between mouth size and diet, the relatively frequent occur- rence (32.6%) of significant DNO amongst the larval fish in Wilson Inlet, when prey abundance was taken into account, was also attributable to the high con- centrations of relatively limited choices of acceptable prey types. The lack of a relation between concen- trations of zooplankton and both feeding prevalence and mean DNO within samples was also probably a result of consistently high concentrations of zooplank- ton. Consequently, significant DNO among larval fish in Wilson Inlet provided no evidence of competition for food. Furthermore, Gaughan and Potter (1995) found that abundances of zooplankton and larval fish were significantly correlated at only two of the four sampling regions in Wilson Inlet. The lack of a rela- 730 Fishery Bulletin 95(4), 1997 tion at the other two regions was due to large fluc- tuations in the abundance of zooplankton between months. These fluctuations did not appear to influ- ence monthly trends in the abundance of fish larvae, probably because concentrations of zooplankton typi- cally remained high(> 100,000/m3). Because in this study we were limited to examin- ing the diets of larval fish, a complete assessment of dietary relations and the potential for competition within the plankton community could not be under- taken. However, other zooplankton taxa (e.g. Sag- itta minima) sufficiently large to have used the same food resources as larval fish were rare in Wilson In- let, contributing less than 0.2% of the total numbers of zooplankton (Gaughan and Potter, 1995). Feeding strategies The diets of the fish larvae from Wilson Inlet may be viewed as representing a spectrum of feeding strategies. The diet of A. suppositus is distinguished from those of the other four species by its broader composition, the larger size of its prey items, and the smaller numbers of prey consumed. At the op- posing end of the spectrum, P tasmanianus larvae consumed large numbers of small prey items. The feeding strategies of the larvae of P. olorum, F. lateralis, and U. carinirostris lay between these ex- tremes; these species consumed many small and in- termediate-size prey which were occasionally supple- mented with larger prey items. Because the trophic character of a species may be influenced by both size and structure as well as be- havior (Lavin and McPhail, 1986), the small influ- ence of mouth width on the size of prey consumed by larval fish in Wilson Inlet indicates that the differ- ent feeding patterns among larval species probably resulted from behavioral differences (Bremigan and Stein, 1994). These patterns, which occurred despite the high concentrations of zooplankton, may enhance survival, and hence recruitment, if marginally low concentrations of zooplankton were present at tem- poral or spatial scales beyond those sampled. Evaluation of methods In this study, we examined a technique for assessing DNO that consists of two parts (accounting for zoop- lankton abundance in the calculation of BNO and objectively assessing significance of DNO with bootstrapping). This technique was substantially more conservative than that which did not consider zooplankton abundance and which did subjectively assess significance. Prevalence of significant DNO thus doubled (32.6% to 65.1%) when zooplankton data were not included in the calculations and an arbitrary cutoff point of 0.6 was used to test for sig- nificance. The less conservative techniques of mea- suring DNO and assessing its significance would have therefore overestimated the degree of DNO among fish larvae in Wilson Inlet. A large overestimation of DNO would likely have led to a different interpretation of the data. For ex- ample, the higher rate of significant overlap may have led to the conclusion that competition for food was sufficiently high to influence markedly the sur- vival rate of fish larvae in Wilson Inlet. In contrast, the lower prevalence of overlap is more consistent with our previous hypothesis that food is unlikely to be limiting for the open-water assemblage of larval fish in Wilson Inlet (Gaughan and Potter, 1995). Although concentrations of potential zooplankton prey are not necessarily directly related to their avail- ability, we suggest that inferences regarding compe- tition for food among larval fish may be misleading if data on the abundance of the zooplankton are not considered when measuring DNO. In studies of other taxa, or even of adult fish, where the abundances of prey in the environment may be very difficult or im- possible to estimate without bias, resource availabil- ity can be estimated in a circular manner with pro- portional-utilization data (see Winemiller and Pianka, 1990). However, because the majority of fish larvae and their potential prey are planktonic, small, and relatively immobile (thus highly susceptible to capture with plankton nets), estimates of prey con- centrations in the environment should be used to calculate DNO for larval fish. Likewise, because a subjective assessment of the significance of DNO is inadequate, bootstrapping techniques may prove to be useful in making an objective examination of di- etary relations, which are typically awkward to ana- lyze statistically (Winemiller and Pianka, 1990; Baltanas and Rincon, 1992). Finally, although the two parts of the technique used in this study, i.e. objectively assessing signifi- cance and accounting for zooplankton abundance, each contributed to the overall result, individually the former had a greater influence ( 18.6% and 20.9%) on the estimated prevalence of significant DNO than the latter ( 11.6% and 13.9%). Even though the direc- tion and magnitude of the differences between the two parts of this technique may apply only to the current study, this finding further suggests that both the incorporation of prey abundance data and an objective assessment of significance need to be con- sidered in an analysis of dietary overlap because ei- ther may have more influence on the apparent preva- lence of significant DNO. Gaughan and Potter: Analysis of diet and feeding strategies within an assemblage of estuarine larval fish 731 Acknowledgments Gratitude is expressed to P. Geijsel, P. Humphries, G. Hyndes, L. Laurenson, and F. Neira for help with sampling, and to W. Fletcher, H. Gill, A. Miskiewicz and M. Platell for helpful comments on the text. The suggestions of two anonymous referees were also appreciated. This work was made possible by an Australian Commonwealth Postgraduate Research Award to D. Gaughan and by a grant from the Fish- eries Department of Western Australia to I. Potter. Literature cited Baltanas, A., and P. A. Rincon. 1992. Application of a cluster-bootstrapping method for identifying the dietary patterns of fish populations. Ecology Freshwater Fish 1:130-139. Bremigan, M. T., and R. A. Stein. 1994. Gape-dependent larval foraging and zooplankton size: implications for fish recruitment across systems. Can. J. Fish. Aquat. Sci. 15:913-922. Cervellini, P. M., M. A. Battini, and V. E. Cussac. 1993. Ontogenetic shifts in the diet of Galaxias maculatus (Galaxiidae) and Odontesthes microlepidotus (Atherini- dae). Environ. Biol. Fish. 36:283-290. Dagg, M. J., M. E. Clarke, T. Nishiyama, and S. L. Smith. 1984. Production and standing stock of copepod nauplii, food items for larvae of the walleye pollock Theragra chalcogramma in the Berring Sea. Mar. Ecol. Prog. Ser. 19:7-16. Fortier, L., and R. P. Harris. 1989. Optimal foraging and density-dependent competition in marine fish larvae. Mar. Ecol. Prog. Ser. 51:19-33. Gaughan, D. J., and I. C. Potter. 1995. Composition, distribution and seasonal abundance of zooplankton in a shallow, seasonally closed estuary in temperate Australia. Estuarine Coastal Shelf. Sci. 41:117-135. Govoni, J. J., D. E. Hoss, and A. J. Chester. 1983. Comparative feeding of three species of larval fishes in the northern Gulf of Mexico: Brevoortia patronus, Leiostomus xanthurus, and Micropogonias undula- tus. Mar. Ecol. Prog. Ser. 13:189-199. Harmelin-Vivien, M. L., R. A. Kaim-Malka, M. Ledoyer, and S. S. Jacob-Abraham. 1989. Food partitioning among scorpaenid fishes in Medi- terranean seagrass beds. J. Fish Biol. 34:715-734. Hartman, K. J., and S. B. Brandt. 1995. Trophic resource partitioning, diets, and growth of sympatric estuarine predators. Trans. Am. Fish. Soc. 124:520-537. Heath, M. R. 1992. Field investigations of the early life stages of marine fish. Adv. Mar. Biol. 28:1-174. Hirst, S. C., and D. R. DeVries. 1994. Assessing the potential for direct feeding interactions among larval black bass and larval shad in two southeast- ern reservoirs. Trans. Am. Fish. Soc. 123:173-181. Hunter, J. R. 1984. Feeding ecology and predation of marine fish larvae. In R. Lasker (ed.), Marine fish larvae: morphol- ogy, ecology and relation to fisheries, p. 34—77. Washing- ton Sea Grant Program, Seattle, WA. Jenkins, G. P. 1987. Comparative diets, prey selection, and predatory impact of co-occurring larvae of two flounder species. J. Exp. Mar. Biol. Ecol. 110:147-170. Kellermann, A. 1990. Food and feeding dynamics of the larval Antarctic fish Nototheniops larseni. Mar. Biol. 106:159-167. Laroche, J. L. 1982. Trophic patterns among larvae of five species of sculpins (Family Cottidae) in a Maine Estuary. Fish. Bull. 80:827-840. Last, J. M. 1980. The food of 25 species of fish larvae in the west-cen- tral North Sea. Ministry of Agriculture, Fisheries and Food (MAFF), Directorate of Fisheries Research, Lowestoft. Fish. Res. Tech. Rep. 60, 44 p. Lavin, P. A., and J. D. McPhail. 1986. Adaptive divergence of trophic phenotype among freshwater populations of the threespine stickleback ( Gasterosteus aculeatus). Can. J. Fish. Aquat. Sci. 43: 2455-2463. Leis, J. M., and T. Trnski. 1989. The larvae of Indo-Pacific shorefishes. New South Wales Univ. Press, Sydney, Australia, 371 p. Neira, F. J., A. G. Miskiewicz, and T. Trnski. In press. Larvae of temperate Australian fishes. Univ. Western Australia Press, Perth, 436 p. Neira, F. J., and I. C. Potter. 1992. The ichthyoplankton of a seasonally closed estuary in temperate Australia. Does an extended period of open- ing influence species composition? J. Fish. Biol. 41:935— 953. Pianka, E. R. 1973. The structure of lizard communities. Ann. Rev. Ecol. Syst. 4:53-74. Vega-Cendejas, M. E., M. Hernandez, and F. Arreguin-Sanchez. 1994. Trophic interrelations in a beach seine fishery from the northwestern coast of the Yucatan peninsula, Mexico. J. Fish Biol. 44:647-659. Ware, D. M., and T. C. Lambert. 1985. Early life history of Atlantic mackerel ( Scomber scombrus) in the southern Gulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 42:577-592. Watson, W., and R. L. Davis Jr. 1989. Larval fish diets in shallow coastal waters off San Onofre, California. Fish. Bull. 87:569-591. Welker, M. T., C. L. Pierce, and D. H. Wahl. 1994. Growth and survival of larval fishes; roles of compe- tition and zooplankton abundance. Trans. Am. Fish. Soc. 123:703-717. Winemiller, K. O., and E. R. Pianka. 1990. Organization in natural assemblages of desert liz- ards and tropical fishes. Ecol. Monogr. 60:27-55. 732 Abstract. -Aspects of the life his- tory of red porgy from the South Atlan- tic Bight (SAB) were examined for four periods (1972-74, 1979-81, 1988-90, and 1991-94), and annual changes in the age and growth of red porgy were described for data collected during 1988-94. The life history of red porgy during 1972-74 was assumed to repre- sent that of an unfished population, al- though this population had been sub- ject to light fishing pressure. From 1972-74 to 1979-81, the back-calcu- lated size-at-age increased slightly for ages 2-8. By 1988-90 and 1991-94, however, the back-calculated size-at- age for the same age classes was sig- nificantly smaller than that in 1979- 81. In addition, size-at-maturity and size-at-sexual-transition occurred at progressively smaller sizes for 1988-90 and 1991-94. The mean size-at-age (ob- served and back-calculated) declined for most ages between 1988 and 1994. Von Bertalanffy growth curves fitted to the mean back-calculated size-at-age for each year showed similar decreas- ing trends. Changes in life history may be a response to sustained 20-year overexploitation that has selectively removed individuals predisposed to- wards rapid growth and larger size. Manuscript accepted 28 May 1997. Fishery Bulletin 95:732-747 (1997). Changes in the life history of red porgy, Pagrus pagrus, from the southeastern United States, 1972-1994* Patrick J. Harris John C. McGovern South Carolina Department of Natural Resources PO Box 12559, Charleston, South Carolina 29422 E-mail address (for P Harris): harrisp@mrd. dnr.state.se. us The red porgy, Pagrus pagrus, is a protogynous sparid distributed throughout the Atlantic Ocean and Mediterranean Sea at depths of 18 to 280 m (Manooch and Hassler, 1978; Vassilopoulou and Papacon- stantinou, 1992). In the South At- lantic Bight (SAB) off the southeast- ern coast of the United States, red porgy are commonly associated with sponge or coral habitat (or both) with rocky outcrops and rocky ledges (Grimes et al., 1982), fre- quently referred to as “live bottom.” Areas of live bottom are distributed patchily throughout the SAB, and patch size can range from square meters to square kilometers (Powles and Barans, 1980). Nevertheless, red porgy in the SAB are thought to constitute a single stock (Manooch and Huntsman, 1977). Red porgy are an important seg- ment of the commercial fisheries of the SAB, averaging 6% of the snap- per-grouper landings since 1978 (SAFMC* 1). Similarly, red porgy make up a considerable portion of the recreational harvest of reef fishes in the SAB (Huntsman et al.2 ). The fishery for red porgy in the SAB has, however, experienced a serious decline in landings since 1982 (Vaughan et al., 1992; Hunts- man et al.2), as well as a decline in fishery-independent catch per unit of effort (CPUE) (Fig. 1). Estimates of stock size derived from virtual population analysis (VPA) showed a peak population size in 1975 and a steady decline through 1992 (Vaughan et al., 1992; Huntsman et al.2). Although estimates of stock size derived from fishery-indepen- dent CPUE for 1993-1995 suggest a slight population recovery (Har- ris, personal obs.), the spawning stock ratio, estimated at 18% in 1993, is still considerably below the 30% level used by the South Atlan- tic Fishery Management Council to define when a species is overfished (Huntsman et al.2). Apart from a size limit instituted in 1992, management of the fishery has remained essentially unchanged, in spite of an apparent continual de- cline of the resource. The ability of fishermen to locate good fishing ar- eas (i.e. patches of live bottom) pre- cisely using LORAN-C and Global Positioning Systems technology and * Contribution number 394 of the South Carolina Marine Resources Center, Charles- ton, SC 29422 1 SAFMC. 1991. Amendment 4, regulatory impact review and final environmental im- pact statement for the snapper grouper fish- ery of the South Atlantic Region. South Atlantic Fishery Management Council, 1 South Park Circle, Charleston, SC, 225 p. 2 Huntsman, G. R., D. S. Vaughan, and J. C. Potts. 1993. Trends in population status of the red porgy Pagrus pagrus in the Atlantic Ocean of North Carolina and South Carolina, USA, 1971-1992. South Atlantic Fishery Management Council, 1 South Park Circle, Charleston, SC 29422. Harris and McGovern: Changes in the life history of Pagrus pagrus 733 an increase in the number of vessels participating in the snapper-grouper fishery in the SAB resulted in a steadily increasing fishing mortality from 1972 through 1993 (Huntsman et al.2). For new management regulations to be con- sidered, current life history data need to be made available. The most recent published discussion of SAB red porgy life history was based on data collected between 1972 and 1974 (Manooch, 1976; Manooch and Huntsman, 1977). It has been shown that age structure, size-at-age, and reproducti ve strategies of a population will change in a predict- able fashion that responds to declining abundance (Lack, 1968; Rothschild, 1986). There is, however, concern over the extent and permanence of these changes (Edley and Law, 1988; Bohn- sack, 1990). The effect of sustained heavy exploitation, combined with cur- rent management strategies in regard to particular size restrictions and quo- tas or bag limits on the life history of a fished stock, is poorly documented. Staff of the Marine Resources Monitoring, Assessment, and Prediction Program (MARMAP), a federally funded program based at the South Carolina Depart- ment of Natural Resources in Charleston, SC, have collected life history data on red porgy since 1979. When combined with data collected from 1972 through 1974 (Manooch, 1976; Manooch and Hunts- man, 1977), data spanning 24 years were available to determine if the life history of the red porgy popu- lation in the SAB had changed. Long-term life history data and the increase in fish- ing pressure provide a mechanism to test the impact of sustained exploitation on the life history of a reef fish species in the SAB. Therefore, the objectives of this paper were to describe temporal changes in the age, growth, and reproduction of red porgy for four periods during 1972-94 and to identify annual changes in age and growth that occurred during 1988-94. Methods Red porgy were collected from 1979 to 1994 during standard MARMAP sampling with chevron traps, hook-and-line gear, Florida traps, and blackfish traps (Collins, 1990; Collins and Sedberry, 1991) in the SAB from Cape Fear, North Carolina, to Cape Canaveral, Florida. Specimens were collected during daylight hours primarily between May and August of each year. MARMAP sampling strategies changed slightly between 1979 and 1994. From 1979 to 1987, samples were collected randomly from four large areas of live bottom (identified by using underwater television) with hook-and-line gear, blackfish traps, and Florida traps to follow trends in the abundance of the various species. Additional sites outside these areas were sampled as time and weather conditions allowed (see Collins and Sedberry, 1991). Traps were baited with cut clupeids, buoyed off the research vessel, and soaked for one to four hours. Hook-and-line gear consisted of bandit reels (commercial bottom-fishing hook-and-line gear) or rod and reel with 6/0 Penn Senator high-speed reels and Electramate electric motors. Terminal gear always consisted of three hooks fished vertically and baited with cut squid or cigar minnow ( Decapterus sp. ). All fishes caught were measured ( mm fork length f FL] ), and the total weight for each species, from each collec- tion, was recorded (g). All red porgy collected during these years were kept for life history studies. In 1988 and 1989, a chevron trap was added to the gear used to sample reef fishes. During these years, the research vessel was anchored over a known live 734 Fishery Bulletin 95(4), 1997 bottom area that was verified with underwater tele- vision. Each of the three trap types (blackfish, Florida, and chevron) was deployed either from the bow, stern, or midships of the research vessel (see Collins, 1990). Hook-and-line collections were taken with rod and reel and with the three-hook terminal rig. Fishes were processed as described for 1979-87. All red porgy collected in 1988 and 1989 were kept for life history studies. Based on the data collected during 1988 and 1989, a decision was made to discontinue the use of black- fish and Florida traps in 1990 because chevron traps sampled a greater species diversity (Collins, 1990). During the late 1980’s, all live bottom locations iden- tified during underwater television surveys and from sampling in previous years were plotted with LO- RAN-C coordinates to the nearest 0.1 ps and included in a sample site database. Currently, there are over 2,500 live bottom sites in the MARMAJP database, from which 300-600 randomly chosen sites have been sampled each year since 1989. In addition, since 1989, the SAB has been stratified on the basis of latitude. Zone 1 includes all sites sampled south of 32°N, zone 2 all sites between 32°N and 33°N, and zone 3 all sites north of 33°N. Buoyed chevron traps were de- ployed from the research vessel and soaked for ap- proximately 90 minutes. Hook-and-line (rod-and-reel) collections were made for 30 minutes at dawn or dusk. All fishes sampled were processed as in previous years. Because of concerns about potential gear selectivity, the length frequency of all red porgy caught by all four gear types during 1988 and 1989 was compared. Since 1989, fork lengths (cm) and total weight (10 g) were recorded for all red porgy sampled in each zone for each year with a Limnoterra FMB-IV elec- tronic fish measuring board and a Toledo electronic scale interfaced with a XT-type personal computer. In 1990 and 1994, all red porgy collected during sam- pling were used for life history studies. In 1991-93, up to 15 fish from each 1-cm size class and all fish larger than 350 mm FL were kept from each zone for life history studies. Red porgy used for life history studies were measured to the nearest mm (total length [TL], FL, and standard length [SL J ) with a Limnoterra FMB-IV electronic fish measuring board interfaced with a XT-type personal computer. Indi- vidual weights were measured to the nearest gram with a triple beam balance. Age and growth Sagittae were removed at sea and stored dry. In the laboratory, the whole right otolith was immersed in cedar wood oil and examined for annuli (one trans- lucent and one opaque zone) (Manooch and Hunts- man, 1977) with a dissecting microscope with incan- descent reflecting light and an ocular micrometer (1979-87) or with a dissecting microscope and re- flected light from a fiber-optic light source (1988- 94). The latter microscope had an attached Hitachi KP-C550 video camera connected to a personal com- puter equipped with a MATROX frame grabber and OPTIMAS image analysis software. The digitized image was viewed on a television monitor, and an- nuli were measured with OPTIMAS software. For both systems, measurements were taken from the core of each otolith to the outer edge of each opaque zone and to the edge of the otolith on a straight line midway between the posterodorsal dome and the most posterior point on the otolith (Frizzel and Dante, 1965). Annuli on this plane were consistently clearer and easier to enumerate, especially for older fish. For years where large numbers of red porgy were col- lected, a minimum of 350 randomly chosen fish were aged per year. All fish larger than 350 mm (FL) were aged for all years. The first reader collected measure- ments from all otoliths, whereas the second reader counted increments from a randomly chosen 35% of otoliths for each year. If agreement between the two counts was less than 90% for any year, the second reader read all otoliths for that year. When counts differed, otoliths were reread by both readers and discarded from further analyses if a difference in readings persisted. Back-calculated lengths-at-age were computed by using the scale proportional hypothesis (Francis, 1990): L( = - (a/b) + ( Lc + alb ) (0; /Oc), where L( = length at the formation of the zth incre- ment; Oi - otolith radius at the formation of the ith increment; Oc - otolith radius at the time of capture; Lc = fish length at the time of capture; a = intercept of otolith radius on fish length regression; b = slope of the otolith radius on fish length regression. Lengths were backcalculated to the most recently formed increment for comparisons of annual growth (1988-94) and to all increments for comparisons be- tween periods (1979-81, 1988-90, and 1991-94). The SigmaPlot curve-fitting module with the Marquardt- Levenburg algorithm was used to fit von Bertalanffy growth curves to the mean back-calculated length- at-age for each year or period (SigmaPlot, 1994). Because red porgy are protogynous sparids, and undergo a size- and behavior-related transition from Harris and McGovern: Changes in the life history of Pagrus pagrus 735 females to males, no comparison of size-at-age or growth rates were undertaken for the sexes sepa- rately. Life history data collected during four peri- ods (1972-74, 1979-81, 1988-90, and 1991-94) were compared. The first study (1972-74) used red porgy sampled from headboats operating from North and South Carolina (see Manooch, 1976; Manooch and Huntsman, 1977). Specimens were collected through- out the year and gonads from 736 fish were exam- ined macroscopically to assess sex and stage of ma- turity (Manooch, 1976). Scales from 3,278 individu- als were examined to determine ages, and 222 fish were aged from whole otoliths ( Manooch and Hunts- man, 1977). Red porgy collected during 1979-81, 1988-91, and 1991-94 were grouped by period. Otolith radius to fork length least-squares regressions were fitted separately for each period (except that of 1972-74) owing to concerns about temporal changes in somatic growth. Von Bertalanffy growth curves (von Ber- talanffy, 1938) were fitted to the mean back-calcu- lated size-at-age for each of the four study periods. Size-at-age was backcalculated for all increments measured. Mean observed and back-calculated sizes- at-age were compared between periods for each age with a single-factor ANOVA. Size and age distribu- tions and size-at-age were compared between the three latitudinal zones sampled with single factor and two-way ANOVA’s. It appeared from observations during sampling that larger fish may be associated with the shelf break; therefore size and age distribu- tions, and size-at-age were also compared for differ- ent depths. Because the shelf break is located at about 48 m, two depth zones — 0 to 45 m and 46 to 90 m — were compared. The same tests were performed in comparing annual data collected between 1988-94. Reproduction The posterior portion of the gonads of red porgy from 1979 to 1994 was removed from the fish and fixed in 10% seawater formalin for 1-2 weeks, then trans- ferred to 50% isopropanol for 1-2 weeks. Gonad samples were processed with an Auto-Technicon 2A Tissue Processor, vacuum infiltrated, and blocked in paraffin. Three transverse sections (6-8 pm thick) were cut from each sample with a rotary microtome, mounted on glass slides, stained with double- strength Gill haematoxylin, and counter-stained with eosin y. Sex and reproductive state were assessed by one reader according to histological criteria (Table 1). Specimens with developing, ripe, spent, or rest- ing gonads were considered sexually mature. For fe- males, this definition of sexual maturity included specimens with oocyte development at or beyond the yolk vesicle stage and specimens with beta, gamma, or delta stages of atresia. Sex ratios, size-at-first- maturity, and the percent of mature females by 20- cm size class were calculated for all functional males and females, 1989-94. Sex ratio, size-at-first-matu- rity, and the percent of mature females were deter- mined by size class for 1979-81, 1988-90, and 1991- 94, and chi-square (%2) analysis was used to deter- mine if there were significant differences in the pro- portion of males to all fish collected during the three periods and if there were differences in size-at-ma- turity between periods. ResuSts 1979-1994 A total of 20,756 ( 13,120 during 1972-74) red porgy were sampled during the four periods, of which 4,503 were aged and 4,293 sexed and staged (Table 2). The mean FL of fish collected from 1979 to 1994 showed a declining trend; however, there was no trend in mean age (Table 2). Increment formation was as- sumed to be annual (Collins et al., 1996; Manooch and Huntsman, 1977). Age and growth The mean observed size-at-age declined markedly from 1972-94 through 1991-94. Except for fishes aged 2-8 yr collected during 1979-81, the mean sizes- at-age for all ages for the three periods between 1979 and 1994 were smaller than those during 1972-94 (Fig. 2). The observed sizes-at-age in 1988-90 and 1991-94 were significantly smaller than those dur- ing 1979-81 (P<0.01) for ages 2 through 8. Red porgy aged 3 through 5 collected during 1991-94 were also significantly smaller than fish of the same age col- lected during 1988-90 (PcO.Ol). We were unable to include data collected by Manooch and Huntsman (1977) in our statistical analyses. The mean back- calculated size-at-age showed trends similar to the mean observed size-at-age (Fig. 3). Fish aged 2-8 were significantly smaller during 1988-90 and 1991- 94 than during 1979-81, and fish aged 2-5 significantly smaller in 1991-94 than in 1979-81 and 1988-90. The von Bertalanffy growth curves derived from mean back-calculated lengths for each period (Fig. 4) showed similar trends. The theoretical mean maxi- mum fork length (LJ declined by 100 mm from 1972- 74 to 1991-94 (Table 3). The theoretical growth rate (&) was higher between 1991 and 1994 than between 1972 and 1974. This difference is a reflection of the large decline in Lto, rather than an increase in growth 736 Fishery Bulletin 95(4), 1997 Table 1 Histological criteria developed by MARMAP (Charleston, SC) to determine reproductive stage in red porgy, Pagrus pagrus (see D’Ancona, 1949, 1950; Wallace and Selman, 1981; Alekseev, 1982, 1983; Hunter etal., 1986; Sadovy and Shapiro, 1987; Matsuyama et al., 1988; West, 1990; Roumillat and Waltz7). Reproductive state Male Female Immature (virgin) No primary males found. Juveniles were either fe- males or, infrequently, simultaneous or transitional (see below). Previtellogenic oocytes only; no evidence of atresia. In comparison with resting female, most previtellogenic oocytes <80 pm, area of transverse section of ovary is smaller, lamel- lae lack muscle and connective tissue bundles and are not as elongate, germinal epithelium along margin of lamellae is thicker, ovarian wall is thinner. Developing Development of cysts containing primary and sec- ondary spermatocytes through some accumulation of spermatozoa in lobular lumina and dorsomedial sinuses. Oocytes undergoing cortical granule (alveoli) formation through nucleus migration and partial coalescence of yolk globules. Running and ripe Predominance of spermatozoa in lobules and dorsomedial sinuses; little or no occurrence of sper- matogenesis. Completion of yolk coalescence and hydration in most advanced oocytes. Zona radiata becomes thin. Postovulatory follicles sometimes present. Developing, recent spawn Not assessed. Developing stage as described above as well as presence of postovulatory follicles. Spent No spermatogenesis; some residual spermatozoa in lobules and sinuses. More than 50% of vitellogenic oocytes with alpha- or beta-stage atresia. Resting Little or no spermatocyte development; empty lob- ules and sinuses. Previtellogenic oocytes only; traces of artresia. In comparison with immature female, most previtellogenic oocytes >80 pm, area of trans- verse section of ovary is larger, lamellae have muscle and connective tissue bundles, lamel- lae are more elongate and convoluted, germi- nal epithelium along margin of lamellae is thinner, ovarian wall is thicker. Mature specimen, stage unknown Mature, but inadequate quantity of tissue or post- mortem histolysis prevent further assessment of reproductive stage. Mature, but inadequate quantity of tissue or postmortem histolysis prevent further assess- ment of reproductive stage. Simultaneous (bisexual) Presence of distinct ovarian and testicular regions in approximately equal amounts and of the same reproductive state. This gonad structure was infrequently observed in both juvenile and adult fish. Transitional Ventrolateral proliferation of active testicular tissue (spermatogonia through spermatozoa) along the outer surface of the ovarian wall in spent or resting ovary (functional protogyny) or immature ovary (juvenile protogyny). As testicular tissue envelopes regressing ovary, ovary collapses laterally and sperm sinuses form within former ovarian wall. 1 Roumillat, W. A., and C. W. Waltz. 1993. Biology of the red porgy, Pagrus pagrus, from the southeastern United States. MARMAP Final Data Report, South Carolina Department of Natural Resources, Charleston, SC, 38 p. rate (i.e. the negative relation between L ^ and k). However, k was highest for the 1979-81 period, when was also still relatively high. Reproduction Our examination of 4,293 gonads (n=l,397, 1979- 81;n=727, 1988-90; n=2, 169, 1991-94) revealed that sexual transition was occurring at smaller sizes in the later periods. There was a significant increase (P<0.001) in the number of males with time (Table 4). However, in 1988-90 and in 1991-94, the propor- tion of males to the total number of fish sexed was significantly greater at smaller sizes than during 1979-81 (Table 4). At 301-350 mm TL, male red porgy made up 24% of the fish that were sexed dur- Hams and McGovern: Changes in the life history of Pagrus pagrus 737 Table 2 Sampling data for the four study periods 1972-74, 1979-81, 1988-90, and 1991-94. Year Fish sampled No. aged Mean fork length (mm) Mean age (years) No. sexed 1972-74 13,120 222 1979-81 1,933 1,177 293 3.07 1,397 1988-90 1,853 1,261 254 2.44 727 1991-94 3,850 1,843 257 3.062 2,169 Total 20,756 4,503 268 2.86 4,293 ing 1991-94, in contrast with 7% at the same size interval during 1979-81 (PcO.OOl; Table 4). In 1979- 81, male red porgy constituted 12% of the fish exam- ined at 351^00 mm TL compared with 32% in 1988- 90 (PcO.Ol) and 49% in 1991-94 (PcO.OOl; Table 4). Size-at-maturity of female red porgy has also changed. Female red porgy became sexually mature at smaller sizes in 1991-94 than in 1979-81. During 1991-94, female red porgy first became sexually mature at 176-200 mm TL (mean age=0.9). In 1979-81, the first mature female was at 201-225 mm TL (mean age=0.9) (Table 5). There were significantly more mature fe- males (54%; PcO.OOl) at 251-275 mm TL (mean age=1.9) in 1991-94 than during 1979- 81 (27%; mean age=1.7). 1988=1994 A total of 2,629 live bottom stations and 5,265 red porgy were sampled May through August 1988-94, of which 4,349 specimens were kept for life history studies (Table 6). The major- ity of the samples were collected with chev- ron traps. During 1988 and 1989, there was no difference between the size range of red porgy collected in chevron traps and the size range of red porgy collected in blackfish traps, Florida traps, hook-and-line gear, or all three of these gear types combined (Fig. 5). Similarly, there was no difference in the size range of red porgy sampled by hook-and- line gear and chevron traps between 1990 and 1994 (Fig. 5). Between 1988 and 1991, however, the mean size of red porgy captured with hook and line was significantly larger each year than the mean size of porgy taken with the remaining gear types (Pc0.05), al- though there was no significant difference be- tween mean size of fish captured with the gear types used in 1993 and mean size of fish captured with the gear types used in 1994. The size of red porgy sampled during 1988-94 ranged from 90 to 501 mm Table 3 Von Bertalanffy growth equation parameters derived from the mean back-calculated fork length for each time period. Parameter 1972-74 1979-81 1988-90 1991-94 k 0.226 0.343 0.273 0.281 459.3 391.4 382.7 356.4 FL. The mean size was 256 mm FL, with the highest frequency occurring at 240 mm FL. The length fre- quency of aged red porgy was very similar to the length frequency of all red porgy sampled. Age (years) Figure 2 The mean size at capture for red porgy collected in 1972-74, 1979- 81, 1988-90, and 1991-94. Ages for red porgy collected between 1972 and 1974 were taken from Manooch and Huntsman (1977). 738 Fishery Bulletin 95(4), 1 997 Figure 3 The mean back-calculated size at age for red porgy collected in 1972- 74, 1979-81, 1988-90, and 1991-94. Back-calculated lengths for red porgy collected between 1972 and 1974 were derived from scale ages. Age and growth Ages were obtained for 2,935 (67%) of the red porgy otoliths collected (Table 6). Agreement between the first and second reader averaged 93%, and was never less than 90% for any year. The mean observed size-at-age declined for most ages between 1988 and 1994 (Table 7), although there was a significant increase in the mean age of red porgy over the study period (Fig. 6; PcO.Ol; r2=0.94). The mean observed size-at-age for red porgy 2 years and older sampled during 1988 and 1989 was sig- nificantly larger than all other years, with the mean observed size-at-age in 1992 and 1993 consistently the smallest (Table 7; Fig. 7). Above age 6, growth rates appeared to taper off sharply for all years (Fig. 7). Age-6 red porgy collected during 1988 had the third highest mean length recorded for all age classes in any year. Similar to the mean ob- served size at age, mean back-calculated size at the most recent annulus was significantly larger for 1988 and 1989 compared with other years for ages 2 and greater and also ap- peared to reach asymptotic size at age 6 for each year (Table 8; Fig. 8). The von Bertalanffy growth curves fitted to the mean back-calculated size at most re- cent age (ages 1-10) for each year demonstrated some differences between years (Fig. 9), with growth curves from 1988 and 1989 showing larger fish at age, and higher Lm and k. Both k and tended to decrease dur- ing the study period, although neither of these trends were significant (linear regression; P>0.05; Fig. 10). No significant differences were apparent in size distribution, age distribution, or size-at-age between Table 4 Percentage of male red porgy relative to the total number of individuals sexed during 1979-81, 1988-90, and 1991-94. A signifi- cant difference (%2; PcO.Ol) in the proportion of males for a particular size class collected during 1979-81 is denoted by “A”. A significant difference (%2; PcO.Ol) in the proportion of males that were collected during 1988-90 is denoted by “B”. Size (mm TL) 1979-81 1988-90 1991-94 Total Males %Males Total Males %Males Total Males %Males <200 19 — — 16 — — 57 — — 200-250 216 — — 140 2 1.43 372 4 1.08 251-300 271 10 3.69 163 5 3.07 491 33 6.72s 301-350 313 21 6.71 226 25 11.06 814 194 23.83^ 351-400 239 29 12.13 136 44 32.35A 371 183 49. 33^ 401-450 160 38 23.75 38 17 44.74A 57 25 43. 86^ 451-500 158 108 68.35 8 4 50.00 6 4 66.67 501-550 18 12 66.67 — — — 1 — — 551-600 2 1 50.00 — — — — — — Total 1,397 220 727 97 2,169 443 Harris and McGovern: Changes in the life history of Pagrus pagrus 739 the three latitudinal zones. However, signifi- cant differences were apparent in the size and age distribution between the two depth zones (P<0.05), with larger and older fish occurring in the deeper zone. There were no significant differences in the size-at-age be- tween these two zones (P>0.05). Discussion Samples were collected from throughout the SAB during 1988-94. However, 69% of the collections and 73% of the aged red porgy were taken from zone 2 (32°N-33°N). Zone 2 was sampled most frequently because it was most accessible from Charleston, South Caro- lina, the base of operations (latitude 32° 45'N). Once settled, red porgy do not appear to move very much (Parker, 1990) and could experience differential growth rates because of differing environments. Therefore the con- centration of sampling in zone 2 could have resulted in biased estimates of size-at-age. However, the comparison of size-at-age of red porgy showed no significant differences be- tween latitudinal or depth zones; therefore, although there may be localized differences in growth rates, perhaps associated with dif- ferent patches of live bottom, the mean growth rate appears to be similar throughout the region. The mean depth and temperature of areas sampled in the MARMAP surveys have not changed significantly since 1987; thus these environmental variables, at least, have not caused the life history changes in red porgy (Fig. 11). There were significant differences in size and age by depth, with larger and older fish occurring in deeper water. This difference may be due to fisher- men operating in deeper water with larger hooks and baits to target groupers, thus reducing the availabil- ity of this gear to red porgy, particularly to smaller Table 5 Percentage of red porgy females that were sexually mature relative to all females collected during 1979-81 and 1991-94. A significant difference (^2; PcO.Ol) in the proportion of mature females is denoted by “A”. Size (mm TL) 1979-81 1991-94 Total females Number mature females Percent mature females Total females Number mature females Percent mature females <200 16 — — 55 2 3.64 200-225 91 1 1.10 182 4 2.20 226-250 85 11 12.94 156 26 16.67 251-275 103 28 27.18 157 85 54.14a 276-300 78 77 98.72 211 189 89.57 >300 512 512 100.00 615 606 98.54 Total 885 62 1,376 912 740 Fishery Bulletin 95(4), 1997 individuals. Therefore, red porgy in deeper water may experience reduced fishing mortality in comparision with those in shallower waters. In shallower water, fishermen reduce hook and bait size to catch smaller fishes, and more red porgy of all sizes are landed. Alternatively, the increase in size and age with depth Table 6 Sampling data for 1988-94, collected from the RV Oregon (1988-89) and the RV Palmetto (1990-94). Hook-and-line Year Trap collections No. porgy collections No. porgy No. processed No. aged 1988 84 294 261 170 427 371 1989 65 248 198 174 388 345 1990 348 957 111 44 997 545 1991 306 830 33 25 519 426 1992 324 1,107 25 1 494 419 1993 414 722 52 45 538 385 1994 370 1,107 38 11 986 444 Total 1,911 5,265 718 470 4,349 2,935 Table 7 Mean observed fork length (mm) at age for red porgy (standard error in parenthesis). Age (yr) 1988 1989 1990 1991 1992 1993 1994 1 191 (2) 203 (3) 197 (3) 200(2) 190(2) 206 (3) 186(1) rc=124 77 = 70 77=78 77=126 77=78 77=70 77=53 2 256(2) 248(2) 234 (2) 237 (2) 228 (2) 245 (3) 253 (2) t? = 107 77=149 77=218 77 = 110 77=119 77=78 77 = 101 3 284 (3) 284 (4) 264(2) 274(3) 261 (3) 267 (3) 264(3) n= 95 77=54 77 = 180 77=71 77 = 72 77 = 104 77=50 4 328 (7) 305 (5) 295 (5) 282 (4) 287 (3) 290 (4) 283 (3) 77= 26 77=39 77 = 26 77=70 77=64 77=59 77 = 106 5 386 (34) 328 (6) 314(10) 303 (8) 305 (4) 308 (5) 297 (3) n= 2 77 = 18 77=16 77 = 17 77=43 77=35 77=86 6 334(7) 305 (34) 317 (13) 335 (7) 310 (7) 305 (5) 313 (5) 71= 4 77=3 77=5 77 = 13 77 = 14 77=25 77=32 7 374 (10) 340 (35) 308 (7) 339 (20) 321 (7) 359 334(5) n= 5 77=3 77=6 77=5 n=8 n= 1 77=9 8 352 (3) 335 307 (9) 322 (5) 328 (5) 348 (25) 324(6) 77=4 n- 1 77=3 77=4 77 = 2 77=6 77=4 9 389 394 (42) 384(17) 362 344 (8) 322 (18) n= 1 77=2 77=2 n- 1 77=6 77=2 10 372 (28) 361 (16) 344(7) 390 77=2 77=4 77=2 n= 1 11 363 72 = 1 12 368 360 354 71 = 1 77=1 n= 1 13 390 (9) 77=2 Harris and McGovern: Changes in the life history of Pagrus pagrus 741 could reflect a gradual movement of red porgy to- wards deeper water as they age. Grimes et al. ( 1982) suggested that red porgy associated with shallow reefs may temporarily move to deep water in response to unusually cold water temperatures. However, tag- ging studies have shown that settled red porgy un- dertake very little long-term movement (Grimes et al., 1982; Parker, 1990). Another reef species, black sea bass ( Centropristis striata), has shown limited movement of larger fish to deeper water (Ulrich and Low3; Harris and McGovern, personal obs.). Fishing mortality (F) of red porgy has increased since 1972, although between 1972 and 1975 it showed a slight decline (Vaughan et al., 1992). The 3 Ulrich, G. F., and R. A. Low. 1992. Movement and utiliza- tion of black sea bass, Centropristis striata, in South Carolina. Final Unpubl. Rep. NOAA Award No. NA90AA-D- FM656, 11 p. F for fully recruited ages (5-9) increased from 0.2 in 1976 to 1.3 in 1991 (Huntsman et al.2, Murphy VPA, M= 0.28). The F for ages 1-4 showed similar trends, although the magnitude of the increase was less (Huntsman et al.2, Murphy VPA, M= 0.28). Owing to the changes in the life history of red porgy since 1972, these estimates of fishing mortality are probably high; yet, the increasing trend in fishing pressure is evident. Except for an increase during 1981-83, land- ings of red porgy have been declining since 1973 (Fig. 1). Similarly, the number of recruits to age 1 have de- clined steadily since 1974 (Huntsman et al.2, Murphy VPA, M= 0.28). An estimate of SSR in 1993 was only 18% (Huntsman et al.2, Murphy VPA, M- 0.28). Again, the changes in the life history of red porgy since 1972 indicate that Huntsman et al.2 may have under- estimated the decline in the abundance of age-1 fish. Concurrent with the fishery becoming increasingly overexploited, there has been corresponding change Mean back-calculated fork length (mm Table 8 ) at age for red porgy (most recent annulus; standard error in parenthesis). Age (yr) 1988 1989 1990 1991 1992 1993 1994 1 163 (2) 180 (3) 173 (4) 166(2) 159 (2) 171 (2) 161 (2) 77=124 n= 70 77=78 77 = 126 oo ii £ 77 = 70 77=53 2 230(2) 233 (2) 218 (2) 219 (2) 210 (2) 227 (3) 237 (2) h=107 n=148 77=218 77 = 110 77=119 OO II e 77 = 101 3 266(3) 273 (3) 252 (2) 259 (3) 246 (3) 257 (3) 253 (3) n= 95 77=54 77 = 180 77 = 71 77=72 77 = 104 77=50 4 316(6) 300 (6) 286 (5) 271 (3) 274 (3) 281 (4) 275 (3) n= 26 77=38 77=26 77 = 70 77=64 77=59 77=106 5 382 (31) 325 (7) 304 (11) 292 (8) 298 (4) 301 (5) 291 (3) n= 2 77=18 77 = 16 77 = 17 77=43 77=35 77=86 6 328(7) 301 (32) 311 (13) 326(6) 299(7) 298 (6) 307 (5) n= 4 77=3 77 = 5 77=13 77 = 14 77=25 77=32 7 369 (10) 339 (34) 294 (7) 331 (18) 313 (6) 354 329 (5) n= 5 77=3 77 = 6 77=5 77=8 77 = 1 77=9 8 347 (3) 335 300(10) 314 (5) 320(4) 351 (20) 320 (6) ti= 4 n- 1 77=3 77 = 4 77 = 2 n-1 77=4 9 389 392 (40) 354 335 (8) 292 n- 1 77 = 2 n= 1 n=6 n= 1 10 371 (27) 378 (17) 350 (16) 336(4) 382 77=2 77 = 2 77=4 n= 2 77 = 1 11 362 n- 1 12 367 356 349 n- 1 n- 1 77 = 1 13 382 (8) 77=2 742 Fishery Bulletin 95(4), 1997 in the life history of red porgy during a 22-year pe- riod ( 1972 to 1994). The first study of age and growth on red porgy (1972-74) (Manooch and Huntsman, 1977) was assumed to describe a stock with the same life history as the virgin population, even though it was subject to light fishing pressure. By the late 1970’s and early 1980’s, the growth pattern of the stock had changed. The mean observed and back- calculated lengths-at-age as well as the von Bertalanffy growth curve for the 1979-81 period showed a larger size at age for ages 2-6 but a lower theoretical maximum size. The increase in growth rate, and resultant increase in size- at-age observed during this period, is consid- ered a typical density-dependent response to an increase in mortality as the depressed population responds to an increased avail- ability of resources (Pitcher and Hart, 1982; Rothschild, 1986). The decrease in theoreti- cal maximum size between 1979 and 1981 may have resulted from the selective removal of larger individuals from the population by fishermen and not from a biological change in the theoretical maximum size that the fish could attain. During the mid 1980’s through the early 1990’s, increasing fishing pressure appar- ently continued the selective removal of larger, faster growing individuals from the population, further exacerbating the changes in the life history of red porgy. By 1988-90, mean back-calculated sizes-at-age were sig- nificantly smaller for all ages except age 1 in comparison with 1979-81. In 1988-90, the values of k and were smaller than during 1979-81 and 1972-74, indicating a reduced growth rate and a lower theoretical maximum attainable size. Mean back-calcu- lated size-at-age for specimens col- lected between 1991 and 1994 were significantly smaller than those col- lected in 1988-90, except for ages 1, 7, and 10. These temporal reduc- tions in the size-at-age and growth rates suggest that many individu- als genetically predisposed towards rapid growth and larger size may have been selectively removed from the population, leaving behind in- dividuals that tend to be slower growing and smaller. Red porgy also responded to the continued removal of larger indi- viduals from the population over many generations by females be- coming sexually mature at smaller sizes during 1991-94 than during 1979-81. Manooch (1976) reported that female red porgy became ma- ture at much larger sizes than those 70 60 50 - S' 40 C u. 30 - 20 - 10 - V/////A Hook and line 62ZZ2 Blackfish trap l l Florida trap nwni Chevron trap iP P 23j 140-159 180-199 220-239 260-279 300-319 340-359 380-399 Size class (mm fork length) Figure 5 Length frequency for all red porgy sampled in MARMAP surveys between 1988 and 1994 by gear type. o ?T ?T £3 3“ 3 3 Year Figure 6 The mean age of red porgy sampled in MARMAP surveys between 1988 and 1994. Hams and McGovern: Changes in the life history of Pagrus pagrus 743 found in the present study. Further- more, the red porgy population has re- sponded to increased fishing pressure by undergoing sexual transition and by producing significantly more males at smaller sizes in recent years than dur- ing 1979-81. Koenig et al. (1996) re- ported that gag, Mycteroperca micro- lepis, a protogynous grouper, was un- dergoing sexual transition at much smaller sizes during 1991-93 in the Gulf of Mexico than were reported by Hood and Schlieder (1992 ) for the same region, during 1977-80. Changes in life history aspects of gag from the Gulf of Mexico were attributed to steadily in- creasing fishing pressure. The decrease in mean size-at-age, growth rates, and size-at-maturity dur- ing 1988-1994 is probably a continua- tion of the changing life history pattern of the population that has resulted from sustained fishing pressure and indi- cates the degree of change that can oc- cur over relatively short periods of time. These relatively rapid changes in size- at-age may reflect the inability of an overfished or depressed population to absorb or respond to further decreases in population size. Apart from the decreases in size-at-age apparent from recent years, the mean age and fork length of the population has increased since 1988. These increases may be due to a decline in the number of younger fish recruiting to the population. The net effect of fewer young fish in the popula- tion (and therefore samples) would be an in- crease in the mean age and size of the sampled fish. Length-frequency data col- lected in MARMAP surveys since 1988 indi- cates no strong recruiting year class (age-1) since 1990. Huntsman et al.2 found that the estimated number of 1-year-old red porgy had declined steadily since 1973 (Murphy VPA, M=0.28); their results also indicate that the population may be experiencing a decline in recruitment. A decline in recruitment may be attributed to several factors that are the result of sus- tained overfishing. First, as the number of fish in the population declines, fewer and fewer females are available to spawn, result- ing in a decline of total potential egg production in fish that are less fecund than larger fish (Manooch, (Vaughan et al., 1992; Huntsman et al.2). Second, 1976). Finally, smaller females may produce eggs that decreases in size-at-maturity and size-at-age result are poorer in quality than those produced by larger i- £ •o 300 mm TL in 1991-94. Thus, many faster growing individuals may reach legal size before they are sexually mature or when they have only had the opportunity to spawn once or twice. Slower growing individuals would take longer to reach the size limit and have a greater chance to spawn be- fore becoming available to the fishery, thus further ex- acerbating the effect of size-selective mortality. The SAB population of red porgy has undergone significant changes in life history, presumably in re- sponse to sustained, long-term size-selective overexploitation. Individuals in the population are smaller, have reduced growth rates, a reduced theo- retical maximum size, and undergo sexual maturity and transition at smaller sizes now than 20 years ago. The selective pressure of fishing mortality may be causing a genetic shift towards a slower growing, smaller population. Unless appropriate management measures are taken, sustained overfishing could re- sult in a permanent genetic shift in the fish or a to- tal collapse of the stock (or both). Acknowledgments The authors wish to thank the crews and scientific parties of the research vessels that made collection of these data possible. Bill Roumillat and Wayne Waltz aged, sexed, and identified maturity stages of red porgy collected between 1979 and 1987, and Oleg Pashuk sexed and staged all red porgy collected since 1987. Three anonymous reviewers provided numerous com- ments that improved the manuscript immeasurably. Literature cited Alekseev, F. E. 1982. Hermaphroditism in porgies (Perciformes, Sparidae). I: Protogyny in scups, Pagrus pagrus (L. ), P. orphus Risso, P. ehernbergi Val. and P. auriga Val. from the West African shelf. J. Ichthyol. 22:85-94. 1983. Hermaphroditism in porgies (Perciformes, Sparidae). II: Sexual structure of the populations, mechanisms of its formation and evolution in scups, Pagrus pagrus, P. orphus, P. ehernbergi and P. auriga. J. Ichthyol. 23:61-73. D’Ancona, U. 1949. II differentziamento della gonade e l’inversione sess- uale degli Sparidae. Arch. Oceanogr. Limnol. 6:97-163. 1950. Determination et differenciation du sexe chez les poissons. Arch. Anat. Microsc. Morphol. Exp. 39(3):274- 294. Bohnsack, J. A. 1990. The potential of marine fishery reserves for reef fish management in the U.S. South Atlantic. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFC-216, 45 p. 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. Collins, M. R., S. B. Van Sant, D. J. Schmidt, and G. R. Sedberry. 1996. Age validation, movements, and growth rates of Harris and McGovern: Changes in the life history of Pagrus pagrus 747 tagged gag ( Mycteroperca microlepis), black sea bass ( Centropristis striata ), and red porgy (Pagrus pagrus). In F. Arraguin-Sanchez, J. L. Munro, M. C. Balgos, and D. Pauly (eds.), Biology, fisheries and culture of tropical grou- pers and snappers, p 161-165. ICLARM Conf. Proc. 48. Collins, M. R., C. W. Waltz, W. A. Roumillat, and D. L. Stubbs. 1987. Contribution to the life history of gag Mycteroperca microlepis (Serranidae), in the South Atlantic Bight. Fish. Bull. 85:648-653. Edley, M. T., and R. Law. 1988. Evolution of life histories and yields in experimental populations of Daphnia magna. Biol. J. Linn. Soc. 34: 309-326. Francis, R. C. C. C. 1990. Back-calculation of fish length: a critical review. J. Fish Biol. 36:883-902. Frizzel, D. L., and J. H. Dante. 1965. Otoliths of some early cenozoic fishes of the Gulf coast. J. Palaeontology 39:687-718. Gilmore, R. G., and R. J. Jones. 1992. Color variation and associated behavior in the epinepheline groupers, Mycteroperca microlepis (Goode and Bean) and M. phenax (Jordan and Swain). Bull. Mar. Sci. 51(1):83-103. Grimes, C. B., C. S. Manooch, and G. R. Huntsman. 1982. Reef and rock outcropping fishes in the outer conti- nental shelf of North Carolina and South Carolina, and ecological notes on the red porgy and vermilion snapper. Bull. Mar. Sci. 32:277-289. Hood, P. B., and R. A. Schlieder. 1992. Age, growth, and reproduction of gag, Mycteroperca microlepis (Pisces: Serranidae), in the eastern Gulf of Mexico. Bull. Mar. Sci. 51:337-352. Hunter, J. R., B. J. Macewicz, and J. R. Sibert. 1986. The spawning frequency of skipjack tuna, Kat- suwonus pelamis, from the South Pacific. Fish. Bull. 84:895-903. Huntsman, G. R., and W. E. Schaaf. 1994. Simulation of the impact of fishing on reproduction of a protogynous grouper, the graysby. North Am. J. Fish- eries Manage. 14:41-52. Koenig, C. C., F. C. Coleman, L.A. Collins, Y. Sadovy, and P. L. Colin. 1996. Reproduction in gag, Mycteroperca microlepis (Pisces: Serranidae), in the eastern Gulf of Mexico and the conse- quences of fishing spawning aggregations. In F. Arraguin- Sanchez, J. L. Munro, M. C. Balgos, and D. Pauly (eds.), Biology, fisheries and culture of tropical groupers and snappers. ICLARM Conf. Proc. 48. Lack, D. 1968. Population studies of birds. Clarendon. Oxford, UK, 341 p. Law, R., and D. R. Grey. 1989. Evolution of yields from populations with age-spe- cific cropping. Evol. Ecol. 3:343-359. Manooch, C. S. 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., and W. W. Hassler. 1978. Synopsis of biological data on the red porgy, Pagrus pagrus (Linnaeus). FAO Fisheries Synopsis 116, 19 p. Manooch, C. S., and G. Huntsman. 1977. Age, growth, and mortality of the red porgy, Pagrus pagrus. Trans. Am. Fish. Soc. 106:26-33. Matsuyama, M., R.T. Lara, and S. Matsuura. 1988. Juvenile bisexuality in the red sea bream, Pagrus major. Environ. Biol. Fishes 21:27-36. Parker, R. O., Jr. 1990. Tagging studies and diver observations of fish popu- lations on live-bottom reefs of the U.S. southeastern coast. Bull. Mar. Sci. 46:749-760. Pitcher, T. J., and P. J. B. Hart. 1982. Fisheries ecology. Croom Helm, London, 414 p. Powles, H., and C. A. Barans. 1980. Groundfish monitoring sponge-coral areas off the southeastern United States. Mar. Fish. Rev. 42(5):21-35. Rothschild, B. J. 1986. Dynamics of marine fish populations. Harvard Univ. Press, Cambridge, MA, 277 p. Sadovy, Y., and D. Y. Shapiro. 1987. Criteria for the diagnosis of hermaphroditism in fishes. Copeia 1987:136-156. Shapiro, D. Y. 1981. Size and maturation and the social control of sex re- versal in the coral reef fish Anthias squamipinis. J. Zool Lond. 193:105-128. SigmaPlot. 1994. Transforms and curve fitting. SigmaPlot Scientific Graphing Software, Jandel Scientific, San Rafael, CA, 249 p. Sutherland, W. J. 1990. Evolution and fisheries. Nature 344:814-815. Van der Elst, R. 1988. A guide to the common sea fishes of southern Africa. Struik, Cape Town, South Africa, 398 p. 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., G. R. Huntsman, C. S. Manooch III, F. C. Rhode, and G. F. Ulrich. 1992. Population characteristics of the red porgy, Pagrus pagrus, stock of the Carolinas. Bull. Mar. Sci. 50:1-20. Von Bertalanffy, L. 1938. A quantitative theory of organic growth. II: Inquir- ies on growth laws. Hum. Biol. 10:181-213. Wallace, R.A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am. Zool. 21:325-343. West, G. 1990. Methods of assessing ovarian development in fishes: a review. Aust. J. Mar. Freshwater Res. 41:199-222. 748 Abstract .—The Exxon Valdez oil spill occurred just prior to the spring migration of Pacific herring, Clupea pallasi, from offshore feeding grounds to nearshore spawning areas in Prince William Sound (PWS), Alaska. Most or all of the life stages of herring in PWS may have been exposed to oil after the March 1989 spill. Delayed impacts from the spill were suspected as one possible cause in the unprecedented crash of the adult herring population in 1993 and stimulated studies to assess reproduc- tive success. In spring 1995, mature herring were collected from four sites in PWS and from three uncontaminated sites in southeast Alaska (SE) to deter- mine if reproductive impairment was evident in PWS herring six years after the spill. Herring were artificially spawned and their eggs were reared in a laboratory until hatching. Observed response parameters included fertiliza- tion success, hatching times, hatching success, as well as larval viability, swimming ability, and spinal abnor- malities. Responses of all year classes combined or those restricted to the same year class did not differ signifi- cantly between regions (P>0.50); the best and worst responses generally oc- curred in the SE. Within each site, re- sponse of the 1989 year class (most likely impacted by the oil spill in PWS) generally did not differ significantly from any other year class. To verify macroscopic observations, a subset of larvae from the 1989 year class was also inspected microscopically for yolk and pericardial abnormalities, and yolk vol- ume was measured — but no significant regional differences were observed for any of these morphological categories. Based on the parameters examined in this study, evidence of reproductive im- pairment of Pacific herring in PWS by the spill was not detected in 1995, and the chances of detecting any oil-related effects against natural background variation appeared to be negligible. Manuscript accepted 28 May 1997. Fishery Bulletin 95:748-761 (1997). Reproductive success of Pacific herring, Clupea pallasi, in Prince William Sound, Alaska, six years after the Exxon Valdez oil spill Scott W. Johnson Mark G. Carls Robert P. Stone Christine C. Brodersen Stanley D. Rice Auke Bay Laboratory, Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 1 1 305 Glacier Highway Juneau, Alaska 99801-8626 E-mail address (for S.W. Johnson): scott.johnson@noaa.gov The Exxon Valdez oil spill (EVOS) in Prince William Sound (PWS), Alaska, occurred just a few weeks prior to the Pacific herring, Clupea pallasi, spawning season. Most or all of the life stages of herring in PWS may have been exposed to oil after the March 1989 spill. Biologi- cally available hydrocarbons were present in the upper water column of PWS for several weeks following the spill (Short and Harris, 1996a), and residual oil may have persisted in some areas into 1990 (Short and Harris, 1996b). An estimated 40- 50% of the egg biomass in PWS was deposited within the oil trajectory (Brown et al., 1996a). The failure of the 1989 year class to recruit to the fishery and the subsequent crash of the 1993 population (Meyers et al., 1994) suggested that the early life stages of herring were impacted ei- ther from exposure of prespawning adults or from direct exposure of eggs and larvae. Thus, as fish ex- posed to oil were recruiting into the fishery (20% by age 3, 80% by age 4, 100% by age 5; Funk1 ), the her- ring population crashed, and recov- ery was minimal through the 1996 season (Wilcock2 ). Genetic damage, physical deformities, and small size were reported for newly hatched larvae following the spill (Brown et al., 1996a; Hose et al., 1996; Nor- cross et al., 1996; Marty et al., in press), but long-term effects remain unknown. In a preliminary study in 1992, Kocan et al. (1996b) observed decreased reproductive success in herring from an oil-contaminated area in PWS compared with an un- contaminated area; results were inconclusive, however, because only two sites were compared. Delayed effects from the spill were suspected as one possible cause of the popula- tion decline and stimulated the 1 Funk, F. In prep. Age-structured as- sessment of Pacific herring in Prince Wil- liam Sound, Alaska and forecast of abun- dance for 1994. Regional Information Report, Alaska Department of Fish and Game, Commercial Fisheries Management and Development Division, PO Box 25526, Juneau, AK 99802-5526. 2 Wilcock, J. 1996. Alaska Department of Fish and Game. Commercial Fisheries Management and Development Division, PO Box 669, Cordova, AK 99574. Per- sonal commun. Johnson et al.: Reproductive success of Clupea pallasl after the Exxon Valdez oil spill 749 Russia Alaska U.S. Southeast Alaska Prince William Sound Canada 100 km Port Chalmers 100 km Ketchikan St. Mathews Bay Rocky Bay Sitka Figure 1 Collection sites of mature Pacific herring in Prince William Sound and southeast Alaska in spring 1995. need for more definitive studies to assess the repro- ductive success of herring. The purpose of this study was to determine if re- productive impairment, a possible result of the spill, was evident in PWS herring six years after the spill. There were two major focuses in the study: 1) a com- parison of reproductive success between regions (PWS and southeast Alaska [SE]) and 2) a compari- son of reproductive success between year classes within sites, particularly the 1989 year class (most likely impacted by the oil spill) with other year classes in PWS. Sites sampled within PWS included all areas where spawning occurred in 1995; spawning was absent in areas that were heavily contaminated with oil in 1989. For example, Naked Island, which was in the middle of the spill trajectory, had 22 km of spawned eggs in 1989 (Brown et al., 1996a) but none in 1995. Although some have speculated that herring home to the same general spawning area each year (Zijlstra, 1963; Hourston, 1982), site fidelity is poorly understood. Thus, the herring we sampled in PWS in 1995 may or may not have been exposed to oil at some earlier time in their life history (as adults, eggs, or larvae). Methods Herring were collected at four sites in PWS and at three sites in SE (Fig. 1); all sites had been used for spawning in previous years. Two of the sites in PWS (St. Mathews Bay and Fish Bay) were not directly contaminated by the oil spill, whereas the other two sites (Port Chalmers and Rocky Bay) were at least lightly contaminated. Shortly after the spill, elevated hydrocarbon levels were detected in mussels at Rocky Bay (Brown et al., 1996b) and in seawater at Port Chalmers (Carls3). Additionally at Port Chalmers, concentrations of oil metabolites in bile of adult her- ring sampled in spring 1990 were similar to metabo- lite concentrations observed in 1989. This finding suggested continued contamination (Brown et al., 1996b). Herring were collected in St. Mathews Bay on 7 April, in Fish Bay on 14 April, at Port Chalmers on 30 April, and in Rocky Bay on 1 May 1995. In SE, herring were collected in waters off Sitka on 29-30 March, in waters near Ketchikan on 11 April, and in Seymour Canal on 13 May 1995. 3 Carls, M. G. 1996. Prince William Sound oil database. Auke Bay Laboratory, National Marine Fisheries Service, NOAA, 11305 Glacier Hwy., Juneau, AK 99801. 750 Fishery Bulletin 95(4), 1997 Mature herring were captured during or just prior to spawning at all sites, sorted by size, and artifi- cially spawned. Capture gear included gill net, cast net, and purse seine. Fish were chilled immediately after capture and transported within two hours to a field laboratory, except Seymour Canal fish, which were transported directly to Auke Bay Laboratory (ABL). To approximate the different age classes present, fish were sorted by sex and size (usually in 10-mm increments; e.g. 220-230 mm fork length). Six or more size classes were usually identified at each site. From each size class, 25 females were ar- tificially spawned with males of the same size; gen- erally 3 males contributed sperm for all 25 crosses. Size classes of fish were processed at random. Each fish was assigned an identification number, mea- sured to the nearest mm (fork length), and weighed to the nearest 0.1 g (wet weight). To determine age, three scales were removed from the left side of each spawned fish near the posterior margin of the dorsal fin, placed on a glass slide, and covered with a sec- ond slide. For spawning, testes were removed, sealed in a plastic bag and maintained in chilled seawater until use; ovarian membranes were cut longitudinally, and eggs were removed with a hydrocarbon-free stain- less steel spatula similar to that used by Brown.4 From each female, approximately 150 eggs were de- posited with a gentle swirling motion onto a 25 x 75 mm glass slide placed on the bottom of a shallow plas- tic dish filled with seawater. Each slide was then placed in a staining rack and suspended in its own 1-L beaker of seawater. Milt was prepared from col- lected testes by cutting sections from each into small segments; segments plus a small amount of seawa- ter were mixed with a spatula. A few milliliters of the milt mixture were added to each beaker contain- ing eggs. Eggs and milt remained in contact for 5 min; the milt was then poured off, and the eggs were gently rinsed in seawater. Slides were kept in stain- ing racks and maintained in ambient seawater with constant aeration until they were transported to ABL by air. To transport the eggs, staining racks were placed in plastic seawater-filled containers, which were then placed in coolers with blue ice. Slides with eggs from each site were randomly dis- tributed among twelve 600-L tanks with flow-through seawater. Slides were suspended from monofilament line attached to a pivoting overhead framework de- signed to cause slow egg movement (1 rpm) through the water. During the first 16-18 days of incubation, 4 Brown, E. D. 1995. Alaska Department of Fish and Game. Commercial Fisheries Management and Development Division, PO Box 669, Cordova, AK 99574. Personal commun. all slides were maintained in the seawater bath. A few days before hatching, each slide was isolated in a 1-L glass jar that contained seawater and that was surrounded by flowing seawater. Lighting was natu- ral, supplemented by overhead fluorescent light dur- ing daylight hours. Seawater flow was approximately 1 L/min at 3.9°C, warming to 7.1°C with normal sea- sonal change. Salinity was 32 ±1 ppt. Reproductive success of female herring was defined as the production of physically and functionally nor- mal larvae. Key reproductive parameters included hatching success and larval viability, swimming abil- ity, and spinal abnormalities. These four parameters were sensitive to oil in laboratory studies (Carls et al.5 ). Other parameters examined included fertility and hatching times. Fertility was not considered a key parameter because it may have been negatively influenced by unavoidable handling conditions at the different sites and by variable periods in the storing of gametes prior to spawning. Hatching times were not considered a key parameter because they were strongly influenced by seasonal increases in water temperature. Fertilization success and stage of development were determined 1 to 10 days after spawning. Ex- cess eggs were removed from all slides by scraping — i.e. those along slide margins susceptible to mechani- cal damage and clumps of eggs not directly exposed to water. This process was accomplished in water with minimal exposure to air. Isolated eggs were inspected every two days to determine onset of hatching. Once hatching began, larvae were counted and assessed daily for swimming ability and gross physical deformities. Without ex- posing eggs to air, we changed the seawater in each jar every two days prior to hatching and daily after hatching began. All hatched larvae were collected, anes- thetized with tricaine methanesulfonate, and preserved in 10% phosphate buffered formalin. Approximately the first and last 10% of larvae hatched from each female were preserved in separate bottles. Live larvae were preserved separately from dead larvae. After hatching was completed, remaining eggs were inspected; infer- tile eggs and dead embryos were counted. A subset of preserved larvae was scored for yolk- sac edema, pericardial edema, and yolk volume. Ten females from the 1989 year class were randomly se- lected from each site, and 10 larvae per female were randomly selected from the central portion of hatched eggs for analysis. At Fish Bay, only five females from 5 Carls, M. G., D. M. Fremgen, J. E. Hose, S. W. Johnson, and S. D. Rice. In prep. Effects of incubating herring (Clupea pallasi) eggs in water contaminated with weathered crude oil. Auke Bay Laboratory, National Marine Fisheries Service, NOAA, 11305 Glacier Hwy., Juneau, AK 99801. Johnson et al.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 751 the 1989 year class were present; therefore the num- ber of larvae analyzed per female was doubled. Sitka and St. Mathews Bay were excluded because of an insufficient number of females from the 1989 year class. Lateral views of larvae were displayed digi- tally, and specimens were rotated to align eyes in order to minimize variance. Yolk shapes were gener- ally elliptical; major and minor axes were measured perpendicular to the body axis. Yolk volume was es- timated from these linear measures according to the method of Hourston et al. ( 1984). Yolk-sac edema was indicated if the anterior margin of the yolk membrane was bounded by an area of clear fluid. Pericardial edema was scored if the pericardium was unusually large or convex ventrally. Data processing and statistics To assess the general health of parent fish, condi- tion factor (K) was calculated for each female accord- ing to the method of Bagenal and Tesch (1978): „ 100(W) A = r 9 FLh where W = somatic wet weight in g; FL = fork length in cm; and b = the value determined by site from length-weight regressions. Gonad weight was substracted from body weight to avoid variation in spawning condition. Times of hatching among sites, which were tem- perature dependent, were compared by using peak hatching times as the estimator. Peak hatching day was defined as the day the most larvae hatched from eggs of a given female; if two hatch peaks of equal magnitude occurred, the first peak was reported. Mean incubation temperature for eggs from each fe- male was calculated by weighting mean water-bath temperatures by the number of eggs hatched daily. This method avoided possible under or over estimates of mean incubation temperature caused by early or late hatching of eggs as seasonal temperature increased. Most observations were expressed as percentages. The denominator used to calculate percentages var- ied by response parameter (Table 1). Percentages of eggs fertile and initially dead were based on the to- tal number of eggs counted near the beginning of the experiment. Percentages of eggs that hatched were based on the total number of hatched larvae plus the number of dead eggs determined at the endpoint. The number of hatched larvae was subdivided into num- ber live, moribund, and dead. Hearts of moribund larvae were beating, but these larvae were incapable Table 1 Description of key response parameters used to evaluate reproductive impairment in Pacific herring collected from Prince William Sound and southeast Alaska. Herring were collected in 1995, artificially spawned, and reared in a labo- ratory until hatching. Moribund larvae were alive (heart beating) but incapable of swimming. Parameter (%) Description Hatch 100 • (total number of eggs that hatched) /(total number of eggs that hatched + total number of dead eggs) Live (viable) 100 ■ (total number of live larvae excluding moribund larvae)/ (total number of eggs that hatched) Effective swimmers 100 ■ (total number of effective swimmers) / (total number of live larvae excluding moribund larvae) Spinal abnormalities 100 • (number of live + moribund larvae with spinal defects) / (total number of live + moribund larvae) of movement. Accordingly, percent live was the num- ber of living larvae (excluding moribund larvae) di- vided by the total number hatched. Swimming of live larvae was categorized as effective, ineffective, or incapable. Effective swimmers were active, fre- quented the water column, and avoided capture. In- effective swimmers were generally more lethargic than effective swimmers and were more likely to be found on jar bottoms. Incapable swimmers were un- able to swim in a straight line and were often ca- pable only of spasmodic twitching. Swimming of moribund and dead larvae was, by definition, non- existent; thus the number of live larvae was used as the denominator for swimming categories. Because larvae quickly became distorted after death, spinal aberrations were assessed only in live and moribund larvae. Percentage of spinal abnormalities, therefore, was the number of larvae with spinal aberrations divided by the sum of live and moribund larvae. One-way analysis of variance (ANOVA) was used to examine differences among sites, among age classes, and between regions. Each reproductive pa- rameter was tested separately by individual age class and for all age classes combined; percentage data were arc-sine transformed and corrected for small n as necessary (Snedecor and Cochran, 1980). To ac- count for variance among sites, the E-test compari- son between regions was: jp IMS between regions IMS 'among sites 752 Fishery Bulletin 95(4), 1997 All year classes Hatching Effective swimmers Figure 2 Mean (±SE) percent hatching, live, effective swimmers, and spinal abnor- malities of larval Pacific herring by site and region in Alaska, 1995. S = Sitka, M = St. Mathews Bay, K = Ketchikan, F = Fish Bay, C = Port Chalmers, R = Rocky Bay, Y = Seymour Canal; sites in chronological order of spawning date. Sample size is shown in hatching graph. No significant differences existed between regions for any reproductive parameter (P>0.50). where MS = mean square. Somatic weight, FL, and K were ana- lyzed similarly. Age-class responses within site were compared because in PWS different age classes were po- tentially exposed to varying levels of oil (Table 2). When the overall ANOVA was significant (P<0.05), a priori multiple comparisons were used to identify which ages differed: pi _ between age classes MSerror Maternal age was used as the stan- dard in all age comparisons because ages frequently differed in the male and female crosses. Age-3 and age-4 herring were not exposed to oil at any life stage in PWS; therefore they were combined as site-specific controls. Few older age fish were captured; thus, ages >age 9 were combined and reported as 9+. Because cold storage of adult fish (mean time of fish capture to mean spawning time) varied among sites (0.7-12.9 h), we also examined the possible effect of storage time on all reproductive parameters with storage time as a covariate in the ANOVA. Storage times up to 7 h did not signifi- cantly affect any of the key reproduc- tive parameters (P>0.376, except P=0.084 for % live larvae). We repeated the ANOVA for regional differences with storage times in the model as a covariate and re- stricted the analysis to include only those fish sampled within the same time period (<7 h). Scored yolk-sac edema was analyzed with the Kruskal-Wallis nonparametric test (SAS Institute Inc., 1989). Yolk-sac edema was also re-expressed as a percentage by female, arcsin-transformed, and analyzed by ANOVA. Yolk volume was analyzed by ANOVA. Results Regional comparison Herring sampled from all sites appeared healthy and showed no obvious external signs of disease. For fish of the same age, there were no significant differences in FL (P>0.09), weight (P>0.09), or condition factor (P>0.41) between regions. For herring in PWS, mean FL ranged from 196 to 260 mm and mean weight from 60.7 to 151.7 g, whereas in SE, mean FL ranged from 198 to 253 mm and mean weight from 65.1 to 140.4 g (Table 3). For all age classes combined, reproductive success of herring did not differ significantly between regions (P>0.50); the best and worst responses generally oc- curred in SE, whereas PWS sites were intermediate (Fig. 2). Statistical power of these tests was high (>0.99) and remained high for most analyses. Re- stricting the analysis to fish stored for <7 h did not alter the overall results; no significant (P>0.39) re- gional differences existed for any reproductive pa- rameter. In SE, mean responses ranged from 63 to 91% for hatching success, 95 to 98% for live larvae, 90 to 96% for effective swimmers, and 1 to 7% for Johnson et a I.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 753 Table 2 Age, year class, and possible oil exposure for Pacific herring collected in Prince William Sound, Alaska, in 1995. The Exxon Valdez oil spill occurred in March 1989. Year Age class Possible oil exposure Year Age class Possible oil exposure 3 1992 no direct oil exposure of any life stage 4 1991 no direct oil exposure of any life stage 5 1990 all life stages possibly exposed to residual oil 6 1989 all life stages likely exposed to oil 7 1988 juveniles at time of spill 8 1987 juveniles or immature at time of spill 9+ 1986 mature — reproductive at time of spill Table 3 Fork length (mm) and somatic weight (g) of mature female Pacific herring captured in southeast (SE) and Prince William Sound (PWS), Alaska, in spring 1995. Values are mean ( x ) and ± standard error; sample size = n. Age (yr) 3 4 5 6 7 8 9 10 11 Fork length SE X 198 211 217 221 236 234 241 236 253 ± 1.9 2.9 2.0 1.2 1.3 3.5 8.4 3.5 7.7 n 86 21 49 94 95 15 3 2 3 PWS X 196 219 225 236 242 260 259 259 260 ± 1.1 2.3 1.1 1.7 1.0 3.2 1.3 2.2 2.9 n 81 18 65 25 149 10 16 7 13 Weight SE X 65.1 79.1 87.2 91.7 112.9 112.7 117.5 108.3 140.4 ± 1.8 3.2 2.1 1.7 1.9 5.1 4.6 6.4 14.7 n 81 21 49 93 94 15 3 2 3 PWS X 60.7 90.3 95.0 107.4 121.0 149.0 147.8 148.9 151.7 ± 1.2 2.2 1.3 2.7 1.5 8.3 3.3 6.8 5.1 n 80 18 65 25 149 10 16 7 13 spinal abnormalities. In PWS, mean responses ranged from 78 to 86% for hatching success, 95 to 96% for live larvae, 92 to 93% for effective swimmers, and 4 to 6% for spinal abnormalities. Among all sites, reproductive success was consistently best at Sitka (e.g. highest hatching success=91% and fewest spi- nal abnormalities=l%) and worst at Seymour Canal or Ketchikan (e.g. lowest hatching success=63% and most spinal abnormalities=7%) (Fig. 2). Of the sites in PWS, reproductive success was usually best at St. Mathews Bay or Port Chalmers (e.g. highest hatch- ing success=86% and fewest spinal abnormali- ties=4%) and worst at Fish Bay (e.g. lowest hatching success=78% and most spinal abnormalities=6%) (Fig. 2). Similarly, when reproductive success was estimated for each age class individually, regional differences were not significant (P>0.50). This find- ing was true for all age comparisons — ages 3 to 9+. For example, age-6 (1989 year class) herring in PWS did not differ significantly from those in SE (Fig. 3). Among all sites where more than four age-6 fish were collected (excluding St. Mathews Bay and Sitka), hatching success ranged from 66% (Seymour Canal) to 91% (Port Chalmers), live larvae from 95% (Fish Bay) to 98% (Port Chalmers), effective swimmers from 93% (Rocky Bay) to 96% (Port Chalmers), and spinal abnormalities from 2% (Port Chalmers) to 6% (Fish Bay). No significant regional differences were observed in progeny of the 1989 year class scored for physical condition. Only one larva of 500 had pericardial edema. Analyzed with the Kruskal-Wallis test, the site with the most yolk-sac edema (Port Chalmers) was significantly different from that with the least (Ketchikan), but there was no regional trend. Per- centages of larvae with yolk-sac edema were low 754 Fishery Bulletin 95(4), 1 997 1989 year class Hatching Effective swimmers 100-i 1 -i i i i i i I i i i i i I r SMKFCRY SMKFCRY Site Figure 3 For the 1989 year class, mean (±SE) percent hatching, live, effective swim- mers, and spinal abnormalities of larval Pacific herring by site and region in Alaska, 1995. The 1989 year class, sampled in Prince William Sound in 1995, was more likely exposed to oil as eggs or larvae than were other year classes. S = Sitka, M = St. Mathews Bay, K = Ketchikan, F = Fish Bay, C = Port Chalmers, R = Rocky Bay, Y = Seymour Canal. Sample size is shown in hatch- ing graph. Progeny of the 1989 year class did not differ significantly between regions for any reproductive parameter (P>0.50). (<16%), and differences among sites and between regions were not signifi- cant (P>0.348) (Fig. 4). Yolk volume in larvae from the 1989 year class did not differ signifi- cantly (P=0.486) between regions but may have been related to incubation temperature. The largest and small- est mean yolk volumes were observed in PWS but closely overlapped those in SE (Fig. 4). Although scatter was high (r2=0.13), yolk volumes declined significantly (P<0.001) as tempera- ture increased. It is possible, how- ever, that site differences and incu- bation temperature were confound- ing factors. Comparison among age classes within sites Reproductive success differed signifi- cantly among some age classes at Sitka, Ketchikan, Port Chalmers, and Rocky Bay but not among age classes at St. Mathews Bay, Fish Bay, and Seymour Canal (Figs. 5-8). The few significant differences we ob- served were highly variable, incon- sistent among sites, and no pattern existed for the 1989 year class. For example, at Rocky Bay, age-3 and age-4 fish had a significantly lower percentage of live larvae than age-5, age-6, and age-7 fish (Fig. 6), whereas at Sitka, age-3 and age-4 fish had a significantly higher percentage of effective swimmers and a signifi- cantly lower percentage of spinal abnormalities than age-7 fish (Figs. 7 and 8). Other parameters Hatching times decreased steadily with increasing incubation temperature (Fig. 9). For Sitka, the first site sampled, peak hatching occurred about 33 d af- ter start of incubation at a mean temperature of about 4.5°C, whereas at Seymour Canal, the last site sampled, peak hatching occurred about 26 d after start of incubation at a mean temperature of about 6.0°C. Fertility did not differ significantly (P>0.50) be- tween regions for all ages combined or when the com- parison was restricted to fish of the same age. For all ages combined, fertility in SE ranged from 80% at Seymour Canal to 96% at Sitka; in PWS, fertility ranged from 88% at Fish Bay to 94% at St. Mathews Bay. Discussion Six years after the spill, reproductive impairment was not detected in PWS herring. This conclusion was reached by comparing reproductive success of fish collected in PWS and SE and among age classes within specific sites. Specifically, hatching success, larval viability, and fertility did not differ signifi- cantly between PWS and SE, including response of the 1989 year class. In fact, discrimination of re- sponses between regions was not possible because the best and worst responses were usually found in Johnson et al.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 755 1 989 year class Yolk-sac edema P=0.348 Figure 4 For the 1989 year class, mean (±SE) per- cent yolk-sac abnormalities (edema) and yolk volume for larval Pacific herring by site and region in Alaska, 1995. Sample size is shown in each bar. S = Sitka, M = St. Mathews Bay, K= Ketchikan, F = Fish Bay, C = Port Chalmers, R = Rocky Bay, Y = Seymour Canal. No significant differ- ences existed between regions for either parameter (P>0.348). one region (SE). Therefore, the chances of detecting any oil-related effects against the natural background variation were negligible when herring were com- pared between regions. Although responses among some age classes within Port Chalmers and Rocky Bay were occasionally significant, these differences were highly variable, did not indicate reproductive impairment of the 1989 year class, and were incon- sistent between sites. Of the four key reproductive parameters we ex- amined, spinal defects were particularly important because exposure of herring eggs to oil frequently causes spinal defects (Linden, 1978; Kocan et ah, 1987; Rice et al., 1987; Pearson et al.6 ), that could 6 Pearson, W. H., D. L. Woodruff, S. L. Kiesser, G. W. Fellingham, and R. A. Elston. 1985. Oil effects on spawning behavior and reproduction in Pacific herring ( Clupea harengus pallasi). Fi- nal Report OF-1742 to American Petroleum Inst., Battelle Ma- rine Res. Lab., Sequim, WA, 108 p. [API publication 4412.] result in reduced swimming ability and long-term survival. Spinal defects, however, can also occur natu- rally as a result of other environmental factors. In our study, herring from an uncontaminated site, Ketchikan, had the highest percentage of spinal de- fects (7%). Ketchikan samples were collected at least 40 km from any urban area, and it is unlikely that these fish were exposed to industrial or other urban pollutants. Whether the incidence of spinal defects at Ketchikan was just random noise or a response to some underlying environmental factor is impossible to determine, but it is evidence that similar results could occur in PWS without implicating oil as a cause. In fact, a 10% incidence of gross abnormalities was observed in PWS herring 23 years prior to the spill (Smith and Cameron, 1979). Reproductive success of herring in PWS was con- sistently better in 1995 than that reported in earlier studies. For example, we observed a mean hatching success of 78-86% compared with 53% in 1976 (Smith and Cameron, 1979), 62%7 in 1989 (McGurk et al.8 ), 85% in 1990 (McGurk et al.9 ), 59-79% in 1991 (Kocan et al., 1996a), and 19-56% in 1992 (Kocan et al., 1996b). The viable hatching10 that we observed in PWS (79%) also exceeded previously reported per- centages; 53%n in 1989 (McGurk et al.8), 57% in 1990 (McGurk et al.9), 35-37%12 in 1991 (Kocan et al., 1996a), and 13-33%12 in 1992 (Kocan et al., 1996b). Incidence of spinal abnormalities in PWS was about 5% in our study compared with 7% in 1989 (McGurk 7 To avoid desiccation effects, and because egg survival was sig- nificantly less in the +1.5-m collections in the McGurk et al.8 data set, these data were not included in this comparison. Es- timated egg survival was 59% when the +1.5-m data were in- cluded. 8 McGurk, M., D. Warburton, T. Parker, and M. Litke. 1990. Early life history of Pacific herring: 1989 Prince William Sound herring egg incubation experiment. Final report, contract number 50ABNC-7-00141, Triton Environmental Consultants LTD., No. 120-13511 Commerce Parkway, Richmond, British Columbia, Canada V6V 2L1. 9 McGurk, M., T. Watson, D. Tesch, B. Mattock, and S. Northrup. 1991. Viable hatch of Pacific herring eggs from Prince William Sound and Sitka Sound, Alaska, in 1990. Re- port number 2060/WP 4269, Triton Environmental Consult- ants LTD., No. 120-13511 Commerce Parkway, Richmond, Brit- ish Columbia, Canada V6V 2L1. 10 To conform with McGurk et al.8,9, % viable hatch was defined as 100 [(no. live larvae - no. abnormal larvae )/(no. hatched eggs)] x (no. eggs hatched/no. eggs total). The value defined by Kocan et al. (1996, a and b) as % viable larvae is nearly syn- onymous with % viable hatch. Our % live larvae (Table 1) in- cluded abnormal larvae, but McGurk et al.8 9 excluded abnor- mal larvae in their definition of % viable larvae (% viable = 100 (no. live larvae - no. abnormal larvae )/no. hatched). 11 As previously, +1.5-m data were not included; estimated % vi- able hatch was 50% when these data were included. 12 Percent viable larvae values reported by Kocan et al. (1996, a and b) should be increased by 2% to approximate percent vi- able hatch. 756 Fishery Bulletin 95(4), 1997 et al.8). Although procedural differences between earlier studies and ours may partially account for dif- ferences in assessment of reproductive success, the re- sponses we observed in 1995 were consistently the best. To interpret the effects of the spill on herring in PWS, it is necessary to understand the life stage ex- posed and the magnitude and duration of exposure. Which life stages were impacted, and to what extent, however, is largely a matter of conjecture. Adult fish may have encountered oil before, during, or after spawning, but determining what percentage of the population was significantly impacted is impossible. Metabolites of aromatic hydrocarbons were detected in adult herring (Haynes et al.13 ), but sample sizes were very low. Nematode prevalence in adult body cavities differed signifi- cantly between contaminated and uncontaminated areas (Moles et al., 1993), also indicating adult exposure. The duration and magnitude of oil exposure of herring eggs and larvae is also unknown. After hatching, herring larvae from both contaminated and uncontaminated sites may have been exposed to oil as they passively tra- versed the spill trajectory. For example, some of the largest concentrations of larvae in June were found in the south- west portion of PWS, well within the oil trajectory (Norcross et al., 1996). By inference, juvenile herring occupying the same nearshore habitat used by juvenile salmonids may have also been exposed to oil: such exposure was docu- mented in juvenile pink and chum salmon (Carls et al., 1996). Response of wild herring to an oil spill can be partially inferred from laboratory studies. For example, ex- posure of mature herring to hydro- carbons in the laboratory did not cause discernible damage in progeny, including fertility, viability, and lar- val swimming, physical, and genetic abnormalities (Rice et al., 1987; Carls et al.14). In contrast, the early life 13 Haynes, E., T. Rutecki, M. Murphy, and D. Urban. 1995. Impacts of the Exxon Valdez oil spill on bottomfish and shellfish in Prince William Sound, Exxon Valdez oil spill state/federal natural resource damage assessment final report (fin/shellfish study no. 18). Auke Bay Laboratory, National Marine Fisheries Service, 11305 Glacier Hwy., Juneau, AK 99801. 14 Carls, M. G., D. M. Fremgen, J. E. Hose, D. Love, and R. E. Thomas. 1995. The im- pact of exposure of adult pre-spawn herring (Clupea harengus pallasi ) on subsequent progeny. Chapter 2 in Carls et al., Exxon Valdez oil spill report, restoration project 94166, annual report; the impact of exposure of adult pre-spawn herring (Clupea harengus pallasi) on subsequent progeny, p. 29- 49. Auke Bay Laboratory, NMFS, NOAA, 11305 Glacier Hwy., Juneau, AK 99801. PWS St. Mathews Bay SE AK P=0.331 Sitka P=0.021 23456789 10 11 Age (years) Figure 5 Mean (±SE) percent hatching of larval Pacific herring by female parent age, site, and region in Alaska, 1995. Sample size is shown in each bar. Overall P- value from ANOVA is listed above each graph. Significant differences were Ketchikan, ages 3 and 4 < age 6, 7, and 8 (P<0.015); Port Chalmers, ages 3 and 4 > age 9+ (P=0.050). Johnson et al.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 757 PWS SE AK Age (years) Figure 6 Mean (+SE) percent live larvae of Pacific herring by female parent age, site, and region in Alaska, 1995. Sample sizes are indicated in Figure 5. Overall P- value from ANOVA is listed above each graph. Significant differences were Rocky Bay, ages 3 and 4 < age 5, 6, and 7 (P<0.047). stages of herring are more suscep- tible to the effects of oil according to laboratory (Linden, 1978; Pearson et al., 1985; Carls, 1987; Kocan et al., 1987; Rice et al., 1987) and field stud- ies (Brown et al., 1996a; Norcross et al., 1996). Abnormal larvae have poor survival potential (Kocan et al., 1996a), and thus the exposure of eggs and larvae to oil in PWS may have resulted in increased mortality. Fur- thermore, the same oil concentra- tions that caused significant genetic damage also caused significant physical damage in developing em- bryos (Carls et al.5); thus early death would likely preclude recruitment of genetically damaged individuals to spawning populations. Although genetic damage was de- tected in larvae collected in the oil- contaminated areas of PWS in 1989 (Hose et al., 1996; Brown et al., 1996a), we did not inspect larvae for genetic damage. Concomitant labo- ratory measurements of larvae that had been artificially contaminated indicated that genetic response was not a more sensitive measure of oil exposure than the parameters we examined (Carls et al.5). In addition, artificial exposure of prespawning adults to relatively high oil concen- trations (58 ppb, initial PAH) did not cause genetic defects in artificially spawned progeny (Carls et al.14). Other defects observed in larvae from PWS in 1989 included physical damage, assessed by scored indices (Hose et al., 1996). Carls et al.5 ob- served that two of these indices, peri- cardial edema and finfold condition, were more sensitive to oil damage than was the genetic response. Be- cause we did not detect significant pericardial abnormalities in larvae from PWS six years after the spill, it is likely that the genetic condition of these larvae has not been adversely affected. The failure of the 1989 year class of herring in PWS to recruit to the spawning population may have been partly attributable to the spill, but it is impossible to separate oil effects from other natural factors. At the sites we sampled in PWS, the 1989 year class usually represented <4.0% of the spawning popula- tion (ADF&G15 ). Larval survival in PWS was reduced an estimated 52% in 1989 as a result of the spill (Brown et al., 1996a); such loss supports inferences of poor survival that are based on laboratory obser- 15 ADF&G (Alaska Department of Fish and Game). 1995. Her- ring test fishery data. Commercial Fisheries Management and Development Division, PO Box 669, Cordova, AK 99574. 758 Fishery Bulletin 95(4), 1997 100 PWS St. Mathews Bay SEAK SO- SO- 40- 20- iff I fMJ.184 Sitka ■ 111 P-0.021 P ill Pi 111 81 Age (years) Figure 7 Mean (±SE) percent effective swimmers of larval Pacific herring by female parent age, site, and region in Alaska, 1995. Sample sizes are indicated in Figure 5. Overall P-value from ANOVAis listed above each graph. Significant differences were Sitka, ages 3 and 4 > age 7 (P=0.011). vation. Natural environmental conditions, however, can also cause a high degree of variability in herring recruitment (Stevenson, 1962; Anthony and Fogarty, 1985). For example, the 1989 year class at Sitka also represented a small proportion of the spawning popu- lation in 1995 (<2%; ADF&G16); therefore factors other than oil are important determinants of cohort size. Whether or not herring in PWS were ever repro- ductively impaired by the EVOS is unknown, but the time lapse between the spill and our study probably precluded any detection of reproductive impairment. 16 ADF&G (Alaska Department of Fish and Game). 1995. Her- ring test fishery data. Commercial Fisheries Management and Development Division, 304 Lake St., Room 103, Sitka, AK 99835. Johnson et a I.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 759 PWS SE AK „ St. Mathews Bay P-0.434 Sitka P-0.005 1 80- 60- 40- 20- U 1 1 1 f " 1 " 1 "'1 1 1 1 1 1 1 1 1 , — - T ■ | 1 Fish Bay P-0.067 Ketchikan P» 0.228 IUU 80- 60- 40- 3? i 20: - ^ C U 1 1 1 I I 1 1 I i i i i i i i » i ' ' t | Port Chalmers P-0.287 Seymour P-0.311 ^ y yy w 75 s 80- Q. to 60- 40- 20- zrr7Y-)rA/}' A TT77M')A fA-h - u i "i r ~ f f i i i r „„„ Rocky Bay P-0.108 i "r r "'i i i i — i — i — i 23456789 10 11 IUU - 1 80- 60- 40- 20- Age (years) u 1 l l 1 1 — 1 — 1 — 1 1 1 23456789 10 11 Age (years) Figure 8 Mean (±SE) percent spinal abnormalities of larval Pacific herring by female parent age, site, and region in Alaska, 1995. Sample sizes are indicated in Figure 5. Overall P-value from ANOVA is listed above each graph. Signifi- cant differences were Sitka, ages 3 and 4 < age 7 (P=0.024). Measurable effects likely declined most rapidly dur- ing the first year as the most adversely affected in- dividuals died. Although oil-related abnormalities were observed in larvae immediately following the spill, both developmental and genetic damage pro- gressively decreased with time (Brown et al., 1996a) and were undetectable in 1990 and 1991 (Hose et al., 1996). The extent of spawning-site fidelity in herring is poorly understood, but unaffected individu- als from other geographic areas have probably joined remaining, less affected spawners, diluting possible residual effects. The disease epidemic observed in PWS in 1993 (Meyers et al., 1994) may have removed additional marginal spill survivors. Thus, it is not particularly surprising that reproductive impairment was not detected in 1995. 760 Fishery Bulletin 95(4), 1 997 1 1 I T'TTt S M K F C R Y Site Figure 9 Peak hatching day and mean (±SE) incu- bation temperature (°C) for eggs spawned from mature Pacific herring by site and region in Alaska, 1995. S = Sitka, M = St. Mathews Bay, K = Ketchikan, F = Fish Bay, C = Port Chalmers, R = Rocky Bay, Y = Seymour Canal; sites are in chronologi- cal order of spawning date. Understanding the long-term implications of ex- posure of Pacific herring to oil in PWS was the prin- cipal objective of this research. Regardless of the life stage of herring and the likelihood of possible oil ex- posure, herring we sampled in PWS in 1995 appeared to be reproductively fit and similar to herring in SE. Although herring stocks are still depressed in PWS, factors other than reproductive impairment are prob- ably limiting recovery. Acknowledgments We thank John Campbell, Jamison Clark, Mary Drew, Dan Fremgen, Chris Hanson, Dave Love, Xiaohong Luan, Nathan Magaret, Joshua Millstein, Catherine Pohl, Jeff Regelin, June Sage, Michele Sleeter, and Justin Stekoll for technical laboratory assistance. We also thank the Alaska Department of Fish & Game Commercial Fisheries Division in Sitka, Ketchikan, Juneau, and Cordova for help in collect- ing herring and for providing laboratory space. The research described in this paper was supported by the Exxon Valdez Oil Spill Trustee Council. The find- ings and conclusions presented by the authors, how- ever, are their own and do not necessarily reflect the view or position of the Trustee Council. Literature cited Anthony, V. C., and M. J. Fogarty. 1985. Environmental effects on recruitment, growth, and vulnerability of Atlantic herring (Clupea harengus harengus) in the Gulf of Maine region. Can. J. Fish. Aquat. Sci. 42 (suppl. 1 ): 158—173. Bagenal, T. B., and F. W. Tesch. 1978. Age and growth. Chapter 5 in T. Bagenal (ed.), Fish production in fresh waters; age and growth, p. 101- 136. Blackwell Scientific Pubis., third ed. (IBP Handbook No. 3). Brown, E. D., T. T. Baker, J. E. Hose, R. M. Kocan, G. D. Marty, M. D. McGurk, B. L. Norcross, and J. Short. 1996a. Injury to the early life history stages of Pacific her- ring in Prince William Sound after the Exxon Valdez oil spill. In S. D. Rice, R. B. Spies, D. A. Wolfe, and B. A. Wright (eds.), Proceedings of the Exxon Valdez oil spill symposium. Am. Fish. Soc. Symp. 18:448-462. Brown, E. D., B. L. Norcross, and J. W. Short. 1996b. An introduction to studies on the effects of the Exxon Valdez oil spill on early life history stages of Pacific her- ring, Clupea pallasi, in Prince William Sound, Alaska. Can. J. Fish. Aquat. Sci. 53:2337-2342. Carls, M. G. 1987. Effects of dietary and water-borne oil exposure on larval Pacific herring (Clupea harengus pallasi). Mar. Environ. Res. 22:253-270. Carls, M. G., A. C. Wertheimer, J. W. Short, R. M. Smolowitz, and J. J. Stegeman. 1996. Contamination of juvenile pink and chum salmon by hydrocarbons in Prince William Sound after the Exxon Valdez oil spill. In S. D. Rice, R. B. Spies, D. A. Wolfe, and B. A. Wright (eds.), Proceedings of the Exxon Valdez oil spill symposium. Am. Fish. Soc. Symp. 18:593-607. Hose, J. E., M. D. McGurk, G. D. Marty, D. E. Hinton, E. D. Brown, and T. T. Baker. 1996. Sublethal effects of the Exxon Valdez oil spill on her- ring embryos and larvae: morphologic, cytogenetic, and histopathological assessments, 1989-1991. Can. J. Fish. Aquat. Sci. 53:2355-2365. Hourston, A. S. 1982. Homing by Canada’s west coast herring to manage- ment units and divisions as indicated by tag recoveries. Can. J. Fish. Aquat. Sci. 39:1414-1422. Hourston, A. S., H. Rosenthal, and H. von Westernhagen. 1984. Viable hatch from eggs of Pacific herring (Clupea harengus pallasi ) deposited at different intensities on a variety of substrates. Can. Tech. Rep. Fish. Aquat. Sci. 1274, 19 p. Kocan, R. M., J. E. Hose, E. D. Brown, and T. T. Baker. 1996a. Pacific herring (Clupea pallasi) embryo sensitivity to Prudhoe Bay petroleum hydrocarbons: laboratory evalu- ation and in situ exposure at oiled and unoiled sites in Prince William Sound. Can. J. Fish. Aquat. Sci. 53:2366- 2375. Johnson et a I.: Reproductive success of Clupea pallasi after the Exxon Valdez oil spill 761 Kocan, R. M., G. D. Marty, M. S. Okihiro, E. D. Brown, and T. T. Baker. 1996b. Reproductive success and histopathology of indi- vidual Prince William Sound Pacific herring 3 years after the Exxon Valdez oil spill. Can. J. Fish. Aquat. Sci. 53:2388-2393. Kocan, R. M., H. von Westernhagen, M. L. Landolt, and G. Furstenberg. 1987. Toxicity of sea-surface microlayer: effects of hexane extract on Baltic herring (Clupea harengus ) and Atlantic Cod (Gadus morhua) embryos. Mar. Environ. Res. 23:291-305. Linden, O. 1978. Biological effects of oil on early development of the Baltic herring Clupea harengus membras. Mar. Biol. (Berl.) 45: 273-283. Marty, G. D., J. E. Hose, M. D. McGurk, and E. D. Brown. In press. Histopathology and cytogenetic evaluation of Pacific herring larvae exposed to petroleum hydrocarbons in the laboratory or in Prince William Sound, Alaska, af- ter the Exxon Valdez oil spill. Can. J. Fish. Aquat. Sci. Meyers, T. R., S. Short, K. Lipson, W. N. Batts, J. R. Winton, J. Wilcock, and E. Brown. 1994. Association of viral hemorrhagic septicemia virus with epizootic hemorrhages of the skin in Pacific herring Clupea harengus pallasi from Prince William Sound and Kodiak Island, Alaska, USA. Dis. Aquat. Org. 19:27-37. Moles, A. D., S. D. Rice, and M. S. Okihiro. 1993. Herring parasite and tissue alterations following the Exxon Valdez oil spill. In Proceedings of the 1993 Oil Spill Conference, American Petroleum Institute Publication 4580, p. 325-328. API, Washington, D.C. 20005. Norcross, B. L., J. E. Hose, M. Frandsen, and E. D. Brown. 1996. Distribution, abundance, morphological condition, and cytogenetic abnormalities of larval herring in Prince William Sound, Alaska, following the Exxon Valdez oil spill. Can. J. Fish. Aquat. Sci. 53:2376-2387. Rice, S. O., M. M. Babcock, C. C. Brodersen, J. A. Gharrett, and S. Korn. 1987. Uptake and depuration of aromatic hydrocarbons by reproductively ripe Pacific herring and the subsequent effect of residues on egg hatching and survival. In W. B. Vemberg, A. Calabrese, F. P. Thurberg, and F. J. Vernberg (eds.), Pollu- tion physiology of estuarine organisms, 139-154. Belle W. Baruch Libr. Mar. Sci. 17, Univ. S.C. Press, Columbia, SC. SAS Institute, Inc. 1989. SAS/STAT-user’s guide, version 6, 4th ed, vol. 2. SAS Inst., Inc., Cary, NC, 846 p. Short, J. W., and P. M. Harris. 1996a. Chemical sampling and analysis of petroleum hy- drocarbons in near-surface seawater of Prince William Sound after the Exxon Valdez oil spill. In S. D. Rice, R. B. Spies, D. A. Wolfe, and B. A. Wright (eds.), Proceedings of the Exxon Valdez oil spill symposium. Am. Fish. Soc. Symp. 18:17-28. 1996b. Petroleum hydrocarbons in caged mussels deployed in Prince William Sound, Alaska after the Exxon Valdez oil spill. In S. D. Rice, R. B. Spies, D. A. Wolfe, and B. A. Wright (eds.), Proceedings of the Exxon Valdez oil spill symposium. Am. Fish. Soc. Symp. 18:29-39. Smith, R. L., and J. A. Cameron. 1979. Effect of water soluble fraction of Prudhoe Bay crude oil on embryonic development of Pacific herring. Trans. Am. Fish. Soc. 108:70-75. Snedecor, G. W., and W. G. Cochran. 1980. Statistical methods, 7th ed. Iowa State Univ. Press, Ames, Iowa, 507 p. Stevenson, J. C. 1962. Distribution and survival of herring larvae ( Clupea pallasi Valenciennes) in British Columbia waters. J. Fish. Res. Board Can. 19(5):735-810. Zijlstra, J. J. 1963. On the recruitment mechanisms of North Sea autumn spawning herring. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 154:198-202. 7 62 Abstract .—We examine the popu- lation to population variability of in- trinsic rate of natural increase, rm, of Atlantic cod, Gadus morhua. The in- trinsic rate of increase is positively re- lated to temperature, contrary to the expectation that rm might increase as the high and low temperature limits of habitability for cod are approached. For the parameter regime considered, rm has a simple dependence on age-at-ma- turity and the number of replacements each spawner can produce at low popu- lation densities (a). It is shown that a. has no significant temperature depen- dence, and thus the covariation of rm and temperature arises from the influ- ence of temperature on age-at-maturity. We demonstrate that our estimates of rm are robust and thus may be of use in estimating the recovery time of de- pleted populations. Manuscript accepted 25 March 1997. Fishery Bulletin 95:762-772 ( 1997). Maximum population growth rates and recovery times for Atlantic cod, Gadus morhua Ransom A. Myers* Gordon Mertz Science Branch Department of Fisheries and Oceans RO. Box 5667 St John's, Newfoundland, Canada A1C 5X1 * Present address: Department of Biology Dalhousie University Halifax, Nova Scotia, Canada B3H 4J1 E-mail address (for R. Myers):Ransom.Myers@Dal.Ca P. Stacey Fowlow Seaborne Ltd. PO. Box 2035 Station C, 200 White Hills Road St. John's, Newfoundland, Canada A I C 5R6 Perhaps the most fundamental of all ecological parameters is the in- trinsic rate of natural increase, r (Cole, 1954; Pimm, 1991). High rm will be selected for in populations that experience frequent excursions to low density (Charlesworth, 1994). Populations subjected to strong en- vironmental variability should evolve toward high rm (Mac Arthur and Wilson, 1967), which will im- part resilience to the population. However, allometric (cross species) comparisons (Fenchel, 1974; Hen- nemann. 1983; Charnov, 1993) sug- gest that rm chiefly depends on metabolic rate or somatic growth rate. Perhaps, the influence of en- vironmental variability on rm can be more readily discerned in cross population comparisons for a single species. In this paper we examine 20 populations of Atlantic cod, Ga- dus morhua , to determine their maximum growth rates. We also discuss how these estimates help predict the recovery times of se- verely overfished populations. Atlantic cod lends itself to a study of this nature because there is a wealth of good quality biological data collected for stock manage- ment purposes. Moreover, these cod populations occupy a broad span of latitudes, including regions that are thought to represent the northern and southern limits of habitability for cod, and there is evidence that population variability increases as these extremes are approached (Myers, 1991). The increase in popu- lation variability at the limits of the range of cod could impose con- straints on r that would mask the m simple dependence of rm on meta- bolic rate or somatic growth rate apparent in cross species compari- sons. In fact, we will show, in what we believe is an unanticipated re- sult, that even for a within-species comparison, there is strong coupling between r and metabolic rate or TYl somatic growth rate (as represented by age at maturity or temperature). Our results have implications for the recovery rates of a number of Myers et at: Population growth rate of Gadus morhua 763 recently collapsed Atlantic cod populations (Hutchings and Myers, 1994). Methods Model For fish populations, reproduction is generally ex- pressed as recruitment, the number of juvenile fish reaching, in a given year, the age of vulnerability to fishing gear. Thus, the reproduction curve (Royama, 1993) for fish is displayed as a spawner-recruitment curve (Ricker, 1954), and rm must be derived from the slope of this curve near the origin (low popula- tion). This derivation will be presented immediately following a brief discussion of the standard popula- tion-recruitment curves. Juvenile fish become vulnerable to fishing gear, that is, they recruit at an age designated as j' . We consider the Ricker spawner-recruitment model which describes the number of recruits at age j' in year t+j\ N t+j\ j\ resulting from a spawning stock biomass (SSB) of Sf. We follow the usual convention in fisheries science of assuming that the number of eggs produced is proportional to the biomass of spawners. The Ricker model has the form E(Nt+r r) = aSte -ps, (1) where a = the slope at the origin (measured perhaps in recruits per kilogram of spawners). Density-de- pendent mortality is assumed to be the product of P multiplied by the spawning biomass (St). For the forthcoming calculations the slope at ori- gin, a, must be standardized. First consider a -a SPR F= 0 ’ where SPRF=Q is the spawning biomass resulting from each recruit (perhaps in units of kg of spawn- ing fish per recruit) in the limit of no fishing mortal- ity (F= 0). This quantity, a , represents, on a lifetime basis, the number of recruits per recruit at very low spawner abundance or, equivalently, the number of spawners produced per spawner (assuming that there is constant survival from recruit to spawner). The quantity, a , required for our calculations is the number of spawners produced by each spawner per year (after a lag of a years, where a is age-at-maturity). If adult survival is p then a = 'L-pJa, or sum- mmg the geometric series If the annual survival fraction for spawners was zero, the population of spawners, Nt, would obey the following equation: N =aN t+a (3) Equation 3 has the solution Nt=na = a nN0, where N0 is the number of spawners at t = 0. It follows that the natural growth rate, per annum, of the popula- tion is rm - (1/ n)loga, (4) for the limit of small pg. The analogous result for the case of overlapping generations is derived below. When adult survival in not zero (ps ^ 0), one has an age-structured spawning population, and rm can- not be derived in the simple manner presented above. Rather, one must solve the Euler-Lotka equation (Charlesworth, 1994) to obtain rm in this situation. The Euler-Lotka equation is l,rn,e (5) where Z = the fraction of animals surviving to age j ; and m- = the number of offspring per animal pro- duced at age j. We now assume that ra • - mQ for fish of age a and older, and also, for j > a, L = la pj ~a, where la is the fraction of juveniles that survive from age zero to age a, and, again, ps is the annual survival fraction of spawners. It follows from Equation 5 that (6) A little manipulation, and the summing of a geomet- ric series, allows Equation 6 to be written as /— r a am 0e =1 1 ~ Pse ^ (7) Since m0 is the number of age-zero fish produced by each spawner, and since la is the fraction of age-zero fish surviving through the juvenile stage to matu- rity, it follows that m0/Q = a , and thus Equation 7 can be expressed as (2) (er" )° - ps(er" )a_1 - a = 0. a = a(l-ps) = a-SPRf=0(l-ps). (8) 764 Fishery Bulletin 95(4), 1997 We have bracketed the em m term to emphasize that Equation 8 is a simple algebraic equation for x= er- . Note that in the limit ps = 0, we recover Equation 4 from Equation 8. Equation 8 is very similar to Equa- tion 1 of Goodman (1984), amounting to a transla- tion into parameters available for fish populations. Equation 8 may also be obtained as the low density limit of the simplified age-structured model of Clark (1976), as modified by Mertz and Myers (1996). It is clear for a moderately large slope at the ori- gin ( a) that age of maturity (a) is the most impor- tant factor in determining rm (Fig. 1). The solid line in Figure 1 shows, for reference, the case ps = 0, for which rm may be calculated from Equation 4. The three broken lines in Figure 1 represent ps, = 0.7, 0.8, 0.9, a range that should encompass all North Atlantic cod stocks (see next section). For this range, survival after reproduction (ps) has only a minor ef- fect on r . m. Data sources and treatment The data we used are estimates obtained from as- sessments compiled by Myers et al. (1995b). Popula- tion numbers and fishing mortality were estimated by using sequential population analysis (SPA) of com- mercial catch-at-age data for most marine popula- tions. Sequential population analysis techniques in- clude virtual population analysis ( VPA), cohort analy- sis, and related methods that reconstruct population size from catch-at-age data (see Hilborn and Walters, 1992) chapters 10 and 11, for description of the meth- ods used to reconstruct the population history). Briefly, the commercial catch-at-age is combined with estimates from research surveys and commercial catch rates to estimate numbers-at-age in the final year and to reconstruct previous numbers-at-age under the assumption that commercial catch-at-age is known without error and that natural mortality- at- age is known and constant. The population boundaries in the North Atlantic generally follow those of the Northwest Atlantic Fish- eries Organization (NAFO) or the International Council for the Exploration of the Sea (ICES) (Fig. 2). Many populations cover more than one NAFO or ICES unit area, e.g. the cod population off Labrador and Northeast Newfoundland, known as “northern” cod, inhabits three NAFO divisions (2J, 3K, and 3L) and is designated as 2J3KL cod. There are three minor populations that are not included in the com- parative analysis: Flemish Cap, Gulf of Maine, and the English Channel. There are no reliable catch data for the Flemish Cap population (NAFO 3M) or the English Channel population (ICES Vlld), and the 1.2 1.0 0.8 <5 0.6 >, Q Q. 4s 0.4 0.2 0.0 Figure 1 Rate of population growth (rm) as a function of slope at the origin ( a ) at minimum population size, age of maturity (a), and adult survival rates ps = 0 (solid line), ps = 0.7 (dotted line),ps = 0.8 (short-dashed line), and ps = 0.9 (long-dashed line) Slope at the origin (a) Myers et al.: Population growth rate of Gadus morhua 765 time series of data for the Gulf of Maine (NAFO 5Y) is not included in the comparative analysis because the time series is too short (less than 10 years) (Myers et al., 1995b). Maturity and weight-at-age were estimated from research surveys carried out for each population. Typically hundreds of fish are measured from a strati- fied random or systematic design each year. Adult natural mortality (M) has been investigated by a variety of methods; for each of the 20 populations, the value of M= 0.2 was accepted by the researchers studying each population. This natural mortality corresponds to adult survival of ps = e~M ~ 0.8 (which will be our working value). The uniformity (across populations) of acceptable values for ps suggests that one can safely assume that the true ps lies in the range 0.7 to 0.9, as shown in Figure 1. Bottom temperature was estimated by Brander ( 1994) for all the populations except the Baltic popu- lations. We used data in Dickson et. al. (1992) to make estimates for the Baltic populations. We estimated independent temperatures for the Northwest Atlan- tic from deYoung et. al. (1994). These changes were small (less than 2°C) and did not have an important effect on our findings. Estimation The slope at the origin, d , and other parameters of the Ricker model were fitted by using maximum-like- lihood estimation and by assuming lognormal vari- ability in the residuals (Hilborn and Walters, 1992); the assumption of gamma variability in model re- siduals resulted in only minor changes in the esti- NORTH ATLANTIC ON AN AZIMUTHAL EQUAL AREA PROJECTION CENTERED AT 40° N AND 35° W Figure 2 Map of the North Atlantic showing the regions used to define fish populations for fisheries management. 766 Fishery Bulletin 95(4), 1997 Labrador / N.E. Newfoundland Kattegat Figure 3 Recruitment versus spawning stock biomass (SSB) for the three representa- tive cod populations. The solid line is the maximum likelihood estimate of the mean for Ricker spawner-recruitment functions under the assumption that the probability distribution for any SSB is given by a lognormal distribution. The dashed line is the median slope at the origin estimated from the six points with the lowest SSB. The straight dotted line is the replacement line with no fishing mortality. mated parameters (Myers et al., 1995b). Difficulties in estimating density-dependent model parameters are well known for insect and bird populations (Holyoak, 1993; Wolda and Dennis, 1993) and have been extensivly studied for exploited fish populations (Hilborn and Walters, 1992). The most important source of statistical bias for exploited fish populations is the nonindependence of spawners and recruitment, i.e. large recruitment usually leads to large spawner abundance (Walters, 1985). Bias in the parameter esti- mates of the spawner recruit function has been extensively studied with simulations for cod populations by Myers and Barrowman (1995) who found minimal bias in the estimates of the a parameter. The slope at the origin can be well estimated for cod populations, in spite of the large variability in re- cruitment, because each population has been reduced to very low levels by overexploitation (Myers et al., 1994). Furthermore, there is no evi- dence that mortality increases at low population size, i.e. depensation or the Allee effect, that would invalidate the assumption of our spawner re- cruitment model (Myers et al., 1995). The disadvantage of using the Ricker model, or any other paramet- ric spawner recruitment model, is that the slope at the origin is influ- enced by observations far from the origin. We investigated an alterna- tive approach: we regressed recruit- ment versus spawner biomass with only six observations with the low- est spawner biomass, forcing the re- gression line through the origin. This simple procedure should be reason- able because all the populations have been reduced to very low levels. ResuSts The Ricker model estimates of the slope at the ori- gin and of the population growth rate were estimated for the 20 spawner recruit data sets (Table 1; Fig. 3). The slope at the origin, a , did not vary enormously among populations (Fig. 4). There is one population, Irish Sea, for which the a is much larger; we believe that this large a is an overestimate (this will be dis- cussed later). It is evident that rm strongly covaries with tem- perature (Fig. 5; Table 2). It is also clear that this behavior does not arise from any dependence of a on temperature, because a is not correlated with temperature (Fig. 5; Table 2). There is a strong de- Myers et al.: Population growth rate of Gadus morhua 767 Table 1 Estimates of rate of population growth (rm), slope at the origin ( a) estimated from the Ricker model, age of maturity (a), bottom temperature, and NAFO/ICES management units for 20 cod populations in the North Atlantic. NAFO/ICES ID no. Stock location management unit rm a a Temperature 1 West Greenland 1 0.23 2.4 6 1.75 2 Labrador/N.E. Newfoundland 2J3KL 0.17 2.3 7 0.00 3 S. Grand Bank 3NO 0.27 3.5 6 1.75 4 N. Gulf of St. Lawrence 3Pn4RS 0.20 3.0 7 1.00 5 St. Pierre Bank 3Ps 0.31 4.6 6 2.50 6 S. Gulf of St. Lawrence 4TVn 0.15 1.9 7 1.75 7 E. Scotian Shelf 4VsW 0.36 9.8 6 3.75 8 S.W. Scotian Shelf 4X 0.36 2.5 3.5 6.75 9 Georges Bank 5Z 0.60 2.2 2 8.00 10 S.E. Baltic 22-24 0.74 8.4 3 7.00 11 Central Baltic 25-32 0.53 3.1 3 5.00 12 Celtic Sea Vllg.f 0.62 5.3 3 11.00 13 Faroe Plateau Vb 0.44 4.2 4 7.40 14 Iceland Va 0.24 4.3 7 5.80 15 Irish Sea Vila 1.03 23.1 3 10.00 16 Kattegat South Ilia 0.53 3.8 3 6.50 17 Barents Sea I 0.26 6.6 7.5 4.00 18 North Sea IV 0.56 9.0 4 8.60 19 Skagerrak North Ilia 0.82 11.2 3 6.50 20 West of Scotland Via 0.80 6.4 2.5 10.00 pendence of rm on age-at-maturity, but there is no corresponding significant relation between a and age-at-maturity (Fig. 6). Consistency demands that there be a relation between age-at-maturity and tem- perature, and, indeed, Table 2 and Figure 7 show that there is a significant correlation between these two variables. We repeated the above analysis with a calculated at the median slope of the six observations with the lowest spawner abundance. The estimates calculated with this robust procedure were generally compa- rable with those estimated from the Ricker model, although the Ricker values were generally higher (Fig. 8). The larger discrepancies in the two meth- ods occurred for the populations in which there were low estimates of recruitment at the largest popula- tion sizes, e.g. Irish Sea cod (Fig. 3). These points, although they are farthest from the origin, resulted in a higher estimate of the slope at the origin be- cause the Ricker model assumes a linear relation between egg-to-recruit mortality and SSB. We repeated the regression analysis of rate of popu- lation growth and slope at the origin ( a ) at mini- mum population size versus bottom temperature with the robust estimate of the slope, and found simi- Table 2 For each of the three variables rm(population growth rate), a (standardized slope of the spawner- recruit curve at the origin), and a (age-at-maturity), the estimated slope pa- rameter of the regression on temperature ( T ) (e.g. rm = a + bT) is presented, labeled b . For rm and a , there are two b values based respectively on the Ricker fit to each spawner- recruit data set and the median of the first six points of each spawner-recruit data set. Also shown are the signifi- cance levels of the regressions on temperature and the cor- responding r2. The results are presented for the Northwest, Northeast, and entire Atlantic. Variable Ricker b Ricker P(b= 0) Ricker r2 Median b Median P(6+0) Median r2 rm West 0.04 0.001 0.63 0.02 0.03 0.51 East 0.09 0.003 0.64 0.04 0.06 0.35 All 0.06 0.00008 0.59 0.04 0.00003 0.62 a West 0.32 0.5 0.06 -0.16 0.3 0.15 East 0.83 0.2 0.16 0.09 0.7 0.02 All 0.67 0.04 0.21 0.13 0.2 0.08 a West -0.62 0.00005 0.92 East -0.45 0.06 0.34 — — — All -0.47 0.00002 0.65 — — — 768 Fishery Bulletin 95(4), 1997 lar results to those using the Ricker model (Table 2; Fig. 9). We conclude that our results are robust in relation to the method used to estimate a . Discussion Perhaps remarkably, our study has revealed that the allometric (cross species) approximate inverse pro- portionality between rm and age-at-maturity holds within the species Atlantic cod. Despite the much narrower range of rm used in our single-species com- parison (in contrast to the wide variation in rm found in cross-species studies), the relation between rm and age-at-maturity prevails over other influences. The expectation that cod populations at the northern and southern extremes of their range should show higher resilience (rm ), because of greater susceptibility to environmental change (Myers, 1991), is not realized. In a similar vein, Roff (1984) suggested that early maturity in fish species may arise through r-selec- tion (in response to extreme environmental variabil- ity); our findings show that age-at-maturity appears to be chiefly explained by ambient temperature. Although we have found a clear relation between rm and temperature, this was not necessarily ex- pected a priori because mortality is positively related to temperature in comparative studies (Pauly, 1980). That temperature dependent (egg to adult) mortal- ity can offset the effect of temperature-dependent growth is emphasized by the temperature indepen- dence of a . More specifically, a depends on both fe- cundity and mortality. Fecundity, being growth dependent (Roff, 1984) increases with temperature; however, mortality also in- creases with temperature and counteracts the influence of temperature-dependent growth, leaving a temperature independent. (Note that many empirical studies of life histories of fish have found that somatic growth rate and survival covary [Beverton and Holt, 1959; Pauly, 1980; Myers and Doyle, 1983; Hut- chings, 1993]). The relation between rm and temperature is presumably a metabolic effect, in that fish growth is strongly influenced by temperature (Taylor, 1958; Pauly, 1980), and implies that a fish in a warm environment will reach the required size for maturity at an early age, which tends to increase r . Although Birch (1948) has noted, for insect populations, that higher rm values do not necessarily corre- spond to higher temperatures, investigators such as Hennemann (1983) and McNab (1980) have suggested that rm should be closely related to metabolic rate, which is closely linked to temperature. Certainly our presentation corroborates this proposed par- allel between metabolism and rm. The importance of the determination of rm has long been known (Lewontin, 1965); how- ever, it is certainly not always the case that age-at-maturity is the dominant factor. For species for which the production of replace- ment adults at low population density, a , is relatively low (e.g. for many mammals and birds), then changes in a or adult survival will have large effects on rm (Fig. 1). How- ever, for cod, and perhaps many fish, a is relatively large (e.g. around 4). In this case, the effects of reasonable changes in adult Figure 4 Estimates of slope at the origin ( a ) and approximate 95% confidence limits for 20 cod populations in the North Atlantic estimated from the Ricker model. Myers et a I.: Population growth rate of Gadus morhua 769 1.0 - 15 0.8 19 20 10 0.6 -i 9 12 11 16 18 13 0.4 - „ 5 7 8 3 4 1 6 17 14 0.2 - 2 1 1 1 1 T Temperature (°C) Figure 5 Rate of population growth (rm) and slope at the origin ( a ) versus bottom temperature. Populations are represented by numbers (See Table 1). 20 15 10 5 0 15 19 10 18 7 20 12 16 9 11 8 13 5 3 1 17 14 4 2 6 1 1 T 2 3 4 5 6 7 Age of maturity (yr) Figure 6 Rate of population growth (r ) and slope at the origin ( a ) versus age of maturity (a ). Populations are represented by numbers (See Table 1). survival (between 0.7 to 0.9) or a have a relatively small effect compared with age-at-maturity (Fig. 1). Our study has presented robust estimates (see Fig. 8) of rm for a variety of Atlantic cod populations, thus establishing recovery times for overfished popu- lations relieved from fishing pressure. At colder tem- peratures, rm is around 18% a year for all popula- tions, independent of how we calculate r . A major source of uncertainty in the SPA estimates of recruitment and SSB used in our analysis is that catches are assumed to be known without error. This assumption is particularly important when estimates of discarding and misreporting are not included in the catch-at-age data used in the SPA.These errors are clearly important for some periods of time for some of the cod stocks (Myers et al., 1997), and these errors will affect our estimates of the number of re- placements each spawner can produce at low popu- lation densities ( a ). However, we have shown that the estimates of rm are not very sensitive to reason- able changes in this parameter. We have carried our estimation of the model pa- rameters separately for each stock. An alternative approach is to analyze simultaneously all stocks in models that include separate estimation error for each stock and a parameter describing the variation among stocks. Myers et al.1 carried out such an analy- sis using variance components models for the data analyzed in this study and found that the optimal estimates of the variation in a was much less than that estimated, e.g. the very high estimate for Irish Sea cod was found to be overestimated. The parameters estimated in this study have man- agement implications that go beyond the estimation of population growth rate. In particular, the number of replacements each spawner can produce at low population densities ( a ), is critical for determining 1 Myers, R., G. Mertz, and N. Barrowman. 1996. Invariants of spawner-recruitment relationships for marine, anadromous, and freshwater species. ICES Council Meeting 1996 /D:ll, 17 p. 770 Fishery Bulletin 95(4), 1997 17 2 4 6 14 3 5 7 13 18 8 11 {§10 15 12 20 9 1 1 0 2 i l l 4 6 8 Temperature (°C) 1 10 Figure 7 Age at maturity (a) versus bottom temperature. Popula- tions are represented by numbers (See Table 1). the limits of exploitation (Cook et al., 1997; Myers and Mertz, in press). A comparative approach, such as the one used here, should help refine our under- standing of the rational exploitation of fisheries. We have used as simple a model as possible to es- timate r . This has the very great advantage that the we can compare the crucial demographic param- eters (i.e. the replacements each spawner can pro- duce at low population densities ( a ), age-at-matu- rity (a), and adult survival (ps) among populations) and can easily study the sensitivity of rm to errors in each of these parameters (Fig. 1). Furthermore, Hutchings and Myers (1994) found similar estimates of rm when they used a fully age-structured model for “northern” cod, i.e. cod in NAFO Division 2J3KL. Previous estimates of the recovery time for de- pleted cod stocks that have not included a detailed analysis of population growth rate have yielded widely different results. For example, it was initially projected that northern cod would recover close to historic levels of spawning biomass after a two-year fishing moratorium (Lear and Parsons, 1993). This projection has since been shown to be incorrect (Myers et al., 1996; Hutchings et al., 1997). Rough- garden and Smith (1996) estimated rm for northern cod to be =1 by simply taking the greatest difference between adjacent estimates of total population abun- dance from annual research vessel surveys. That is, they assumed that any change in the estimates rep- resented an increase in population abundance. However, this change in estimated abundance was almost entirely estimation error (Myers and Cadigan 1995, a and b) and had nothing to do with an in- crease in abundance. At present (March 1997) there are six Canadian cod populations (2 to 7 in Table 1) that are currently protected by a fishing morato- rium because the populations have been greatly re- duced by overfishing (Myers et al., 1997). Unfortu- nately, our results indicate that recovery could re- quire a long period. Under average environmental conditions, our results suggest a doubling time of about 4 years. Given the severe depletion of these populations (some are less than 5% of their maxi- mum observed levels (Myers et al., 1996) recovery to Myers et al.: Population growth rate of Gadus morhua 771 O) o Cl (/} 8 - 15 6 - 18 717 10 4 - 5 14 19 12 2 3 1 CO CD 20 2 - 4 6 9 8 0 - n r 0 2 4 6 8 10 Temperature (°C) Figure 9 Rate of population growth (rm) and slope at the origin (a) versus bottom temperature with estimated from the me- dian slope of the six observations with the lowest SSB. Populations are represented by numbers (See Table 1). desired levels of spawning biomass should not be expected for at least a decade of minimal mortality caused be fishing. Acknowledgments We thank the Northern Cod Science program for fi- nancial assistance. We thank N. Barrowman, K. Brander, J. Hutchings, J. Morgan, and S. Walde for helpful suggestions. Literature cited Beverton, R. J. H., and S. J. Holt. 1959. A review of the lifespans and mortality rates of fish in nature, and their relation to growth and other physi- ological characteristics. In G. Wolstenholme and M. O’Connor (eds.), CIBA Foundation Colloquia on Ageing 5, p. 142-174. Little, Brown and Co., Boston, 324 p. Birch, L. C. 1948. The intrinsic rate of increase of an insect popula- tions. J. Anim. Ecol. 17:15-26. Brander, K. 1994. Patterns of distribution, spawning, and growth in North Atlantic cod: the utility of inter-regional compari- sons. ICES Mar. Sci. Symp. 198:406-413. Charlesworth, B. 1994. Evolution in age-structured populations, second ed. Cambridge Univ. Press, Cambridge, 306 p. Charnov, E. L. 1993. Life history invariants: some explorations of symme- try in evolutionary ecology. Oxford Univ. Press, Oxford, 167 p. Clark, C. W. 1976. A delayed-recruitment model of population dynam- ics, with an application to baleen whale populations. J. Math. Biol. 3:381-391. Cole, L. C. 1954. The population consequences of life history phe- nomena. Q. Rev. Biol. 29:103-137. Cook, R. M., A. F. Sinclair, and G. Stefansson. 1997. Potential collapse of north sea cod stocks. Nature 358:521-522. deYoung, B., F. Perry, and R. Greatbach. 1994. Objective analysis of hydrographic data in the North- west Atlantic. Can. Data Rep. Hydrogr. Ocean Sci. 130: 1-93. Dickson, R. R., P. Malkki, G. Radach, R. Saetre, and M. P. Sissenwine (eds.). 1992. Hydrobiological variability in the ICES area, 1980- 1989. ICES Mar. Sci. Symp. 195, 514 p. Fenchel, T. 1974. Intrinsic rate of natural increase: the relationship with body size. Oecologia 14:317-326. Goodman, D. 1984. Statistics of reproductive rate estimates, and their implications for population projection. Rep. Int. Whaling Comm. Spec. Issue 6:161-173. Hennemann, W. W. 1983. Relationship among body mass, metabolic rate and the intrinsic rate of natural increase in mammals. Oecolo- gia 56:104-108. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dy- namics and uncertainty. Chapman and Hall, New York, NY, 570 p. Holyoak, M. 1993. New insights into testing for density dependence. Oecologia 93:435-444. Hutchings, J. A. 1993. Adaptive life histories effected by age-specific sur- vival and growth rate. Ecology 74:673-684. Hutchings, J. A., and R. A. Myers. 1994. What can be learned from the collapse of a renew- able resource? Atlantic cod, Gadus morhua , of Newfound- land and Labrador. Can. J. Fish. Aquat. Sci. 51:2126- 2146. Hutchings, J. A., C. J. Walters, and R. L. Haedrich. 1997. Is scientific inquiry compatible with government in- formation control? Can. J. Fish. Aquat. Sci. 54:1198-1210. Lear, W. H., and L. S. Parsons. 1993. History and management of the fishery for northern cod in NAFO divisions 2J, 3K and 3L. In L. Parsons and W. Lear (eds.), Perspectives on Canadian marine fisheries management, p. 55—89. Can. Bull. Fish. Aquat. Sci., vol. 226 772 Fishery Bulletin 95(4), 1 997 Lewontin, R. C. 1965. Selection for colonizing ability. In H. Baker and G. Stebbins (eds.), The genetics of colonizing species, pp. 77- 91. Acad. Press, New York, NY. MacArthur, R. H., and E. O. Wilson. 1967. The theory of island biogeography. Princeton Univ. Press, Princeton, NJ, 203 p. McNab, B. K. 1980. Food habits, energetics, and the population biology of mammals. Am. Nat. 116:106-124. Mertz, G., and R. A. Myers. 1996. An augmented dark model for stability of popula- tions. Math. Biosci. 131:157-171. Myers, R. A. 1991. Recruitment variability and range of three fish species. NAFO Sci. Counc. Stud. 16:21-24. Myers, R. A., and N. J. Barrowman. 1995. Time series bias in the estimation of density-depen- dent mortality in stock-recruitment models. Can. Tech. Rep. Fish. Aquat. Sci. 52:223-232. Myers, R. A., N. J. Barrowman, J. A. Hutchings, and A. Rosenberg. 1995a. Population dynamics of exploited fish stocks at low population levels. Science (Wash. D.C.) 269:1106-1108. Myers, R. A., J. Bridson, and N. Barrowman. 1995b. Summary of worldwide stock and recruitment data. Can. Tech. Rep. Fish. Aquat. Sci. 2024:iv + 327. Myers, R., and N. G. Cadigan. 1995a. The statistical analysis of catch-at-age data with correlated errors. Can. J. Fish. Aquat. Sci. 52:1265-1273. 1995b. Was an increase in natural mortality responsible for the collapse of northern cod? Can. J. Fish. Aquat. Sci. 52:1274-1285. Myers, R. A., and R. Doyle. 1983. Predicting natural mortality rates and reproduction- mortality trade-offs from fish life history data. Can. J. Fish. Aquat. Sci. 40:612-620. Myers, R. A., J. A. Hutchings, and N. J. Barrowman. 1996. The decline of cod in the North Atlantic. Mar. Ecol. Prog. Ser. 138:293-308. 1997. Why do fish stocks collapse? the example of cod in eastern Canada. Ecol. Appl. 7:91-106. Myers, R. A., and G. Mertz. In press. The limits of exploitation: a precautionary approach. Ecol. Appl. Myers, R. A., A. Rosenberg, P. Mace, N. J. Barrowman, and V. Restrepo. 1994. In search of thresholds for recruitment overfishing. ICES J. Mar. Sci. 51:191-205. Pauly, D. 1980. On the interrelationships between natural mortal- ity, growth parameters and mean environmental tempera- ture in 175 fish stocks. J. Cons. Cons. Int. Explor. Mer 39:175-193. Pimm, S. L. 1991. The balance of nature. Univ. Chicago Press, Chi- cago, IL, 434 p. Ricker, W. E. 1954. Stock and recruitment. J. Fish. Res. Board Can. 11:559-623. Roff, D. A. 1984 The evolution of life history parameters in teleosts. Can. J. Fish. Aquat. Sci. 41:989-1000. Roughgarden, J., and F. Smith. 1996. Why fisheries collapse and what to do about it. Proc. Natl. Acad. Sci. USA 93:5078-5083. Royama, T. 1993. Analytical population dynamics. Chapman and Hall, London, 371 p. Taylor, C. C. 1958. Cod growth and temperature. J. Cons. Cons. Int. Explor. Mer 23:366-370. Walters, C. J. 1985. Bias in the estimation of function relationships from time series data. Can. Tech. Rep. Fish. Aquat. Sci. 42: 147- 149. Wolda, H., and B. Dennis. 1993. Density dependence test, are they? Oecologia 95:581-591. 773 Abstract.-Sagittal otoliths were used to determine age and growth of 605 larval and juvenile Atlantic croaker, Micropogonias undulatus, collected in the Middle Atlantic Bight and estua- rine waters of Virginia. This study is the first to use age-based analysis for young Atlantic croaker collected in this region. A Laird-Gompertz model (^=0.95) was used to describe the growth of Atlantic croaker up to 65 mm standard length (SL) and 142 days (t): SL[t) = 2.657 exp 14.656 [1-exp (-0.008 It )]); where SL{t) = standard length at day t. Spatial and temporal patterns in the size and age of Atlantic croaker showed a pattern of in- shore immigration from offshore spawn- ing grounds, and faster early-season growth compared with late-season growth. Back-calculated hatching dates of Atlantic croaker collected in Virginia estuaries indicated a protracted spawn- ing period over 8 months, from early July 1987 to early February 1988, with at least 82% of spawning occurring from August to October. Regression analysis indicated that early-spawned larvae (July through August) grew more than 39% faster than late-spawned larvae (September through February). Lapillar and sagittal otoliths were compared with light microscopy; ages were under- estimated with lapillar otoliths, which were particularly inadequate in deter- mining the age of older juveniles. The relation between SL and sagittal otolith maximum diameter was best described by a fourth order polynomial (r2=0.99) and faster-growing Atlantic croaker had larger otoliths ( 12%) than the same size slower-growing fish. Manuscript accepted 12 March 1997. Fishery Bulletin 95:773-784 ( 1997). Age and growth of larval and juvenile Atlantic croaker, Micropogonias undulatus, from the Middle Atlantic Bight and estuarine waters of Virginia Stephen W. Nixon* Cynthia M. Jones Applied Marine Research Laboratory, Department of Biological Sciences Old Dominion University Norfolk, Virginia 23529 * Present address: Department of Zoology PO Box 761 7, North Carolina State University Raleigh, North Carolina 27695 E-mail address (for S W. Nixon):swnixon@umty.ncsu.edu Atlantic croaker, Micropogonias un- dulatus, range from New York to Florida and along the western and northern Gulf of Mexico (Ross, 1988; Atlantic States Marine Fisheries Commission, 1993). Historically, Atlantic croaker have ranked as one of the top five species in the com- mercial catch of finfishes in the middle Atlantic region (McHugh and Conover, 1986), although re- cruitment is highly variable in the species. In Virginia, annual com- mercial landings have varied by as much as threefold and have appar- ently declined overall since 1937 (Chesapeake Bay Program, 1988). Similarly, recreational landings have varied by as much as twofold over two years (U.S. Department of Commerce, 1991). In species with substantial an- nual recruitment variability, change in the survival rate of prerecruits (larvae and juveniles) is a key fac- tor in determining adult abundance (Houde, 1987). Determination of survival rates of prerecruits relies, in part, on daily age and growth information (Jones, 1992), and al- though otolith daily increment analysis has become common prac- tice (Jones, 1992), there are few published age and growth studies on the early life history of Atlantic croaker. Furthermore, there are no age-based estimates of growth for larval and juvenile Atlantic croaker for the Middle Atlantic Bight (MAB: shelf waters from Long Island, NY, to Cape Hatteras, NC) and estua- rine waters of Virginia. Comparable age-based studies on the early life history of Atlantic croaker have con- centrated on larvae collected in coastal waters south of Cape Hat- teras, North Carolina (Warlen, 1982), or the northern Gulf of Mexico (Co- wan, 1988). North Carolina larvae (Warlen, 1982), collected south of Cape Hatteras at Beaufort Inlet, show a twofold decline in length-at-age be- tween early- and late-season collec- tions. Likewise, Cowan (1988) shows a similar slow growth rate for late-season larvae collected in the northern Gulf of Mexico. Warlen’s (1982) back-calculated hatching dates indicate a broad spawning season from September to February, with peak spawning in October and November. On the basis of the pat- tern of progressive increase in mean size and age from the shelf to the estuary and on the basis of seasonal 774 Fishery Bulletin 95(4), 1997 variability in age of larvae entering the estuary, Warlen ( 1982) postulates two offshore spawning locations for Atlantic croaker entering Beaufort. Warlen’s conclusions imply potential differences in spawning source be- tween larvae entering Chesapeake Bay and some of those entering Beaufort Inlet. The purpose of the present study was to examine age and growth of larval and juvenile Atlantic croaker from the MAB and estuarine waters of Chesapeake Bay by using daily growth rings on otoliths. Specifically, we investigated the variability of size, size and age of entry into Chesa- peake Bay, calculated hatching-date distributions to estimate spawning periodicity, and estimated temporal and spatial differences in growth rates. In addition, we determined if there were significant differences in age counts between lapillar and sag- ittal otoliths and in size and age counts between left and right sagit- tal otoliths. Finally, we compared the relation between otolith growth and somatic growth for field-captured At- lantic croaker with results presented in the literature. Materials and methods Sampling regime Figure 1 Station locations in estuarine waters of Virginia for collection of larval and juvenile Atlantic croaker from 21 September 1987 to 30 March 1988. Larval Atlantic croaker were col- lected in the MAB (from Cape Hen- lopen, Delaware, to Cape Hatteras, North Carolina; Fig. 1) from 3 November to 14 No- vember 1987 from the shore to the 91-m (50-fm) con- tour with a stratified grid system illustrated in the MARMAP Plankton Survey Manual (Jossi and Marak, 1983). Seven additional stations at 2-km in- tervals were sampled along a transect across the mouth of the Chesapeake Bay. Larvae were sampled in oblique tows with a 60-cm bongo sampler contain- ing 505-prn mesh. Larval and juvenile Atlantic croaker sampled by Norcross and Hata ( 1990) were collected monthly from 29 September 1987 to 10 March 1988 at three inshore stations at Virginia’s Eastern Shore (Wachapreague, Sand Shoal, and Occohannock Channel) and at two stations at the mouth of the York River (Guinea and Tue Marshes; Fig. 1) with two 4.9-m otter trawls (one lined, one unlined) towed simultaneously. The lined net had a 6.4-mm mesh and a 3.2-mm mesh liner, and the un- lined net a 15.9-mm mesh. Additional larval and ju- venile Atlantic croaker were collected monthly from 21 September 1987 to 1 February 1988 at 8.1-km intervals along a 40. 2-km transect running from the mouth of the York River to the mouth of the Chesa- peake Bay (Fig. 1) with an otter trawl with a 9.1-m lined net containing 15.9-mm mesh and a 6.4-mm mesh liner. Finally, juvenile Atlantic croaker sampled by Dameron et al.1 were collected monthly from 25 January to 30 March 1988 in the channels of the York and James Rivers at 8. 1-km intervals from the mouth of the two rivers to 56.3 km upstream (Fig. 1) with Nixon and Jones: Age and growth of larval and juvenile Micropogonias undulatus 775 an otter trawl with a 9. 1-m lined net containing 15.9- mm mesh and a 6.4-mm mesh liner. Otter trawls were towed at 1.0 to 1.5 m/s over the bottom for five min- utes. Specimens were preserved in 70% ethanol im- mediately upon collection. The ethanol was changed within 24 hours and again after two days. Otolith processing and data analysis Standard length (SL) measurements were made on fish to the nearest 0.1 mm with an image analysis system for individuals <20 mm SL or with vernier calipers for individuals >20 mm SL. Sagittal and lapillar otoliths were extracted from at least 30 fish chosen at random from each station and sampling date (rc=605, 40 from the MAB and 565 from estua- rine waters). Otoliths were extracted from all indi- viduals when samples contained less than 30 fish. Only 40 of the 126 larvae collected in the MAB were available for age analysis owing to inadequate pres- ervation. Otolith maximum diameter (OMD) was measured on sagittae from rostrum to postrostrum to the nearest 0.1 mm with an image analysis sys- tem— 39 otoliths from larvae collected in the MAB and 143 otoliths from randomly selected estuarine larvae and juveniles were measured. The right sagittal otolith was used in age and growth analyses except when lost or damaged; then the left otolith was used. Procedures for the prepa- ration of otoliths that required sectioning and pol- ishing followed Epperly et al. (1991) — in short, otoliths were sectioned longitudinally, ground, and then polished to the primordia on both sides. Gener- ally, otoliths from fish <15 mm SL did not require grinding or polishing to distinguish daily increments; they were placed directly on glass slides and embed- ded in Euparal. Otoliths were read at l,000x, under cross-polarized, transmitted light on a monitor with an image analysis system. All specimens were aged without knowledge of fish size or collection date. Three independent age counts were averaged to de- termine final ages. Age counts were estimated by adding 5 days to the number of daily increments in the otoliths by assuming that increment deposition begins at 5 days posthatching as in spot (Leiostomus xanthurus', Peters et al.1 2 ). 1 Dameron, J. C., P. J. Geer, C. F. Bonzek, and H. M. Austin. 1994. Juvenile finfish and blue crab stock assessment pro- gram, bottom trawl survey. Annual data summary report se- ries vol. 1987. Special Scientific Report 124, Virginia Insti- tute of Marine Science, College of William and Mary, Gloucester Pt„ VA. 2 Peters, D. S., J. C. Devane Jr., M. T. Boyd, L. C. Clements, and A. B. Powell. 1978. Prebminary observations of feeding, growth, and energy budget of larval spot ( Leiostomus xanthurus). In Annu. Rep. NMFS, Beaufort, NC, p. 377-397. Paired /-tests were used to determine if there were significant differences in age estimates between lapillar and sagittal otoliths (n=32) and in size and age counts between left and right otoliths (rc=30). Also, the precision of sagittal age counts by the pri- mary reader and a secondary reader were compared (ti=50 ) with the indices of average percent error (Beamish and Fournier, 1981), coefficients of varia- tion, and index of precision (Chang, 1982). A paired /- test also was used to determine if there was a signifi- cant difference in mean age counts between readers. To generalize comparisons of mean growth rates and size-at-age of Atlantic croaker across capture sites, stations were grouped geographically into re- gions. These regions were designated as MAB, Chesa- peake Bay, seaside Eastern Shore (includes the Wachapreague and Sand Shoal Channel stations), bayside Eastern Shore (includes the Occohannock Channel station), marshes (includes the Tue and Guinea marsh stations), and rivers (includes the James and York river transects). The length and age of fish were compared among regions sampled with similar gears with independent, two-sample /-tests. Linear regression comparisons (Rawlings, 1988) were used to compare growth rates (slopes) and size at day 0 (y-intercepts) between early- (September through October) and late-captured (November through March) larvae <15 mm SL and <80 d. The analysis was restricted by size because larger, older juveniles were not available during early-season col- lections. Linear regression comparisons (Rawlings, 1988) were also used 1) to compare growth between early- (July through August) and late-season (Sep- tember through February) spawned larvae (<19 mm SL) and 2) to compare growth between early- and late-season spawned juveniles (19.1-65 mm SL). Early- and late-spawned larvae and juveniles were analyzed separately so that linear growth patterns could be described for the two life stages. We also used ANCOVA to compare mean size between early- and late-spawned juveniles. A Laird-Gompertz growth model (Laird et al., 1965) was used to describe the growth of Atlantic croaker larvae and juveniles <50 mm SL and <142 d: SLU) = SL(0)/expj[A(0)/a][l - exp (-a/)]); SL{t) = standard length at day /; SL(0) = assumed standard length at hatching (/=0); A(0) = specific growth rate at hatching (Z=0); and a = rate of exponential decay of the specific growth rate. The model was fitted by an iterative, nonlinear least- squares procedure. Age-specific growth rates were subsequently calculated as 776 Fishery Bulletin 95(4), 1997 A(t) = -A(o) exp (-at). Finally, ANCOVA was used to compare mean otolith size between early-captured, fast-growing Atlantic croaker with late-captured, slower-growing Atlantic croaker between 11 and 37 mm SL. Results Otolith analysis Age counts in sagittal and lapillar otoliths were sig- nificantly different (/-test, P=0.002). For older age fish, lapillar counts underestimated sagittal counts, and the disparity increased with increasing age (Fig. 2). Also, no differences were found among size (Gtest, P=0.49) and age counts (£-test, P= 0.85) between left and right sagittal otoliths (n=30). Sagittal age counts by the primary reader showed good precision — the average percent error (APE) of counts was 4.8%, with a coefficients of variation (CV) and index of precision (D) of 6.4% and 3.8%, respec- tively. For the second reader’s age counts these indi- ces were 8.4% (APE), 11.5% (CV), and 6.7% (D). Al- though the second reader’s age counts had relatively low precision, there was no significant difference in mean age counts between readers (Gtest, P=0.27). Size and age distributions Monthly length- and age-frequency distributions showed that size and age of fish generally increased from September to January (Fig. 3). Size and age appeared to decline in February (although sample sizes were small) and mostly represented fish col- lected in the rivers after January. Also, length dis- tributions were highly variable in comparison with respective age distributions (Fig. 3). For example, fish collected in November had two distinct length modes, whereas their age frequencies clearly had only one mode. This pattern was also evident for fish col- lected in October and January; thus size does not appear to be a good predictor of age in these fish. Generally, smaller and younger fish were found in the seaside Eastern Shore region compared with the bay- side Eastern Shore or marsh regions (Mests, P<0.001) over the entire sampling season. This pattern was evi- dent regardless of gear type. Significantly smaller (P=0.02) and younger (P<0.001) fish were found in the mainstem Chesapeake Bay compared with the rivers inland of the Bay. However, the rivers were sampled only during the later half of the sampling period. The age of larval Atlantic croaker entering Virginia nursery grounds was examined from specimens col- lected at the mouth of the Chesapeake Bay (the most seaward station along the Chesapeake Bay transect) and at Wachapreague and Sand Shoal Channels. The youngest larvae (24 d) entered the mouth of the Chesapeake Bay on 21 September 1987 and mea- sured 6.1 to 7.6 mm SL (n= 3). Fish collected at Wachapreague and Sand Shoal Channels were prob- ably better representatives of the age of larvae that enter Virginia nursery grounds because smaller mesh nets (with 3.2-mm mesh liner which sampled smaller larvae more effectively) were used at these stations. The youngest larvae observed at Wachapreague Channel were collected on 29 September 1987 with a mean size and age of 7.3 mm SL and 26 d, respec- tively, with the youngest individuals (20 d) measur- ing 5.4 and 6.1 mm SL (n= 2). The youngest larvae observed at Sand Shoal Channel were collected on 30 September 1987 with a mean size and age of 8.3 mm SL and 29 d, respectively, with the youngest in- dividuals (23 d) measuring 6.1 and 7.3 mm SL (n= 2). In conclusion, it appeared that the youngest larvae entered Virginian estuaries at an approximate age of 20 to 26 d, measuring 5.4 to 7.6 mm SL. Hatching-date distributions Hatching-date distributions indicated a protracted spawning period of 8 months from 5 July 1987 to 10 February 1988 and with 82% of spawning limited to Nixon and Jones: Age and growth of larval and juvenile Micropogonias undulatus 111 >. o c 0) C T 0) Sep 87 x =7.7(04) n = 1,620 r n Dec 87 x= 15 3(0 6) - 1 n = 103 | 90 80 70 60 \0f o L 50 40 30 20 10 0 50 p 40 - 30 - 20 - 10 - 0 50 p 40 - 30 - 20 10 0 x = 28 1 (17) n = 93 x = 44.6 (1.3) n= 161 x = 65 5 (1.9) n = 88 x = 70 1 (1 4) n = 63 x = 99 0(3 0) n = 85 x = 91 1 (3 4) n= 31 x = 83 5 (2 8) n = 44 20 160 Standard length (mm) Age (days) Figure 3 Monthly length- and age-frequency distributions for larval and juvenile Atlantic croaker collected from 21 September 1987 to 30 March 1988 in estuarine waters of Virginia. Also given are mean standard length and standard error in parentheses, ages and stan- dard error in parentheses, and sample size (n). August through October. Fish spawned earlier in the season are underrepresented in samples because they have experienced greater cumulative mortality than later spawned fish (Campana and Jones, 1992). Es- timates of size- and age-specific mortality are needed to predict hatching-date distributions more accu- rately, and these were not available. However, the result of greater accumulated mortality on early- spawned fish is to minimize the height of the esti- mated spawning peak. Hence, our results establish a lower bound of 82% of surviving juveniles spawned from August to October. 778 Fishery Bulletin 95(4), 1997 Growth Mean growth rates (mean SL divided by mean age) varied from 0.18 mm/d in the MAB to 0.41 mm/d in Chesapeake Bay (Table 1). Furthermore, early- spawned fish experienced considerably faster growth than late-spawned fish (Table 1). Early-captured larvae grew 37% faster than late- captured larvae for individuals <15 mm SL and <80 d (Fig. 4A). Early-captured individuals were larger at age than late-captured individuals because of differ- ent growth rates and not because of larger size at hatching, as indicated by tests of regression coeffi- cients that showed highly significant differences be- tween slopes (P<0.001) but not between intercepts (P=0.16). Early-spawned larvae grew 39% faster than late- spawned larvae (Fig. 4B), and once these size differ- ences were established, they persisted through the juvenile stage (Fig. 40. There was a significant dif- ference between slopes (PcO.OOl) for early- and late- spawned larvae, whereas, early- and late-spawned juveniles experienced similar growth according to tests of regression coefficients (P=0.75). Furthermore, Tabie 1 Mean standard length (mm) and standard error (SE), age (d) (SE), and growth rate (mean SL/mean age) for larval and juvenile Atlantic croaker collected in the Middle Atlantic Bight and estuarine waters of Virginia from 21 September 1987 to 10 March 1988 by hatching month, region, and gear type. Hatching month Sample size Mean standard length ± SE Mean age ± SE Mean growth Middle Atlantic Bight (MAB)i Sep 19 7.9 ±0.2 43.7 ± 0.9 0.18 Oct 21 5.8 ± 0.4 29.7 ± 1.5 0.20 Seaside Eastern Shore (SES: Wachapregue and Sand Shoal Channel)2 Aug 17 13.2 ± 1.5 41.0 ± 3.8 0.32 Sep 115 11.0 ±0.5 42.7 ± 1.8 0.26 Oct 6 13.6 ± 0.8 56.5 ± 1.1 0.24 Bayside Eastern Shore (BES: Occohannock Channel)2 Aug 1 28.3 ± 0.0 94.0 ± 0.0 0.30 Sep 32 14.4 ± 0.7 67.1 ± 1.2 0.21 Oct 15 11.0 ± 0.2 57.5 ± 0.7 0.19 Chesapeake Bay3 Jul 7 25.8 ± 6.5 77.3 ± 11.2 0.33 Aug 74 28.9 ±2.0 70.5 ± 3.8 0.41 Sep 21 27.8 ± 3.4 78.4 ± 8.0 0.36 Oct 21 35.4 ± 1.7 94.2 ± 2.7 0.38 Nov 7 20.4 ± 1.3 71.6 ± 1.0 0.29 Marshes (Guinea and Tue Marshes)2 Jul 10 24.0 ± 2.4 71.2 ±3.0 0.38 Aug 39 16.5 ± 1.2 48.6 ± 2.2 0.34 Sep 75 9.8 ±0.3 47.0 ± 2.2 0.21 Oct 4 10.6 ±0.6 60.0 ± 1.0 0.18 Jan 7 12.7 ±0.4 54.7 ± 2.0 0.23 Rivers (York and James River)3 Sep 13 48.5 ± 2.2 124.3 ± 1.7 0.39 Oct 25 32.1 ± 2.1 105.3 ± 2.2 0.30 Nov 28 27.6 ± 2.0 91.5 ± 2.8 0.30 Dec 32 31.8 ±2.2 89.5 ± 2.9 0.36 Jan 15 24.4 ± 1.7 74.4 ± 1.4 0.33 Feb 1 10.6 ± 0.0 49.0 ± 0.0 0.22 1 Gear type used was oblique 60-cm bongo nets with 505-gm mesh. 2 Gear type used was a 4.9-m lined trawl net with a 6.4-mm mesh and 3.2-mm mesh liner and a 4.9-m unlined net with a 15.9-mm mesh. The lined and unlined nets were towed simultaneously. 3 Gear type used was a 9.1-m lined trawl net with a 15.9-mm mesh and 6.4-mm mesh liner. Nixon and Jones: Age and growth of larval and juvenile Micropogonias undulatus 779 early-spawned juveniles were significantly larger (18%) than late-spawned juveniles when mean size was adjusted by age (ANCOVA, PcO.OOl, Table 2B). A Laird-Gompertz growth model fitted the entire range of Virginia data well (r2=0.95), al- though variance in size increased with age (Fig. 5). Standard length at hatching (SL(0)) es- timated from the Laird-Gompertz growth model (Fig. 5) was 2.7 ±0.3 mm SL and was similar to size-at-hatching estimates for laboratory- spawned Atlantic croaker from the Chesapeake Bay (2.0 mm SL; Middaugh and Yoakum, 1974) and North Carolina (2.4 mm SL; Jones3 ), but considerably higher than Warlen’s (1982) esti- mate of 0.9 mm SL for wild-captured Atlantic croaker larvae from North Carolina. The rate of exponential decay of the specific growth rate (a) was estimated at 0.0081 ± 0.0012 (Fig. 6). Changes in age-specific growth (At, a function of the rate of exponential decay of specific growth in time) indicated that larvae experi- enced a decline of daily growth rate from 3.2% at day 20 to 2.3% by day 60 (Fig. 6). Standard length and otolith maximum diameter (OMD) relation The relation between sagittal OMD and SL was best described by a fourth order polynomial (Fig. 7). A simple linear model also described the OMD and SL relation fairly well (SL = 13.5 (OMD) + 4.2, r2=0.98); however, there were strong patterns in the residuals. Otolith growth was similar between early-captured (fast grow- ers captured from September to October) and late-captured (slow growers captured from No- vember to March) groups when compared by slope (P= 0.50). However, a significant difference was observed between the two groups when otolith size was adjusted for fish size (ANCOVA, P<0.001, Table 3A). Size-adjusted means indi- cated that otoliths from the early-captured group were almost 13% larger than otoliths from the late-captured group (Table 3B). Plots of in- dividual otoliths showed very little overlap be- tween groups (Fig. 8). Discussion Otolith analysis We were unable to obtain known-age Atlantic croaker to validate the assumption that incre- Larvae Juveniles 40 60 80 100 120 140 160 Age (days) Figure 4 (A) Growth comparison between early- (September through Oc- tober, r2= 0.78, n=199) and late-captured (November through March, r2= 0.77, n=132) Atlantic croaker <15 mm standard length (SL) and <80 d. (B) Growth comparison between early- (July through August, r2=0.92, n=77) and late-spawned (September through February, r2=0.84, n =314) Atlantic croaker up to 19 mm SL. (C) Growth comparison between early- (r2=0.92, n= 71) and late-spawned (r2=0.86, n = 143) Atlantic croaker from 19.1 to 65 mm SL and from 51 to 142 d. 3 Jones, C. J. 1995. Applied Marine Research Laboratory, Old Dominion University, Norfolk, VA 23529. Unpubl. data. 780 Fishery Bulletin 95(4), 1 997 ments form daily in larvae. However, daily growth increments have been validated in the otoliths of lar- val spot, Leiostomus xanthurus, a sciaenid relative of Atlantic croaker (Peters et al.2), and we assumed, therefore, that increment deposition is daily in At- lantic croaker. Peters et al.2 also demonstrated that the first daily increment forms in the sagittae of spot at the time of first feeding, which occurs in spot about 5 d after hatching at 18 and 20°C (Powell and Chester, 1985); thus, we added 5 days to our increment counts. Sagittal otoliths of Atlantic croaker begin to in- crease growth along their anterior and posterior axes during the late-larval stage, whereas lapilli main- tain concentric growth through the juvenile stage, potentially making lapilli preferable when using otoliths to backcalculate growth. In examining the potential for using lapillar counts as a surrogate to sagittal counts, we found under light microscopy that lapillar counts increasingly underestimated sagittal counts as fish increased in age. The microstructure in sagittal otoliths also had better defined incre- ments, leading us to assume that sagittal counts were more accurate predictors of age than lapillar counts. No differences in the size and age counts between left and right sagittal otoliths warranted the replacement of lost or damaged right otoliths with left otoliths. Size and distribution Offshore spawning and subsequent estuarine migra- tion of Atlantic croaker have been thoroughly docu- mented by studies with egg and larval size distribu- tions (Hildebrand and Cable, 1930; Wallace, 1940; Haven, 1957; Colton et al., 1979; Morse, 1980; Lewis Table 2 Growth comparison between early- (July through August) and late-spawned (September through February) Atlantic croaker from 19.1 to 65 mm SL and from 51 to 142 d with an analysis of covariance (ANCOVA) of standard length (SL) of fish (mm), with age (d) as the covariate. Size ad- justed means equals the mean SL of fish, adjusted for the effects of age. ANCOVA A df F-value P-value Covariate age Main effect Early-spawned vs. 1 743.6 P<0.001 late-spawned Residual sums 1 89.9 P<0.001 of squares (d) r2 5,489.2 (211) 0.78 B Size adjusted means (mm) (SE) Early-spawned: 39.5 (0.6) Late-spawned: 32.3 (0.4) Nixon and Jones: Age and growth of larval and juvenile Micropogonias undulatus 781 Table 3 Otolith comparison between faster-growing early-captured (September through October) Atlantic croaker and slower- growing late-captured (November though March) Atlantic croaker between 11 and 37 mm standard length (SL) with an analysis of covariance (ANCOVA) of the otolith maxi- mum diameter (OMD) of sagittae (mm), with SL of fish (mm) as the covariate. Size-adjusted means equals the mean OMD of sagittae, adjusted for the effects of SL of fish. ANCOVA A df F-value P-value Covariate standard length 1 1,126.9 P<0.001 Main effect Early-captured vs. late-captured 1 36.2 PcO.001 Residual sums of squares (d) 1.16 (66) r2 0.95 B Size adjusted means (mm) (SE) Early-captured: 1.54 (0.02) Late-captured: 1.25 (0.02) and Judy, 1983; Warlen and Burke, 1990). However, only two studies (Warlen, 1982; this study) used daily age information from otoliths to support such find- ings. Daily age information is critical because size- at-age is highly variable in this species, and age- based data provide reliable confirmation of cross- shelf transport of larvae. Warlen (1982) found a gen- eral increase in the age of fish entering the Beaufort estuary as the season progressed, and suggested this increase in age was an effect of variable transport distance or rates to the estuary (or both). Seasonal trends observed in this study may be attributed to similar processes. Mean ages generally increased over time, lagging about 10 d between monthly sampling dates, sug- gesting constant recruitment over the entire sam- pling season. However, mean ages in the rivers de- clined after December. Our samples collected in Janu- ary along river transects show a gradient of smaller, younger individuals upstream and of larger, older individuals downstream. Bottom waters in the York River experience a winter temperature gradient, with the lowest temperatures occurring in upper reaches of the river (Chao and Musick, 1977; Dameron et al.1; Land et al.4) and the higher temperatures in the Bay mainstem. This bottom water temperature gradient coupled with the increase in size of Atlantic croaker 4 Land, M. F., P. J. Geer, C. F. Bonzek, and H. M. Austin. 1994. Juvenile finfish and blue crab stock assessment program. Bot- 3.0 r 0 0 1 1 1 1 1 1 1 10 15 20 25 30 35 40 Standard length (mm) Figure 8 The relation between otolith maximum diameter (OMD) (mm) and standard length (SL) of fish (mm) illustrating otolith and somatic growth relation between faster-grow- ing, early-captured (September through October) Atlantic croaker (n=36) and slower-growing, late-captured (Novem- ber through March) Atlantic croaker (n=32) between 11 and 37 mm SL. tom trawl survey. Annual data summary report series. Volume 1988. Spec. Sci. Rep. Va. Inst. Mar. Sci., College of William and Mary, Gloucester Point, VA, 171 p. 782 Fishery Bulletin 95(4), 1997 downstream may indicate movement of larger, older individuals into deeper, warmer waters of the mainstem Chesapeake Bay. Atlantic croaker’s sensi- tivity to low temperatures (Massmann and Pacheco, 1960; Joseph, 1972) may further explain the move- ment of older and larger fish from the rivers into warmer Bay waters. Hatching-date distributions Our estimate of a protracted spawning season from early July 1987 to February 1998, with peak spawn- ing in September, is similar to earlier reports in stud- ies that used the presence of eggs or early larvae to estimate spawning. These studies suggest a pro- tracted spawning period from August through De- cember with peak spawning from August to October (Wallace, 1940; White and Chittenden, 1977; Johnson, 1978; Colton et al., 1979; Morse, 1980; Chittenden et al.5 ). Although our observation that spawning may begin as early as July has not been reported elsewhere, ovaries containing postovulatory follicles recently have been observed in Atlantic croaker from the Chesapeake Bay as early as July (Barbieri et al., 1994). Furthermore, because sexu- ally mature adults do not begin to migrate out of the Chesapeake Bay until early July, and mainly in Au- gust and September (Wallace, 1940), limited spawn- ing of Atlantic croaker may occur in proximal coastal waters as suggested by Haven (1957). Growth When comparing temporal patterns in growth, it is best to analyze differences between groups by spawn- ing date, rather than by capture date; otherwise older fish are under represented because of their greater accumulated mortality (Campana and Jones, 1992). Temporal growth variability was observed when the data were analyzed both by capture date and spawn- ing date, although a 6% greater difference was ob- served when the data were analyzed by spawning date. Seasonal variability in the growth of larval and juvenile Atlantic croaker may result from higher water temperatures or increased food in July and August (or both) (Alden et al.6) or from improved 5 Chittenden, M. E., C. M. Jones, L. R. Barbieri, S. J. Bobko, and D. E. Kline. 1990. Initial information on the Atlantic croaker, a final report on development of age determination methods, life history-population dynamics information, and evaluation of growth overfishing potential for important recreational fishes. Final Rep. to Virginia Mar. Res. Comm. VMRC I, Va. Inst. Mar. Sci., College of William and Mary, Gloucester Point, VA, 88 p. survival of larger, faster-growing fish (Miller et al., 1988; Isley and Grimes, 1996). Hatching dates of faster-growing, early-spawned fish coincided with peak mean surface water temperatures in July and August (26.3° and 26.7°C, respectively; U.S. Depart- ment of Commerce7) and with peak plankton abun- dances in the Chesapeake Bay which typically occur in July (Alden et al.6). Furthermore, hatching dates of slower-growing fish coincided with increased patchiness and falling plankton abundances which typically begin in September and October (Alden et al.6). Warlen (1982) reported similar seasonal growth patterns for North Carolina Atlantic croaker larvae and speculated that slow growth observed in late- captured fish (mid- January to mid- April) might be attributed to colder ocean temperatures and low food availability in mid- to late-winter, or less likely, to smaller egg size of late-spawned larvae. Recently immigrated larvae collected in Virginia estuaries in this study were larger at age than North Carolina larvae. Monthly mean growth rates (mean SL/mean age) of estuarine collected larvae (26-65 d) in this study ranged from 0.26 to 0.40 mm/d and were considerably higher than weekly mean growth rates (0.16-0.27) for similar age (32-64 d) North Carolina larvae collected in estuarine waters (see Warlen 1982, Table 1). Because this study and Warlen’s (1982) were conducted in different years, we cannot eliminate the real possibility that these differences may be tempo- ral, year-to-year changes. Further inter-year studies within Virginia and North Carolina, showing consis- tent patterns of growth variability, are needed to con- clude that there are regional growth differences. How- ever, whether spatial or temporal, or a combination of both, within-season patterns among the two studies are similar, whereas growth rates themselves differ. Apparently, Atlantic croaker larvae immigrating into estuaries of Virginia and North Carolina can be categorized as early-spawned, fast growers or as late- spawned, slow growers. These seasonal growth dif- ferences, coupled with a spatially and temporally extended spawning season suggest that Atlantic croaker encounter variable environmental factors that may affect their survival. Identifying factors that may enhance survival or affect mortality rates of these spatially and temporal explicit groups are of major interest and are worthy of further study. 6 Alden, R. W., Ill, R. S. Birdsong, D. M. Dauer, H. G. Marshall, and R. M. Ewing. 1992. Virginia Chesapeake Bay water qual- ity and living resources monitoring programs: executive report, 1985-89. Applied Marine Research Laboratory, Old Domin- ion University, Norfolk, VA 23529, Report 849, 33 p. 7 U.S. Department of Commerce, National Ocean Service, NOAA, Ocean and Lake Level Division Database, Rockville, MD 20874, June 1992. Nixon and Jones: Age and growth of larval and juvenile Micropogonias undulatus 783 Standard length and otolith maximum diameter (OMDJ relation We found that in wild-caught larval Atlantic croaker, faster-growing individuals have larger otoliths than similar-size slower-growing individuals. Our results differ from results found for laboratory-reared gup- pies ( Poecilia reticulata) ( Reznick et al., 1989), pond- reared striped bass ( Morone saxatilis) (Secor and Dean, 1989), red seabream ( Pagrus major), and spot ( Leiostomus xanthurus) (Secor et al., 1989). In these studies, where food ration was controlled, slower- growing individuals had larger otoliths than simi- lar-size faster-growing individuals. However, in Arc- tic char ( Salvelinus alpinus), otolith growth rate has been found to continue increasing while somatic growth remains constant when exposed to tempera- tures above 13°C (Mosegaard et al., 1988). In our study, the early-captured, faster-growing Atlantic croaker, may have experienced otolith growth that exceeded their maximum somatic growth rate. Unfortunately, there are no quantitative data that can be tested to determine what factors influenced the growth of Atlantic croaker in our samples and what were the subsequent effects on the otolith growth-somatic growth relation. The underlying is- sue, however, is to examine how the fish and otolith- size relation is affected by temperature responses of somatic growth rate at particular food levels (Mosegaard et al., 1988) and to determine its signifi- cance when backcalculating growth from increment widths (Campana and Jones, 1992). Acknowledgments We thank Hassan Lakkis for statistical advise and guidance, and Joseph E. Hightower, Allyn Powell, and the late Ray S. Birdsong for critical reviews of this manuscript. Additional thanks are given to all those at the Applied Marine Research Laboratory at Old Dominion University who assisted on this project. We owe a special debt of gratitude to Brenda L. Norcross for working with us and for providing the Atlantic croaker samples used in this study, which were obtained from projects funded by the National Sea Grant Program (sub-contract # 5-29532) to the Virginia Institute of Marine Science (VIMS), and by the National Marine Fisheries Services, Northeast Fisheries Center to B. Norcross (NA85-EAH 00026) and to VIMS by the Virginia Council on the Envi- ronment. Other specimens used were collected in part of an expansion of the VIMS river trawl survey. This study was funded by Virginia Sea Grant in 1990 and 1991 to C. Jones and B. Norcross (grant R/CF- 25VGMSC) and is based on a thesis submitted by the senior author in partial fulfillment of the require- ment for the M.S. degree, Department of Biology, Old Dominion University, Norfolk, VA 23519. Literature cited Atlantic States Marine Fisheries Commission. 1993. Proceedings of a workshop on spot ( Leiostomus xan- thurus) and Atlantic croaker ( Micropogonias undulatus). Spec. Rep. 25, Dec. 1993, 160 p. Barbieri, L. R., M. E. Chittenden Jr., and S. K. Lowerre-Barbieri. 1994. Maturity, spawning, and ovarian cycle of Atlantic croaker, Micropogonias undulatus, in the Chesapeake Bay and adjacent waters. Fish. Bull. 92:671-685. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determination. Can. J. Fish. Aquat. Sci. 38:982-983. Campana, S. E., and C. M. Jones. 1992. Analysis of otolith microstructure data. In D. K. Stevenson and S. E. Campana (eds.), Otolith microstruc- ture examination and analysis, p. 73-100. Can. Spec. Publ. Fish Aquat. Sci. 117. Chang, W. B. 1982. A statistical method for evaluating the reproducibil- ity of age determinations. Can. J. Fish. Aquat. Sci. 39:1200-1210. Chao, L. N., and J. A. Musick. 1977. Life history, feeding habits, and functional morphol- ogy of juvenile sciaenid fishes in the York River estuary, Virginia. Fish. Bull. 75:657-702. Chesapeake Bay Program. 1988. Chesapeake Bay Stock Assessment Plan. Agr. Comm. Rep., Jun. 1988, 66 p. Colton, J. B., Jr., W. G. Smith, A. W. Kendall Jr., P. L. Berrien and M. P. Fahay. 1979. Principal spawning areas and times of marine fishes, Cape Sable to Cape Hatteras. Fish. Bull. 76:911-915. Cowan, J. H. 1988. Age and growth of Atlantic croaker, Micropogonias undulatus, larvae collected in the coastal waters of the northern Gulf of Mexico as determined by increments in saccular otoliths. Bull. Mar. Sci. 42:349-357. Epperly, S. P., D. W. Ahrenholz, and P. A. Tester. 1991. A universal method for preparing, sectioning and polishing fish otoliths for daily aging. U.S. Dep. Commer., NO AA Tech. Memo. NMFS-SEFSC-283, 15 p. Haven, D. S. 1957. Distribution, growth and availability of juvenile croaker, Micropogonias undulatus, in Virginia. Ecology 38:88—97. Hilderbraed, S. F., and L. E. Cable. 1930. Development and life history of fourteen teleostean fishes at Beaufort, North Carolina. Bull. U.S. Bur. Fish. 46:383-488. Houde, E. D. 1987. Fish early life dynamics and recruitment vari- ability. In R. D. Hoyt (ed.), 10th Annual Larval Fish Con- ference, p. 17-29. Am. Fish. Soc. Symp. 2. Isley, J. J., and C. B. Grimes. 1996. Influence of size-selective mortality on growth of Gulf menhaden and king mackerel larvae. Trans. Am. Fish. Soc. 125:741-752. Johnson, G. D. 1978. Micropogonias undulatus (Linnaeus), Atlantic croaker. In Development of fishes of the Mid-Atlantic Bight: an at- 784 Fishery Bulletin 95(4), 1 997 las of egg, larval and juvenile stages, vol. IV, Carangidae through Ephippidae, p. 227-233. U.S. Fish Wildl. Serv. FWS/OBS-78/12. Jones, C. M. 1992. Development and application of the otolith increment technique. In D. K. Stevenson and S. E. Campana (eds.), Otolith microstructure examination and analysis, p. 1- 11. Can. Spec. Publ. Fish. Aquat. Sci. 117. Joseph, E. B. 1972. The status of the sciaenid stocks of the middle At- lantic coast. Chesapeake Sci. 13:87-100. Jossi, J. W., and R. R. Marak. 1983. MARMAP plankton survey manual. U.S. Dep. Commer., NOAATech. Memo. NMFS-F/NEC-21, 258 p. Laird, A. K., S. A. Tyler, and A. D. Barton. 1965. Dynamics of normal growth. Growth 29:233-248. Lewis, R. M., and M. H. Judy. 1983. The occurrence of spot, Leiostomus xanthurus , and Atlantic croaker, Micropogonias undulatus, larvae in Onslow Bay and Newport River estuary, North Caro- lina. Fish. Bull. 81:405-412. Massmann, W. H., and A. L. Pacheco. 1960. Disappearance of young Atlantic croaker from the York river, Virginia. Trans. Am. Fish. Soc. 89:154-159. McHugh, J. L., and D. O. Conover. 1986. History and condition of food finfisheries in the middle Atlantic region compared with other sections of the coast. Fisheries 11:8-13. Middaugh, D. P., and R. L. Yoakum. 1974. The use of chorionic gonadotropin to induce labora- tory spawning of the Atlantic croaker, Micropogonias undu- latus, with notes on subsequent embryonic development. Chesapeake Sci. 15:110-114. Miller, T. J., L. B. Crowder, J. A. Rice, and E. A. Marschall. 1988. Larval size and recruitment mechanisms in fishes: toward a conceptual framework. Can J. Fish. Aquat. Sci. 45: 1957-1670. Morse, W. W. 1980. Maturity, spawning and fecundity of Atlantic croaker, Micropogonias undulatus, occurring north of Cape Hatteras, North Carolina. Fish. Bull. 78:190-195. Mosegaard, H., H. Svedang, and K. Taberman. 1988. Uncoupling of somatic and otolith growth rates in Arctic char (Salvelinus alpinus) as an effect of differences in temperature response. Can. J. Fish. Aquat. Sci. 45: 514-1524. Norcross, B. L., and D. Hata. 1990. Seasonal composition of finfish in waters behind the Virginia Barrier Islands. Va. J. Sci. 41(4A): 441-461. Powell, A. B., and A. J. Chester. 1985. Morphometric indices of nutritional condition and sensitivity to starvation of spot larvae. Trans. Am. Fish. Soc. 114:338-347. Rawlings, J. O. 1988. Applied regression analysis: a research tool. Wads- worth, Inc., Belmount, CA, 553 p. Reznick, D., E. Lindbeck, and H. Bryga. 1989. Slower growth results in larger otoliths: an experi- mental test with guppies (Poecilia reticulata). Can. J. Aquat. Sci. 46:108-112. Ross, S. W. 1988. Age, growth, and mortality of Atlantic croaker in North Carolina, with comments on population dynamics. Trans. Am. Fish. Soc. 117:461-473. 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 saxatilis. Can. J. Fish. Aquat. Sci. 46:113-121. Secor, D. H., J. M. Dean, and R. B. Baldevarona. 1989. Comparison of otolith growth and somatic growth in larval and juvenile fishes based on otolith length/fish length relationships. Rapp. R-V. Reun. Int. Explor. Mer 191:431— 438. U.S. Department of Commerce. 1991. Marine recreational fishery statistic survey, Atlan- tic and Gulf coasts, 1987-1989. U.S. Dep. Commer., NOAA, Natl. Mar. Fish. Serv. Curr. Fish. Stat. 8904, 363 p. Wallace, D. H. 1940. Sexual development of the croaker, Micropogonias undulatus, and distribution of the early stages in Chesa- peake Bay. Trans. Am. Fish. Soc. 70: 475-482. Warlen, S. M. 1982. Age and growth of larvae and spawning time of At- lantic croaker larvae in North Carolina. Proc. Annu. Conf. SE Assoc. Fish Wildl. Agency 34:204-214. Warlen, S. M., and J. S. Burke. 1990. Immigration of fallAvinter spawning marine fishes into a North Carolina estuary. Estuaries 13:453-461. White, M. L., and M. E. Chittenden. 1977. Age determination, reproduction, and population dynamics of the Atlantic croaker, Micropogonias undu- latus. Fish. Bull. 75:109-123. 785 Abstract .—Three species of the stromateoid genus Peprilus have been found to occur in the northwest Atlan- tic: P. triacanthus (butterfish), P. burti (gulf butterfish), and P. alepidotus (harvestfish). Peprilus triacanthus and P. alepidotus reportedly spawn from May through August and June through July, respectively. Peprilus burti spawns twice yearly: February through May and September through November. Collec- tions of larvae and juveniles of Peprilus spp. from the northern South Atlantic (SAB) and Mid- Atlantic (MAB) Bights during both the spring and summer of 1988 and 1989 suggest that either a combination of species was spawning or that reported spawning dates were suspect. Species identification of Peprilus in these collections was deter- mined with morphometric, meristic, and pigment character analyses. Speci- mens sampled had counts for caudal vertebrae (18-19) and ventral midline melanophores (11-16) consistent with those found for P. triacanthus in previ- ous studies. By analyzing otoliths, we estimated larval and juvenile growth rates to be approximately 0.23 mm/day. Backcalculation of hatch dates suggests either two spawning events for P. triacanthus , February through mid- April and mid-May through late July, or one extended spawning period begin- ning in late February and ending in late July. This study reveals that P. tria- canthus spawns for a much longer period than previously thought. It is possible that P. triacanthus spawns during the spring in the SAB and summer in the MAB as a strategy to extend the dura- tion of its spawning period. This strat- egy is one used by other north-south migrating species and warrants further study. Manuscript accepted 7 March 1997. Fishery Bulletin 95:785-799 (1997). Temporal and spatial spawning patterns of the Atlantic butterfish, Peprilus triacanthus, in the South and Middle Atlantic Bights* Teresa Rotunno Robert K. Cowen Marine Sciences Research Center State University of New York Stony Brook, New York 1 1 794-5000 E-mail address (for T Rotunno): trotunno@ccmaifsunysb.edu Three species of the stromateoid genus Peprilus (order: Perciformes) are reported to occur in the west- ern North Atlantic: butterfish, P triacanthus (Peck), gulf butterfish, P burti (Fowler), and harvestfish, P. alepidotus (Linnaeus). Although the ranges of all three species ex- tend along the eastern coast of North America and the West Indies, there is some question as to which species of Peprilus are regular resi- dents in the South Atlantic Bight (SAB) and Mid-Atlantic Bight (MAB) regions (Caldwell, 1961; Haedrich, 1967; Horn, 1970; Persch- bacher et al., 1979). This question is extended by the suggestion of possible hybridization between P. triacanthus and P. burti in the northern portion of the SAB (Horn, 1970; Perschbacher et al., 1979). Reproduction in P. triacanthus and P burti is seasonal and appar- ently associated with annual migra- tion patterns (Horn, 1970). In the summer and fall, P. triacanthus migrates northward and inshore, where it reportedly spawns from late May through August, with a peak in June (Horn, 1970). During winter, P triacanthus migrates off- shore and becomes horizontally re- stricted (Horn, 1970). Movement by P burti is somewhat opposite to that ofP. triacanthus. Peprilus burti mi- grates offshore during late spring through early fall, then onshore to- wards shallow bays and inlets dur- ing the winter and early spring (Horn, 1970). Spawning by P. burti is reported to occur during two dis- tinct periods, February through May and September through November (Murphy, 1981). Unlike these other two species, P. alepidotus does not exhibit seasonal migration patterns, remaining in shallow waters throughout the year where it spawns during June and July (Horn, 1970). During the spring and summer seasons of 1988 and 1989, we con- sistently collected larval and juve- nile Peprilus from the South Atlan- tic and Mid-Atlantic Bights. The combination of their location and dates of capture raised the question as to which species were present in our samples. Although most abun- dant within the MAB region, P. triacanthus is reported to spawn during the summer only (Horn, 1970). According to time of capture, the spring-collected larvae in our SAB samples should have been P burti because they were collected be- fore P triacanthus and P. alepidotus supposedly begin to spawn (Horn, 1970; Murphy, 1981). However, their occurrence within the South Atlantic and Mid-Atlantic Bights suggests that they were either P. alepidotus or P. triacanthus. Thus, the overall aim of this study was to identify the spe- cies in our samples and to back- * Contribution 1061 of the Marine Sciences Research Center, State University of New York, Stony Brook, New York 11794-5000. 786 Fishery Bulletin 95(4), 1997 Figure 1 Study area from Cape Hatteras, North Carolina, to Long Island, New York. Dotted and barred areas represent spring and summer sampling areas, respectively. calculate hatching dates by using validated otolith-increment analysis. Materiafs and methods Collections Peprilus larvae and juveniles were collected during April and May of 1988 and April of 1989 in the northern SAB, offshore of Cape Hatteras, North Carolina, and June-August of 1988 and 1989 in the MAB, from Long Is- land, New York, to Cape May, New Jersey (Fig. 1; Table 1). Fish were collected with a 1-m2 Tucker trawl and a 5-m2 Frame trawl. The Tucker trawl had three opening-closing 505-pm mesh nets. Five-minute tows were taken at three primary depth intervals: 0-5 m, 5-10 m, and 10-15 m. For the purpose of this study, all depths were combined. The Frame trawl was fitted with a 2-mm mesh net and towed at the surface for 10 minutes. A flow- meter was attached to each net to estimate the volume of water sampled. Tucker trawl samples were split in half with a Folsom plankton splitter. One half of each sample was preserved in 5% buffered formaldehyde and used for identification and length and body depth measurements. The other half was preserved in 95% ethanol and used for otolith analysis. Frame samples were preserved in 95% ethanol. Samples were sorted in the laboratory with a dissecting mi- croscope for eggs, larvae, and juveniles. Morphometries Determination of a seasonal difference in body size was accomplished by analyzing body depth (BD) and standard length (SL) for all specimens collected. Measurements were made to the nearest 0.1 mm with either a video-enhanced digitizing system (Optical Pat- tern Recognition System, BIOSONICS, Inc.) or an ocu- lar micrometer. Standard length was measured from tip of the snout to the tip of notochord. Body depth was measured perpendicular to the longitudinal body axis at the anterior margin of the pectoral base (Ditty, 1981). To examine differences in body depth between spring and summer seasons, linear regressions of body depth on standard length were calculated for Tucker and Frame cruises, 1988 and 1989. Slopes of regression lines were tested for homogeneity. Allom- etric effects of growth on body depth were examined by calculating regressions of body depth on standard length as a function of standard length. Meristics To determine the species composition of the spring- and summer-collected Peprilus larvae, two meristic characters were examined: number of caudal verte- brae and number of ventral midline melanophores. Character counts were compared with published data for each of the three possible species (caudal verte- brae: Ditty, 1981; ventral midline melanophores: Ditty and Truesdale, 1983). Subsamples of specimens were either cleared and stained (Taylor, 1967; Wassersug, 1976; Dingerkus and Uhler, 1977; Potthoff, 1984) or x-rayed (Gosline, 1948; Miller and Tucker, 1979; Tucker and LaRoche 1984; Kosenko et al., 1987). Photomicrographs of cleared and stained specimens were taken and slides developed for further Rotunno and Cowen: Temporal and spatial spawning patterns of Peprilus triacanthus 787 Table 1 List of 1988 and 1989 cruises, sampling dates, number of fish collected, and locations. Atlantic Bight. SAB = South Atlantic Bight; MAB = Mid- Net Cruise Date No. of fish captured Location 1988 DEL88-5-2 24 April-1 May 71 SAB Frame ATLANTIC TWIN 21 May-23 May 29 SAB DEL88-7-1 1 June-3 June 3 MAB DEL88-7-2 11 June-16 June 22 MAB DEL88-7-3 6 July-9 July 4 MAB DEL88-7-4 16 July-22 July 99 MAB DEL88-7-5 29 July-2 August 50 MAB DEL88-7-6 8 August-12 August 110 MAB Tucker DEL88-7-3 6 July-9 July 188 MAB DEL88-7-4 16 July-22 July 1,191 MAB DEL88-7-5 29 July-2 August 514 MAB 1989 DEL88-7-6 8 August-12 August 1,063 MAB Frame FE1-89 25 April-29 April 158 SAB ONR1-89 30 May-2 June 8 MAB ONR2-89 6 June-7 June 4 MAB FE2-89 7 July-10 July 2 MAB ONR4-89 18 July-2 August 22 MAB ONR5-89 14 August-18 August 170 MAB Tucker FE1-89 25 April-29 April 201 SAB FE2-89 7 July- 10 July 445 MAB ONR4-89 18 July-2 August 89 MAB ONR5-89 14 August- 18 August 146 MAB inspection. Fish were x-rayed with a Kodak Faxitron at settings of 40 kilovolts, 20 milli-amperes and 30 sec- onds, which gave the best results with Kodak Industry X negatives. These negatives were placed in a Kodak GBX developer and replenisher solution for 2 to 3 min- utes, rinsed in water, placed in a Kodak GBX fixer so- lution for 5 minutes, rinsed for 15 minutes in water, and dried overnight. Caudal vertebrae were counted with the aid of a dissecting scope and included those vertebrae attached to the first fully formed hemal spine and extending to the urostyle (Gosline, 1960). Caudal vertebrae counts were made for larvae se- lected to cover the range of sizes (all were larger than 7 mm, the size at which Ditty (1981) found most fish to have adult characters), body depth-standard length ratios (shallow and deep-bodied), and dates of capture (spring and summer) encountered. Caudal ver- tebrae were counted for a subsample of 34 cleared-and- stained fish and 64 x-rayed fish from the 1988 Frame trawls and 76 x-rayed fish from the 1989 Frame trawls. Ventral midline melanophores were counted for a subsample of 50 fish (25 from the spring and 25 from the summer) smaller than 4 mm SL (the size at which Ditty based his observations). These fish were ran- domly chosen from 1989 formalin-preserved collec- tions. Ventral midline melanophores were considered to be those located between the hindgut and noto- chord tip (Ditty, 1981). Otolith marking experiment Otoliths of 22 fish were marked with oxytetracycline hydrochloride (OTC) in August of 1991 to determine if Peprilus larvae and juveniles deposit daily rings. Sizes of fish ranged from 10 to 31 mm SL. Fish were acclimated for two days in a five-gallon bucket with built-in screening that allowed water to flow-through the bucket when it was placed in an aerated seawa- ter bath. Fish were fed twice daily with either live Artemia nauplii or live field-captured zooplankton during both acclimation and experimental periods. Experimental conditions were maintained as follows: temperature fluctuated between 12°C and 24°C, sa- linity ranged from 28 to 30 %o, pH varied from 7.6 to 8.0 (however, during marking the pH dropped to 6.5 and 6.6), and the photoperiod remained constant at a 14-h light and 10-h dark cycle. The marking procedure followed those for mass- marking larvae and juveniles (Hettler, 1984; Tsuka- moto, 1985; Muth et al., 1988). Briefly, fish were im- mersed in a 450 mg/L concentration of OTC for a six- hour period. Fish were covered during the marking period to decrease the amount of light because light may interfere with the effectiveness of tetracycline (Secor et al.1 ). While immersed in the marking solu- tion, four fish died. The remaining 18 fish were trans- ferred from the marking solution and placed inside 788 Fishery Bulletin 95(4), 1997 two five-gallon screened buckets. Two additional fish died during the transfer process. Fifteen of the re- maining 16 fish survived until the end of the experi- ment. At four days after marking, four fish were sac- rificed. An additional four fish were sacrificed five days after marking. Three fish were sacrificed eight days after marking. To add a second mark, the remaining four fish were immersed a second time (12 days after the initial marking) and were fed amphipods and brine shrimp that had been immersed in a 450mg/L OTC seawa- ter solution for five hours. These fish were held and fed in a fresh 450 mg/L concentration of OTC for 16 hours. Ten days after the second immersion the re- maining four fish were sacrificed. All specimens were placed in 95% ethanol for pres- ervation. Otoliths were dissected from fish and placed in immersion oil on glass slides. Otoliths were ground with a size-600 carborundem grit and mineral oil slurry. The tetracycline mark was viewed under a Zeiss ultraviolet microscope with a Zeiss FITC acri- dine-orange excitation filter set. Increments were counted blindly by one reader, a minimum of three times. If the number of increments between counts differed by more than one, the otolith was not used. The number of increments present after the mark was regressed against the number of days since marking; the regression coefficient was compared with unity by using a Ctest. Otolith ageing Otolith ageing was performed to determine a length- age relationship for our combined Peprilus spp. samples. The presence of daily increments in tem- perate and tropical water fishes has been noted by Pannella (1971, 1974). Fish fixed and preserved in 95% ethanol and ranging in size from 6 to 28 mm SL were aged from 1988 spring (n=30) and summer (n=33) samples to determine seasonal growth rates. Sagittae and lapilli were removed from each fish fol- lowing the techniques of Brothers (1984). Otoliths were placed in type-B immersion oil and left to clear for one month. Sagittae were analyzed with a video- enhanced digitizing system (Optical Pattern Recog- nition System) viewed at 250x with an oil immer- sion lens. Increments were counted blindly by one reader a minimum of two times. If counts differed between 1 Secor, D. H., E. D. Houde, and D. M. Monteleone. 1995. Development of otolith-marking methods to estimate survival and growth of early life stages of natural and hatchery-produced striped bass in the Patuxent River in 1991. Maryland Depart- ment of Natural Resources, Chesapeake Bay Research and Monitoring Division, CBRM-GRF-94-1, 145 p. readings by more than three increments, they were not used. Counts were made along the longest radial axis whenever possible (Brothers, 1980). When otoliths were too thick to see increments clearly, they were ground with size-600 carborundem grit and im- mersion oil to create thinner sections (Brothers, 1980). A length-at-age relationship was determined by regressing standard length on age. Because most fish deposit increments on their otoliths either at hatch- ing or yolk-sac absorption (Brothers et al., 1976), a y-intercept of hatch size (1.72 mm; Colton and Honey, 1963) was assigned to the regression. The slope of the spring regression, i.e. growth rate, was compared with the slope of the summer regression by means of a homogeneity of slopes test. To determine date of hatching, an age-on-length relation was first deter- mined. The y-intercept for this equation was deter- mined by the best fit of the data. This regression equation was solved for age by using standard lengths of all 1988 and 1989 Tucker- and Frame- caught fish (size range of 6.0-28.0 mm SL). Hatch- ing dates were backcalculated by subtracting age at capture from the capture date. Hatching date distri- butions were then plotted to determine spawning date distributions for all 1988 and 1989 Frame- trawl-caught and Tucker-trawl-caught butterfish. Results Collection During 1988, 3,980 Peprilus were collected in the Tucker trawl (4 cruises; 196 hauls), and 388 in the Frame trawl (8 cruises; 404 hauls). In 1989, 880 Tucker-trawl-caught (4 cruises; 128 hauls) and 364 Frame-trawl-caught (6 cruises; 158 hauls) Peprilus were collected. In 1988, the greatest number of lar- vae and juveniles were caught in Tucker trawls in the MAB, July (n=l,191; 24.3 per haul) through mid- August (az= 1063; 21.7 per haul). There were no Tucker trawl collections during the spring of 1988 in the SAB. The highest numbers of Peprilus were collected in 1989 cruises from mid-July through early August in the MAB (n=445; 17.8 per haul). In general, the number of fish collected in the Frame trawl was less than in the Tucker trawl. In the 1988 Frame trawls, Peprilus were most numerous during late May {n= 29; 1.8 per haul) in the SAB and from mid-July (n=99; 2.0 per haul) through mid-August (n=110; 2.2 per haul) in the MAB. Frame trawls for 1989 had the great- est abundances of larvae during late April (n=158; 4.3 per haul) in the SAB and from mid- July through early August in the MAB (n=170; 4.4 per haul). Rotunno and Cowen: Temporal and spatial spawning patterns of Peprilus triacanthus 789 1000 - 1000 - Tucker 1988 (summer only) Tucker 1989 800 - 1 1 Summer 800 - Spring n = 2,984 1 1 Summer 600 - 600 - n = 881 400 - 400 - 200 - 200 - 1 C D 0 - J 1 1 PPpPrrr-j-r i I , p i rrp 0 - J llWrrrrrr-rirrMMMMMMiMMiM.MM -rrrrrrr i rr t 0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55 Standard length (mm) Figure 2 Length-frequency histograms for Peprilus triacanthus collected by net during the spring and summer of 1988 and 1989. Morphometries A wide range of sizes of Peprilus were collected dur- ing both seasons of 1988 and 1989. Because size-fre- quency distributions had the same mode on all cruises within a season, length frequencies were combined for all fish sampled with a particular gear during each sea- son each year. Spring-collected 1988 fish ranged in size from 6 to 43 mm SL (Frame; Fig. 2A). Similar size dis- tributions of fish were found for the 1989 spring cruises: 4.7 to 37.5 mm SL (Frame; Fig. 2B) and 2 to 6 mm SL (Tucker; Fig. 2D). The 1988 summer trawls collected fish between 5 and 29 mm SL (Frame; Fig. 2A) and 1.3 to 36.0 mm SL (Tucker; Fig. 2C). Summer 1989 fish ranged in size from 6.0 to 50.0 mm SL (Frame; Fig. 2B) and 1.7 to 49.0 mm SL (Tucker; Fig. 2D). Deep- and shallow-bodied fish were identified from both the spring and summer samples in both years. Fish had BD-SL ratios ranging from 0.119 to 0.750 (Fig. 3, A and B). There was an indication of two modes during both spring and summer of 1989. These BD:SL ranges overlap with ranges reported by Horn (1970) and Ditty (1981) for all species (Table 2). Fur- ther analysis of body depth: standard length demon- strated that body depth increases allometrically with respect to standard length up to a size of 10 to 15 mm SL (Fig. 4). Body depth subsequently remained about 50% of length for each year. Meristics Ninety-nine percent of the fish collected with the Frame net in 1988 and 1989 had either 18 or 19 cau- dal vertebrae (of those, approximately 90% had 19 caudal vertebrae; Table 3). Counts of either 18 or 19 caudal vertebrae are consistent with those reported for P. triacanthus (Table 3). The number of caudal vertebrae was not related to body depth. According to Ditty and Truesdale ( 1983), juvenile P. burti and P alepidotus have mean body depth-standard length ratios of greater than 0.557 and caudal vertebrae counts of 17-18. However, our findings indicate that fish with high body depth-standard length ratios also had 19 caudal vertebrae. The fish we sampled had between 8 and 16 ventral midline melanophores. Ditty ( 1981) reported ranges of 4 to 8 ventral midline melanophores for B burti and 11 to 17 for P triacanthus. Seventy-six percent of our speci- 790 Fishery Bulletin 95(4), I 997 1200 1000 800 600 400 200 o -Q £ 0 z 1200 1000 800 600 400 200 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Body depth to standard length ratio - 1988 Spring 1 1 Summer n = 3,372 _l 1 1 1 11 i r i i i i - 1989 [■1 Spring 1 1 Summer n = 1 ,245 t B Figure 3 Body depth to standard length-frequency histograms for Peprilus triacanthus collected during the spring and summer of 1988 and 1989. Frame- and Tucker-trawl collections were combined. mens sampled had 11 to 17 ventral midline melano- phores, and 94% had 10 to 16 melanophores. These data suggest the presence of P. triacanthus in our samples. On further analysis of pigment patterns, we found no difference among specimens (Rotunno, 1992). These data corroborate our caudal vertebral counts. OtoSith marking experiment Daily increment formation was a valid indicator of age of Peprilus spp.. The presence of subdaily incre- ments was also noted in these otoliths. The number of increments observed after the tetracycline mark regressed against the number of days since marking showed a 1:1 correspondence (Fig. 5). Ay-intercept value of zero was assigned to the regression. The slope of the regression (0.921) did not differ signifi- cantly from unity (/-test: t= 0.82, P=0.42). Growth and hatching date Spring and summer growth rates were estimated from 1988 collections of Peprilus at 0.233 mm/day and 0.219 mm/day, respectively. These growth rates were not significantly different (F=2.071, P=0.155). Therefore, seasons were combined and an overall regression was computed that yielded an average growth rate of 0.225 mm/day (Fig. 6). The regression equation (Age = 3.433(SL) + 8.270, r2= 0.93) was used to backcalculate hatching dates from length frequen- cies. This age-length relation was also applied to 1989 collections for back-calculation purposes. Rotunno and Cowen: Temporal and spatial spawning patterns of Peprilus triacanthus 791 0.8 0.6 0.4 0 2 0.2 JZ O) c © ■g 1 o.o c (0 1 08 .c Cl d> T3 E 0.6 co 0.4 0.2 0.0 0 10 20 30 40 50 60 Standard length (mm) Figure 4 Body depth to standard length versus standard length for Peprilus triacanthus collected during the spring and summer of 1988 and 1989. Frame- and Tucker-trawl collections were combined. Hatching for P triacanthus in our samples occurs from February through at least July and appears to be concentrated in two peaks that occur during early spring and summer (Fig. 7, A and B). The earliest hatching date recorded for fish caught in 1988 was 19 January 1988 and the latest was 22 July 1988. The winter-spring spawning appears to begin in January and continue through late April, with an apparent peak in March in the SAB (Fig. 7A). A de- crease in spawning occurs at this time, although some spawning continues at a low level through May (Fig. 7A). Spring-summer spawning occurs during June and July in the MAB with a peak in late June. The relative strength of these two spawnings is not clear because sampling effort was not equivalent during each season. Fish caught in 1989 had a similar spawning pat- tern to that of fish collected in 1988; the earliest hatching date calculated was 14 February 1989, the latest was 29 July 1989 (Fig. 7B). Spawning peaks occurred in late March to early April in the SAB and early to mid-June in the MAB. There appears to be a reduction in spawning during the month of May along the U.S. Atlantic coast. Discussion Peprilus triacanthus is the most common species of Peprilus found along the Atlantic coast of the United States. Within the New York Bight, spawning and larval presence for P triacanthus is reported to oc- 792 Fishery Bulletin 95(4), 1997 Number of days since marking (D) Figure 5 Number of increments observed after oxytetracycline hydro- chloride mark versus the number of days since marking for a subsample of 1988 Frame-caught Peprilus triacanthus. Num- bers in parentheses represent the number of fish. *n = 3, but one otolith was not readable. Table 2 Body depth-standard length ratios reported by Horn (1970), Ditty and Truesdale surveys. (1983), and from our Frame and Tucker trawl Body depth-standard length range Standard length range (mm) n Horn (1970) P. burti 0.460-0.640 7.80-167.00 232 P. triacanthus 0.364-0.600 10.60-198.00 202 P. alepidotus 0.565-0.877 18.22-222.00 205 Ditty (1981) P. burti 0.241-0.579 2.16-19.82 160 P. triacanthus 0.235-0.546 2.04-20.86 159 P. alepidotus 0.205-0.750 1.85-18.92 80 1988 and 1989 Frame and Tucker Spring 0.187-0.750 2.01-43.00 496 Summer 0.119-0.727 1.29-50.00 4,121 Spring and Summer 0.119-0.750 1.29-50.00 4,617 cur during summer only (Wilk et al., 1990). Fahay (1975) suggested however, on the basis of the range of larval lengths collected in 1967-68, that spawn- ing of P triacanthus may occur throughout the year in the SAB. The occurrence of larval and pelagic ju- venile Peprilus within our spring samples collected in the northern SAB suggests two possible scenarios: 1) a species other than the most commonly found P triacanthus spawns within the Atlantic (e.g. P burti ) or 2 ) P triacanthus or P alepidotus has a more pro- tracted spawning season than previously thought. Peprilus larvae and juveniles in our samples proved to be P triacanthus; therefore, our results demon- strate an extended spawning period for P. triacanthus Rotunno and Cowen. Temporal and spatial spawning patterns of Peprilus triacanthus 793 30 - SL = 1.72 + 0.225(Age) *• 25 - r2 = 0'9 m * • n = 60 • • ® • 20 - •* t « 15 - 10 - £ 5 - 0 ' T 1 1 1 1 1 1 1 1 1 T 0 20 40 60 80 100 120 Number of increments (age in days) Figure 6 Standard length versus the number of increments for a sub- sample of 1988 Frame-caught Peprilus triacanthus. Table 3 Vertebral counts for Peprilus collected by Collette (1963), Horn 1970), Ditty (1981), and in this study (Frame). Standard length (mm) range 16 17 17-18 18 18-19 19 20 n Collette (1963) P. burti — 2 92 0 2 0 0 0 96 P. triacanthus — 0 4 0 36 0 138 2 180 Horn (1970) P. burti 6.0-115.0 5 262 0 6 0 0 0 273 P. triacanthus 6.0-115.0 0 7 0 62 0 208 2 279 P. alepidotus 6.0-115.0 3 176 0 3 0 0 0 182 Ditty (1981) P. burti 7.14-14.45 0 13 9 0 0 0 0 22 P. triacanthus 7.14-14.73 0 0 0 1 8 10 0 19 P. alepidotus 7.74-11.01 0 5 0 0 0 0 0 5 Frame Spring 1988 8.0-26.0 0 0 0 4 0 48 0 52 Summer 1988 8.0-26.0 0 1 0 6 0 39 0 46 Spring 1989 8.52-37.5 0 0 0 2 0 28 0 30 Summer 1989 8.0-38.0 0 0 0 6 0 39 1 46 Total Frame 8.0-38.0 0 1 0 18 0 154 1 174 in the Atlantic. According to back-calculated hatch- ing dates, P. triacanthus spawns from late January through at least July. We did not observe any sea- sonal differences in body depth. However, geographic differences in body depth need to be further analyzed before discounting the existence of either a polymor- phic P triacanthus or a hybrid of P triacanthus and P burti. 794 Fishery Bulletin 95(4), 1 997 Species identification According to the reported ranges of body depth-stan- dard length ratios (Horn, 1970; Ditty and Truesdale, 1983), our larval specimens could have been any one of the Atlantic or Gulf coast species of Peprilus. How- ever, two points must be considered when interpret- ing these data. First, we found a strong allometric relation between body depth and standard length for individuals smaller than 15 mm SL. Because the col- lections of Horn (1970) include fish ranging in size from 6 to 222 mm SL, this allometric relation could have confounded his conclusions. Ditty and Truesdale (1983), however, apparently recognized this problem and therefore presented their data by size class. Sec- ond, our other analyses (caudal vertebrae and mel- anophore counts) supported the contention that P. triacanthus was the dominant species of Peprilus in our samples. Our sample was predominantly com- posed of individuals with 19 caudal vertebrae, which was consistent with findings of previous authors for P triacanthus (Collette, 1963; Horn, 1970; Ditty and Truesdale, 1983). The ventral midline melanophore counts were also similar to those used for P triacan- thus by Ditty (1981). The overall results of the above morphometric, meristic, and pigment analyses lead us to conclude that P triacanthus is the dominant and probably the only species of Peprilus collected in our samples. The two most definitive characters for identification were Rotunno and Cowen: Temporal and spatial spawning patterns of Peprilus triacanthus 795 the number of caudal vertebrae and ventral midline melanophores. Age and growth Ages of larval and juvenile P. triacanthus can be de- termined by counting otolith increments. Validation was necessary because of the prevalence of subdaily increments in this species. Secondary nuclei (multinucleation) were also noted in Peprilus otoliths. The cause and timing of the formation of secondary nuclei are not presently understood (Campana and Neilson, 1985), although these secondary nuclei have been demonstrated to form during metamorphosis in bluefish (Hare and Cowen, 1994). Secondary nu- clei, in butterfish otoliths, do not seem to follow a consistent pattern in the development of the fish; we found that two sagittal bones from one fish frequently contained differing numbers of secondary nuclei. Al- though secondary nuclei may form during metamor- phosis (Campana and Neilson, 1985; Hare and Cowen, 1994), we found secondary nuclei in larvae that were several millimeters smaller than the size at which metamorphosis occurs (16 mm). Larval and early juvenile P. triacanthus from 6.0 to 28.0 mm SL grew at a rate of 0.227 mm/day. Al- though ages of young butterfish have not been re- corded previously, Colton and Honey (1963) gave sizes of P. triacanthus from hatching to six days of age. Based on their estimates, growth rates ranged from 0.01 to 0.55 mm/day and decreased with the age of the fish. Specimens of young P. burti have been aged with modal length-frequency analysis; growth rates of P. burti ranged from 0.25 to 0.56 mm/day (Murphy and Chittenden, 1990). Peprilus burti may be ex- pected to have a higher growth rate than P. triacanthus owing to the fact that it spawns in warmer waters of the Gulf of Mexico. Our analysis of hatching-date distributions dem- onstrated that P. triacanthus has a more extensive spawning season than previously reported. Their spawning effort seems to be focused into two cohorts (spring: February-March; and summer: June-July) although evidence is presented that suggests at least some fish spawn during the interim period between seasonal peaks (i.e. late April to early June). Kawa- hara2 speculated that P. triacanthus may begin spawning in April. It should be noted that Kawahara ( 1977) based his spawning estimate on adult growth 2 Kawahara, S. 1977. Age and growth of butterfish, Peprilus triacanthus (Peck), in ICNAF Subarea 5 and Statistical Area 6. ICNAF Res. Doc. 77/VI/27. June 1977 Annual Meeting. Far Seas Fishery Laboratory, Shimizu, Japan. rates, which are higher than our estimate for larvae and juveniles and, therefore, would have underesti- mated the spawning duration. A bimodal hatching-date distribution in P. tria- canthus is similar to that reported for another north- south migrating species within the western Atlan- tic, the bluefish, Pomatomus saltatrix (Kendall and Walford, 1979; Nyman and Conover, 1988). However, Hare and Cowen (1993) and Smith et al. (1994) have proposed that P. saltatrix spawns continually dur- ing its north-south migration and that the apparent bimodal hatching date distribution may result from advective processes acting on the larval distributions and from sampling artifact. The presence of an apparent bimodal spawning in P. triacanthus , with spring and summer peaks, may be an artifact of our sampling. Because we did not sample from April to June in each year and because our sampling locations were spatially distinct be- tween spring and summer, it is possible that we did not collect larvae spawned during May and early June. However, we could have collected older fish in our samples that were spawned early in June. Data collected monthly by the National Marine Fisheries Services (NMFS) as part of their Monitoring, Assess- ment, and Prediction (MARMAP) surveys suggest that larvae are indeed present from April to August in the MAB, thus it is likely that spawning occurs continually (Fig. 8). Moreover, MARMAP data indi- cate a northward progression of larvae from near Cape Hatteras during March or April (or both) into the entire MAB by mid-summer and until October. This spatio-temporal pattern is consistent with the possibility of spawning associated with a seasonal northward migration of adult P triacanthus. Horn ( 1970) has speculated that butterfish movements are highly influenced by temperature (and salinity to a lesser degree). Temperature-related movements by butterfish (and bluefish) would correspond with the observed northward progression of young larvae from the SAB in the spring into the MAB in the summer in association with seasonal warming. The extent of north-south migration by P. triacanthus , however, requires further study. In conclusion, this study adds to our current knowl- edge of the early life history and spawning seasonal- ity of butterfish, P. triacanthus. Our finding of a more protracted spawning season and of a seasonal differ- ence between spawning locations should be of value in reassessing management plans for this species. Current management plans are based on conclusions that P. triacanthus spawns during summer months only. The apparent similarity of spawning periodic- ity of butterfish to that of bluefish (Cowen et al., 1993; Hare and Cowen, 1994, Smith et al., 1994) suggests 796 Fishery Bulletin 95(4), 1997 Figure 8 MARMAP monthly averaged collections (March through October) of larval Peprilus triacanthus during the years (1977-87). that a possible adaptive strategy may be shared by these two seasonally migrating pelagic species. Per- haps with closer inspection, other seasonally migrat- ing species may be found that share this spawning strategy (Hare and Cowen, 1996). Further study into the basis of this pattern is warranted. Acknowledgments We would like to thank all the people who assisted during the cruises and in the laboratory. In addition, comments on earlier drafts were provided by Jeff Buckel, Dave Conover, Steve Morgan, Michael Murphy, Rotunno and Cowen: Temporal and spatial spawning patterns of Peprilus triacanthus 797 Figure 8 (continued) and two anonymous reviewers. Jon Hare provided help with all phases of larval identification and otolith analy- sis. Mike Fahay generously provided Figure 8. This work is a result of research sponsored by the NOAA Office of Sea Grant, U.S. Department of Commerce, under Grants #NA86AA-D-SG045 and #NA90AA-B- SG078 to the New York Sea Grant Institute. Literature cited Brothers, E. B. 1980. Methodological approaches to the examination of otoliths in aging studies. In R. C. Summerfelt and G. E. Hall (eds.), Age and growth offish, p. 319-330. Iowa State Univ. Press, Des Moines, IA. 798 Fishery Bulletin 95(4), 1 997 1984. Otolith studies. In Moser et al. (eds.), Ontogeny and systematics of fishes, p. 50-57. Spec. Publ. Am. Soc. Ichthyol. Herp. 1. Brothers, E. B., C. P. Mathews, and R. Lasker. 1976. Daily growth increments in otoliths from larval and adult fishes. Fish. Bull. 74( 1 ): 1—557. Caldwell, D. K. 1961. Populations of butterfish, Poronotus triacanthus (Peck), with systematic comments. Bull. S. Cal. Acad. Sci. 60:19-31. Campana, S. E., and J. D. Nielson. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1033. Collette, B. B. 1963. The systematic status of the Gulf of Mexico butter- fish, Poronotus burti (Fowler). Copeia 1963(3):582-583. Colton, J. B, Jr., and K. A. Honey. 1963. The eggs and larval stages of the butterfish, Porono- tus triacanthus . Copeia 1963(2):447-450. Cowen, R. K., J. A. Hare, and M. P. Fahay. 1993. Beyond hydrography: Can physical process explain larval fish assemblages within the Mid-Atlantic Bight? Mar. Ecol. Prog. Ser. 53(2):567-587. Dingerkus, G., and L. D. Uhler. 1977. Enzyme clearing of alcian blue stained whole small vertebrates for demonstration of cartilage. Stain Tech. 52(4):229-232. Ditty, J. G. 1981. Comparative morphological development of Peprilus burti, Peprilus triacanthus, and Peprilus paru (Pisces: Stromateidae) from the Western North Atlantic. M.S. thesis, Louisiana State Univ., Baton Rouge, LA, 39 p. Ditty, J. G., and F. M. Truesdale. 1983. Comparative larval development of Peprilus burti, Peprilus triacanthus, and Peprilus paru (Pisces: Stroma- teidae) from the Western North Atlantic. Copeia 1983(2): 397-406. Fahay, M. P. 1975. An annotated list of larval and juvenile fishes cap- tured with surface-towed meter net in the South Atlantic Bight during four RV Dolphin cruises between May 1967 and February 1968. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-685, 39 p. Gosline, W.A. 1948. Some possible uses of x-rays in ichthyology and fish- ery research. Copeia 1948( 1):58-61. 1960. Contributions toward a classification of modern isospondylous fishes. Bull. Br. Mus. (Nat. Hist.) Zool. 6:327-365. Haedrich, R. L. 1967. The stromateoid fishes: systematics and a classifi- cation. Bull. Mus. Comp. Zool. 135(2):31-139. Hare, J. A., and R. K. Cowen. 1993. Ecological and evolutionary implications of the lar- val transport and reproductive strategy of bluefish Pomato- mus saltatrix. Mar. Ecol. Prog. Ser. 98:1-16. 1994. Ontogeny and otolith microstructure of bluefish Pomatomus saltatrix (Pisces: Pomatomidae). Mar. Biol. (Berl.) 118:541-550. 1996. Transport mechanisms of larval and pelagic juvenile bluefish (Pomatomus saltatrix) from the South Atlantic Bight spawning grounds to the Middle Atlantic Bight nurs- ery habitats. Limnol. Oceanogr. 41:1264-1280. Hettler,' W. F. 1984. Marking otoliths by immersion of marine fish larvae in tetracycline. Trans. Am. Fish. Soc. 113:370-373. Horn, M. H. 1970. Systematics and biology of the stromateoid fishes of the genus Peprilus. Bull. Mus. Comp. Zool. 140(5):165- 262. Kendall, A. W., Jr., and L. A. Walford. 1979. Sources and distribution of bluefish, Pomatomus saltatrix, larvae and juveniles off the east coast of the United States. Fish. Bull. 84:91-103. Kosenko, N. A., L. A. Nikolaychuk, and A. A. Vronskiy. 1987. Applications of electroradiography to studies of skel- etons of fishes. J. Ichthyol.. 27( 1): 126—128. Miller, J. M., and R. W. Tucker Jr. 1979. X-radiography of larval and juvenile fishes. Copeia 1979(31:535-538. Murphy, M. D. 1981. Aspects of the life history of the Gulf butterfish, Peprilus burti. M.S. thesis, Texas A&M Univ., Galveston, TX, 76 p. Murphy, M. D., and M. E. Chittenden Jr. 1990. Reproduction, age and growth, and movements of the Gulf butterfish, Peprilus burti. Fish. Bull. 89:101-116. Muth, R. T., T. P. Nesler, and A. F. Wasowicz. 1988. Marking cyprinid larvae with tetracycline. Am. Fish. Soc. Symp. 5: 89-95. Nyman, R. M., and D. O. Conover. 1988. The relation between spawning season and the re- cruitment of young-of-the-year bluefish, Pomatomus saltatrix, to New York. Fish. Bull. 6(2):237-250. Pannella, G. 1971. Fish otoliths: daily growth layers and periodical patterns. Science (Washington, D.C.) 173:1124-1127. 1974. Otolith growth patterns: an aid in age determina- tion in temperate and tropical fishes. In T. Bagenal ( ed. ), Ageing of fish, p. 28-39. Unwin Bros. Ltd., London. Perschbacher, P. W., K. J. Sulak , and F. J. Schwartz. 1979. Invasion of the Atlantic by Peprilus burti (Pisces: Stromateoidae) and possible implications. Copeia 1979:538-541. Potthoff, T. 1984. Clearing and staining techniques. In Moser et al. (eds.), Ontogeny and systematics of fishes, p. 35-37. Spec. Publ. Am. Soc. Ichthyol. Herp. 1. Rotunno, T. K. 1992. Species identification and temporal spawning pat- terns of butterfish, Peprilus spp., in the south and mid- Atlantic Bights. M.S. thesis. State Univ. New York, Stony Brook., NY, 77 p. Smith, W., P. Berrien, and T. Potthoff. 1994. Spawning patterns of bluefish, Pomatomus saltatrix, in the northeast continental shelf ecosystem. Bull. Mar. Sci. 54:8-16. Taylor, W. R. 1967. An enzyme method of clearing and staining small vertebrates. Proc. U.S. Natl. Mus. 122(3596):1-17. Tsukamoto, K. 1985. Mass-marking of Ayu eggs and larvae by tetracycline tagging of otoliths. Bull. Jap. Soc. Sci. Fish. 51(6):903— 911. Tucker, J. W., and J. L. Laroche. 1984. Radiographic techniques in studies of young fishes. In H. G. Moser et al. (eds). Ontogeny and system- atics of fishes, p. 37-39. Spec. Publ. Am. Soc. Ichthyol. Herp. 1. Wassersug, R. J. 1976. A procedure for differential staining of cartilage and bone in whole formalin-fixed vertebrates. Stain Tech. 51(2):131— 134. Rotunno and Cowen: Temporal and spatial spawning patterns of Peprilus triacanthus 799 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 from the New ork Bight. Fish. Bull. 88: 775-786. 800 Abstract .—Three types of genetic markers were used to determine ge- netic relations among four spawning populations of orange roughy off New Zealand. Eleven allozyme loci were tested in starch and cellulose acetate gels. Restriction fragment length poly- morphisms were tested in two regions of the mitochondrial DNA amplified with the polymerase chain reaction. Random amplified polymorphic DNA (RAPD) products were generated with 10-base oligonucleotide primers and separated in agarose gels. There was a significant heterogeneity among all four populations, at 5 out of 11 allozyme loci, at 2 of 29 RAPD primer fragments, and in the frequency of mtDNA haplo- types. There was no significant differ- ence between the two northern spawn- ing populations for any marker, but there were significant differences be- tween all other pairwise population com- parisons with allozymes and RAPD’s, in- dicating the presence of three genetic stocks. The mtDNA analysis revealed less genetic subdivision than did allozymes and RAPD’s. Manuscript accepted 14 May 1997. Fishery Bulletin 95:800-811 (1997). A comparison of three genetic methods used for stock discrimination of orange roughy, Hoplostethus atlanticus: allozymes, mitochondrial DNA, and random amplified polymorphic DNA Peter J. Smith Peter G. Benson S. Margaret McVeagh National Institute of Water and Atmospheric Research Ltd (NIWA) PO Box 14,901, Wellington, New Zealand E-mail address: p.smith@niwa.cri.nz The orange roughy, Hoplostethus atlanticus, is a deepwater species with wide distribution in the Atlan- tic, Indian, and South Pacific Oceans. Around New Zealand the species supports a fishery which peaked at 50,000 tons per annum in the mid 1980’s but which has sub- sequently declined owing to quota restrictions. There are several geo- graphically isolated spawning popu- lations of orange roughy which are the major targets of fishing within the New Zealand Exclusive Eco- nomic Zone (EEZ). A basic prerequisite of fisheries management is the identification of production units or stocks of a spe- cies; inadequate knowledge of stock structure may lead to over- or un- der-exploitation. Orange roughy occur at depths of about 1,000 m and therefore tag and release stud- ies to estimate movements between areas are impracticable. There have been several other approaches to stock identification of orange roughy with differing results. Stud- ies of parasite distribution (Lester et al., 1988), morphometric charac- ters (Linkowski and Liwoch, 1986; Haddon and Willis, 1995), and trace element composition of otoliths (Edmonds et al., 1991) have dem- onstrated regional subdivisions in Australasian orange roughy. An allozyme study revealed a high level of genetic variation but only mar- ginally significant differences be- tween the fishing areas around New Zealand (Smith, 1986). Genetic evi- dence for discrete stocks off South Australia and eastern Australia based on an allozyme survey (Black and Dixon1 ) was not supported by a larger-scale study (Elliott and Ward, 1992). Restriction fragment length polymorphism (RFLP) analy- ses of mitochondrial (mt)DNA have indicated genetic subdivision of or- ange roughy around Australia (Smo- lenski et al., 1993) and New Zealand (Smith et al., 1996). The development of the poly- merase chain reaction (PCR), which amplifies DNA, enables genetic analyses to be carried out on small tissue samples and provides a range of methods for the population biolo- 1 Black, M., and R I. Dixon. 1989. Pop- ulation structure of orange roughy (Hoplo- stethus atlanticus) in Australian waters. Internal Report, Centre for Marine Sci- ence, University of New South Wales, Kensington, Australia, 22 p. Smith et a I.: A comparison of three genetic methods for stock discrimination of Hoplostethus atlanticus 80 i gist without need for cloning and sequencing. PCR amplification of specific regions of mtDNA and di- gestion with restriction enzymes (PCR-RFLP) has been used as a fisheries tool for the differentiation of various fish species (Chow et al., 1993; Chow and Inoue, 1993) and for stock identification of albacore tuna (Chow and Ushiama, 1995), anchovies (Bembo et al., 1995), and salmonids (Cronin et al., 1993; Hall and Nawrocki, 1995; O’Connell et al., 1995; Hansen and Loeschcke, 1996). Mitochondrial DNAis mater- nally inherited and has a higher evolutionary rate relative to protein coding loci (Brown, 1983) and con- sequently has become a useful stock discrimination tool. Random amplified polymorphic DNA (RAPD) uses PCR to amplify fragments of DNA with primers with random nucleotide sequences (Welsh and McClelland, 1990; Williams et al., 1990). Most fisheries applications of RAPD’s have been at the species level (Dinesh et al., 1993; Bardakci and Skibinksi, 1994; Takagi and Taniguichi, 1995), although Macaranas et al. (1995) used RAPD’s to distinguish populations of the fresh- water red claw crayfish, Cherax quadricarinatus, in northern Australia, and a population specific RAPD marker was found in the marine shrimp Penaeus vannamei (Garcia et al., 1996). In this paper we used three methods (allozymes, mtDNA, and RAPD’s) to determine the genetic rela- tions among orange roughy collected from four spawning sites off the east and south coasts of New Zealand. Materials and methods Tissue samples were collected on the RV Tangaroa from four spawning sites off the east and south coasts of New Zealand (Fig. 1). These sites were chosen be- cause they are isolated by distances beyond the likely limit of larval drift (Zeldis et al., 1994). Each site supports significant fisheries, although the Waitaki fishery is relatively small and has declined quickly since development in the early 1990’s (Annala and Sullivan2 ). Heart, liver, and muscle tissues were dis- 2 Annala, J. H., and K. J. Sullivan. 1996. Report from the fish- ery assessment plenary, April-May 1996: stock assessments and yield estimates. Unpubl. Rep., Ministry of Fisheries, Greta Point Library, Wellington, New Zealand. I' I 1 I 1 I 1 165” i i,,rn_Ti 1 i mvw i ' m, i 1 i 1 i 1 i 1 i 1 n r1 r 40° T-T—r 175° W 45° S Pacific Ocean k 1 ' *i' • Puysegur Figure 1 Location of orange roughy spawning sites around New Zealand sampled for genetic analyses. The dotted line represents the 1,000-m isobath. 802 Fishery Bulletin 95(4), 1 997 sected from 100 specimens at three sites and from 50 specimens at Waitaki (Fig. 1). Tissue samples were frozen in liquid nitrogen at sea and stored at -70°C in the laboratory. Allozyme electrophoresis Eight enzyme systems were tested in heart, liver, and muscle tissues of orange roughy with cellulose acetate and starch gel electrophoresis following the methods in Smith (1986), except that BDH (British Drug House Chemicals Ltd, Poole, England) starch was substituted for Electrostarch (Electrostarch Company, USA). DNA extraction DNA was extracted from liver tissue of 50 orange roughy from each site. For each sample, 0.5 g of tis- sue was homogenized with 750 pL 4M guanidinium isothiocyanate in 8M urea and 2% sodium dodecyl sulfate (SDS) (Turner et al., 1989). DNA was ex- tracted by mixing with an equal volume of phenol chloroform and centrifugation at 13,000 rpm for 5 min. The phenol-chloroform extraction was repeated and the aqueous fraction mixed with an equal vol- ume of chloroform-isoamyl alcohol (24:1). Following centrifugation at 13,000 rpm, the aqueous fraction was mixed with two volumes of ethanol and the DNA allowed to precipitate at -20°C overnight. The DNA pellet was washed in 70% ethanol, air dried, and re- suspended in 40 pL of sterile deionised water. mtDNA amplification and restriction enzyme digestion Three primer pairs were used to amplify the mtDNA. Amplification reactions were performed in 50-pL volumes in a Perkin Elmer Cetus DNA thermocycler: protocols followed those of Palumbi et al. (1991) and Cronin et al. (1993). The nucleotide sequences of the primers were the following: D-loop 5’-ATAGTGGGGTATCTAATCCCA-3' 5’-RCRCCCAAAGCTRRRRTTCTA-3' (Palumbi et al., 1991); cytochrome b 5'-CCCTCAGAATGATA- TTTGTCCTCA-3' 5’-TGACCTGAARAACCA- YCGTTG-3' (Palumbi et al.,1991); and ND 5/6 5'-AATAGTTTATCCA- GTTGGTCTTAG-3' 5 ' -TTAC AACGATGGTTTTTC A- TAGTCA-3' (Cronin et al., 1993) Twelve restriction endonucleases recognizing 4-base sites (Bfa I, BstU I, Cfo I, Hae III, Hpa II, Mse I, Msp I, Nla III, Rsa I, Sal I, Sau 3A, and Taq I) were used to digest the D-loop primer amplification products. Eleven restriction endonucleases recognizing 4-base sites (Alu I, Bfa I, Cfo I, Hpa II, Msp I, Nar I, Rsa I, Sal I, Sau 3A, Taq I, and Tru I) were used to digest the cytochrome b primer amplification products. The ND 5/6 primers produced between 1 and 3 amplifi- cation products in different specimens, therefore no restriction digests were undertaken with the PCR products. For each primer pair and restriction enzyme, 24 fish were tested, 6 from each area. The restriction enzymes that showed polymorphisms were used to test 50 fish from each site. The amplified and digested DNA products were separated in 1.4% agarose gels and detected with ethidium bromide under a UV light (312 nm). RAPD amplification and separation Six individuals from each sample site were ampli- fied with 24 RAPD primers. Each sample was am- plified separately with a 10-base oligonucleotide primer from Operon (OperonTechnologies, Alameda, CA). These primers were randomly selected from Operon series A, D, E, and H primers, but all have a G+C content of 60-70%. Amplification reactions were performed in 50-pL volumes in a Perkin Elmer Ce- tus DNA thermocycler. Serial dilutions of DNA samples were tested initially to determine optimum DNA concentration for amplification (Fig. 2). The DNA concentration in each sample was estimated fluorometrically and appropriate volumes were used for amplification. Each reaction contained approxi- mately 50 ng DNA in 10 mM Tris HC1 (pH8.3), 30 ng single 10-base primer, 50 mM KC1, 2 mM MgCl2, 100 mM each of dATP, dCTP, dGTP, and dTTP, and 1 unit Taq DNA polymerase in Perkin Elmer PCR buffer. The reaction was overlaid with mineral oil and amplified. The thermocycler was programmed for 40 cycles of 1-min duration at 94°C, 1 min at 36°C, and 2 min at 72°C. Amplification products were sepa- rated in 1.4% agarose gels and detected with ethidium bromide staining under a UV light (312 nm). A DNA size-ladder was included in each gel. Control samples were amplified without a DNA tem- plate. Those primers that yielded variable fragment patterns were retested in the same fish. Primers pro- ducing repeatable fragment patterns in the initial six fish from each site were tested in 50 fish from each site. Polymorphisms were scored by the pres- ence or absence of an amplification product at spe- cific positions in the gel. Smith et al.: A comparison of three genetic methods for stock discrimination of Hoplostethus atlanticus 803 Figure 2 (A) Random amplified polymorphic DNA (RAPD) profiles in orange roughy generated with the primers A16 and E19. Lane 1 contains a DNA size-ladder (2,072-100 bp), lanes 2-7 represent orange roughy amplifed with primer A16, lane 8 contains no DNA template, and lanes 9-13 represent orange roughy amplifed with primer E19. Each amplified sample of orange roughy contained approximately 50 ng of DNA. (B) RAPD profiles in orange roughy generated with the primers A16 and E19 at different concentra- tions of DNA template. Lane 1 contains a DNA size-ladder (2,072-100 bp), lane 2 no DNA, lanes 3 and 4 contain 12.5 ng DNA amplified with E19, lanes 5 and 6 contain 50 ng DNA amplified with E19, and lanes 7 and 8 contain 200 ng DNA amplified with E 19, lane 9 no DNA, lanes 10 and 11 contain 50 ng DNA amplified with A16, and lanes 12-14 contain 200 ng DNA amplified with A16. (C) RAPD profiles in orange roughy generated with the primer A14, lanes 1-3 contain 50 ng DNA, and lanes 4-6 represent the same samples at a concentration of 200 ng DNA. Statistical analyses Allozyme genotypes Genotypic frequencies were tested for Hardy- Weinberg equilibrium; weakly poly- morphic loci (frequency of most common allele >0.95) were excluded. Rare heterozygotes were pooled with their nearest electrophoretic neighbor to reduce the number of cells with less than five observations. Al- lele frequencies were tested for heterogeneity among populations with contingency %2 tests with the BIOSYS software program (Swofford and Selander, 1981). To test for geographic structure, contingency X2 tests were undertaken on all pairwise combina- tions of populations. Probability levels were modi- fied by the Bonferroni procedure for multiple tests according to Rice (1989). The proportion of allozyme variation due to differ- entiation among populations was estimated with Nei’s gene-diversity statistic CST( Nei, 1973), which is a multiallele estimator of Wright’s FgT statistics (Wright, 1951). Gene diversity is equal to (Ht-Hs)/Ht where HT = the total genetic diversity of all popu- lations; and Hs = the mean genetic diversity per popula- tion, calculated from the average ex- pected heterozygosities. Sampling error will produce differences in allele fre- quencies, even when samples are drawn from the same population, therefore a randomization test was used to test for differences due to sampling error (Elliott and Ward, 1992). One thousand random- izations were used, and the probability was estimated from the number of randomizations that were equal to or greater than the observed Gsr Gene diversity, GST allows an estimation to be made of the number of migrants exchanged between populations per generation from the relation Nem = (1/Gst—1)/4, 804 Fishery Bulletin 95(4), 1997 where Ne = the effective population size; and m = the rate of gene flow per generation. It is assumed that m«l and that population differen- tiation is due to genetic drift and migration with no selec- tion. Gene diversity was corrected to a “true” estimate by subtracting the GSTnuU due to sampling error, derived from a randomization test (Elliott and Ward, 1992). mtDIMA Heterogeneity in haplotype frequencies in the total data was tested by the x2 randomization test described by Roff and Bentzen (1989) with the REAP package (McElroy et al., 1992). This method overcomes the problem of a large number of observed haplotypes at low frequency, by comparing x2 values in 1,000 random rearrangements of the data. In ad- dition the %2 randomization test was applied to pairwise comparisons of all populations to test for geographic structure. Probabilities were estimated from the number of randomizations that were equal to or greater than the observed x2 value. The propor- tion of haplotype variation due to differentiation be- tween populations was estimated by GST from the haplotype frequencies, as for allozymes. The num- ber of migrants exchanged per generation was esti- mated from the relation Nrrif = ( 1 /Gst - l)/2, where m^= female migration, modified to account for the maternal inheritance of mtBNA. RAPD Standard genetic calculations are not imme- diately applicable to RAPD data because the frag- ments are dominant: individuals carrying two cop- ies of an allele cannot be distinguished from indi- viduals carrying one copy of the allele. Black ( 1995) has provided a set of programs for analyzing RAPD population data but points out that a number of as- sumptions have to be made. First, the observed frag- ments are dominant alleles and the absent fragments are recessive alleles. Second, the genotypes are in Hardy- Weinberg equilibrium and each observed poly- morphism is biallelic: all the absent observations are produced by the same recessive allele and all the present observations are produced by a single domi- nant allele with or without the recessive allele. Each primer was scored for the presence or absence of frag- ments in the gel. Each fragment, regardless of primer, was treated as an independent locus. In most RAPD studies, fragments have been found that vary in staining intensity; we scored only fragments that were intensely stained, following Black (1993). Random amplified polymorphic DNA allele fre- quencies were calculated from the presence or ab- sence observations with the RAPDBIOS software pro- gram (Black, 1995) and then used in the BIOSYS software program (Swofford and Selander, 1981) for calculation of heterogeneity in allele frequencies as for allozyme data. The gene-diversity statistic GgT (=Fst) was calculated with the RAPDFST software program (Black, 1995); probabilities were calculated according to Workman and Niswander (1970). An es- timation of the number of migrants exchanged per generation, Ngm, was estimated as for the allozyme data. Results Allozymes Eleven enzyme loci were resolved in the four popu- lations and allele frequencies are given in Appendix Table 1. Eight loci were sufficiently polymorphic (P<0.95) for Hardy- Weinberg tests. One out of a pos- sible 32 tests (8 loci x 4 populations) showed a sig- nificant departure from Hardy- Weinberg equilibrium when a Bonferroni modified probability level was applied (Idh-1* Puysegur, %2=13.99, 1 df, P<0.001). The polymorphic loci were tested with a contin- gency x2 test. There was a significant heterogeneity among all four populations at 5 loci, Est-1*, Gpi-2*, Idh-1* , Idh-2*, and Ldh-1* , with a Bonferroni-modi- Table 1 Results of comparisons of allele frequencies at eleven loci and mtDNA haplotypes in four populations of orange roughy. df = degrees of freedom; P = probability value; and Gst= gene diversity . * = significant at the 5% level with a Bonferroni-modified P for multiple tests. Locus l2 df P Gst P Cck-1* 7.93 6 0.243 0.006 0.277 Est-1* 63.09 12 <0.001* 0.030 <0.001* Gpi-1* 4.70 9 0.860 0.002 0.889 Gpi-2* 25.09 6 <0.001* 0.021 0.002* Idh-1 * 26.91 9 0.001* 0.026 <0.001* Idh-2* 46.08 9 <0.001* 0.065 <0.001* Ldh-1* 19.02 3 0.003* 0.025 0.004* Ldh-2* 12.09 6 0.061 0.008 0.153 Mdh-1* 11.20 6 0.082 0.012 0.066 Mpi-1* 7.56 9 0.581 0.003 0.653 Pgm-1* 2.58 6 0.860 0.001 0.839 all loci 226.2 81 <0.001 0.020 <0.001 mtDNA haplotypes 45.51 0.001 0.057 0.001 Smith et al.: A comparison of three genetic methods for stock discrimination of Hoplostethus atlanticus 805 Table 2 Heterogeneity x2 pairwise comparisons for allozyme loci, mtDNA haplotypes, and random amplified polymorphic DNA (RAPD) fragments, among four populations of orange roughy. For the allozyme and RAPD data only those loci and fragments that were significant applying a Bonferroni-modified probability level are given. mtDNA haplotype RAPD primer fragments Pair Allozyme loci and probabilities probablities and probabilities Ritchie and Box NS NS NS Ritchie and Waitaki Est-1* PcO.001 0.001 E19-3 P<0.001 Ritchie and Puysegur Gpi-2* PcO.001, Idh-2* P<0.001 NS A16-1 P<0.001 Box and Waitaki Est-1* P< 0.001, Idh-l*P= 0.001 NS E19-3 P<0.001 Idh-2* PcO.001 Box and Puysegur Idh-2* P<0.001 NS A16-1 P<0.001 Waitaki and Puysegur Est-1* PcO.001 0.003 E19-3 P<0.001 fled P for 11 loci (Table 1). To test for geographic struc- ture, additional x2 tests were carried out on all pairwise combinations of populations. There was a significant heterogeneity for at least one locus be- tween all population pairs, except Ritchie Bank and Box (Table 2). The heterogeneity in the total data was confirmed by the gene diversity analysis (Table 1). When a Bonferroni-modified P is applied, the 5 loci show a Gst significantly greater than that due to sampling error. Over all eleven loci GST was 0.020 (Table 1), indicating that around 2% of the observed genetic variation was due to differences among populations. From this estimate of GST, and by subtracting the G sTnull ’ ^e minimum number of effective migrants per generation ( Nem ) was 13.2 (Table 3). Individual pairs of Nrn varied from 15.7 (Box and Waitaki) to 124 (Ritchie and Box). mtDNA The estimated size of the PCR amplified B-loop was 1,500 base pairs and that of the cytochrome b was 500 bp. Four restriction enzymes, BstU I, Cfo I, Msp I, and Nla III, produced two or more fragment pat- terns with the D-loop primers (e.g. Fig. 3) and were tested in all fish. For each area, a few fish samples failed to produce an amplification product; the same fish samples also failed to produce an amplification product with the RAPD primers. Four restriction enzymes, Alu I, Bfa I, Rsa I, and Taq I, showed varia- tion in the first 24 fish tested with the cytochrome b primers, but the variation was limited to a single individual with each restriction enzyme. No further amplifications were undertaken with this set of prim- ers. The numbers of haplotypes observed at each site are shown in Appendix Table 2. There is a signifi- cant heterogeneity in the total data (P=0.001), with only 1 out of 1,000 randomizations exceeding the Table 3 The estimated number of migrants exchanged per genera- tion ( Njn ) for allozyme, mtDNA, and random amplified polymorphic DNA (RAPD) data sets of orange roughy. Population Allozyme mtDNA RAPD Ritchie and Box 124.0 277.8 75.0 Ritchie and Waitaki 14.7 7.2 7.7 Ritchie and Puysegur 19.4 35.7 18.0 Box and Waitaki 15.7 18.3 7.8 Box and Puysegur 25.3 36.5 16.0 Waitaki and Puysegur 75.5 9.6 6.5 Total 13.2 9.8 7.0 original x2 value (Table 1). In pairwise comparisons of the four spawning populations (Table 2), signifi- cant differences were found between Ritchie Bank and Waitaki (P<0.001) and between Waitaki and Puysegur (P=Q.Q01), but not in the other pairwise comparisons. Gene diversity was estimated to be 0.057 (Table 1), which is significantly greater than that due to sampling error, and indicates that around 6% of the observed genetic variation is due to differences among populations. From this estimate of GgT, and by subtracting the GSTnul[, the minimum number of female migrants per generation (Afem^) among the four populations was estimated to be 9.8 (Table 3). The pairwise values varied from 7.2 (Ritchie and Waitaki) to 277.8 (Ritchie and Box). RAPD Seven primers tested in 24 orange roughy produced clear DNA fragments and the same profiles in re- peat tests. The primers (and their sequences 5' to 3') 806 Fishery Bulletin 95(4), 1997 were A14 (TCTGTGCTGG), A15 (TTCCGAACCC), A16 (AGCCAGCGAA), A17 (GACCGCTTGT), D15 (CATCCGTGCT), E19 (ACGGCGTATG), and H17 (CACTCTCCTC). The number of scored fragments varied from 1 to 6 per primer, and the size of the fragments from 0.6 to 2.8 kb. Fragments that could be scored were numbered in decreasing order of elec- trophoretic mobility (e.g. primer A14 fragment 1 = A14-1); each individual fish was scored for the pres- ence or absence of each fragment. Repeat tests on some individuals did not produce repeatable patterns for some weakly staining fragments, therefore pres- ence or absence of each fragment was not scored for these fragments. Omitting the DNA template from the PCR reaction (i.e. negative control) failed to pro- duce fragments. The amount of DNA in the initial extractions varied tenfold between samples. Excess DNA, 250 ng, produced different fragment patterns with some primers (Fig. 2), therefore all amplifica- tions were optimized to contain a 50-ng template of DNA. The estimated allele frequencies are given in Ap- pendix Table 3. Two out of 29 primer fragments (A16- 1, E19-3) revealed a significant heterogeneity among populations when a heterogeneity %2 test with a modi- fied probability for multiple tests was applied (Table 4). Pairwise comparisons showed significant differ- ences between all pairs of populations except Ritchie Bank and Box at these two primer fragments (Table 2). The heterogeneity in the total data set was con- firmed by the GST tests (Table 4). The effective num- ber of migrants per generation was estimated to be 7.0 from the overall GST, minus GSTnu!j due to sam- pling error (Table 3), as described for allozymes. Pairwise values varied from 6.5 (Waitaki and Puysegur) to 75 (Ritchie and Box). Discussion There was significant heterogeneity in the allozyme data set at five loci (Table 1) which indicated that the population samples had not been taken from a single panmictic stock. Pairwise comparisons showed that there were significant differences between all pairs of spawning populations except the two north- ern populations at Ritchie Bank and Box (Fig. 1). The RAPD data also showed a significant heteroge- neity that indicated that the population samples had been taken from more than one genetic unit stock. There were no area-specific RAPD fragments in or- ange roughy, as have been reported in marine prawns (Garcia et al., 1996) and freshwater crayfish (Mac- aranas et ah, 1995), but there were differences in Smith et al.: A comparison of three genetic methods for stock discrimination of Hoplostethus atlanticus 807 frequencies of two primer fragments (A16-1, E19-3, Table 4). As with the allozyme data, there were dif- ferences between all pairwise comparisons, with the exception of those samples taken at Ritchie Bank and Box (Table 2). The mtDNA data also showed a significant het- erogeneity in the total data set but demonstrated less genetic differentiation than the allozyme and RAPD data sets, with only two pairwise comparisons show- ing a significant difference (Table 2). However all three methods, which have measured different parts of the genome, gave similar results of low genetic exchange among the four populations (Table 3). None of the estimates of Ngm are true estimates because our data sets are biased in favor of polymorphic mark- ers, which will tend to inflate the GST estimate; such estimates of Nem (Table 2) can be used to compare only relative levels of geneflow between areas (Fer- guson, 1994). In this respect there is 8-10 times as much gene flow between Ritchie Bank and Box than between these sites and Waitaki, when measured with allozymes and RAPD’s, and 15-38 times as much with mtDNA (Table 3). Significant genetic dif- ferences between spawning groups provides evidence of genetic isolation, and thus the data reveal three genetic groups: 1) Puysegur, 2) Waitaki, and 3) Ritchie Bank and Box (Table 2). There are problems with RAPD analyses that may preclude them from use as stock markers for orange roughy. Because RAPD markers are dominant, a number of assumptions have to be made to analyze the data (Lynch and Milligan, 1994). Some of these assumptions, in particular that fragments with the same electrophoretic mobility are genetically identi- cal and that absent fragments represent the same DNA fragment, may not be valid. Fragments that ex- hibited weak staining activity were not scored, so that there is a subjective element when scoring RAPD gels. In the absence of breeding studies, the allelic na- ture of presence or absence of RAPD fragments may be suspect. Garcia and Benzie (1995) reported an extra RAPD fragment in prawn larvae that was ab- sent in adults, although they found Mendelian in- heritance of other RAPD markers. Unlike the other two genetic methods, there are no internal checks that can be used to fit RAPD phenotypes to a genetic model: with allozymes there is an expected gel phe- notype for each enzyme and all alleles are equally expressed; with mtDNA the size of the restricted frag- ments should add up to the size of the undigested fragment. Some primers produced weak fragments that were not repeatable in reamplifications. These weak frag- ments may be produced by excessive PCR cycles; Bell and DeMarini (1991) have shown that by increasing Table 4 Heterogeneity x2 tests and gene diversity ( GST) for seven random amplified polymorphic DNA (RAPD) primers in four populations of orange roughy. df = degrees of freedom; P = probability value; GST = gene diversity. (* = significant at Bonferroni-modified P for multiple tests). Primer and fragment Y2 (3 df) P Gst (3 df) P A14-1 2.820 0.422 0.010 0.415 A14-2 3.306 0.347 0.012 0.341 A14-3 9.089 0.028 0.032 0.031 A15-1 1.543 0.672 0.006 0.671 A15-2 8.189 0.042 0.029 0.044 A15-3 3.593 0.309 0.013 0.299 A15-4 2.138 0.544 0.008 0.540 A16-1 16.921 <0.001* 0.079 <0.001* A16-2 1.473 0.688 0.005 0.685 A16-3 6.874 0.076 0.024 0.087 A16-4 5.072 0.167 0.019 0.156 A16-5 0.513 0.916 0.002 0.915 A17-1 2.138 0.544 0.008 0.540 A17-3 6.715 0.082 0.025 0.073 D15-1 5.436 0.143 0.018 0.178 D15-2 2.114 0.549 0.008 0.513 D15-3 3.862 0.277 0.014 0.280 D15-4 1.678 0.642 0.007 0.615 E19-1 8.080 0.044 0.300 0.043 E19-2 9.283 0.026 0.032 0.033 E19-3 23.079 <0.001* 0.082 <0.001* E19-4 6.558 0.087 0.026 0.071 E19-5 2.680 0.444 0.010 0.423 E19-6 0.823 0.844 0.002 0.887 H17-1 3.824 0.281 0.014 0.268 H17-2 3.119 0.374 0.012 0.343 H17-3 0.776 0.855 0.003 0.863 H17-4 2.155 0.541 0.007 0.604 H17-5 0.500 0.919 0.002 0.913 Total (87 df) 164.66 <0.001 0.019 <0.001 the number of PCR cycles above 30, nonspecific DNA products can be obtained. However, in our prelimi- nary amplifications in extracting DNA from frozen tissue samples, less than 40 cycles produced faint fragment patterns for most primers; thus 40 cycles were used as a standard. The RAPD technique has been shown to be very sensitive to changes in con- centration of primer, concentration of template, an- nealing temperature, and the concentration of mag- nesium ions, all of which can affect the number and intensity of bands (Devos and Gale, 1992; Ellsworth et al., 1993; Patwary et al., 1993; Penner et al., 1993). We sought to avoid these problems by standardizing DNA quantities prior to amplification, performing all amplifications on the same thermocycler, and using the same batch of chemicals. Tissue samples from the four spawning sites were collected and stored under similar conditions. 808 Fishery Bulletin 95(4), 1997 Given the technical problems with RAPD’s, we would recommend them only when other genetic methods have failed to reveal polymorphisms. Tech- niques such as PCR-RFLP of mtDNA, or allozymes, yielded fewer polymorphisms per unit of laboratory time than did RAPD’s but still produced sufficient polymorphisms to detect population structure in or- ange roughy. Our allozyme data set indicated a higher level of genetic subdivision than that found with mtDNA in orange roughy. This result is sur- prising in view of the relatively higher rate of evolu- tion of mtDNA (Brown, 1983), and it is possible that other regions of the mitochondrial genome, or use of additional restriction enzymes, might reveal more genetic variation. Several studies of marine organ- isms have detected greater genetic subdivision with mtDNA than with allozyme markers (e.g. Reeb and Avise, 1990), although there are examples of the re- verse in the fisheries literature (Grewe et al., 1994; Ward et al., 1994). It is possible that the allozyme markers are under selection (Koehn et al., 1980) and are responding to short-term population events rather than to historical events due to reproductive isolation. Acknowledgments We are grateful to Malcolm Clark and Di Tracey, for collection of orange roughy tissues samples, and to John Patton and two anonymous referees for con- structive comments on the manuscript. This research was supported by funds from The Exploratory Fish- ing Company (ORH3B) Limited and the New Zealand Ministry of Fisheries, project number FBOROl. Literature cited Bardakci, F., and D. O. F. Skibinski. 1994. Application of the RAPD technique in tilapia fish: species and subspecies identification. Heredity 73:117- 123. Bell, D. A., and D. M. DeMarini. 1991. Excessive cycling converts PCR products to random length higher molecular weight fragments. Nucleic Ac- ids Res. 19:5079. Bembo, D. G., G. R. Carvalho, M. Snow, N. Cingolani, and T. J. Pitcher. 1995. Stock discrimination of European anchovies, Engrau- lis encrasicolus, by means of PCR-amplified mitochondrial DNA analysis. Fish. Bull. 94:31-40. Black, W. C. 1993. PCR with arbitrary primers: approach with care. Insect Mol. Biol. 2: 1-6. 1995. FORTRAN programmes for the analysis of RAPD- PCR markers in populations. Colorado State Univ., Ft. Collins, CO, 6 p. Brown, W. M. 1983. Evolution of animal mitochondrial DNA. In M. Nei, R. K. Koehn (eds.), Evolution of genes and proteins, p. 62- 88. Sinauer, Sunderland. Chow, S., M. E. Clarke, and P. J. Walsh. 1993. PCR-RFLP analysis on thirteen western Atlantic snappers (subfamily Lutjanidae): a simple method for spe- cies and stock identification. Fish. Bull. 91:619-627. Chow, S., and S. Inoue. 1993. Intra- and interspecific restriction fragment length polymorphism in mitochondrial genes of Thunnus tuna species. Bull. Nat. Res. Inst. Far Seas Fish. 30: 207-225. Chow, S., and H. Ushiama. 1995. Global population structure of albacore ( Thunnus alalunga ) inferred by RFLP analysis of the mitochondrial ATPase gene. Mar. Biol. 123: 39-45. Cronin, M. A., W. J. Spearman, R. L. Wilmot, J. C. Patton, and J. W. Bickham. 1993. Mitochondrial DNA variation in chinook (Onco- rhyncus tshawytscha ) and chum (O. keta ) detected by re- striction enzyme analysis of polymerase chain reaction (PCR) products. Can. J. Fish. Aquat. Sci. 50:708-715. Devos, K. M., and M. D. Gale. 1992. The use of random amplified DNA markers in wheat. Theor. App. Genet. 84:567-572. Dinesh, K. R., T. M. Lim, K. L. Chua, W. K. Chan, and V. P. E. Phang. 1993. RAPD analysis; an efficient method of DNA finger- printing in fishes. Zool. Sci. 10:849-854. Edmonds, J. S., N. Caputi, and M. Morita. 1991. Stock discrimination by trace element analysis of otoliths of orange roughy ( Hoplostethus atlanticus) a deep- water marine teleost. Aust. J. Mar. Freshwater Res. 42:383-389. Elliott, N. G., and R. D. Ward. 1992. Enzyme variation in orange roughy ( Hoplostethus atlanticus) samples from southern Australian and New Zealand waters. Aust. J. Mar. Freshwater Res. 45:51-67. Ellsworth, D. L., K. D. Rittenhouse, and R. L. Honeycutt. 1993. Artifactual variation in randomly amplified polymor- phic DNA (RAPD) banding patterns. BioTechniques 14: 214-217. Ferguson, A. 1994. Molecular genetics in fisheries: current and future perspectives. Rev. Fish Biol. Fish. 4: 379-383. Garcia, D. K., and J. A. H. Benzie. 1995. RAPD markers of potential use in penaeid prawn ( Penaeus monodon) breeding programs. Aquaculture 130:137-144. Garcia, D. K., A. K. Dhar, and A. Alcivar- Warren. 1996. Molecular analysis of a RAPD marker (B20) reveals two microsatellites and differential mRNA expression in Penaeus vannamei . Mol. Mar. Biol. Biotechnol. 5:71—83. Grewe, P. M., A. J. Smolenski, and R. D. Ward. 1994. Mitochondrial DNA diversity in jackass morwong ( Nemadactylus macropterus : Teleostei) from Australian and New Zealand waters. Can. J. Fish. Aquat. Sci. 51:1101- 1109. Haddon, M., and T. J. Willis. 1995. Morphometric and merisitic comparison of orange roughy (Hoplostethus atlanticus'. Trachichthyidae) from the Puysegur Bank and Lord Howe Rise, New Zealand, and its implications for stock structure. Mar. Biol. (Berl.) 123:19-27. Hall, H. J., and L. W. Nawrocki. 1995. A rapid method for detecting mitochondrial DNA Smith et al.: A comparison of three genetic methods for stock discrimination of Hoplostethus atlanticus 809 variation in the brown trout, Salmo trutta. J. Fish Biol. 46:360-364. Hansen, M. M., and V. Loeschcke. 1996. Genetic differentiation among brown trout popula- tions, as detected by RFLP analysis of PCR amplified mi- tochondrial DNA segments. J. Fish Biol. 48:422-436. Koehn, R. K., R. I. E. Newell, and F. Immerman. 1980. Maintenance of an aminopeptidase allele frquency dine by natural selection. Proc. Natl. Acad. Sci. 77:5385-5389. Lester, R. J. B., K. B. Sewell, A. Barnes, and K. Evans. 1988. Stock discrimination of orange roughy Hoplostethus atlanticus by parasite analysis. Mar. Biol. 99:137-143. Linkowski, T. B., and M. Liwoch. 1986. Variations in morphology of orange roughy, Hoplo- stethus atlanticus (Trachichthyidae) otoliths from New Zealand waters. Pr. Morsk. Inst. Ryb. Gdyni. 21:43-59. Lynch, M., and B. G. Milligan. 1994. Analysis of population genetic structure with RAPD markers. Mol. Ecol. 3:91-99. Macaranas, J. M., P. B. Mather, P. Hoeben, and M. F. Capra. 1995. Assessment of genetic variation in wild populations of the redclaw crayfish (Cherax quadricarinatus, von Mar- tens 1868) by means of allozyme and RAPD-PCR markers. Mar. Freshwater Res. 46:1217-1218. McElroy, D., P. Moran, E. Bermingham, and I. Kornfield. 1992. REAP: an integrated environment for the manipula- tion and phylogenetic analysis of restriction data. J. Hered. 83:157-158. Nei, M. 1973. Analysis of gene diversity in subdivided popula- tions. Proc. Nat. Acad. Sci. U.S.A. 70:3321-3323. O’Connell, M., D. O. F. Skibinski, and J. A. Beardmore. 1995. Absence of restriction site variation in the mitochon- drial ND5 and ND6 genes of Atlantic salmon amplified by the polymerase chain reaction. J. Fish Biol. 47:910-913. Palumbi, S. R., A. Martin, S. Romano, W. O. McMillan, L. Stice, and G. Grabowski. 1991. The simple fool’s guide to PCR. Department of Zo- ology, Univ. Hawaii, Honolulu, Hawaii, 23 p. Patwary, M. U., R. M. Mackay, and J. P. van der Meer. 1993. Revealing genetic markers in Gelidium vagum (Rhodophyta) through the random amplified polymorphic DNA (RAPD) technique. J. Phycol. 29:216-222. Penner, G. A., A. Bush, R. Wise, W. Kim, L. Domier, K. Kasha, A. Laroche, G. Scoles, S. J. Molnar, and G. Fedak. 1993. Reproducibility of random amplified polymorphic DNA ( RAPD) analysis among laboratories. PCR methods and applications 2:341-345. Reeb, C. A., and J. C. Avise. 1990. A genetic discontinuity in a continuously distributed species: mitochondrial DNA in the Amercian oyster, Crassostrea virginica. Genetics 124:397-406. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225. Roff, D. A., and P. Bentzen. 1989. The statistical analysis of mitochondrial DNA poly- morphisms: x2 and the problem of small samples. Mol. Biol. Evolut. 6:539-545. Smith, P. J. 1986. Genetic similarity between samples of the orange roughy Hoplostethus atlanticus from the Tasman Sea, South-west Pacific Ocean and North-east Atlantic Ocean. Mar. Biol. (Berl.) 91:173-180. Smith, P. J., S. M. McVeagh, and A. Ede. 1996. Genetically isolated stocks of orange roughy (Hoplostethus atlanticus), but not of hoki ( Macruronus novaezelandiae) in the Tasman Sea and southwest Pacific Ocean around New Zealand. Mar. Biol. (Berl.) 173:783- 793. Smolensk!, A. J., J. R. Ovenden, and R. W. G. White. 1993. Evidence of stock separation in southern hemisphere orange roughy (Hoplostethus atlanticus), (Trachichthyidae) from restriction-enzyme analysis of mitochondrial DNA. Mar. Biol. 116:219-230. Swofford, D. L., and R. B. Selander. 1981. BIOSYS-1 a fortran programme for the comprehen- sive analysis of electrophoretic data in population genet- ics and systematics. J. Heredity 72:281-283. Takagi, M., and N. Taniguichi. 1995. Random amplified polymorphic DNA(RAPD) for iden- tification of three species of Anguilla, A. japonica, A . aus- tralis, and A. bicolor. Fish. Sci. 61:884—885. Turner, B. J., J. F. Elder, and T. F. Laughlin. 1989. DNA fingerprinting of fishes: a general method us- ing oligonucleotide probes. DNA Fingerprinting News 4:15-16. Ward, R. I)., N. G. Elliot, P. M. Grewe, and A. J. Smolenski. 1994. Allozyme and mitochondrial DNA variation in yel- lowfin tuna (Thunnus albacares) from the Pacific Ocean. Mar. Biol. (Berl.) 118:531-539. Welsh, J., and M. McClelland. 1990. Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Res. 18:7213-7218. Williams, J. G. K., A. R. Kubelik, K. J. Livak, J. A. Rafalski, and S. V. Tingey. 1990. DNA polymorphisms amplified by arbitrary primers are useful genetic markers. Nucleic Acids Res. 18:6531- 6535. Workman, P. L., and J. D. Niswander. 1970. Population studies on southwestern Indian tribes. 11. Local genetic differentiation in the Papago. Am. J. Hum. Genet. 22:24-49. Wright, S. 1951. The gene tical structure of populations. Ann. Eugen. 15:323-354. Zeldis, J. R., P. J. Grimes, and J. K. V. Ingerson. 1994. Ascent rates, vertical distribution, and a thermal history model of development of orange roughy, Hoplo- stethus atlanticus, eggs in the water column. Fish. Bull. 93:373-385. 810 Fishery Bulletin 95(4), 1997 Appendix Appendix Table 1 Allele frequencies for 11 allozyme loci tested in four populations of orange roughy. Locus Allele (rc=no. of fish) Ritchie Box Waitaki Puysegur Locus Allele (n=no. of fish) Ritchie Box Waitaki Puysegur Cck-1* 1 0.494 0.500 0.596 0.551 Idh-2* 1 0.417 0.442 0.180 0.196 2 0.489 0.500 0.404 0.444 2 0.583 0.548 0.809 0.793 3 0.017 0.000 0.000 0.005 3 0.000 0.005 0.011 0.011 n 90 81 47 99 4 0.000 0.005 0.000 0.000 Est-1* i 0.051 0.160 0.093 0.152 n 96 94 47 94 2 0.222 0.229 0.372 0.250 Ldh-1* i 1.000 0.983 0.979 0.931 3 0.709 0.606 0.430 0.585 2 0.000 0.017 0.021 0.069 4 0.006 0.005 0.006 0.013 n 96 90 47 99 5 0.013 0.000 0.000 0.000 Ldh-2* i 0.000 0.000 0.011 0.000 n 79 94 43 99 2 1.000 0.980 0.989 0.980 Gpi-1* i 0.515 0.551 0.553 0.517 3 0.000 0.020 0.000 0.020 2 0.232 0.253 0.245 0.265 n 96 99 47 99 3 0.253 0.191 0.191 0.214 Mdh-1* i 0.625 0.729 0.656 0.745 4 0.000 0.006 0.001 0.004 2 0.375 0.271 0.344 0.250 n 99 89 47 99 3 0.000 0.000 0.000 0.005 Gpi-2* i 0.015 0.033 0.042 0.058 n 96 94 48 98 2 0.970 0.944 0.948 0.860 Mpi-1* i 0.037 0.027 0.052 0.057 3 0.015 0.022 0.010 0.081 2 0.957 0.968 0.948 0.943 n 99 90 48 99 3 0.000 0.005 0.000 0.000 Idh-1* i 0.030 0.011 0.106 0.083 4 0.005 0.000 0.000 0.000 2 0.970 0.979 0.883 0.897 n 94 94 48 97 3 0.000 0.005 0.011 0.021 Pgm-1* i 0.116 0.144 0.117 0.126 4 0.000 0.005 0.000 0.000 2 0.879 0.850 0.883 0.874 n 99 94 47 99 3 0.005 0.006 0.000 0.000 n 99 90 47 99 Appendix Table 2 Numbers of composite mtDNA D-loop haplotypes observed in four populations of orange roughy. The composite haplotypes are based on the restriction enzymes BstU I, Cfo I, Msp I, and Nla III. Haplotype Ritchie Box Waitaki Puysegur AABA 21 19 5 14 BBBA 11 18 21 7 AC BA 0 0 2 0 AAAA 2 0 0 0 AABB 1 1 1 3 AABC 1 0 0 0 AACA 1 0 0 0 ABBA 4 4 6 10 BABA 2 6 6 6 Smith et a I.: A comparison of three genetic methods for stock discrimination of Hoplostethus atlanticus 81 1 Appendix Table 3 Random amplified polymorphic DNA(RAPD) fragment frequencies, calculated by assuming a biallelic system in Hardy -Weinberg equilibrium, in four populations of orange roughy. Locus Allele (n=no. of fish) Ritchie Box Waitaki Puysegur Locus Allele (n=no. of fish) Ritchie Box Waitaki Puysegur A14-1 1 0.00 0.021 0.00 0.021 D15-2 1 0.426 0.542 0.500 0.489 2 1.00 0.979 1.00 0.979 2 0.574 0.458 0.500 0.500 n 44 48 42 40 n 44 48 42 40 A14-2 i 0.454 0.417 0.417 0.330 D15-3 i 0.629 0.708 0.583 0.745 2 0.546 0.583 0.583 0.670 2 0.361 0.292 0.417 0.255 n 44 48 42 40 n 44 48 42 40 A14-3 i 0.306 0.188 0.167 0.394 D15-4 i 0.083 0.063 0.042 0.043 2 0.694 0.813 0.833 0.596 2 0.907 0.938 0.958 0.957 n 44 48 42 40 n 44 48 42 40 A15-1 i 0.009 0.00 0.00 0.00 E19-1 i 0.269 0.292 0.142 0.330 2 0.991 1.00 1.00 1.00 2 0.731 0.708 0.858 0.670 n 44 48 42 40 n 43 48 42 38 A15-2 i 0.806 0.792 0.708 0.638 E19-2 i 0.148 0.271 0.042 0.149 2 0.194 0.208 0.292 0.362 2 0.852 0.729 0.958 0.851 n 44 48 42 40 n 43 48 42 38 A15-3 i 0.028 0.042 0.042 0.00 E19-3 i 0.639 0.646 0.125 0.521 2 0.972 0.958 0.958 1.00 2 0.361 0.354 0.875 0.479 n 44 48 42 40 n 43 48 42 38 A15-4 i 0.009 0.021 0.00 0.00 E19-4 i 0.093 0.021 0.083 0.021 2 0.991 0.979 1.00 1.00 2 0.898 0.979 0.917 0.979 n 44 48 42 40 n 43 48 42 38 A16-1 i 0.49 0.46 0.33 0.74 E19-5 i 0.056 0.042 0.00 0.021 2 0.51 0.54 0.67 0.26 2 0.944 0.958 1.00 0.979 n 43 48 42 38 n 43 48 42 38 A16-2 i 0.019 0.00 0.00 0.021 E19-6 i 0.769 0.708 0.708 0.745 2 0.981 1.00 1.00 0.979 2 0.231 0.292 0.292 0.255 n 43 48 42 38 n 43 48 42 38 A16-3 i 0.278 0.271 0.125 0.149 H17-1 i 0.046 0.00 0.00 0.021 2 0.722 0.729 0.875 0.851 2 0.954 1.00 1.00 0.979 n 43 48 42 38 n 44 48 42 38 A16-4 i 0.694 0.708 0.583 0.564 H17-2 i 0.806 0.708 0.708 0.713 2 0.306 0.292 0.417 0.436 2 0.194 0.292 0.292 0.287 n 43 48 42 38 n 44 48 42 38 A16-5 i 0.019 0.021 0.00 0.021 H17-3 i 0.769 0.792 0.708 0.745 2 0.981 0.979 1.00 0.979 2 0.231 0.208 0.292 0.255 n 43 48 42 38 n 44 48 42 38 A17-1 i 0.009 0.021 0.00 0.00 H17-4 i 0.731 0.708 0.708 0.638 2 0.991 0.979 1.00 1.00 2 0.269 0.292 0.292 0.362 n 43 48 42 38 n 44 48 42 38 A17-3 i 0.009 0.00 0.00 0.053 H17-5 i 0.019 0.021 0.042 0.021 2 0.991 1.00 1.00 0.947 2 0.981 0.979 0.958 0.979 n 43 48 42 38 n 44 48 42 38 D15-1 i 0.111 0.188 0.042 0.074 2 0.889 0.813 0.958 0.926 n 44 48 42 40 812 Abstract .—To characterize the im- pact of spring floods on the survival of juvenile chinook salmon in the un- stable, braided rivers on the east coast of New Zealand’s South Island, I exam- ined correlations between spring and summer flows in the mainstem of the Rakaia River and fry-to-adult survival for chinook salmon spawning in a head- water tributary. Flow parameters that were investigated included mean flow, maximum flow, and the ratio of mean to median flow (an index of flow vari- ability), calculated during peak down- river migration of ocean-type juveniles (August to January). Survival was uncorrelated with mean or maximum flow but was positively correlated with the ratio of mean to median flow dur- ing spring (October and November). The correlation suggests that pulses of freshwater entering the ocean during floods may buffer the transition of fin- gerlings from fresh to saline waters and thus partly compensate for the lack of an estuary on the Rakaia River. A posi- tive correlation between spring flow variability and the proportion of ocean- type chinook in relation to stream-type chinook is also consistent with this hy- pothesis. All correlations were rela- tively weak, reinforcing earlier results that production is primarily controlled by marine influences. These findings further demonstrate the considerable ability of chinook salmon to adapt to new habitats. Manuscript accepted 7 May 1997. Fishery Bulletin 95:812-825 (1997). Survival of chinook salmon, Oncorhynchus tshawytscha, from a spawning tributary of the Rakaia River, New Zealand, in relation to spring and summer mainstem flows Martin J. Unwin National Institute of Water and Atmospheric Research (NIWA) PO Box 8602, Christchurch, New Zealand E-mail address: m.unwin@niwa.cri.nz To understand the population dy- namics of anadromous Pacific salmonids ( Oncorhynchus spp.), it is important to isolate and charac- terize the influence of varying en- vironmental factors on annual pro- duction. In the course of their life cycle salmon inhabit a succession of freshwater and marine environ- ments, where prospects for survival depend on prevailing conditions. Spawning and incubation success may be adversely affected by sub- strate disturbance during floods; the suitability of riverine waters as habitat for rearing juveniles is de- pendent on both flow and tempera- ture and may be reduced by flows that are too low or too high; and adult survival within the marine environment is at least partly de- termined by environmentally con- trolled factors such as oceanic wa- ter masses and the availability of suitable prey. Numerous studies have demonstrated significant cor- relations between environmental variables and indices of survival and growth, at scales ranging from local to global. Although correlation analysis in fisheries studies has been criticized for its potential for misuse and for a propensity to pro- duce weak results of little practical value (Walters and Collie, 1988), other authors have noted that pro- vided the method is used with dis- cretion, biologically meaningful re- sults can be derived (Kope and Bots- ford, 1990). Despite the importance of in- stream habitats for rearing juvenile chinook salmon (O. tshawytscha), the relation between flow and brood year survival has received compara- tively little attention. Interannual trends in the abundance of chinook salmon in the Fraser River, British Columbia, have been linked to flow variations in the mainstem (Beamish et al., 1994 ) and in the Nechako River tributary (Bradford, 1994). In the former study, annual production was inversely related to mean an- nual discharge, whereas in the lat- ter study, juvenile survival in the upper Nechako appeared to decline as a result of flow diversion for hy- droelectric generation, and the pro- portion of spawning fish using the upper river appeared to be nega- tively correlated with August flows. However, in the Nechako River study, as in some other studies link- ing downriver migration to river flows (e.g. Kjelson et al., 1982), low flows were often associated with increased water temperatures, making it difficult to differentiate between flow-related and tempera- ture-related effects. Williams and Matthews (1995) found that sur- vival of Snake River spring and summer chinook salmon juveniles was reduced during low flow condi- tions but concluded that these Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River, New Zealand 813 losses were primarily due to problems with passage through hydroelectric dams rather than to low dis- charge per se. The effects of flow variability on sur- vival have also received little attention. Increased downstream movement of newly emerged salmonid fry following sudden increases in discharge has been well documented (e.g. Irvine, 1986; Saltveit et al., 1995), but only rarely has flow variability been used as a predictor variable in population studies (Berggren and Filardo, 1993). Whenever brood year survival is estimated from stock-recruitment or similar data, a search for cor- relations between river flow and survival will usu- ally involve deriving a single flow index, such as the annual mean, for each cohort. Most such studies con- ducted to date have used flow averaged over periods from three months (Kope and Botsford, 1990) to one year (Beamish et al., 1994), but it is by no means obvious that these are the most informative or bio- logically meaningful parameters to use. A single cata- strophic flood during the incubation period may cause large-scale destruction of redds and loss of alevins through bed scour (Montgomery et al., 1996) with- out having much effect on the mean annual flow. Prior to smolting, fry may be susceptible to short- term floods that carry them prematurely into sea- water, when the same floods a few months later would have little impact. In addition, mean flow is not nec- essarily the most relevant statistic for characteriz- ing flow regimes; it is possible that in the two ex- amples given above, some other parameter (such as maximum flow or the coefficient of variation) might be more informative (e.g. Hvidsten and Hansen, 1988). For example, in New Zealand, where high flow variability is a defining characteristic of riverine eco- systems (Biggs, 1995), statistics such as the propor- tion of the time the flow exceeds three times the median (Clausen and Biggs, in press) and the ratio of mean flow to median flow ( Jowett, 1990), have been successful in elucidating relations between flow re- gime and biological parameters. To explore fully the relation between flow and survival, therefore, it is necessary to consider not only the type of flow sta- tistic that is likely to be of interest but also the dura- tion and seasonal timing of the period over which the statistic is to be calculated. Since the introduction of fall-run Sacramento River stock to New Zealand in the early 1900’s (McDowall, 1994a; Quinn et al., 1996), chinook salmon have maintained self-sustaining populations in all major rivers on the east coast of the South Island (McDowall, 1990; Quinn and Unwin, 1993). Like most New Zealand rivers, these rivers (whose wide, braided shingle beds drain steeply mountainous catchments on the South Island main divide) are characterized by highly vari- able flows (Jowett and Duncan, 1990), flooding quickly whenever snow and ice melt in the headwaters is augmented by heavy orographic rainfall. These floods occur at any time of year but are particularly common in spring. In rivers such as the Rakaia they cause mas- sive bed movement ( Ibbitt, 1979) and reduce the abun- dance and diversity of invertebrate fauna (Sagar, 1986) for up to one month afterwards. The impact of these events on seaward-migrating juvenile chinook has gen- erated some debate. Several authors have remarked that survival of New Zealand chinook fry may be ad- versely affected during floods (McDowall, 1990; Flain1). Other studies suggest that, although some fry may be lost during extreme floods, flow fluctuations in a more typical season do not have a serious negative impact on migration (Hopkins and Unwin, 1987). Chinook salmon spawning populations in Glen- ariffe Stream, a headwater spawning tributary of the Rakaia River (Fig. 1), have been monitored since 1965 by means of an upstream counting fence (Quinn and Unwin, 1993). In this study I analyzed brood year survival, for chinook spawning in Glenariffe Stream, in relation to Rakaia mainstem discharge during spring and summer (August to January). My primary objectives were to examine various flow statistics as possible correlates with survival and to determine the sensitivity of any resulting correlations to changes in the interval used to calculate each statistic. A second- ary objective was to examine evidence that spring floods were detrimental to brood year survival. Chinook salmon in New Zealand New Zealand chinook salmon are broadly similar to their Sacramento ancestors in terms of both their general life history (Unwin, 1986) and genetic make up (Quinn et al., 1996). Present day stocks comprise a mixture of ocean- and stream-type fish (Gilbert, 1913; Healey, 1983), corresponding to juveniles that spend 3-6 mo or 12-15 mo in fresh water before entering the ocean (Unwin and Lucas, 1993). In the Rakaia River, the most thoroughly studied of the major salmon pro- ducing rivers, ocean-type fish make up about two-thirds of the returning adults (Quinn and Unwin, 1993). The migration patterns of age-0+ juvenile chinook salmon in the Rakaia River and a key spawning tribu- tary, Glenariffe Stream (Fig. 1), have been studied in some detail, and are relatively well understood (Unwin, 1986; Hopkins and Unwin, 1987). From 1 Flain, M. 1982. Quinnat salmon runs, 1965-1978, in the Glenariffe Stream, Rakaia River, New Zealand. Occasional Publ 28, N.Z. Ministry of Agriculture and Fisheries, Fisheries Research Div., 22 p. [Copy held at NIWA, Christchurch, New Zealand.] 814 Fishery Bulletin 95(4), 1997 The South Island of New Zealand and the Rakaia River catchment, showing geographical features referred to in the text, the four main headwater tributaries used by spawning chinook salmon (1-4), and the 20-m and 30-m isobaths off the Rakaia mouth. August to October (late winter to mid spring), large numbers of newly hatched fry emerge from the gravel in Glenariffe Stream and other spawning areas and begin to move downstream within 24 h of emergence. This migration appears to be driven by population pressure; the rearing capacity of Glenariffe Stream has been estimated at less than 100,000 fry, whereas annual fry production can exceed 3.7 million (Unwin, 1986). A second wave of larger fry, representing indi- viduals remaining in their natal stream for up to 3 months, enters the upper river from November to January, but in Glenariffe Stream these fry repre- sent less than 10% of the total production. This pat- tern appears to be typical for chinook populations within their native range (e.g. Lister and Walker, 1966; Reimers, 1973; Healey, 1991). Within the upper reaches of the Rakaia River, fry quickly take up residence along the margins of the braided channels, where there is an abundance of suitable rearing habitat (Glova and Duncan, 1985). Aquatic invertebrates, predominantly Deleatidium spp., are the primary prey in spring, but in summer the diet of fry is dominated by terrestrial species and chironomids (Sagar and Glova, 1987). From mid- August fingerlings gradually disperse downriver, growing steadily as the season progresses and reach- ing the lower river in mid-October at about 60-80 mm fork length (FL) (Hopkins and Unwin, 1987). Fingerlings remain abundant in the lower river un- til early February but show little tendency to increase in size; thus there appears to be a steady emigration of 90-day fingerlings into marine waters with con- tinual replacement from upriver (Hopkins and Unwin, 1987). Similar patterns of movement have been observed in other New Zealand stocks (Davis and Unwin, 1989), in their ancestral Sacramento River (Kjelson et al., 1982), and elsewhere in North America (Healey, 1991). Mean FL at seawater entry Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River, New Zealand 815 for these fingerlings is consistent with the back-cal- culated mean FL at seawater entry for ocean-type adults of Rakaia origin (Unwin and Lucas, 1993), confirming the importance of springtime mainstem rearing for ocean-type fry. Freshwater residence pat- terns of juvenile stream-type fish are less well un- derstood, but there is some evidence that an initial period of tributary rearing lasting 3-6 mo is followed by mainstem rearing for the remainder of the first year (Unwin, 1986). In the absence of a commercial marine fishery for salmon, very little is known about the marine phase of the salmon life cycle. Adult chinook are piscivo- rous and appear to feed opportunistically within the pelagic zone, although prey diversity is low and food availability is potentially limited by annual fluctua- tions in prey abundance (James and Unwin, 1996). Brood year survival rates for both naturally and hatchery-produced chinook of Rakaia origin vary by up to two orders of magnitude and appear to be pre- dominantly related to marine influences (Unwin, in press). However, these results do not preclude the possibility that survival may also be partly influenced by conditions within the freshwater environment. Data sources and methods The Rakaia River The Rakaia River is a large, braided, glacier-fed river draining a 2,910 km2 catchment that spans 70 km of the Southern Alps and rises to 2,800 m. Apart from a gentle 5-km gorge where the flow is briefly confined to a single channel, the river occupies an unstable, highly braided shingle bed up to 5 km wide (see Fig. 2 of Glova and Duncan, 1985). After collecting water from two major headwater tributaries (the Mathias and Wilberforce), the lower section (90 km) of the river flows directly into the Pacific Ocean with no significant tributary input. All major salmon spawn- ing waters are located upstream of the Wilberforce confluence (Fig. 1). River gradient is virtually con- stant below this point, averaging 4.5 m/km, and the river discharges into the ocean via a small freshwa- ter lagoon extending inland from the open sea for less than 100 m. The term “lagoon” is used in prefer- ence to “estuary” because there is no ebb and flow of the tide (although there is a tidal backup of fresh water) and because the area supports few, if any, predominantly estuarine life forms.2 A detailed de- 2 Eldon, G. A., and A. J. Greager. 1983. Fishes of the Rakaia Lagoon. Fisheries Environmental Report 30. N. Z. Ministry of Agriculture and Fisheries, Fisheries Res. Div., Christchurch, 65 p. [Copy held by NIWA, Christchurch, New Zealand.] scription of the river and its catchment is given by Bowden.3 4 5 6 7 Continuous flow data for the Rakaia River have been collected since 1959 by means of recorders at the downstream end of the gorge, 62 km above the mouth. For the purposes of this study, all flow sta- tistics were calculated from the daily mean discharge (Q). Discharge (annual mean 200 m3/s) shows a mod- erately seasonal pattern, monthly means varying from 127 m3/s in July to 265 m3/s in November.3 The mean annual flood (the average of the annual maxi- mum flow) is 1 448 m3/s, with a peak instantaneous discharge of 5,600 m3/s (estimated to have a return period of 60 yr) recorded in January 1994. The bank- full discharge (the instantaneous flow which results in complete inundation of the river bed as individual braids coalesce) ranges from 800 m3/s just below the gorge to about 2,500 m3/s in the lower river. Flood waters move rapidly downriver, typically reaching the mouth 8-12 h after passing through the gorge, although peak velocity increases with flood inten- sity, and travel times as short as 3.5 h over 40 km have been observed.4,5 Over the months relevant to this study, daily mean water temperatures in the lower river (23 km above the mouth) range from 6°C in August to 16°C in January (Unwin, 1986). Glenariffe Stream spawning runs Glenariffe Stream is a spring-fed tributary joining the Rakaia River 100 km above its mouth at an alti- tude of 430 m (Fig. 1). The flow regime is exception- ally stable, with a mean discharge of 3.4 m3/s and a maximum recorded discharge (over a seven-year pe- riod) of 16 m3/s. Chinook salmon spawning runs in Glenariffe Stream have been monitored annually by means of a counting fence installed in 1965 (Quinn and Unwin, 1993; Plain1 ). The modal age-at-matu- rity is three years, with smaller numbers of 2-year and 4-year-olds and very few 5-year-old fish. The angler interception rate varies little between years, typically ranging from 30% to 40%. 6,7 Since 1980, 3 Bowden, M. J. 1983. The Rakaia River and catchment — a resource survey, vol. 2. North Canterbury Catchment Board and Regional Water Board, Christchurch, New Zealand, 101 p. [Copy held by NIWA, Christchurch, New Zealand.] 4 1997. Unpubl. data, NIWA, Christchurch, New Zealand. 5 Horrel, G. 1997. Canterbury Regional Council, PO Box 345, Christchurch, New Zealand. Personal commun. 6 Unwin, M. J., and S. F. Davis. 1983. Recreational fisheries of the Rakaia River. Fisheries Environmenal Report 30. New Zealand Ministry of Agriculture and Fisheries, Fisheries Re- search Division, Christchurch, New Zealand. [Copy held by NIWA, Christchurch, New Zealand.] 7 Millichamp, R. 1997. North Canterbury Fish and Game Council, 3 Horatio Street, Christchurch, New Zealand. Per- sonal commun. 816 Fishery Bulletin 95(4), 1 997 spawning stocks have been supplemented by hatch- ery releases, but scale-pattern analysis and coded- wire tag recoveries of spawning fish intercepted at the fence allow each run to be partitioned into hatch- ery-reared and naturally spawning components (Unwin and Glova, 1997). The stability of the flow regime ensures that pre-emergence mortality of ova and alevins is not flow-dependent and is reflected by the lack of interannual variation in egg to fry sur- vival for naturally spawning fish (Unwin, 1986; Unwin, in press). Over five years (1973-76, and 1992) of record egg-to-fry survival ranged from 38 to 52% and averaged 48%. On this basis, annual production can be consistently expressed in terms of the num- ber of fry leaving Glenariffe Stream (Unwin, 1997). For this study, I used the data set in Table 1 of Unwin (1997), summarizing fry-to-adult survival (S) for the 26 years from 1965 to 1990, expressed as live adult spawners reaching Glenariffe Stream (summed over all year classes for each cohort) per 10,000 fry. These data are reproduced here as Table 1. Survival ranged from 1.3 (for the 1971 brood year) to 117 (for the 1973 brood year), with an annual mean of 8 spawners per 10,000 fry (0.079%). These data were log-normally distributed (Unwin, 1997); therefore I used log-transformed values for all calculations (see also Bradford, 1995). Data analysis For each year from 1965 to 1990, 1 calculated mean flows ( Q ) for each calendar month, for the two three- month periods August-October (“spring”) and No- vember-January (“summer”), and for the full six months. As indices of flow variability, I determined the maximum flow ( Q ), and the ratio of the mean to the median flow ( Q ), for the same monthly, three- monthly and six-monthly intervals. Although other measures of flow variability are possible, such as skewness, coefficient of variation (CV), and baseflow index (which measures the ratio of the volume of base flow to the volume of total runoff), these all tend to be highly correlated and there is no one measure which represents the “best” index (Jowett and Duncan, 1990). I chose Q because it is more robust (i.e. insensitive to extreme outliers) than statistics Table 1 Spawning population size, fry production, adult returns, and fry-to-adult survival (adults per 10,000 fry, rounded to the nearest integer) for naturally spawning Glenariffe Stream chinook, 1965-90. Fry production for 1973-76 was estimated from trapping records; all other figures are based on a mean egg-to-fry survival of 48% (Unwin, 1997). Brood year Number of female spawners Estimated fry production (thousands) Number of returning adults Adults per 10,000 fry 1965 1,278 2,988 4,676 16 1966 573 1,513 1,334 9 1967 746 1,760 505 3 1968 1,781 4,310 2,642 6 1969 1,286 3,249 2,760 8 1970 248 655 474 7 1971 1,084 2,573 330 1 1972 1,618 3,418 1,731 5 1973 160 275 3,207 117 1974 173 426 1,242 29 1975 799 1,834 2,045 11 1976 1,522 3,436 2,943 9 1977 778 1,614 1,341 8 1978 863 1,810 1,787 10 1979 1,413 2,138 546 3 1980 481 826 548 7 1981 856 1,978 1,097 6 1982 276 643 2,815 44 1983 326 784 1,766 23 1984 772 2,037 3,037 15 1985 1,800 4,722 950 2 1986 970 2,136 493 2 1987 669 1,400 1,151 8 1988 838 2,074 526 3 1989 391 767 615 8 1990 295 676 504 7 Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River, New Zealand 817 such as CV or skewness, which involve raising flows to the second and third power, respectively. I also calculated several indices related to the incidence of flood peaks, including the number of days when the daily mean discharge exceeded 500 m3/s, 1,000 m3/s, and 1,500 m3/s, and the mean of the ten highest flows over the six months from August to January. The complete set of flow parameters used is summarized in Table 2. For each parameter, I looked for evidence of a re- lation with the log-transformed Glenariffe Stream survival data by calculating the correlation coeffi- cient for the paired data sets over the 26 years of record. I examined bivariate scatter plots and re- sidual plots for any data sets showing a significant relation (P<0.05). For these preliminary results I did not correct for the effect of multiple tests (i.e. the possibility of finding an artificially inflated correla- tion with one of the 32 flow parameters purely by chance); therefore P-values for each correlation over- estimated their true significance (Walters and Col- lie, 1988; Kope and Botsford, 1990). For these pa- rameters, my next level of analysis was to recalcu- late the appropriate flow statistic for periods rang- ing in duration from one week to four months, dat- ing from 1 June to 31 January (representing 805 periods in total), and to recalculate the correlation with the survival data for each choice of date and duration. I then constructed contour plots depicting variations in the correlation coefficient as a function of starting date and duration and examined these “surfaces of correlation” for the presence of local ex- trema. My motivation for this analysis was not to identify a single period that maximized the correla- tion, but rather to gauge the sensitivity of the corre- lation to small changes in interval, and hence to iden- tify seasonal periods for which significant correla- tions between flow indices and survival persisted over biologically meaningful time scales. All statistical calculations were performed with version 6.0 of SYSTAT software (Wilkinson, 1996). Contour and surface fits were accomplished with version 6 of SURFER for Windows’ implementation of Kriging smoothing (Keckler, 1994) applied to a grid of correlation coefficients calculated at 5-day inter- vals on both the date and period axes. Results River flows Over the period covered by this study (August 1965 to January 1991), monthly mean discharge ranged from 146 m3/s to 306 m3/s (Table 2). Individual Table 2 Flow parameters used for correlation analysis, together with their mean and range over the period 1965 to 1990. Parameter Symbol Mean Range Measures of flow volume (m3/s) Mean annual flow, Feb-Jan QAnnua, 209 156-277 Mean flow, Aug-Jan Q Spring/Summer 237 157-329 Mean flow, Aug-Oct Q Spring 190 86-348 Mean flow, Nov-Jan Q Summer 283 186-456 Mean flow, Aug Q Aug 146 80-293 Mean flow, Sep Q Sep 181 71-570 Mean flow, Oct Q Oct 243 106-493 Mean flow, Nov Q Nov 282 131-457 Mean flow, Dec Q Dec 306 160-589 Mean flow, Jan Q Jan 262 148-416 Measures of flood peaks (flows in m3/s) Maximum flow, Aug— Jan Q Spring/Summer 1,488 630-2,800 Maximum flow, Aug-Oct Q Spring 975 167-2,470 Maximum flow, Nov-Jan Q Summer 1,279 540-2,800 Maximum flow, Aug Q Aug 385 86-2,470 Maximum flow, Sep Q Sep 502 83-2,230 Maximum flow, Oct QOct Q Nov 755 133-2,030 Maximum flow, Nov 783 203-1,960 Maximum flow, Dec Q Dec 1,006 282-2,800 Maximum flow, Jan Q Jan 806 222-2,470 Mean of 10 highest flows, Aug-Jan Q lOmax 854 443-1,410 Number of days Q > 500 m3/s, Aug-Jan N 1 o00 13 3-26 Number of days Q > 1,000 m3/s, Aug-Jan N A> 1,000 2 0-8 Number of days Q > 1,500 m3/s, Aug-Jan N -^1,500 1 0-4 Measures of flow variability Mean/median, Aug-Jan Q Spring/Summer 1.33 1.04-1.73 Mean/median, Aug-Oct Q Spring 1.32 1.01-1.92 Mean/median, Nov-Jan Q Summer 1.29 1.09-1.52 Mean/median, Aug Q Aug 1.14 0.98-2.25 Mean/median, Sep Q Sep 1.15 1.00-1.91 Mean/median, Oct Qoct 1.22 0.94-1.64 Mean/median, Nov Q Nov 1.23 1.02-1.75 Mean/median, Dec Q Dec 1.27 1.04-2.11 Mean/median, Jan Q Jan 1.27 1.02-2.03 818 Fishery Bulletin 95(4), 1997 monthly means varied from 71 m3/s (Septem- ber 1977) to 734 m3/s (December 1979). The mean maximum August to January flow (effec- tively the mean annual spring and summer flood) was 1,488 m3/s; values for individual sea- sons ranged from 630 m3/s in 1980 to 2,800 m3/ s in 1979. Flow variability was lowest during August and September, although highly vari- able flows ( Q > 1.6) were recorded in all months. Hydrographs for 1977 (a low-flow season) and 1984 (an above average season) illustrate the sharp peaks and rapid recession typical of Rakaia floods (Fig. 2). In the November 1984 event, dis- charge increased by a factor of 10 (from 172 m3/s to 1,710 m3/s) over 48 h and then fell from 1,960 m3/s to 455 m3/s over 72 h. Despite the contrast between the two seasons, both hydrographs also show a common period of low and relatively stable base flows in August and September, followed by more frequent floods as the season progresses. Correlation analysis Of the 32 flow parameters listed in Table 2, 29 showed no correlation with survival rates for Glenariffe Stream chinook (Table 3). Correlation coefficients for these indices ranged from -0.223 to 0.283, none of which differed significantly from zero (P>0.16 in all cases). The three exceptions were the mean flow Q , the maximum flow Q , and the ratio of the mean to median flow Q , for the month of November. All three measures were positively correlated with survival (QN0V,P = 0.045; QN0V,P = 0.003; QN0V, P = 0.006). By contrast, there was no correlation between sur- vival and the same set of flow variables for the adja- cent months of October and December. The strongest and most consistent set of correla- tions involved the ratio of mean to median flow, cal- culated over periods of 30 to 90 d duration centered on or about November 1 (Fig. 3 A). Over much of this range the correlation between Q and log S exceeded 0.5, with a maximum value of 0.667 for Q calculated over the 40-day period from 9 October to 17 Novem- ber. More generally, survival tended to be high for broods corresponding to years when flow variability during October, November, and early December was high. For periods centered on the four weeks between Table 3 Correlations between log-transformed brood year survival rates for Glenariffe Stream chinook and 32 indices of Rakaia mainstem flows, 1965-90. See Table 2 for definitions of symbols. Asterisks denote correlations significant at the 95% level (*) and 99% level Flow parameter Period Q Q Q ^500 ^1,000 2, © o Q 10max February-January (year) -0.091 August-January (spring and summer) -0.105 0.109 0.182 -0.062 0.261 0.165 0.164 August-October (spring) -0.195 -0.140 -0.218 November-January (summer) 0.046 0.205 0.269 August -0.068 -0.054 -0.066 September -0.082 -0.191 -0.159 October -0.223 -0.053 0.283 November 0.397* 0.558** 0.520** December -0.214 -0.112 -0.088 January -0.070 0.053 0.141 Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River, New Zealand 819 18 October and 15 November, of five to nine weeks duration, the correlation be- tween Q and log S averaged 0.493. Taken as an isolated result, this correlation cor- responds to an average significance level of P < 0.01 and an average coefficient of determination (r2) of 0.243. There was some evidence of a weaker and more tran- sient period of negative correlation be- tween Q and log S earlier in the season. For Q calculated over periods of seven to nine weeks duration and centered on the fortnight from 6-19 September, the correlation with log S averaged -0.404, corresponding to an r2 of 0.16. The sym- metric upright “V” shape apparent in the contours of Fig. 3 A, centered on the be- ginning of November, is an artifact caused by the tendency for data sets rep- resenting Q over periods of similar du- ration centered on the same date to be highly correlated. Maximum flow ( Q ) and mean flow ( Q) were generally only weakly correlated with survival, and the few periods dur- ing which correlations stronger than ±0.4 were recorded showed little tendency to persist over temporal scales of more than a few days (Fig. 3, B and C). In addition, these correlations tended to become stronger at shorter time scales, suggest- ing that for Q , and possibly for Q , the cor- relations apparent in Table 3 were a tran- sient effect of a fortuitous sequence of flood events. By contrast, the persistence of a positive correlation between Q and log S for flows averaged over periods of up to 90 days suggests that the relation is much more robust and hence likely to be of biological significance. The correla- tions between Q , Q , and Q are also con- sistent with this interpretation. Whereas Qnov and Qnov were highly correlated (r=0.84), Qnqv was only moderately cor- related with both parameters (r=0.61 in both cases), confirming that Q captured information on flow variation not mea- sured by either Q or Q. The relation between Q and log S was moderately influenced by the 1973 brood (for which the survival rate was unusu- ally high) but was not dependent on it. The 1965, 1974, 1982, and 1983 broods (for which survival was also relatively high) and the 1985, 1986, and 1988 A B C Figure 3 Correlations between log-transformed survival data for Glenariffe Stream chinook salmon and Rakaia River mainstem discharge expressed as the ratio of mean to median discharge (A), maximum discharge (B), and mean discharge (C), as a function of the period used to calculate each flow statistic. For each point, the locations on the x- and y-axes represent the mid-point of the period, and the duration, respectively. The points “a” and “b” correspond to the two intervals represented in Figure 4. 820 Fishery Bulletin 95(4), 1997 broods (for which survival was low) conform to the general trend irrespective of the duration of the pe- riod used to calculate Q (Fig. 4). With the exception of the 1969 brood, the value of Q during October and November was not highly sensitive to the choice of interval. Discussion Correlations and flow parameters This study shows that, although fry-to-adult survival of Glenariffe Stream chinook salmon is correlated with the occurrence of springtime flood events in the Rakaia River, both the magnitude and the direction of the observed correlation depend strongly on the flow parameter used and the period over which this parameter is calculated. Survival was most strongly correlated with flow variability (as measured by the ratio of mean to median flow), the correlation being moderately negative for flows averaged over the pe- riod from mid-August to mid-October, and rather more strongly positive from mid-October to Novem- ber. Similar but weaker correlations were apparent between survival and maximum flow, but mean flow was a poor predictor of survival irrespective of the time interval used. The correlations reported here have two key fea- tures. First, although quite strong in a biological con- text, they are nevertheless relatively weak, account- ing for at most 25-30% of the observed variation in log survival. Even if the maximum positive correla- tion (r=0.667) is taken at face value, its predictive power allows years to be categorized only as “above average” or “below average,” at best (Prairie, 1996). This result is consistent with an earlier finding that annual survival rates for New Zealand chinook are primarily determined by marine rather than fresh- water influences (Unwin, 1997; see also Bradford, 1995). Second, the tendency for survival to be posi- tively correlated with flow variation but uncorrelated with mean flow suggests that increased flow vari- ability (in the sense illustrated in Fig. 5) at the ap- propriate time of year is beneficial to survival. This is in sharp contrast to the generally held view that spring floods have a detrimental impact on chinook fry in the Rakaia and other New Zealand rivers (Waugh, 1980; McDowall, 1990; Flain1). The three key flow parameters used in this study — mean flow, maximum flow, and ratio of mean flow to median flow — -are by no means the only ones pos- sible and can only capture some of the information contained in the hydrograph for a given time period. Mean flow essentially measures the total volume of water passing through the river over the interval in question, without conveying any information about the magnitude or distribution of floods. For example, a 90-day mean of 200 m3/s could arise from 90 con- secutive days at exactly 200 m3/s each, or from 89 days at 180 m3/s and one day at 1,980 m3/s. Maximum flow characterizes peak flood intensity, but not flow vari- ability, so that a flow period with one 2,000 m3/s flood will outrank another period with ten 1,900 m3/s floods. Mean flow/median flow Figure 4 The relation between log-transformed survival data for Glenariffe Stream chinook salmon and the ratio of mean to median discharge in the Rakaia River mainstem for two periods near local maxima in Figure 3A. Twelve points corresponding to broods with extreme survival or flow indices for at least one of the two periods shown are identified by year. Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River. New Zealand 821 1973 I I 1974 Sep Oct Nov 1983 Sep Oct Nov Figure 5 Hydrographs for the Rakaia River from September to November for three years of high mean flow to median flow (as calculated over the period indicated by dashed lines) and high survival (1973, 1974, 1983) and three years of low mean flow to median flow and low survival (1985, 1986, 1988). All hydrographs are drawn to a common vertical scale of 0- 2,000 m3/s. The ratio of mean to median flow is a direct mea- sure of flow variability, but it does not necessarily follow that low values of Q correspond to low and stable flows. Flow regimes where Q is close to one (i.e. mean flow differs little from the median) can arise in two different ways, only one of which corre- sponds to an extended period of low flows (Fig. 5). The other possibility is a period of highly variable flows superimposed on a high base flow, so that even though flows vary greatly from day to day, the flow distribution is only moderately skewed. By contrast, high values of Q are more consistently associated with highly variable flows, tending to correspond to intervals when a lengthy period of low, stable base flows is punctuated by a relatively small number of short and sharp flood events. Both the 1973 and 1974 seasons ( Qgoct-nNov^-^ and 1.55, respectively) were characterized by low and stable base flows dur- ing September, followed by two or three relatively short-lived flood events during October and Novem- ber. The 1985 and 1986 seasons ( Q9Oct-i7Nov = l-00 in both cases) were characterized by uniformly low and stable flows, with no floods over the six weeks from 1 October. The effect of high base flows on flow vari- ability is illustrated by comparing the 1983 and 1988 seasons ( Q9oct-i7Nov = l-46 and 1.15, respectively). Whereas base flows were low throughout 1983, so that a 2,000 m3/s event in late October generated a high figure for Q , the absence of a major flood dur- ing October and November 1988 produced a lower figure for Q despite high and rapidly fluctuating base flows over most of the period shown. The other flow parameters examined during the initial phase of this study, such as the number of days from August to January when flows exceeded a cer- tain level, showed no correlation with survival. Al- though in principle it would have been possible to subject these parameters to the same level of analy- sis used for Q , Q , and Q , this analysis becomes progressively less meaningful at shorter time scales. For example, given that Nx 000 ranged from 0 to 8 over a six month period (Table 2), the same statistic calculated over monthly intervals would typically take on only a few discrete integer values and would be inappropriate for correlation analysis. Mechanisms The topography of the correlations between Q and survival, as illustrated by the qualitative features of Fig. 3 A, coincides to a striking degree with several key events in the migration patterns of juvenile ocean-type fry in the Rakaia River. The period of highest positive correlation between flow variability 822 Fishery Bulletin 95(4), 1997 and survival, mid-October to November, coincides with the period when fingerlings first become abun- dant in the lower river and begin their transition to oceanic waters. The period centered on mid-Septem- ber, when there is some evidence of a negative corre- lation between flow variability and survival, corre- sponds to the earlier time when fry are still migrat- ing downriver and have yet to grow to the point where they are able to withstand the transition to salt wa- ter. Survival is not correlated, either positively or negatively, with flow variability at the beginning of the season (August, when most fry have yet to hatch) or at the end of the season (January, by which time most fingerlings have left the river). The correlation also tends to disappear when flow variability is av- eraged over more than about 100 days, a period that is consistent with the 90-day freshwater residence period of Rakaia fingerlings. The tendency for survival to be positively corre- lated with flow variability rather than flow volume (as measured by Q), and the short duration of each individual flood peak, suggest that these flood pulses must be an integral part of any plausible linking mechanism. Maximum survival appears to result when stable flows prior to mid-October are followed by a few (perhaps two or three) large floods over the next four to six weeks. By contrast, seasons when there are no major floods during October and No- vember seem to result in poor survival irrespective of flows earlier in the season. Although any discus- sion based solely on the present results is specula- tive, I suggest that sudden increases in Rakaia dis- charge coinciding with peak emigration of fingerlings from the river mouth may increase survival by buff- ering the transition from fresh to saline waters in the vicinity of the offshore plume. If so, these pulses may help to compensate for the lack of an estuary at the Rakaia mouth and the low importance of the la- goon as a salmon rearing habitat,2 one of the key features distinguishing the Rakaia (and other New Zealand salmon-producing rivers) from the extensive tidal basins below the Sacramento River mouth (Kjelson et al., 1982). The effect may be compounded by the well-documented tendency for downstream migration rates to increase with flow, both in New Zealand (Irvine, 1986) and North American popula- tions (e.g. Kjelson et al., 1982; Berggren and Filardo, 1993), so that each flood pulse increases both the number of fingerlings leaving the river and their prospects for successful acclimation to salt water. Outflow of turbid flood waters may also increase sur- vival by reducing visibility, and hence decreasing losses to inshore marine predators such as kahawai (. Arripis trutta), although reduction in visibility is likely to be no more than a secondary effect (cf. St. John et al., 1992). A positive correlation between survival of hatchery-reared Atlantic salmon ( Salmo salar ) and maximum river discharge during the first seven days after release has been attributed to re- duced predation at higher flows (Hvidsten and Han- sen, 1988). A distinguishing feature of chinook salmon com- pared with other species of Oncorhynchus is their gradual acquisition of seawater tolerance while still in fresh water, without the sudden transition associ- ated with smoltification in species such as coho (O. kisutch) or steelhead (O. mykiss) (Hoar, 1976). By early November, Rakaia fingerlings are of an age and size close to the generally accepted minima for suc- cessful transfer to seawater (Clarke and Shelbourn, 1985; Franklin et al., 1992), and water temperatures in the Rakaia River (12-15°C; Unwin, 1986) and at sea ( 12-14°C )8 appear to be within the optimal range for juvenile chinook salmon reported by these stud- ies. However, acclimation to seawater also depends on the rate of transition. A gradual transfer to saline waters allows even very young fish to acclimatize successfully (Hoar, 1976), as evidenced by the abun- dance of chinook fry in low salinity estuarine waters in North America (Reimers, 1973; Healey, 1980; Levy and Northcote, 1981), including the lower Sacra- mento River (Kjelson et al., 1982). Changes in es- tuarine ecology during low-flow seasons in the Snake- Columbia River system have been suggested as a factor contributing to reduced survival of Snake River chinook (Williams and Matthews, 1995). In the Strait of Georgia, where the Fraser River plume forms a well-developed halocline at a depth of 5-10 m, juve- nile salmonids showed a preference for surface wa- ters of low salinity (10-15 ppt) in the plume, com- pared with the more saline waters (25-30 ppt) in other regions of the Strait (St. John et al., 1992). The Georgia Strait study also reported a tendency for plankton and larval fish to align with the plume boundary, providing enhanced feeding opportunities. There have been no studies on salinity gradients off the Rakaia mouth, but nearshore salinity off Otago Harbour (on the east coast of the South Is- land 220 km south of the Rakaia) is inversely corre- lated with discharge from the Clutha River a fur- ther 100 km to the south (Jillett, 1969). The coastal shelf off the Rakaia River has a very gentle slope, with the 20-m isobath 5-10 km offshore (see Fig. 1). Consequently, peak Rakaia outflows should have a substantial impact on inshore salinity. For example, a daily mean discharge of 1 200 m3/s over 24 h (which is less than the mean annual spring flood) represents a total volume of fresh water of 0. 1 km3, equivalent 8 1997. NIWA, Christchurch, New Zealand. Unpubl. data. Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River, New Zealand 823 to a 10 km2 surface layer that is 10 m deep. Bevolve- ment of the resulting halocline would presumably require considerable input of wave energy, and un- der calm conditions, a surface layer of low salinity water may persist for some days until dispersed to the north-east under the influence of the Southland current (Heath, 1972). Implications The analysis described in this study relates only to ocean-type fry. Because ocean-type fish make up ap- proximately two thirds of the Rakaia adult popula- tion (Quinn and Unwin, 1993), fluctuations in their abundance would have a significant impact on brood year survival. However, one third of Rakaia adult chinook have a stream-type juvenile life history, and there is some evidence that the emergence of stream- type behavior in what was originally a fall-run stock represents an adaptive response to the lack of es- tuarine waters on New Zealand’s salmon-producing rivers (Unwin and Glova, 1997). The freshwater habi- tat preferences and migration patterns of stream- type chinook in the Rakaia River are poorly under- stood, and their sensitivity to flow variations is un- known. However, the possibility that survival of ocean-type fry may increase in seasons of variable flow could provide a compensating selective force that would help to establish a balance between the incidence of the two phenotypes. If so, the ratio of ocean- to stream-type fry in any one annual cohort should also be positively correlated with the variability of spring flows into the Rakaia River during the first year of life. To explore this hypothesis further, and hence to provide an independent test of the plausibility of the mechanisms outlined in the previous section, I ex- amined archival material from NIWA’s scale collec- tions, aspects of which are summarized in Tables 2, 3, and 4 of Quinn and Unwin (1993). These records include scale samples from salmon taken in the Rakaia sports fishery for nine annual cohorts be- tween 1965 and 1984. These scale samples permit- ted returning fish from each cohort to be classified according to juvenile life history. The incidence of ocean-type fish among 3-year-old adult chinook was positively correlated with mainstem flow variability over the period 9 October to 17 November (r2=0.54, P=0.Q25; Fig. 6) during the year in which they emi- grated as juveniles. Of particular interest is the high incidence of ocean-type fish in the 1973 and 1982 cohorts, which were also the broods for which sur- vival was highest. With regard to the New Zealand salmon fishery, which is managed purely for recreational anglers 90 n 80- JD O' CO i 70- 03 <1) f 60- CD O O 50- 40- 1. Mean flow/median flow Figure 6 The relation between the incidence of ocean-type adults among 3-year-old chinook salmon caught by Rakaia River anglers over nine seasons from 1965 to 1984, and the ratio of mean to median discharge in the Rakaia mainstem for the corresponding juvenile cohort, for the same time interval (9 October to 17 November) as that represented in Figure 4B. Sample sizes averaged 127, and ranged from 36 to 373. 73 + 80 + I I I 1 1 1 ,0 1.1 1.2 1.3 1.4 1.5 1.6 (McDowall, 1994b), the main conclusion to be drawn from this study is that despite the observed correla- tions between survival and flow variability, interan- nual variation in survival remains at best weakly predictable. However, spring flow variability is easy to analyze on a season by season basis, and the pos- sibility of making at least a broad prediction identi- fying years of high or low survival up to two seasons in advance suggests the effort is worth making. The New Zealand Fish and Game Council (which is re- sponsible for management of the sports fishery) have recently instituted annual monitoring programs for key spawning runs in all major salmon rivers,9 and these will eventually enable comparisons between survival and flows to be made for other major east coast catchments. My results also suggest that any reduction in flow variability resulting from develop- ment of storage impoundments for hydroelectric or irrigation purposes would have a significant nega- tive impact on the fishery over and above any losses caused by barriers to upstream passage. The high bed load of major braided rivers such as the Rakaia makes them unattractive candidates for impound- ment, but declines in salmon runs in two other riv- 9 Webb, M. 1997. Central South Island Fish and Game Coun- cil, PO Box 150, Temuka, New Zealand. Personal commun. 824 Fishery Bulletin 95(4), 1 997 ers (the Clutha and Waitaki) following hydroelectric development (McDowall, 1990) may be partly linked to a reduction in the magnitude and frequency of spring floods. A positive association between salmon production and large offshore plumes is also consis- tent with the general distribution of salmon in east coast rivers (McDowall, 1990), with the largest popu- lations confined to major rivers draining the main divide. Traditionally this distribution has been at- tributed to the presence of stable headwater spawn- ing tributaries such as Glenariffe Stream, but this explanation is not fully convincing. Many minor east coast rivers support self-sustaining stocks of brown trout (Jowett, 1990; McDowall, 1990), and spawning requirements for chinook are not dissimilar. From an evolutionary standpoint, the present re- sults help to shed further light on the processes by which chinook salmon have been able to succeed in New Zealand waters. In addition to the emergence of stream-type fish as a significant component of modern New Zealand stocks, several other recent studies have noted differences between present-day New Zealand and Sacramento chinook at both phe- notypic and genotypic levels (Quinn and Unwin, 1993; Quinn et al., 1996), suggesting that the pro- cess of adaptation may be ongoing. The somewhat unusual mechanisms which have apparently enabled New Zealand chinook salmon to establish the only self-perpetuating stocks outside their native range (Harache, 1992) underscore the great phenotypic plasticity of the species, and the value of the New Zealand populations as a laboratory for studying this plasticity. Acknowledgments I thank all staff associated with the Glenariffe programme for technical assistance; Greg Kelly and Mark Weatherhead for help with drafting the fig- ures; Bob McDowall, Charles Pearson, Ian Jowett, Tom Quinn, and three anonymous referees for com- ments on earlier drafts of the manuscript. Kathy Walter and Graeme Davenport extracted Rakaia River flow data from the National Water Resources Archive, and the study was funded by the Founda- tion for Research, Science and Technology through Contracts C01417 and CO1501. Literature cited Beamish, R. J., M. N. Chrys-Ellen, B. L. Thomson, P. J. Harrison, and M. St. John. 1994. A relationship between Fraser River discharge and interannual production of Pacific salmon ( Oncorhynchus spp.) and Pacific herring ( Clupea pallasi) in the Strait of Georgia. Can. J. Fish. Aquat. Sci. 51:2843-2855. Berggren, T. J., and M. J. Filardo. 1993. An analysis of variables influencing the migration of juvenile salmonids in the Columbia River basin. N. Am. J. Fish. Manage. 13:48-63. Biggs, B.J. 1995. The contribution of flood disturbance, catchment ge- ology and land use to the habitat template of periphyton in stream ecosystems. Freshwater Biol. 33:419-438. Bradford, M. J. 1994. Trends in the abundance of chinook salmon ( Onco- rhynchus tshawytscha ) of the Nechako River, British Columbia. Can. J. Fish. Aquat. Sci. 51:965-973. 1995. Comparative review of Pacific salmon survival rates. Can. J. Fish. Aquat. Sci. 52:1327-1338. Clarke, W. C., and J. E. Shelbourn. 1985. Growth and development of seawater adaptability by juvenile fall chinook salmon ( Oncorhynchus tshawytscha) in relation to temperature. Aquaculture 45:21-31. Clausen, B., and B. J. F. Biggs. In press. Relationships between benthic biota and flow indices in New Zealand rivers. Freshwater Biol. Davis, S. F., and M. J. Unwin. 1989. Freshwater life history of chinook salmon ( Onco- rhynchus tshawytscha) in the Rangitata River catchment, New Zealand. N.Z. J. Mar. Freshwater Res. 23:311-319. Franklin, C. E., W. Davison, and M. E. Forster. 1992. Seawater adaptability of New Zealand’s sockeye (Oncorhynchus nerka) and chinook salmon (O. tshawy- tscha): physiological correlates of smoltification and sea- water survival. Aquaculture 102:127-142. Gilbert, C. H. 1913. Age at maturity of the Pacific coast salmon of the genus Oncorhynchus. Bull. U.S. Bur. Fish. 32:1-23. Glova, G. J., and M. J. Duncan. 1985. Potential effects of reduced flows on fish habitats in a large braided river, New Zealand. Trans. Am. Fish. Soc. 114:165-181. Harache, Y. 1992. Pacific salmon in Atlantic waters. ICES Mar. Sci. Symp. 194:31-55. Healey, M. C. 1980. Utilization of the Nanaimo River estuary by juvenile chinook salmon, Oncorhynchus tshawytscha. Fish. Bull. 77:653-668. 1983. Coastwide distribution and ocean migration patterns of stream- and ocean-type chinook salmon, Oncorhynchus tshawytscha. Can. Field-Nat. 97(41:427-433. 1991. Life history of chinook salmon (Oncorhynchus tshawytscha). In C. Groot and L. Margolis (eds.) Pacific salmon life histories, p. 311-393. Univ. British Colum- bia Press, Vancouver, B.C. Heath, R. A. 1972. The Southland current. N.Z. J. Mar. Freshwater Res. 6:497-534. Hoar, W. S. 1976. Smolt transformation: evolution, behavior, and physiology. J. Fish. Res. Board Can. 33:1234—1252. Hopkins, C. L., and M. J. Unwin. 1987. River residence of juvenile chinook salmon ( Onco- rhynchus tshawytscha) in the Rakaia River, South Island, New Zealand. N.Z. J. Mar. Freshwater Res. 21:163-174. Hvidsten, N.A., and L.P. Hansen. 1988. Increased recapture rate of adult Atlantic salmon, Unwin: Survival of Chinook salmon in relation to spring and summer mainstem flows of the Rakaia River, New Zealand 825 Salmo salar L., stocked as smolts at high water dis- charge. J. Fish Biol. 32:153-154. Ibbitt, R. P. 1979. Flow estimation in an unstable river illustrated on the Rakaia River for the period 1958-1978. J. Hydrol. N.Z. 18:88-108. Irvine, J. R. 1986. Effects of varying discharge on the downstream move- ment of salmon fry, Oncorhynchus tshawytscha Wal- baum. J. Fish. Biol. 28:17-28. James, G.D., and M.J. Unwin. 1996. Diet of Chinook salmon Oncorhynchus tshawytscha in Canterbury coastal waters. New Zealand. N.Z. J. Mar. Freshwater Res. 30:69-78. Jillett, J. B. 1969. Seasonal hydrography of waters off the Otago Pen- insula, south-eastern New Zealand. N.Z. J. Mar. Fresh- water Res. 3:349-375. Jowett, I. G. 1990. Factors related to the distribution and abundance of brown and rainbow trout in New Zealand clear-water rivers. N.Z. J. Mar. Freshwater Res. 24:429-440. Jowett, I. G., and M. J. Duncan. 1990. Flow variability in New Zealand rivers and its rela- tionship to in-stream habitat and biota. N.Z. J. Mar. Freshwater Res. 24:305-317. Keckler, D. 1994. SURFER for Windows user’s guide. Golden Soft- ware, Inc., Golden, CO. Kjelson, M., P. S. Raquel, and F. W. Fisher. 1982. Life history of fall-run juvenile chinook salmon, Oncorhynchus tshawytscha, in the Sacramento-San Joaquin estuary, California. In V. S. Kennedy (ed. ), Es- tuarine comparisons, p. 393-412. Academic Press, New York, NY. Kope, R. G., and L. W. Botsford. 1990. Determination of factors affecting recruitment of chinook salmon Oncorhynchus tshawytscha in central California. Fish. Bull. 88:257-269. Levy, D. A., and T. G. Northcote. 1981. The distribution and abundance of juvenile salmon in marsh habitats of the Fraser River estuary. Westwater Research Centre Technical Report 25, Univ. British Co- lumbia, Vancouver, 117 p. Lister, D. B., and C. E. Walker. 1966. The effect of flow control on freshwater survival of chum, coho, and chinook salmon in the Big Qualicum River. Can. Fish. Cult. 37:3-25. McDowall, R. M. 1990. New Zealand freshwater fishes: a natural history and guide. Heinmann Reed, Auckland, 553 p. 1994a. The origins of New Zealand’s stocks of chinook salmon. Mar. Fish. Rev. 56:1-7. 1994b. Gamekeepers for the nation. Canterbury Univ. Press, Christchurch, 508 p. Montgomery, D. R., J. M. Buffington, N. P. Peterson, D. Schuett-Hames, and T. P. Quinn. 1996. Stream-bed scour, egg burial depths, and the influ- ence of salmonid spawning on bed surface mobility and embryo survival. Can. J. Fish. Aquat. Sci. 53:1061-1070. Prairie, Y. T. 1996. Evaluating the predictive power of regression models. Can. J. Fish. Aquat. Sci. 53:490-492. Quinn, T. P., and M. J. Unwin. 1993. Life history patterns of New Zealand chinook salmon (Oncorhynchus tshawytscha) populations. Can. J. Fish. Aquat. Sci. 50:1414-1421. Quinn, T. P., J. L. Nielsen, C. Gan, M. J. Unwin, R. L. Wilmot, C. M. Guthrie, and F. M. Utter. 1996. Origin and genetic structure of chinook salmon (Oncorhynchus tshawytscha ) transplanted to New Zea- land. Fish. Bull. 94:506-521. Reimers, P. E. 1973. The length of residence of juvenile fall chinook salmon in Sixes River, Oregon. Res. Rep. Fish Comm. Oregon 4(2), 42 p. Sagar, P. M. 1986. The effects of floods on the invertebrate fauna of a large, unstable braided river. N.Z. J. Mar. Freshwater Res. 20:37-46. Sagar, P. M., and G. J. Glova. 1987. Prey preferences of a riverine population of juvenile chinook salmon, Oncorhynchus tshawytscha (Walbaum). J. Fish. Biol. 31:661-673. St. John, M. A., J. S. Macdonald, P. J. Harrison, R. J. Beamish, and E. Choromanski. 1992. The Fraser River plume: some preliminary observa- tions on the distribution of juvenile salmon, herring, and their prey. Fish. Oceanogr. 1:153-162. Saltveit, S. J., T. Bremnes, and O. R. Lindas. 1995. Effect of sudden increase in discharge in a large river on newly emerged Atlantic salmon (Salmo salar ) and brown trout (Salmo trutta) fry. Ecol. Freshwater Fish 4(4):168- 174. Unwin, M. J. 1986. Stream residence time, size characteristics, and mi- gration patterns of juvenile chinook salmon (Oncorhynchus tshawytscha) from a tributary of the Rakaia River, New Zealand. N.Z. J. Mar. Freshwater Res. 20:231-252. 1997. Fry-to-adult survival of naturally and hatchery pro- duced chinook salmon (Oncorhynchus tshawytscha) from a common origin. Can. J. Fish. Aquat. Sci. 54:1246-1254. Unwin, M. J., and G. J. Glova. 1997. Changes in life history parameters in a wild popula- tion of chinook salmon (Oncorhynchus tshawytscha) asso- ciated with releases of hatchery-reared fish. Can. J. Fish. Aquat. Sci. 54:1235-1245. Unwin, M. J., and D. H. Lucas. 1993. Scale characteristics of wild and hatchery chinook salmon (Oncorhynchus tshawytscha) in the Rakaia River, New Zealand, and their use in stock identification. Can. J. Fish. Aquat. Sci. 50:2475-2484. Walters, C. J., and J. S. Collie. 1988. Is research on environmental factors useful to fish- eries management? Can. J. Fish. Aquat. Sci. 45:1848- 1854. Waugh, J. D. 1980. Salmon in New Zealand. In J. E. Thorpe (ed.), Salmon ranching, p. 277-303. Academic Press, New York, NY. Wilkinson, L. 1996. SYSTAT 6.0 for Windows, 4 vols. SPSS Inc., Chi- cago, IL, var. pag. Williams, J. G., and G. M. Matthews. 1995. A review of flow and survival relationships for spring and summer chinook salmon, Oncorhynchus tshawytscha, from the Snake River Basin. Fish. Bull. 93:732-740. 826 Abstract.™ From the mid-1970’s to the mid-80’s, Stellwagen Bank was an important humpback whale feeding area with sand lance (Ammodytes spp.) as the major prey. Between 1988 and 1994, however, the number of hump- back whales we identified each year on Stellwagen declined from a high of 258 (1990) to 7 (1994), and the mean num- ber of whales identified per day fell from 17.7 (1988) to 0.9 (1994). Adult whales decreased steadily after 1988; juveniles decreased rapidly after 1991. Echo-sounder data from Stellwagen showed that prey trace levels declined from 19.1% of the vertical water column in 1990 to 2.8% in 1992 (no readings were taken in 1988-89, or 1993-94). Si- multaneously, the number of whales identified on Jeffreys Ledge, north of Stellwagen Bank, increased dramati- cally beginning in 1992. Sixty-four per- cent of the whales identified on Jeffreys in 1992-94 were seen on Stellwagen Bank in 1988 and 1989. We hypothesize that humpback whales shift their dis- tribution in order to prey upon recov- ering herring populations, their pri- mary source of food. Manuscript accepted 21 April 1997. Fishery Bulletin 95:826-836 (1997). A shift in distribution of humpback whales, Megaptera novaeangliae, in response to prey in the southern Gulf of Maine Mason Weinrich IVSaScoIm Martin* Rachel Griffiths Jennifer Bove Mark Schilling Cetacean Research Unit, PO Box 1 59 Gloucester, Massachusetts 0 1 930 'Present address: Department of Biology Rutgers University, Brunswick, New Jersey E-mail address (for M. Weinrich): mason@cetacean.org Humpback whales, Megaptera novae- angliae, migrate seasonally between low-latitude breeding grounds and high-latitude feeding areas (Kellogg, 1929; Mackintosh, 1965; Katona, 1986). In the western North Atlan- tic, whales that winter in Caribbean waters migrate to feeding grounds in New England (the Gulf of Maine), in the Gulf of St. Lawrence, and in waters off Newfoundland, Green- land, Iceland, and Norway (Katona and Beard, 1990). The whales us- ing each feeding area appear to con- sist of extended matrilines (Baker et al., 1990; Clapham et al., 1992). Within feeding areas, prey distribu- tion has been a primary influence on the local distribution and micro- movements of all baleen whales ex- amined to date (Whitehead and Carscadden, 1985; Payne et al., 1986, 1990; Piatt et al., 1989). Studies of humpback whale move- ment, ecology, demography, behavior, and social organization on their feeding grounds in the Gulf of Maine have been ongoing since the mid-1970’s, (Payne et al., 1986; Clapham and Mayo, 1987, 1990; Weinrich, 1991; Weinrich and Kuhl- berg, 1991; Clapham et al., 1992; Weinrich et al., 1992; Katona et al.1). During this period, several shifts in the distribution of hump- back whales have been reported. Payne et al. (1986) showed that humpback whales in the late 1970’s had moved from primary abun- dance on Georges Bank and in the waters of the northern Gulf of Maine to the inshore southwestern Gulf of Maine, especially Stell- wagen Bank and the Great South Channel. They attributed this shift to a fishery-induced collapse of her- ring (Clupea harengus) populations (Anthony and Waring, 1980; Gross- lein et al., 1980) and a correspond- ing increase in sand lance ( Ammo- dytes spp.) (Meyer et al., 1979; Sherman et al., 1981, 1988; Sher- man 1986; Sissenwine 1986). Both species are known prey for hump- back whales (Mitchell, 1973; Over- holtz and Nicholas, 1979; Kawa- mura, 1980). These fish species are potential ecological competitors (Reay, 1970; Meyer et al., 1979; Sherman et al., 1981); moreover, herring are known predators of 1 Katona, S. K., P. Harcourt, J. S. Perkins, and S. D. Kraus. 1980. Humpback whales: a catalog of individuals identified by fluke photographs. College of the At- lantic, Bar Harbor, ME, var. pagination. Weinrich et al.: A shift in distribution of humpback whales, Megaptera novaeangliae, in reponse to prey 827 sand lance (Fogarty et al., 1991). Sight- ings of humpback whales off the Maine coast, where herring were the primary whale prey, decreased dramatically dur- ing the late 1970’s (Payne et al., 1986; Mullane and Rivers2 ). Sand lance fre- quently use shallow areas with sandy bottoms, such as Stellwagen Bank in the southern Gulf of Maine (Meyer et al., 1979). This shift in distribution, and corresponding change in primary prey type, may have also led to changes in feeding behavior (Weinrich et al., 1992). Humpback whales remained abundant in the southwestern Gulf of Maine throughout the 1980’s, with a brief de- crease in some areas during 1986-87 (Payne et al., 1990; Cetacean Res. Unit3). We documented a gradual but con- tinuous decrease in the use of Stell- wagen Bank by humpback whales dur- ing 1988-94. Our data suggest that whales have returned to a distribution similar to that documented until the late 1970’s. We hypothesize that this re- turn is due to the recovery of herring stocks in the Gulf of Maine and to a cor- responding decrease in available prey for humpback whales on Stellwagen Bank and in other areas favored by sand lance in the southwestern Gulf of Maine. Methods Survey methods From 1 May to 30 October, 1988 to 1994, daily ship- board surveys were carried out aboard commercial whale- watching boats. These departed from Gloucester and Boston, Massachusetts, and were typically 4-5 hours in duration. There were usually two cruises per vessel per day. A typical cruise included 90-120 min- utes in areas where whales were often observed, as well as 2-3 hours of transit time. Whale watches usually emphasized the northern half of Stellwagen Bank. On occasion, whale watches surveyed the southern half of Jeffreys Ledge to the northeast of 2 Mullane, S. J., and A. Rivers. 1982. Mt. Desert Rock, Maine. Annual Report, 27 p. [Available from Allied Whale, College of the Atlantic, Bar Harbor, ME.] 3 Cetacean Research Unit. 1980-89. Cetacean Research Unit, PO Box 159, Gloucester MA 01930. Unpubl. data. Figure 1 The study area in the Gulf of Maine. Cape Ann (Fig. 1). This effort is detailed in Table 1. Within each whale-watching trip, protocol and typi- cal amount of observation time were consistent on all vessels. Whale-watching cruises were supplemented by oc- casional day-long ( 7-13 h) excursions on research ves- sels. These took place 1 April to 15 November of each year, with emphasis on April and October-Novem- ber, as well as during periods of significant whale concentration from May to September. During each cruise, a specific attempt was made to conduct a com- prehensive photo-identification survey of a specific area (i.e. northern Stellwagen Bank, southern Jeffreys Ledge, etc.). As time allowed, coverage was devoted to a larger portion of the entire geographic feature (either Stellwagen Bank or Jeffreys Ledge). Specific areas were determined by recent sightings of whale aggregations, reliable reports of whale sightings from local boaters, or a determination that an area had not been recently surveyed. Jeffreys 828 Fishery Bulletin 95(4), 1997 Table 1 Study effort by both number of survey days and number of survey trips for both Stellwagen Bank and Jeffreys Ledge. “JLSN days” represent the total number of survey days represented by the Jeffreys Ledge Sighting Network (JLSN), established after the 1992 season (see text for fur- ther details). Year Stellwagen days Stellwagen trips Jeffreys days Jeffreys trips JLSN days 1988 145 558 16 44 0 1989 151 550 17 20 0 1990 166 516 32 37 0 1991 160 460 31 36 0 1992 171 506 34 37 69 1993 106 364 48 79 119 1994 86 141 86 141 138 Ledge was the destination for just under half of the dedicated cruises from 1988 to 1992, all but four in 1993, and all but two in 1994. Beginning in 1990, sighting and photo-identifica- tion data were also collected from a whale-watching boat operating out of Kennebunk, Maine, to obtain information from the northern end of Jeffreys Ledge. Observer coverage was for one trip per day, 3-5 days per week. Because of the unusually large number of whales first observed on our dedicated cruises to Jeffreys Ledge in 1992, a photo-identification net- work (consisting of three whale-watching boats work- ing on Jeffreys Ledge for one trip per day) was for- malized in fall 1992 (after the completion of field ef- forts), and existing 1992 data were obtained. Begin- ning in 1993, data collection from these vessels was standardized to be directly comparable with Stell- wagen Bank whale-watching data. Because 1993 represented the first year in which Jeffreys Ledge data were collected in any kind of standardized fashion, oc- currence and occupancy (defined below) were not cal- culated for Jeffreys Ledge humpback whale sightings. Study areas Stellwagen Bank, now a National Marine Sanctu- ary, is a sandy glacial deposit approximately 32 km long with depths from 18 to 37 m (Fig. 1). It borders the eastern margin of Massachusetts Bay and is lo- cated approximately halfway between Cape Ann and Cape Cod, Massachusetts. Jeffreys Ledge is a more complex, winding, shallow ledge, with typical depths of 45 to 61 m and with a length of approximately 54 km. Its substrate is a mixture of rocky and muddy bottoms. The southern edge of Jeffreys Ledge is 9 km northeast of Rockport, Massachusetts, whereas the northern end lies 36 km east of York, Maine. Stellwagen Bank and Jeffreys Ledge are separated by 21.6 km at their closest point. Field methods Individual humpback whales were identified from photographs of distinctive pigment patterns on the ventral surface of their tail flukes or from the shape of and scarring on the dorsal fin (or by both features) (Katona and Whitehead, 1981). Two observers col- lected data on each whale or group of whales. One observer was responsible for photographing each whale, while the second recorded the whale’s loca- tion (by means of LORAN-C), group affiliations, and behavior. This observer also recorded which photo- graphs were taken of each whale, as dictated by the photographer. Each group of whales in an area was usually observed for 1-30 minutes; most, if not all, whales in a single location (3-5 km radius) were iden- tified during each observation period. Field methods were consistent on all vessels. Age class and sex determination Individuals were identified by comparing photo- graphs with those of a catalog of humpback whales maintained at the Cetacean Research Unit (CRU), Gloucester, MA. Details on cataloging methods and contents of the catalog were given in Weinrich ( 1991), Weinrich and Kuhlberg (1991), and Weinrich et al. (1992) and are based on procedures outlined by Katona and Whitehead (1981). Whales were sexed by photographing them while belly up at the surface (and by noting the presence or absence of a small lobe immediately posterior to the genital slit [Glockner, 1983]), by observing a female with calf, or by using molecular techniques (Baker et al., 1991). Individuals were assigned to age classes (juvenile or adult) based on known age (first observation as a calf) or based on the consensus among all experienced CRU observers of an animal’s relative size at first sighting. The accuracy of the latter technique was confirmed by estimating the age class of animals of unknown iden- tity in the field and by finding that these estimates matched (photographically) animals of known age. No incorrect classifications were made (n=51). For the purposes of this paper, an animal was classified as an adult if it was known to be at least five years old, an age at which 50% or more of the population is mature (Chittleborough, 1965; Clapham and Mayo, 1990). Prey density In 1990-92, a SITEX HE-358 50-kHz echo-sounder and chart recorder aboard a whale-watching vessel Weinrich et a I.: A shift in distribution of humpback whales, Megaptera novaeangliae, in reponse to prey 829 were used to record prey density on Stellwagen Bank in the immediate area where whales had been ob- served. The echo-sounder was used for 83 days dur- ing 1990 (9 May to 20 October; 153 total hours), 98 days during 1991 (9 May to 28 September; 221 total hours), and 69 days during 1992 (24 April to 24 Oc- tober; 60 total hours). Clear readings throughout the water column (i.e. with no interference present) were obtained for 69 hours in 1990, 166 hours in 1991, and 60 hours in 1992. An echo-sounder operating at this frequency is likely to detect the presence of fish but unlikely to detect plankton unless it is present in extreme densities (Dolphin4 ). The echo-sounder was started as the boat slowed to begin whale obser- vations and turned off when the vessel left the ob- servation area to return to port. Because echo- sounder tracings were obscured by noise when the vessel was moving at cruising speed (e.g. moving from one group of whales to the next), tracings performed at cruising speed were eliminated from analysis. A timing mark was placed simultaneously on both the echo-sounder chart and the data sheets by the sec- ond observer at 10-min intervals. The echo-sounder chart was later sampled at 2- min intervals by interpolating between the 10-min marks. For each sampling point, prey presence was scored visually in 3.3 m ( 10 ft) vertical increments from the surface to the bottom, with a sliding score of zero (for no prey) to 10 (prey throughout that 3.3 m interval). From these readings mean values for vertical bait density were calculated for each quar- ter of the water column and the total water column. Mean depth in which readings were taken was 38.4 m (SD=15.1 m). No echo-sounder data were recorded on Jeffreys Ledge. Although such data give an idea of the availabil- ity of prey in the immediate vicinity of whales, they do not reflect an area where whales were not present. Hence, there could have been very similar or differ- ent prey concentrations very nearby, without that information ever being recorded. However, since each year’s data set came from numerous days and con- tained data points from several different locations (albeit within a 3-4 mile radius) within each day’s observations, we feel they at least give a crude over- view to overall prey densities in the vicinity of whales. Data management and analysis Both daily whale sighting data and prey density data were stored in PC-based computer files and analyzed with commercially available statistical software 4 Dolphin, W. F. 1994. Department of Biomechanical Engineer- ing, Boston University, Boston, MA02215. Personal commun. (SPSS/PC+, Kinnear and Gray, 1992). For daily sight- ing data, an Xbase program was written to isolate the sightings of each whale and to calculate statis- tics summarizing that individual’s within-year sight- ing history (including occurrence and occupancy — see below) in each part of the study area. These val- ues were then stored in a separate data file and ana- lyzed with the same statistical software. Temporal trends were analyzed with least-squares regression (Snedecor and Cochran, 1967) of individual data points with the year of observation as the indepen- dent variable, although only annual means are pre- sented in our tables for occurrence and occupancy scores. The slope of the regression line (B) and the probability value ( P ) from a test of the null hypoth- esis that the slope did not differ from zero are pre- sented for each test. Calves were eliminated from these analyses because we assumed that a calf is merely following the mother in her choice of habitat. Definitions “Occurrence” is defined as the number of days on which an individual whale was photographed in a single year. “Occupancy” is the number of days elapsed from the first to the last recorded sighting of an individual whale within a year. These definitions are consistent with those used by Clapham et al. (1992). Results Stellwagen Bank Total number of humpback whales identified per year The number of humpback whales identified in any single year on Stellwagen Bank ranged from 258 ( 1990) to a low of 7 ( 1994), with a mean of 153.6 (SD=88.4) ( Fig. 2). These values show a statistically significant declining trend (B=-32.82, P=0.033). When the total number of whales was broken into age class, differences in annual trends were appar- ent. Numbers of adult whales identified on Stell- wagen ranged from 173 (1990) to 3 (1994; mean= 102.8, SD=60.4). These values also showed a statis- tically significant declining trend ( B=-0.84, P=0.018). Number of juveniles identified in each year varied from 85 (1990) to 4 (1994; mean=50.71, SD=29.2). These also showed a downward trend, although not statistically significant (B=-23.50, P=0.099). The ratio of identified adult whales to identified juveniles varied from 2.5:1 (in 1988) to 0.75:1 (in 1994). Num- bers of cow-calf pairs throughout the study period 830 Fishery Bulletin 95(4), 1997 200 „ 150 0 03 _C £ ° 100 50 1988 1989 1990 1991 1992 1993 1994 Year 1.2 1 0.8 0.6 0.4 0.2 0 I j Stellwagen adults O Stellwagen juveniles ■ Jeffreys adults B Jeffreys juveniles Figure 2 The number of individual humpback whale adults and juveniles identified per year on Stellwagen Bank and Jeffreys Ledge, 1988-94. Note the rapid annual decrease of adults on Stellwagen Bank starting in 1991, and the corresponding increase on Jeffreys Ledge beginning in 1992. Juveniles started a rapid decrease on Stellwagen Bank in 1992. Stellwagen Jeffreys Total cow-calf pairs Figure 3 The number of cow-calf pairs on Stellwagen Bank, Jeffreys Ledge, and in the entire Gulf of Maine, 1988-94. The percentage of known mother-calf pairs that were sighted on Stellwagen began to decrease dramatically in 1992, the same year that the per- centage mother-calf pairs began to increase on Jeffreys Ledge. showed no significant trend in the absolute number seen on Stellwagen (B=-1.214, P= 0.424). Numbers of cows and calves began in 1991 to decline sharply, especially when com- pared with the total number of cow-calf pairs in the Gulf of Maine. By the last year of the study no cow-calf pairs were seen (Fig. 3). Occurrence and occupancy Mean occurrence of humpback whales on Stellwagen Bank within a single season ranged from 13.1 days (1989, n= 147) to 6.6 days (1993, n= 69) (Table 2; B=-0.30, P=0.501). Adults showed a within-year mean oc- currence of 6.4 days (SD=4.8, n=720), with a statistically sig- nificant declining trend through the study period (B=-1.98, P<0.001). Compared with adults, juveniles showed a higher mean within-year occur- rence (mean= 14.5 days, SD = 4.2, n= 352), which signifi- cantly increased throughout the study period (B = 1.63, P=0.030). Occupancy of individual whales within years declined significantly from a mean of 61.8 days (1989, n=147) to 21.6 days (1994, n=l) (Table 3; B=- 7.07, P=0.002). Again, age classes showed different trends. Adults had a mean occupancy period of 39.3 days (SD = 23.56, n=720) throughout the study period, with a significant declin- ing trend (B=-10.65, PcO.001). In contrast, juveniles had a mean occupancy period of 55.0 days (SD=13.21, n=352), with no significant trend apparent (B=-2.82, P=0.296). Although juveniles showed no significant trend in occu- pancy and had occurrence val- ues that actually increased throughout the period, a com- parison of median values for Weinrich et al. : A shift in distribution of humpback whales, Megaptera novaeangliae, in reponse to prey 831 Table 3 The mean occupancy (in days) of humpback whales on Stellwagen Bank, 1988-94. Year Adults Juveniles Combined total 1988 67.9 47.1 60.4 1989 56.5 71.6 61.8 1990 52.3 55.0 51.8 1991 47.5 73.1 54.6 1992 32.4 53.3 42.2 1993 17.2 48.4 25.7 1994 1.3 36.8 21.6 Table 2 The mean occurrence (in days) of humpback whales on Stellwagen Bank, 1988-94. Year Adults Juveniles Combined total 1988 13.5 7.2 11.2 1989 12.2 15.2 13.1 1990 7.9 12.5 9.6 1991 6.2 17.0 10.7 1992 7.2 19.6 12.0 1993 3.1 15.7 6.6 1994 1.3 19.8 11.9 1988 1989 1990 1991 1992 1993 1994 Year | | Stellwagen adults 1 Stellwagen juveniles CH Jeffreys whales Figure 4 The mean number of whales identified per day in each age class and year on Stellwagen Bank and Jeffreys Ledge, 1988-94. Jeffreys Ledge juveniles were not included because of their low numbers. Note the rapid decrease among adults on Stellwagen Bank beginning after 1988, and the decrease among juveniles on Stellwagen beginning in 1992. Jeffreys Ledge values were highest in the final three years, after the general decrease on Stellwagen Bank. each of these variables portrays a trend more similar to that seen from adults. From 1992 through 1994, prolonged residency of a few juveniles skewed occurrence and occupancy values. During 1991-93, median occurrence of juveniles fell from seven days to three, whereas median occu- pancy periods fell sharply, from 59.5 days to 15 days. In 1994, so few juveniles were seen (four) that the relatively high values of two individuals severely skewed the results for that year. Median values of adult occurrence and oc- cupancy showed the same trends as those portrayed from the re- gression analyses. Number of whales per day One of the clearest indicators of habitat use is the number of iden- tified humpback whales sighted on Stellwagen Bank each day. This measure incorporates two of the above components — the number of whales identified as well as how often they were sighted in the area. Throughout the study period, a mean of 12.7 (SD=11.31, n=l,072) whales were identified per day, ranging from an annual high of 17.7 (SD= 15.30, n=153 days) in 1988 to a low of 0.9 (SD=0.76, n= 97 days) in 1994 (Fig. 4). Adults and juveniles again showed different trends. Adults per day declined steadily from 14.4 in 1988 to 0.1 in 1994 (B=-2.21, P<0.001), whereas juveniles showed no clear trend, with a high of 8.8 in 1991 and a low of 0.8 in 1994 (B=-0.44, P=0.501). Juvenile values showed a clear peak in 1990- 91 as compared with other years (Fig. 4). Vertical prey density Mean overall vertical prey density decreased from 19.1% with prey traces in 1990 to 2.8% with prey traces in 1992 (B=-0.38, P<0.001) (Table 4). Similar significant decreases were seen in each vertical quarter of the water column (Table 4). Although it was impossible to determine prey type from traces alone, catches of groundfish (mainly At- lantic cod [ Gadus morhua] and haddock [ Melano - grammus aeglefinus ]), and bluefish ( Pomatomus saltatrix) in the immediate area of trace recordings 832 Fishery Bulletin 95(4), 1997 Tab!e 4 Percentage of the water column with echo-sounder prey traces by year in each quarter. Mean depth was 38.4 meters. Quarter of the water column Year Top 25% 2nd 25% 3rd 25% 4th 25% Total 1990 17.2% 15.0% 16.8% 24.3% 19.1% 1991 3.1% 4.5% 7.3% 12.7% 7.9% 1992 1.4% 0.3% 0.8% 1.1% 2.8% by party-fishing boats indicated that sand lance were the predominant fish prey in stomach contents of hump- back whales; some small mackerel ( Scomber scombrus ) and herring were also observed in stomachs in much lower frequencies. Herring were more prominent in Oc- tober during each field season, when only a small num- ber of echo-sounder data points were recorded. Jeffreys Ledge Total number of humpback whales identified The number of humpback whales we identified on Jeffreys Ledge increased from a low of 35 (in 1988) to a high of 138 (in 1992) (B=19.57, P=0.004; Fig. 2). Although there was a generally increasing trend, there was a sudden increase from 58 in 1991 to 138 in 1992. The increase among adult whales also showed a significant increase across years (B=17.25, P=0.003). Although juveniles increased steadily throughout the period, and suddenly from 1991 to 1992, they did not do so at a significant rate (B=1.357, P=0.201). (The same analysis without 1992 data, where there were an unusually high number of juveniles, does show a statistically significant increasing trend among ju- veniles LB=0.914, P=0.006]). Cow-calf pairs also showed a significantly increasing trend (B= 1.429, P=0.049). In each year, identified humpback whales on Jeffreys Ledge were biased toward adults. No more than 17 juveniles were photographed on Jeffreys Ledge in any year, and the number of juveniles photographed exceeded 10 in only a single season ( 1992). The ratio of adult to juvenile whales ranged from a high of 34.0:1 in 1988 to 7.1:1 in 1992, higher in all cases than the adult:juvenile ratios on Stellwagen Bank. Number of whales per day The mean number of whales per day ranged from a low of 2.9 (SD=1.9, n= 22) in 1989 to a high of 9.2 (SD=7.7, n=138) in 1994 (Fig. 4; B=0.98, P=0.022). In 1993 and 1994, the only years with coverage comparable to Stell- wagen Bank levels, means of 6.2 (SD=6.9, n- 116) and 9.2 (SD=7.7, n=138) whales were identified on each day of coverage, respectively. The pattern of humpback abundance on Jeffreys Ledge showed surprising seasonal consistency throughout the study. Sightings were sporadic dur- ing May, June, and early July, with few, if any, con- centrations of whales observed. In all years, concen- trations increased from late July through Septem- ber, with whales still abundant in three of the seven Octobers observed (1988, 1989, 1993). Identification comparison To determine whether the whales using Jeffreys Ledge were the same as those previously inhabiting Stellwagen Bank, we examined how many of the 210 humpback whales identified on Jeffreys Ledge in 1992-94 had been previously sighted on Stellwagen Bank. Of this group, 123 (58.5%) were photographed on Stellwagen Bank during 1988-89. When the 17 animals that had not yet been born in 1988-89 were also discounted from the Jeffreys population, 63.7% of all animals were found to have been seen previously on Stell- wagen. By comparison, only 35 ( 16.6%) of the Jeffreys Ledge whales were also seen on Stellwagen Bank during the 1992-93 period, or 16.6% of the total Jeffreys Ledge population. Discussion Humpback whales, especially adult and cow-calf pairs, decreased their use of Stellwagen Bank dras- tically between 1988 and 1994. The decreased use is reflected in decreased numbers of whales identified, decreased numbers of whales (regardless of age class) per day, and decreased adult occurrence and occu- pancy. The decline led to a virtual abandonment in 1994, when only seven humpback whales were seen on Stellwagen, and only two of those had occupancy periods longer than ten days. The decline in whale use corresponds with a decline in the amount of echo- sounder prey traces at the sites on Stellwagen Bank where whales were found over three years during the study. Although adults showed a clear decreas- ing trend on Stellwagen Bank, juvenile whales showed a less clear pattern. However, even juveniles showed a rapid decrease in use from 1991 to 1994. The increase in juvenile whales on Stellwagen Bank during 1990-91 while adult use decreased may also be a more subtle indicator of a shift in habitat quality. Previous work has shown that juvenile humpback whales are often found in areas where prey density is lower than in areas where adults predominate (Weinrich and Kuhlberg, 1991; Belt et al.5 ), and may, therefore, be considered suboptimal Weinrich et al.: A shift in distribution of humpback whales, Megaptera novaeangliae, in reponse to prey 833 habitat for the species. The vertical distribution of prey has also been reported to be different between concentration areas of the two age classes. Adults are found where prey is concentrated in the upper reaches of the water column (Belt et al.5) where a humpback whale’s bubble and cooperative feeding strategies are most effective (Hain et al., 1982; D’Vincent et al., 1985; Weinrich et al., 1992; Weinrich et al.6 ) or where foraging is most efficient because energy expenditures associated with diving are low- est (Dolphin, 1987). Juveniles appear to concentrate more often in areas where prey are predominantly subsurface, often feeding on or near the sea floor (Swingle et al., 1993; Hain et al., 1995; Belt et al5; Weinrich et al.6). In the years where juvenile use in- creased while adult use decreased (1990-91), echo- sounder data showed that prey were most concen- trated in the bottom 25% of the water column. Even within the year 1990, prey traces were found to be more common in the upper portions of the water col- umn on days when more adult whales than juveniles were present (Belt et al.5). These findings suggest that there are multiple ways of assessing habitat quality for whales. Past reports of population trends have included only the number of whales sighted per unit of effort as a guide to habitat quality (Payne et al., 1986, 1990; Piatt et al., 1989). However, indicators such as independent trends in occurrence and occupancy of individual whales, the number of individuals identified over a given time period, and even the age class of individu- als, may also be important indicators of habitat qual- ity. Although all of these measures (except the last) are factors of sightings per unit of effort, these indi- vidual components may be illuminating in detailed studies of a particular area. Prey type, for instance, could influence factors such as occurrence or occu- pancy (or both). In this case, a relatively nonmi- gratory prey species, such as sand lance (which are tied to areas of particular bottom substrate and to- pography) could lead to residency extremes (with whales staying in an area for prolonged periods or avoiding the area altogether), while a less habitat- restricted prey (such as herring) could lead to highly variable intraseason distribution patterns. 5 Belt, C. R., M. T. Weinrich, and M. R. Schilling. 1991. Effects of prey density on humpback whale (Megaptera novaeangliae) distribution in the Southern Gulf of Maine. P. 6 in Abstracts of the 9th biennial conference on the biology of marine mammals. Society for Marine Mammalogy, Chicago, IL. 6 Weinrich, M. T., C. R. Belt, M. R. Schilling, and M. E. Cappellino. 1985. Habitat use patterns as a function of age and reproductive status in humpback whales. Abstract in Abstracts of the 6th biennial conference on the biology of ma- rine mammals. Society for Marine Mammalogy, Lawrence, KS. Although the number of whales on Stellwagen Bank showed a dramatic decrease, the number of whales photographed on Jeffreys Ledge more than doubled in the last three years of the study. The cor- responding increase in observer effort during the same period no doubt had some effect on the dra- matic increase in both the number of identified indi- viduals and the mean number of whales identified per day. However, existing opportunistic data were collected following the 1992 season because of the increased use of the area suggested from our dedi- cated vessel surveys, where methods remained stan- dard across years. Further, captains of whale watch- ing boats and naturalists who had worked on Jeffreys Ledge since the mid-1980’s unanimously agreed that there was a sudden, dramatic increase in daily whale sightings beginning in 1992. Therefore, we fully be- lieve that an increase in effort is not the sole, or even the primary, cause for any increase in humpback whale numbers reported beginning in 1992. Our data show that the sudden increase in hump- back whale abundance on Jeffreys Ledge was pri- marily the result of whales seen on Stellwagen Bank earlier in the study relocating for much or all of their summer feeding season. What is perhaps more sur- prising is the relatively small number of whales that appeared in both areas during 1992 and 1993, de- spite the relative nearness of these areas to each other. Most of those whales photographed in both areas were seen on Stellwagen Bank for a brief pe- riod in October 1993, when herring stocks are known to migrate through the area (Fogarty and Clark7 ). The consistent timing of whale aggregations on Jeffreys Ledge in each year (starting in early summer) corresponds with both the major influx of herring onto the Ledge and the start of their spawning season (USDC, 1991; Fogarty and Clark7). The biomass of the Georges Bank herring population (of which this is a segment — Stephenson and Komfeld, 1990; Fogarty and Clark7) has increased dramatically over the past de- cade and, by 1991, was comparable to that of its pre- exploitation size (Stephenson and Kornfeld, 1990; Sherman, 1992; NMFS8 ). Echo-sounder data, obser- vation of surface prey, and catches of local fishing boats all indicated that herring were common on Jeffreys Ledge at the same time and location as aggregations of 7 Fogarty, M. J., and S. H. Clark. 1983. Status of herring stocks in the Gulf of Maine region for 1983. Woods Hole Laboratory Reference Document 83-46, NMFS, NOAA, 33 p. [Available from Northeast Fisheries Center, Woods Hole, MA.] 8 NMFS (National Marine Fisheries Service). 1992. Report of the thirteenth Northeast regional stock assessment workshop (13th SAW). Northeast Fisheries Science Center Document 92- 02, Northeast Fisheries Center, NMFS/NOAA, Woods Hole MA. 71 p. 834 Fishery Bulletin 95(4), 1997 whales. The area is also a primary location for seine fishing for herring off New England. Herring seiners were observed fishing or transiting to or from areas of whale aggregation daily during summers 1992-94. Although herring stocks were increasing, our data indicated that prey available for whales on Stell- wagen showed a marked decrease, corresponding to a decrease in sand lance populations throughout the Northeast ecosystem (Sherman, 1992). This decrease in prey would be expected given the documented in- verse relation between sand lance and herring or mackerel stocks, primarily due to direct predation (Fogarty et al., 1991). Although we cannot assign a definitive prey type to our echo-sounder traces from Stellwagen, the documented importance of sand lance as a prey for whales on Stellwagen Bank through observations of prey in the mouths of feeding whales (Hain et al., 1982; Weinrich et al., 1992), the direct observation of sand lance on Stellwagen Bank (Hain et al., 1995), prey in fish stomachs, and the lack of other suitable prey records throughout the years suggest that sand lance remained the predominant prey type for whales in that location. We propose that humpback whales feeding in the Gulf of Maine ecosystem have shifted from their pri- mary distribution of the mid-1970s through the late- 1980’s as a result of a shift in the abundance of avail- able prey. Although we have considered only a small portion of the Gnlf of Maine habitat, our findings correspond with other data from the same period. In the western side of the Great South Channel (an important area for whales from 1979 to 1991 where sand lance were the primary prey [Kenney and Winn, 1986; Payne et al., 1990 J) humpback whale sightings were sporadic after July 1991 (Francis9 ; Clapham10; Mattila* 11 ). OffMt. Desert Rock, Maine, where hump- backs were virtually absent throughout the 1980’s, numbers of whale sightings increased to levels far above those of the mid-1970’s (Fernald12). Surveys conducted in 1993 on Georges Bank by researchers from the YONAH (Years of the North Atlantic Hump- back) project also sighted large numbers of hump- backs, including many animals photographed on Stellwagen Bank in previous years (Clapham10). If a resurgence of herring is responsible for shifts in distribution and in primary prey type, it suggests that the distribution of humpback whales through the late 1970’s and 1980’s may have been a human- 9 Francis, L. 1995. Atlantic Cetacean Research Center, PO Box 1413, Gloucester, MA 01930. Personal commun. 10 Clapham, P. 1995. Smithsonian Institution, Washington, DC. Personal commun. 11 Mattila, D. 1995. Center for Coastal Studies, PO Box 1036, Provincetown, MA 02657. Personal commun. 12 Fernald, T. 1995. Allied Whale, College of the Atlantic, Bar Harbor, ME 04609. Personal commun. induced consequence. The “explosion” of sand lance in the mid- to late 1970’s is thought to be primarily the result of the virtual elimination of herring due to overfishing. If this is true, we hypothesize that our observed distribution of whales from 1992 to 1994 should remain relatively stable over the course of a fairly long period because the current situation would be closer to a “natural” ecosystem. Alternatively, fluctuations in primary prey may occur naturally, and may take place regardless of human interference. If this is true, we hypothesize that whale distributions will show fluctuations that may be cyclical. New England ground-fishermen have for years talked of regular cycles in sand lance abun- dance, although there are no scientific data to sup- port this often-made contention. Regardless of which hypothesis, if either, proves true, our data show a shift in both distribution and primary prey type for humpback whales in southern New England waters in recent years. Because this shift has been so complete, it will be interesting and illustrative to see whether, and how, other potentially prey-dependant humpback whale life history param- eters, such as reproductive patterns, social behav- ior, and demographics of whales, all well-documented during a period of explosive sand lance abundance (Clapham and Mayo, 1990; Weinrich, 1991; Weinrich and Kuhlberg, 1991; Clapham et al., 1992), change in response to these ecosystem alterations. Acknowledgments We would like to thank Cindy Belt, Tina Laidlaw, Joy Lapseritis, Laine Gonzales, Nancy Miller, Liz Phinney, Leo Axtin, Karla Bristol, and many interns at the Cetacean Research Unit for their assistance in gathering sighting data for this study. Staff of the Center for Coastal Studies, Provincetown, MA, and the Atlantic Cetacean Research Center, Gloucester, MA, were generous in sharing data on ages and sightings of individuals, especially mother-calf pairs. Portions of this study were funded by the Bruce J. Anderson Founda- tion, the Sudbury Foundation, the Fund for Preserva- tion of Wildlife and Natural Areas, the Norcross Wild- life Foundation, and the American Cetacean Society — Los Angeles Chapter. This work was conducted under Marine Mammal Research Permit no. 681. Literature cited Anthony, V. C., and G. Waring. 1980. The assessment and management of the Georges Bank herring fishery. Rapp. P.-V. Reun. Int. Explor. Mer 177:72-111. Weinrich et a I.: A shift in distribution of humpback whales, Megaptera novaeangliae, in reponse to prey 835 Baker, C. S., R. H. Lambertsen, M. T. Weinrich, J. Calambokidis, G. Early, and S. J. O’Brien. 1991. Molecular genetic identification of the sex of hump- back whales t Megaptera novaeangliae). Rep. Int. Whal. Comm. (spec, issue 13): 105-111. Baker, C. S., S. R. Palumbi, R. H. Lambertsen, M. T. Weinrich, J. Calambokidis, and S. J. O’Brien. 1990. Influence of seasonal migration on geographic parti- tioning of mitochondrial DNA haplotypes in humpback whales. Nature 344:238-240. Chittleborough, R. G. 1965. Dynamics of two populations of the humpback whale, Megaptera novaeangliae (Borowski). Aust. J. Mar. Fresh- water Res. 16:33-128. Clapham, P. J., L. S. Baraff, C. A. Carlson, M. A. Christian, D. K. Mattila, C. A. Mayo, M. Murphy, and S. Pittman. 1992. Seasonal occurrence and annual return of humpback whales, Megaptera novaeangliae , in the southern Gulf of Maine. Can. J. Zool. 71:440-443. Clapham, P. J., and C. A. Mayo. 1987. Reproduction and recruitment of individually iden- tified humpback whales, Megaptera novaeangliae , observed in Massachusetts Bay, 1979-1985. Can. J. Zool. 65:2853- 2863. 1990. Reproduction of humpback whales ( Megaptera novaeangliae) observed in the Gulf of Maine. Rep. Int. Whal. Comm, (special issue 12 ): 17 1—175. Dolphin, W. F. 1987. Ventilation and dive patterns of humpback whales, Megaptera novaeangliae, on their Alaskan feeding grounds. Can. J. Zool. 65:83-90. D’Vincent, C. G., R. M. Nilson, and R. E. Hanna. 1985. Vocalization and coordinated feeding behavior of the humpback whale in southeast Alaska. Sci. Rep. Whales Res. Inst. 36:41-47. Fogarty, M. J., E. B. Cohen, W. L. Michaels, and W. W. Morse. 1991. Predation and the regulation of sand lance popula- tions: an exploratory analysis. ICES Mar. Sci. Symp. 193:120-124. Glockner, D. A. 1983. Determining the sex of humpback whales in their natural environment. In R. Payne (ed. ), Communication and behavior of whales, p. 447-464. American Academy for the Advancement of Science (AAAS) Selected Sympo- sium 76, Westview Press, Boulder, CO. Grosslein, M. D., R. W. Langton, and M. P. Sissenwine. 1980. Recent fluctuations in pelagic fish stocks of the North- west Atlantic Georges Bank region, in relation to species interactions. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 177:377-404. Hain, J. H. W., G. Carter, S. Kraus, C. Mayo, and H. Winn. 1982. Feeding behavior of the humpback whale in the west- ern North Atlantic. Fish. Bull. 80(2):259-268. Hain, J. H. W., S. L. Ellis, R. D. Kenney, P. J. Clapham, B. K. Gray, M. T. Weinrich, and I. G. Babb. 1995. Apparent bottom feeding by humpback whales on Stellwagen Bank. Mar. Mamm. Sci. 11:464-479. Katona, S. K. 1986. Biogeography of the humpback whale, Megaptera novaeangliae , in the western North Atlantic. In A. C. Pierrot-Bults, S. van der Spoel, B. J. Zahuranec, and R. K. Johnson (eds.), Pelagic biogeography, p. 166-171. UNESCO Technical Papers in Marine Science 49, Paris. Katona, S. K., and J. A. Beard. 1990. Population size, migrations, and feeding aggregations of the humpback whale ( Megaptera novaeangliae ) in the western North Atlantic Ocean. Rep. Int. Whaling Comm, (spec, issue 12):295-306. Katona, S. K., and H. P. Whitehead. 1981. Identifying humpback whales using their natural markings. Polar Rec. 20:439-444. Kawamura, A. 1980. A review of food of balaenopterid whales. Sci. Rep. Whales Res. Inst. 32:155-197. Kellogg, R. 1929. What is known of the migration of some of the whale- bone whales. In Anonymous (ed.), Smithsonian Institu- tion Annual Report, 1928, Publ. 2981, p. 467-494. Smith- sonian Press, Washington, D.C . Kenney, R. D., and H. E. Wine. 1986. Cetacean high-use habitats of the northeast United States continental shelf. Fish. Bull. 84:345-357. Kinnear, P. R., and Gray, C. D. 1992. SPSS/PC+. Lawrence Erlbaum Associates, East Sussex, U.K., 210 p. Mackintosh, N. A. 1965. The stocks of whales. Fishing News (Books), Ltd., London. 232 p. Meyer, T. L., R. A. Cooper, and R. W. Langton. 1979. Relative abundance, behavior, and food habits of the American sand lance, Ammodytes americanus, from the Gulf of Maine. Fish. Bull. 77:243-253. Mitchell, E. A. 1973. Draft report on humpback whales taken under spe- cial scientific permit by eastern Canadian land stations, 1969-71. Rep. Int. Whaling Comm. 23: 138-154. Overholtz, W. J., and J. R. Nicolas. 1979. Apparent feeding of the fin whale, Balaenoptera physalus, and the humpback whale, Megaptera novaean- gliae, on the American sand lance, Ammodytes americanus, in the northwest Atlantic. Fish. Bull. 77:285-287. Payne, P. M., J. R. Nicolas, L. O’Brien, and K. D. Powers. 1986. The distribution of the humpback whale, Megaptera novaeangliae, on Georges Bank and in the Gulf of Maine in relation to densities of the sand eel, Ammodytes americanus. Fish. Bull. 84:271-277. Payne, P. N., D. Wiley, S. 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 prey abundance. Fish. Bull. 88:687-696. Piatt, J. F.» D. A. Methven, and A. E. Burger. 1989. Baleen whales and their prey in a coastal environ- ment. Can. J. Zool. 67:1523-1530. Reay, P. J. 1970. Synopsis of biological data on North Atlantic sand eels of the genus Ammodytes. FAO Fisheries Synopsis 82, Food and Agriculture Organization of the United Nations, Rome, var. pagination. Sherman, K. 1986. Measurement strategies for monitoring and forecast- ing variability in large marine ecosystems. In K. Sherman and L. M. Alexander (eds.), Variability and management of large marine ecosystems, p. 203-236. American Acad- emy for the Advancement of Science ( AAAS ) Selected Sym- posia 99, Westview Press, Boulder CO. 1992. Where have all the fish gone? Nor ’easter 4(2): 14—19. Sherman, K., M. Grosslein, D. Mountain, D. Busch, J. O’reily, and R. Theroux. 1988. The continental shelf ecosystem of the northeast coast of the United States. In H. Postema and J. J. Zijlstra 836 Fishery Bulletin 95(4), 1997 (eds.). Ecosystems of the world 27: continental shelves, p. 279-337. Elsevier Press, Amsterdam. Sherman, K., C. Jones, L. Sullivan, W. Smith, P. Berrin, and L. Ejstmont. 1981. Congruent shifts in sand eel abundance in western and eastern North Atlantic ecosystems. Nature 291:486-489. Sissenwine, M. 1986. Perturbation of a predator-controlled continental shelf ecosystem. In K. Sherman and L. M. Alexander (eds.), Variability and management of large marine eco- systems, p. 55-85. American Academy for the Advance- ment of Science (AAAS) Selected Symposia 99, Westview Press, Boulder, CO. Snedecor, G. W., and W. G. Cochran. 1967. Statistical methods. Iowa State Univ. Press, Ames, IA, 593 p. Stephenson, R. L., and I. Kornfield. 1990. Reappearance of spawning Atlantic herring (Clupea harengus harengus) on Georges Bank: population resurgence not recolonization. Can. J. Fish. Aquat. Sci. 47:1060-1064. Swingle, W. M., S. G. Barco, T. D. Pitchford, W. A. McLellan, and D. A. Pabst. 1993. Appearance of juvenile humpback whales feeding in the nearshore waters off Virginia. Mar. Mamm. Sci. 9:309-315. USDC (United States Department of Commerce). 1991. Status of fishery resources off the northeastern United States for 1991. U.S. Dep. Commer., NOAATech. Memo. NMFS-F/NEC-86, 132 p. Weinrich, M. T. 1991. Long term stability in grouping patterns of hump- back whales ( Megaptera novaeangliae) in the southern Gulf of Maine. Can. J. Zool. 69:3012- 3019. Weinrich, M. T., and A. K. Kuhlberg. 1991. Short-term association patterns of humpback whale (Megaptera novaeangliae ) groups on their southern Gulf of Maine feeding ground. Can. J. Zool. 69:3005-3011. Weinrich, M. T., M. R. Schilling, and C. R. Belt. 1992. Evidence for acquisition of a novel feeding behavior: lobtail feeding in humpback whales, Megaptera novaean- gliae. Anim. Behav. 44:1059-1072. Whitehead, H. P., and J. E. Carscadden. 1985. Predicting inshore whale abundance — whales and capelin off the Newfoundland coast. Can. J. Fish. Aquat. Sci. 42:976-981. 837 Temporal variation in sexual maturity and gear-specific sex ratio of the vermilion snapper, Rhomboplites aurorubens, in the South Atlantic Bight* Boxian Zhao John C. McGovern South Carolina Department of Natural Resources P O. Box 12559, Charleston, South Carolina 29422-2559 E-mail address (for B. Zhao): zhaob@mrd.dnr.state.scus Abstract .—Percentages of mature male and female vermilion snapper, Rhomboplites aurorubens, based on to- tal length (TL) and age were calculated for five three-year periods during 1979- 93. Males and females collected during 1982-87 became sexually mature at a smaller size and younger age than in- dividuals collected during 1979-81. The median TL at maturity for females de- creased from 160 mm in 1979-81 to 151 mm in 1985-87. The median TL at maturity for males was 145 mm dur- ing 1979-81. During 1985-87 all males were mature at 140 mm. The temporal shift toward a smaller size at maturity was more pronounced in males than in females. The percentage of mature males at age 1 significantly increased from 63.6% in 1979—81 to 100% in 1985-87 and afterwards. More than twice as many females at age 1 were mature in 1985-87 (48.6%) as in 1979- 81. The decline in size and age at ma- turity may have been caused by fish- ing pressure that gradually increased during the 1980’s. The sex ratio of vermilion snapper was dependent upon latitude and gear type but was generally independent of water depth, fish length, and sampling years. Although the sex ratios were sig- nificantly different among latitudes, there were no trends among latitudes 31°N, 32°N, and 33°N. The percentage of females was 72.1%, 68.0%, and 59.9% for vermilion snapper caught by trap, hook-and-line, and trawl, respectively. Reasons for the difference in sex ratio among gear types are unclear, suggest- ing that caution must be used when interpreting sex ratios estimated for any fish species collected by various gear types. Manuscript accepted 30 April 1997. Fishery Bulletin 95: 837-848 ( 1997). The spawning potential ratio (SPR) has been widely used by U.S. fish- ery management councils to define overfishing of a fish stock (Good- year, 1993; Rosenberg et al., 1994; SAFMC* 1). To estimate SPR, life his- tory characteristics (e.g. growth and reproduction) are required and are generally assumed constant among years (Gabriel et al., 1989). How- ever, these parameters, particularly maturity schedules, are not static. They can change in response to fish- ing pressure, predator and prey abundance, stock composition, and other biotic and abiotic environmen- tal factors (Wootton, 1990). Igno- rance of temporal changes in life history parameters may result in the use of incorrect data by fishery managers and therefore may be a reason why fish stocks fail to be pro- tected from overfishing (Rosenberg et al., 1994). Vermilion snapper, Rhomboplites aurorubens, from the South Atlan- tic Bight (SAB) occur in shelf and upper-slope waters between depths of 26 and 183 m (Grimes, 1978). This species spawns multiple times dur- ing a prolonged spawning season (April through September: Grimes and Huntsman, 1980; Cuellar et al., 1996). Vermilion snapper have been of extreme commercial and recre- ational importance along southeast- ern states since the early 1980’s. Total landings in this region have increased over the years with a peak in 1991 (Zhao and McGovern2). However, recent studies have sug- gested that vermilion snapper are overfished. The stock abundance estimated by virtual population analysis (VPA) has declined since 1984 (Zhao and McGovern2). The relative abundance represented by catch per unit of effort (CPUE) markedly declined during 1988-93. There has also been a significant decrease in mean length of vermil- ion snapper caught by fishery-inde- pendent surveys and by the head- boat and commercial fishery (Zhao and McGovern2). Changes in life history characteristics induced by intense harvesting have been re- ported for vermilion snapper. Zhao et al. (1997) validated the ageing * Contribution 391 of the South Carolina Ma- rine Resources Center, 217 Fort Johnson Rd., Charleston, South Carolina 29412. 1 SAFMC ( South Atlantic Fishery Manage- ment Council). 1993. Amendment 6, regulatory impact review, initial regula- tory flexibility analysis and environmental assessment for the snapper grouper fishery of the south Atlantic region. South Atlan- tic Fishery Management Council, Charles- ton, SC, 155 p. 2 Zhao, B., and J. C. McGovern. 1996. Population characteristics of the vermil- ion snapper from the southeastern United States. In preparation 838 Fishery Bulletin 95(4), 1 997 method of otolith sections and demonstrated that size-at-age decreased with time. Collins and Pinckney (1988) reported preliminary evidence that vermilion snapper caught in 1978—80 from SAB became reproductively mature earlier in life than those caught in 1972-74. Grimes and Huntsman (1980) deter- mined that vermilion snapper were gonochorists, fe- males (62.5%) significantly outnumbered males, and the sex ratio was dependent on fish length. However, Nelson (1988) reported that the sex ratio of vermil- ion snapper from the Gulf of Mexico differed from 1:1 in favor of males (54.5%). He also reported that area and season had a significant effect on this ra- tio. Although what may have caused the difference in sex ratios of two studies was unknown, limited sample sizes made their comparisons among areas, lengths, or seasons less convincing (Grimes and Huntsman, 1980: n= 874; Nelson, 1988: n=881). In this paper, we investigated the percentage of mature vermilion snapper at each length class and age for each sex and examined the temporal change in maturity schedules during 1979-93. We also de- termined the sex ratio of vermilion snapper accord- ing to depth and latitude of sampling sites, fish length, sampling period, and types of fishing gear used. Materials and methods Data were collected from the SAB during 1979-93 by reef fish surveys of the Marine Resources Moni- toring, Assessment, and Prediction (MARMAP) pro- gram. Most vermilion snapper were captured by stan- dardized hook-and-line and trapping gear during the spring and summer of 1979-93 (Collins, 1990; Collins and Sedberry, 1991). In addition, many vermilion snapper were obtained from the MARMAP trawling program that was terminated in 1988. In the field, fish were measured to the nearest mm (TL and FL) and weighed to the nearest gram (whole body weight). Water depth, latitude, and longitude of sampling sites were recorded. The posterior portion of the gonad was removed and preserved in 10% formalin buffered with seawater. In the laboratory, sex and maturity stages were determined by examining the stained gonad sections according to standard MARMAP histologi- cal criteria (Cuellar et al., 1996). All gonad samples taken in 1987-93 were examined with histological methods. During 1979-86, sex of specimens was de- termined and a reproductive condition was assigned in the field by using clearly defined visual staging criteria. The definitions of maturity stages by gross visual inspection were confirmed to be accurate by histological examination during 1978-80 (Collins and Pinckney, 1988). The codes used for defining matu- rity stages were consistent between histological and macroscopic methods. Seven reproductive stages were used: 1) immature, 2) developing, 3) running ripe, 4) spent, 5) resting, 6) developing with evidence of spawn- ing in the previous week, and 7) mature, but stage- unknown. Mature fish included those in stages 2-7. Percentages of mature males and females based on total length (TL) and age were calculated for five periods, each corresponding to a three-year interval, i.e. 1979-81, 1982-84, 1985-87, 1988-90, and 1991- 93. Because the monthly distribution of samples was not necessarily similar among periods, it was essen- tial to define a standard month(s) if comparison of maturity schedules among years was to be meaning- ful. In addition, the reproductive condition was most discernible and the sizes and ages at first maturity could be determined during and immediately prior to the spawning season. Therefore, we used the data only from May and June that corresponded to the time of slow somatic growth. Age assignments of ver- milion snapper were based on the number of annuli on otolith sections (Zhao et al., 1997). We took into consideration knowledge of the time of annulus for- mation, the relative growth of the otolith margin, the date of sampling, and used a January 1 hatching date. Because only a part of the samples was aged, some sexed and maturity-determined samples did not have age information. Although the variation in gear type, latitude, and depth of sampling sites should have been included in the maturity analysis before the data were pooled, limited sample sizes of small (TL<170 mm) and young (age-1) vermilion snapper prevented us from com- paring maturity schedules by area (latitude) or by gear type. However, 97% of the males and females smaller than 170 mm were collected by trawl in con- sistent depth ranges during 1979-90. A majority of these small fish were collected from similar areas, with 88% males and 82% females from latitude 31°N, and 12% males and 18% females from 32°N. Thus, gear selectivity and geographical distribution can be disregarded as sources of bias in comparison of ma- turity schedules among periods. As recommended by Trippel and Harvey (1991), the G-test was used to compare maturity schedules of each sex among periods with the same length class or the same age (e.g. age 1). Maturity schedules be- tween sexes at age 1 were compared for the periods of 1979-81 and 1985-87 respectively, when sample sizes of both sexes were sufficient. When conditions of the G-test were not met, Fisher’s exact test was used (Zar, 1984). When the percentages of mature fish of each sex in 20 mm TL intervals exhibited a successive increase with length, the median length Zhao and McGovern: Variation in sexual maturity and sex ratio of Rhomboplites aurorubens 839 at sexual maturity (TL50) was calculated by probit analysis following the recommendation of Tripple and Harvey (1991). The likelihood ratio test at a signifi- cance level of 0. 10 was used to determine if the probit model could be fitted to the observations of matu- rity-at-length (SAS Institute, 1990). Because sex ratio may change with water depth and latitude of sampling sites, fish length, sampling year, gear, or season (Grimes and Huntsman, 1980; Nelson, 1988), all sexed samples, including mature and immature individuals between 1979 and 1993, were split into the same five periods as described above for maturity analysis. The time frame was in- creased to May through August to increase the sample size. We compared the percentages of females among varying water depths (midpoints: 25, 35, 45, and 55 m), latitudes (31°N, 32°N, and 33°N), length classes (TL=100 - 450 mm, with 50-mm intervals), periods, and gear types (traps, hook-and-line, and trawl). Latitudes of 31°N, 32°N, and 33°N refer to 31°00'-31°59'N, 32°00'-32°59’N, and 33°00’-33°59’N, respectively. When the independence between sex ratio and one of above factors was tested, other fac- tors were kept consistent. The chi-square test and Fisher’s exact test were used to decide the indepen- dence. As a reference, Bonferroni’s method was used to adjust the significance level, i.e. a'=0.05/m, where m = the number of cases (Sokal and Rohlf, 1995). First, we compared the percentage of female vermil- ion snapper taken from varying depths with the same length classes (TL=2QQ-249 mm and TL=250-299 mm respectively), same latitude (32°N), same peri- ods, and same gear type. Seventy-three percent of the vermilion snapper were collected from latitude 32°N during 1979-93, and eighty percent of them were between 200 and 299 mm TL. If the hypothesis 840 Fishery Bulletin 95(4), 1997 of independence between sex ratio and depth was not rejected, the data from all depth classes were pooled to compare further the percentage of females among latitudes, other factors (length, period, and gear) being consistent. If the hypothesis was rejected, data from a certain depth class were chosen for fur- ther comparison. In a similar fashion, we compared the percentage of females among length classes, pe- riods, and gear types. The statistical methods available in SAS were used to analyze data of maturity and sex ratio (SAS Insti- tute, 1990). Rejection of the null hypothesis was based on a significance level of 0.05, unless other- wise noted. ResuSts Maturity schedules Males and females collected in 1982-84 and 1985- 87 became sexually mature at a smaller size than individuals collected in 1979-81 (Fig. 1). For in- stance, 31% of male vermilion snapper collected in 1979-81 were mature at 140 mm. However, 100% of males taken at the same size during 1982-90 were mature. All fish taken in May and June of 1991-93 were larger than 180 mm and mature, and therefore were not included in our analysis. There was a sig- nificant temporal increase in the percentage of ma- ture males among 1979-81, 1982-84, and 1985-87 at 120 mm (Fisher’s exact test: two-tailed P<0.01), and at 140 mm (two-tailed P<0.005). The percent- age of mature females also showed a significant in- crease with time for 140-mm individuals collected between 1979-81 and 1985-87 (two-tailed P<0.005). The observed differences in the percentage of ma- ture females at 160 mm collected during 1979-81, 1985-87, and 1988-90 were not statistically signifi- cant (Fig. 1). Males larger than 140 mm and females larger than 160 mm were not tested because all fish were mature. The likelihood ratio tests indicated that the probit model could be used to describe the maturity at length during 1979-81 for both males (likelihood ratio x2=2.721, P>0.10) and females (likelihood ratio %2=1.407, P>0.10), and for females during 1985-87 (likelihood ratio ^2=4.647, P>0.10). The median TL at maturity of males was 145 mm (95% limits: 135- 203 mm) during 1979-81. The TL50 of females was 160 mm (95% limits: 155-164 mm) in 1979-81 and 151 mm (95% limits: 143-156 mm) in 1985-87. There was a significant increase in the percentage of mature age-1 males with time between 1979-81, 1982-84, and 1985-87 (Fisher’s exact test: two-tailed P=0.013) (Table 1). The percentage of mature age-1 females increased (G=5.318,P=0.021) between 1979- 81 and 1985-87. More than twice as many age-1 fe- males were mature in 1985-87 (48.6%) as in 1979- 81 (23.1%). The median age at sexual maturity could not be calculated because of the abrupt transition from immature to mature. Males matured at a smaller size and younger age than females (Fig. 1; Table 1). During 1979-81, TL50 for males ( 145 mm) was smaller than that for females (TL50=160 mm). Although TL50 for males in 1985-87 could not be calculated, it was observed that the TL50 of males declined with time faster than that of fe- Tab!e 1 Percentages of sexually mature vermilion snapper caught in May and June of 1979-93. Numbers of fish in each category are given in parentheses. Blanks indicate no data available for that category. There were significant (P<0.05) differences in percent mature of age-1 fish among periods for each sex. Period Age 1 Age 2 Age 3 Ages 4+ Males 1979-81 63.6 (11) 100.0(15) 100.0(12) 100.0 (13) 1982-84 85.7 (7) 100.0 (28) 1985-87 100.0 (19) 100.0 (12) 100.0(10) 100.0(55) 1988-90 100.0 (2) 100.0 (4) 100.0 (4) 100.0 (47) 1991-93 100.0(1) 100.0 (32) Females 1979-81 23.1 (39) 91.3 (23) 100.0 (9) 100.0(9) 1982-84 100.0(4) 100.0 (6) 100.0 (67) 1985-87 48.6 (35) 95.8 (24) 100.0 (18) 100.0(135) 1988-90 100.0 (2) 100.0 (8) 100.0 (10) 100.0(104) 1991-93 100.0 (3) 100.0 (86) Zhao and McGovern: Variation in sexual maturity and sex ratio of Rhomboplites aurorubens 841 males (Fig. 1). During 1979-81 and 1985-87, the percentage of mature males at age 1 was significantly larger than that of females at the same age (Table 1; 1979-81: Fisher’s exact test, two-tailed P=0.024; 1985-87: G= 20.252, P<0.001). Sex ratios More vermilion snapper were caught by traps and hook-and-line in the depth range of 40-49 m than in other depth classes (Table 2). The trawl was gener- ally deployed in shallower water (20-39 m) than were traps and hook-and-line. These three gear types were deployed most often in latitude 32°N than in other areas (Table 2). To exclude the effects on sex ratios from varying latitude, fish length, years, and gear types, we used the data collected from the same lati- tude (32°N), length (200-249 mm or 250-299 mm TL), period, and gear when the independence be- tween sex ratio and depth was tested. Sample sizes in all categories were not always sufficient for a chi- square test. If the expected frequency of a depth-class was unacceptably low, that data was discarded from the contingency table. When sample sizes of depth Table 2 Numbers of sexed vermilion snapper collected from May through August during 1979-93 by depth and latitude of sampling sites. Blanks indicate that no samples were available for that category. Period Depth midpoint (m) Traps Hook-and-line Trawl 30°N 31°N 32°N 33°N 34°N Total 31°N 32°N 33°N Total 31°N 32°N Total 1979-81 15 3 3 25 26 26 27 10 37 5 97 102 35 1 1 5 10 27 42 242 120 362 45 8 8 326 326 55 65 Total 26 9 35 32 346 27 405 247 220 467 1982-84 15 25 21 21 6 42 48 29 217 236 35 50 50 31 31 28 28 45 298 298 214 214 55 100 100 27 27 65 7 7 Total 469 469 6 321 327 19 245 264 1985-87 15 129 129 25 13 13 283 22 305 35 1 1 173 173 45 20 367 387 145 145 55 95 95 17 114 131 65 Total 149 462 611 30 260 290 283 195 478 1988-90 15 25 44 111 80 235 50 53 27 130 35 3 13 70 86 22 12 34 45 406 23 429 226 226 55 2 153 155 118 118 65 1 1 Total 5 57 741 103 906 72 409 27 508 1991-93 15 9 9 25 155 69 81 10 315 1 1 35 104 84 155 6 349 10 4 14 45 3 124 127 35 35 55 18 286 304 14 14 65 4 4 Total 21 259 576 236 16 1,108 1 59 4 64 842 Fishery Bulletin 95(4), 1997 classes were similar to one another and were small relative to the size of a contingency table, Fisher’s exact test was used instead of a chi-square test (Zar, 1984). All tested cases supported the null hypoth- esis of independence between sex ratio and depth, except for the samples of 200-249 mm TL fish caught with traps during 1982-84 (Table 3). After allowance for multiple-testing (a'=0. 05/18=0. 0028), none of the cases was statistically significant. Since there were no significant differences among depth classes, we pooled the data from all depth classes to compare the percentage of females among Table 3 Comparison of percentages of females among water depth-classes with the same latitude (32°00'-32°59' N), length ranges (TL=200- 249 mm and 250-299 mm), period, and gear type. % = female percent, n = the total number of male and female fish. The null hypothesis ( H0 ) = sex ratio is independent of depth. Blanks indicate no or few samples available for comparison, df = degrees of freedom. Period Depth midpoint (m) Traps Hook-and-line Trawl TL=200-249 TL=250-299 % n % n TL=200-249 TL=250-299 % n % n TL=150- % -199 n TL=200- % -249 n 1979-81 25 54.6 55 53.9 39 35 66.7 48 59.7 62 chi-square 1.572 0.333 P 0.210 0.564 df 1 1 Reject H0 No No 1982-84 25 83.3 18 100.0 21 66.7 15 57.7 26 58.9 35 91.7 36 84.6 13 64.7 17 80.0 10 65.2 45 67.1 152 69.1 97 74.4 43 71.6 109 55 80.0 45 75.6 41 75.0 8 40.0 15 chi-square 11.28 1.708 Fisher’s2 Fisher’s2 0.314 P 0.010 0.426 0.859 0.056 0.575 df 3 2 1 Reject H0 Yes No No No No 1985-87 25 57.9 35 62.2 45 86.9 206 72.4 134 68.6 51 63.9 72 55 76.7 60 82.4 34 61.4 44 71.2 59 chi-square 3.727 1.414 0.550 0.783 0.132 P 0.054 0.234 0.458 0.376 0.716 df 1 1 1 1 1 Reject H0 No No No No No 1988-90 25 82.7 81 73.7 19 61.8 34 35 77.1 48 79.0 19 45 72.5 189 67.1 176 65.1 126 60.0 45 55 64.9 74 75.0 76 70.0 60 71.8 39 chi-square 6.808 2.484 0.745 1.286 P 0.078 0.478 0.689 0.257 df 3 3 2 1 Reject H0 No No No No 1991-93 25 80.0 30 84.6 132 35 74.5 47 80 20 45 70.0 30 77.9 68 55 64.4 160 73.5 102 chi-square 3.910 0.649 P 0.271 0.723 df 3 2 Reject H0 No No 1 The row was discarded from the contingency table because of its excessively low expected frequency. 2 When conditions for a chi-square test were not met, Fisher’s exact test was used. Zhao and McGovern. Variation in sexual maturity and sex ratio of Rhomboplites aurorubens 843 latitudes with other factors (length, period, and gear) remaining consistent. Six of the 14 tested cases indi- cated that sex ratio was dependent on latitude (Table 4). Even after allowance for multiple-testing ( ct'=0.05/ 14=0.0036), there were still two cases indicating sig- nificant difference (1979-81, trawl, 150-199 TL and 1988-90, traps, 200-249 TL). Because the latitudi- nal distribution of samples was not similar among periods for any gear type (Table 2), we used only data from the latitude of 32°N in subsequent analyses. There were no significant differences in percent- age of females caught by traps among length classes 200-249, 250-299, and 300-349 mm during 1982- 84, 1988-90, or 1991-93 (Table 5). During 1979-81, the hypothesis of independence between sex ratio and length was rejected for fish caught by hook-and-line within the length range of 150-449 mm. However, it was not rejected within 200-299 mm (Table 5). Thus, the common TL range of vermilion snapper caught by hook-and-line was 200-299 mm, within which the sex ratio was independent of length during 1979- 93. There were no significant differences in percent- ages of females caught by trawl within 150-249 mm during 1979-81 and 1985-87 (Table 5). After allow- ance for multiple-testing ( a'=0. 05/12=0. 004), none of the 12 cases showed significant difference. We compared the percentage of females among periods by pooling data for each gear type within the common TL range and periods when the sex ratio was independent of length. There were no signifi- Table 4 Comparison of percentages of females among latitudes with the same length ranges (TL=200-249 mm and TL=250-299 mm), period, and gear type. All data were pooled from various water depth classes. % = female percent, n - the total number of male and female fish. The null hypothesis ( H0 ) = sex ratio is independent of latitude. Blanks indicate no or few samples available for comparison. Traps Hook-and-line Trawl Latitude TL=200-249 TL=250-299 TL=200-249 TL=250-299 TL= 150-199 TL=200-249 Period (°N) % n % n % n % n % n % n 1979-81 31 32 33 chi-square P df Reject H0 1985-87 31 82.6 32 84.6 chi-square 0.234 P 0.629 df 1 Reject H0 No 1988-90 31 63.9 32 73.7 33 91.8 chi-square 11.96 P 0.003 df 2 Reject H0 Yes 1991-93 31 80.0 32 68.8 33 77.7 chi-square 9.509 P 0.009 df 2 Reject H0 Yes 57.1 14 87.5 73.6 106 67.6 87.5 16 50.0 Fisher’s1 2 Fisher’s2 0.178 0.352 No No 115 50.0 20 72.7 11 93.3 266 74.4 168 65.6 96 67.2 5.258 Fisher’s2 Fisher’s2 0.022 0.747 0.039 1 Yes No Yes 36 100.0 4’ 392 70.5 291 61 76.7 30 0.511 0.475 1 No 235 57.1 7 276 76.8 207 206 60.0 20 3.925 0.141 2 No 8 36.5 115 57.1 14 105 60.2 103 57.4 101 4 12.205 0 <0.001 0.984 1 1 Yes No 15 39.4 221 52.3 44 131 58.3 36 61.7 154 4.571 1.260 0.033 0.262 1 1 Yes No 1 The row was discarded from the contingency table because of its excessively low expected frequency. 2 When conditions for a chi-square test were not met, Fisher’s exact test was used. 844 Fishery Bulletin 95(4), 1997 Table 5 Comparison of percentages of females among length classes with the same latitude (32°00'-32°59' N), period, and gear type. All data were pooled from various depth classes. % = female percent, n = the total number of male and female fish. The null hypoth- esis (H0) = sex ratio is independent of length. Blanks indicate no or few samples were available for comparison. Period TL (mm) Traps Hook-and -line Trawl % n % n % n 1979-81 90-149 X2 = 14.079 78.6 14 X2 = 2.293 150-199 38.9 18 P = 0.015 60.2 103 P = 0.318 200-249 73.6 106 df = 5 57.4 101 df = 2 250-299 67.6 105 100.0 21 300-349 54.6 77 350-399 72.4 29 >399 54.6 11 Reject H0 Yes No 1982-84 150-199 66.7 37 X2 = 1.597 80.0 51 X2 = 0.449 41.0 100 X2 = 8.847 200-249 74.1 251 P = 0.450 71.1 83 P = 0.930 60.0 135 P = 0.012 250-299 72.6 153 df = 2 67.1 161 df = 3 40.0 10 df = 2 300-349 65.3 49 67.3 52 350-399 80.0 107 66.7 18 >399 100.0 31 100.0 21 Reject H0 No No Yes 1985-87 90-149 100.0 l1 X2 = 6.844 0 l1 X2 = 0.515 66.7 37 X2= 0.138 150-199 100.0 51 P = 0.033 75.0 41 P = 0.773 58.3 36 P = 0.710 200-249 84.6 266 df = 2 65.6 96 df = 2 61.7 154 df = 1 250-299 74.4 168 67.2 131 100.0 21 300-349 81.0 21 73.1 26 350-399 100 l‘ 100.0 21 >399 Reject H0 Yes No No 1988-90 90-149 X2 = 1.048 33.3 3’ X2 = 1.547 150-199 100.0 51 P = 0.592 59.7 62 P= 0.461 200-249 73.7 392 df = 2 67.1 228 df = 2 250-299 70.5 291 68.8 93 300-349 75.0 44 41.2 177 350-399 57.1 77 25.0 41 >399 50.0 21 100.0 21 Reject H0 No No 1991-93 150-199 50.0 41 X2 = 6.035 100 21 Fisher’s2 200-249 68.8 276 P = 0.110 64.5 31 P= 1.0 250-299 76.8 207 df = 3 64.3 14 300-349 63.4 71 25.0 47 350-399 71.4 14 60.0 57 >399 50.0 41 100.0 37 Reject H0 No No 1 The row was discarded from the contingency table because of its excessively low expected frequency. 2 When conditions for a chi-square test were not met, Fisher’s exact test was used. cant differences in percentages of females among periods for either gear type (Table 6). The overall chi-square test indicated a significant lack of inde- pendence between sex ratio and gear type (Table 7 A). We subdivided the data into 2x2 contingency tables formed by any two gear types. Significant differences in the percentage of females were found between trawl and traps (x2=22.642, P<0.001), between trawl and hook-and-line (%2=8.424, P=0.004), and between traps and hook-and-line (%2=5.166, P=0.023). There- fore, we concluded that traps caught more females than did the other two gear types, and hook-and-line caught more females than trawl. This conclusion was supported by additional analysis with various length Zhao and McGovern: Variation in sexual maturity and sex ratio of Rhomboplites aurorubens 845 Table 6 Comparison of percentages of females among periods with the same latitude (32°00'-32°59' N) and gear type. All data were pooled from various depth classes. % = female percent. n = the total number of male and female fish. The null hypothesis (H0) = sex ratio is independent of period. Blanks indicate few samples available for comparison or the null hypothesis of independence between sex ratio and length was rejected in Table 5. df = degrees of freedom. Traps Hook-and-line Trawl (TL=200-349) (TL=200-299) (TL=150- 249) Period % n % n % n 1979-81 70.6 211 58.8 204 1982-84 72.6 453 68.4 244 1985-87 66.5 227 61.1 190 1988-90 72.5 727 67.6 321 1991-93 71.1 554 64.4 45 chi-square 0.382 1.199 0.204 P 0.826 0.878 0.652 df 2 4 1 Reject H0 No No No Table 7 Comparison of percentages of females among gear types with the same latitude (32°00'-32°59' N). All data were pooled from various depth classes. % = female percent, n = the total number of male and female fish. The null hypoth- esis (H0) = sex ratio is independent of gear type. Data were pooled from periods that were used in Table 6. Data with different length ranges were used in A, B, and C. (A) TL ranges were the same as used in Table 6 for each gear; (B) TL = 200-249 mm for all gear types; (C) TL = 186-540 mm, mean TL = 254 mm for traps, TL = 142-560 mm, mean TL = 261 mm for hook-and-line, and TL = 112-256 mm, mean TL = 203 mm for trawl. Gear A B C % n % n % n Traps 72.1 1734 72.4 919 72.1 1786 Hook-and-line 68.0 1048 68.6 544 66.2 1395 Trawl 59.9 394 60.0 255 61.0 415 chi-square 23.460 14.558 24.938 P <0.001 0.001 <0.001 df 2 2 2 Reject H0 Yes Yes Yes ranges (Table 7B: a single TL range for all gear types; Table 7C: all data available were used with full TL ranges). Vermilion snapper caught by all gear types had an unequal sex ratio that was female-biased (traps and hook-and-line: P« 0.001, trawl: P<0.005). Discussion Maturity schedules Although data used for maturity analysis were lim- ited to those collected in the same season (May and June) of each period, growth may occur, and stages of maturity may change, within two months. An im- mature fish in May may become mature in June. If more fish were collected in June of recent years than in 1979-81, the percent mature at a certain age would be overestimated for recent years. The monthly dis- tribution of observations, however, was similar among periods with more than 80% of the fish smaller than 170 mm being collected in May. Relative differ- ences in maturity schedules among periods remained valid throughout the study. The essential underlying assumption of the matu- rity analysis is that length, age, and reproduction conditions are correctly measured or determined. In 1985, 1986, 1987, and 1988, only fork lengths (FL) were measured. These FL were converted to TL with TL (mm) = 1.115PL - 0.254 (Zhao et al., 1997). For other years, observed TL was available. Because the method for ageing vermilion snapper by means of otolith sections has been validated by Zhao et al. ( 1997) and because the persons who read otoliths in Zhao et al.’s study also read otoliths in our study, incorrect ageing of vermilion snapper is not consid- ered a source of bias. During 1979-86, gonads were examined macroscopically by experienced biologists by means of clearly defined gross morphological stag- ing criteria. These criteria were confirmed to be ac- curate by histological examination during 1978-80. All gonads, since 1987, were examined by using reli- able histological techniques (Cuellar et al., 1996). Therefore, it is believed that sex and maturation were correctly determined. In general, sex and maturity stages can be more reliably determined during the spawning season than in off-seasons. Data used for maturity analysis were only from May and June, the first part of the spawning season, during which er- rors in maturity determination were not expected. Furthermore, in the present study, all maturity stages of mature fish were pooled in only one matu- rity state, i.e. mature. As long as immature and ma- ture fish could be distinguished, inaccurate classifi- cation of mature substage would not introduce a bias in estimates of age and size at maturity. This study indicated that both age and length at sexual maturity of vermilion snapper declined over time. This decline may have resulted from increased fishing pressure, because the total landings consis- 846 Fishery Bulletin 95(4), 1997 tently increased during the 1980’s (Zhao and McGovern2). The demonstration that the harvest of a fish stock can lead to declines in length or age at maturity has been reported for many fishes, includ- ing northeast Arctic cod (Jprgensen, 1990), Pacific salmon (Ricker, 1981), and California halibut (Love and Brooks, 1990). Changes in size or age at matu- rity may be the result of a density-dependent re- sponse to decreased stock abundance, selective re- moval and incomplete replacement of later-matur- ing fish by the fishery, or genetic change within a population (Nelson and Soule, 1987). Jprgensen (1990) attributed a decline in median age-at-matu- rity in northeast Arctic cod to an increase in length- at-age (i.e. faster growth) coincident with declining stock density, an idea that implicitly assumes a mini- mum threshold for size-at-maturity. If the scenario of Jprgensen ( 1990) is correct, declines in length and age should not occur concurrently. Furthermore, Zhao et al. ( 1997) indicated that the size-at-age of vermil- ion snapper has decreased with time. Therefore, changes in maturity schedules of vermilion snapper are not part of a density-dependent compensatory response to harvesting, but quite likely a result of the selective removal and incomplete replacement of faster-growing, later-maturing fish by the fishery. If intensive fishing pressure continues, and the early- maturing trait is heritable, length and age at matu- rity in the population will decrease with time. Life history theory predicts that genetic changes in life history characteristics will occur following increased mortality (Roff, 1992). Harvesting can reverse the relative fitness of genotypes, because an inferior genotype (e.g. slow-growing and early-maturing) in an unexploited population may be more fit under increased fishing pressure (Bergh and Getz, 1989). Early-maturing genotypes reproduce before being fully recruited to the fishery, whereas genotypes that mature at larger sizes or older ages tend to be re- moved before reproduction. This process would ex- plain the decreasing abundance of larger, immature fish with time and would account for declines in both size and age at maturity. The long-term impacts of size-selective fish harvests may have caused the de- cline in size-at-age of vermilion snapper through dis- proportionate harvesting of fast-growing individuals (Zhao et al., 1997). Similarly, it may be that late- maturing genotypes were removed from the vermil- ion snapper population in the 1980’s when fishing pressure was intensive. Maturity schedules of vermilion snapper collected during 1972-74, prior to heavy exploitation, were investigated by Grimes and Huntsman (1980). They used a gonadosomatic index and indicated that “most fish attain sexual maturity during their third or fourth years of life (186-256 and 256-324 mm TL), but a few precocious individuals may mature in their second year (100-186 mm TL) at about 150 mm TL.” It is not rigorous to compare the percent mature, based on age, between Grimes and Huntsman ( 1990) and the present study because an obvious discrep- ancy in size-at-age exists between Grimes (1978) and Zhao et al. (1997). It is meaningful, however, to com- pare maturity schedules based on length between these two studies. The maximum-likelihood esti- mates from the probit analysis of data from the present study predicted that 50% of males and 15% of females matured by 150 mm during 1979-81 and that 50% females at 150 mm matured during 1985- 87. All males and females at 180 mm were mature in the present study. Differences between previous (Grimes and Huntsman, 1980) and present results could be partially due to differences in methods used to determine maturity (Collins and Pinckney, 1988) or may truly reflect the changes in maturity that occurred in the 1970’s. The increase in percentage of mature females at 150 mm was faster during the 1980’s than during the 1970’s (i.e. an increase of 35% in six years from 1979-81 to 1985-87, versus an in- crease of less than 15% in seven years from 1972-74 to 1979-81). The degree of exploitation may account for the differing rates of change in maturity while the fishery for vermilion snapper was initiated in the 1970’s, but heavy exploitation did not occur until the 1980’s. Sex ratios The chi-square analysis did not suggest significant differences in percentages of females among months (May-August) for any gear type. This information supported the notion that pooling data between May through August would not bias the comparison of sex ratios. Seasonal comparisons of sex ratios could not be done because little sampling was done in fall or winter. This study showed that the sex ratio of vermilion snapper was dependent on area (latitude) and gear type, but independent of depth of sampling sites, fish length, or sampling years. The reason for the signifi- cant differences among latitudes is unknown. How- ever, only 2 of the 14 cases showed a significant dif- ference according to Bonferroni’s method (Table 4), and no trend was observed between latitudes. In addition, relatively small sample sizes collected from latitudes other than 32°N may have induced errors in comparison. Therefore, we attribute the difference in sex ratio between latitudes to chance. Although significantly different, the percentages of females of vermilion snapper collected by traps Zhao and McGovern: Variation in sexual maturity and sex ratio of Rhomboplites aurorubens 847 and hook-and-line were similar to each other (traps: 72.1%; hook-and-line: 68.0%) but differed from that for trawl capture (59.9%). Trawls caught smaller fish from shallower waters when compared with traps and hook-and-line (Tables 2 and 7C). However, present results indicated that sex ratios were not affected by water depth or fish length. With traps and hook-and-line gear, baits were used to attract fish. If female vermilion snapper were more aggres- sive in pursuing bait than males, the percentage of females in the catch of traps and hook-and-line could be higher than that in the population. In contrast, no baits were used for trawling, and therefore males and females might be caught with the same prob- ability. If the difference in feeding behavior between sexes can account for the difference in sex ratio be- tween gear, then the sex ratio of vermilion snapper in the population may be correctly represented by the trawl catch. Watanuki et al. ( 1993) reported that basket traps caught the greatest ratio of female cuttlefish among three types of gear (basket traps, jigs, and trammel nets). More females may be at- tracted to traps for spawning, but Watanuki et al. indicated that there are probably other unknown factors governing the entry of cuttlefish into traps. Because information on spawning behavior of ver- milion snapper is unavailable, we cannot evaluate how the spawning activity of vermilion snapper may affect its vulnerability to different gear types. We pooled data from all gear types and calculated the overall sex ratio by period. The percentage of fe- males gradually increased from 62% in 1979-81 to 70% in 1991-93. The temporal increase in the per- centage of females proved to be an artifact of unequal distribution of catch by gear among periods. Reasons for the difference in sex ratios among gear types are unknown. Caution must be used when evaluating the sex ratios of any fish species collected by various gear types. Our conclusion of independence between sex ra- tios and lengths differs from previous studies. Grimes and Huntsman (1980) concluded that the sex ratio of vermilion snapper was dependent on fish length, with the percentage of females increasing in larger size classes. However, the percentage of females within the range of 551-600 mm TL (89.3%, n= 32) was obviously higher than those for other length ranges. Thus, it is suspected that the significant chi- square calculated by Grimes and Huntsman (1980) was probably due to this length range. We used the original data published in Table 4 of Grimes and Huntsman (1980) but excluded the data with length greater than 550 mm TL (n= 32). We found that sex ratio was independent of length (x2=13.105, P=0.108, n=841, df=8, TL=101-55Q mm) and thus was in agreement with the conclusion of the present study. A further 2x2 contingency table analysis formed by the TL range of 551-600 mm versus all other length ranges rejected the null hypothesis of independence between sex ratio and length (%2=11.732, P=0.001, n= 873, df=l). Thus, we confirmed that the sex ratio within 551-600 mm TL is significantly different from those of other length ranges. Because our data had relatively few vermilion snapper larger than 450 mm TL, the conclusion of independence between sex ra- tio and length may be limited to 450 mm TL or less. However, the similar size-at-age and the same lon- gevity of male and female vermilion snapper do not suggest the percentage of females would increase with length even beyond 450 mm TL (Zhao et al., 1997). Acknowledgments We thank W. A. Roumillat and D. M. Wyanski for their advice and help with data analysis. We are grateful to the many individuals who participated in the field effort and to W. A. Roumillat, G. R. Sedberry, D. S. Vaughan, and to two anonymous re- viewers for helpful comments on an earlier draft of this paper. This research was supported by the Na- tional Marine Fisheries Service, MARMAP contract Number 50WCNF006002, and the South Carolina Department of Natural Resources. Literature cited Bergh, M. <)., and W. M. Getz. 1989. Stability and harvesting of competing populations with genetic variation in life history strategy. Theoretical Population Biology 36:77-124. Cuellar, N., G. R. Sedberry, and D. M. Wyanski. 1996. Productive seasonality, maturation, fecundity, and spawning frequency of the vermilion snapper, Rhomboplites aurorubens, off the southeastern United States. Fish. Bull. 94:635-653. Collins, M. R. 1990. A comparison of three fish trap designs. Fish. Res. 9:325-332. Collins, M. R., and J. L. Pinckney. 1988. Size and age at maturity for vermilion snapper (Rhomboplites aurorubens) (LUTJANIDAE ) in the south Atlantic bight. Northeast Gulf Science 10:51-53. 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. 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. Man- age. 9:383-391. Goodyear, C. P. 1993. Spawning stock biomass per recruit in fisheries man- 848 Fishery Bulletin 95(4), 1997 agement: foundation and current use. In S. J. Smith, J. J. Hunt, and D. Rivard (eds.), Risk evaluation and biologi- cal reference points for fisheries management, p. 67- 81. Can. Spec. Publ. Fish. Aquat. Sci. 120. Grimes, C. B. 1978. Age, growth, and length-weight relationships of ver- milion snapper, Rhomboplites aurorubens, from North Carolina and South Carolina waters. Trans. Am. Fish. Soc. 107:454-456. Grimes, C. B., and Gene R. Huntsman. 1980. Reproductive biology of the vermilion snapper, Rhomboplites aurorubens, from North Carolina and South Carolina. Fish. Bull. 78:137-145. Jorgensen, T. 1990. Long-term changes in age at sexual maturity of Northeast Arctic cod ( Gadus morhua L.). J. Cons. Cons. Int. Explor. Mer 46:235-248. Love, M. S., and A. Brooks. 1990. Size and age at first maturity of the California hali- but, Paralichthys californicus, in the southern California bight. Calif. Dep. Fish and Game, Fish Bull. 174:167- 174. Nelson, R. S. 1988. A study of the life history, ecology, and population dynamics of four sympatric reef predators (Rhomboplites aurorubens, Lutjanus campechanus, Lutjanidae; Haemulon melanurum, Haemulidae; and Pagrus pagrus, Sparidae) on the East and West Flower Garden Banks, Northwest- ern Gulf of Mexico. Ph.D. diss., North Carolina State Univ., Raleigh, SC, 197 p. Nelson, K., and M. Soule. 1987. Genetical conservation of exploited fishes. In N. Ryman and F. Utter (eds.), Population genetics and fish- ery management, p. 345-368. Univ. Washington, Seattle, WA, 420 p. Ricker, W. E. 1981. Changes in the average size and average age of Pa- cific salmon. Can. J. Fish. Aquat. Sci. 38:1636-1656. Roff, D. A. 1992. The evolution of life histories. Chapman and Hall, Ltd., New York, NY, 535 p. Rosenberg, A., and 14 coauthors. 1994. Scientific review of definitions of overfishing in U.S. fishery management plans. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-F/SPO-17, 205 p. SAS Institute, Inc. 1990. SAS/STAT® user’s guide, version 6, 4th ed. SAS Inst., Inc., Cary, NC, 1686 p. Sokal, R. R., and F. J. Rohlf. 1995. Biometry. W. H. Freeman and Company, New York, NY, 887 p. Trippel, E. A., and H. H. Harvey. 1991. Comparison of methods used to estimate age and length of fish at sexual maturity using populations of white sucker (Catostomus commersoni). Can. J. Fish. Aquat. Sci. 48:1446-1459. Watanuki, N., T. Iwashita, and G. Kawamura. 1993. Sex composition and sexual maturity of Sepia esculenta captured in cuttlefish basket traps. Nippon Suisan Gakkaishi 59:919-924. Wootton, R. J. 1990. Ecology of teleost fishes. Chapman and Hall, New York, NY, 404 p. Zar, J. H. 1984. Biostatistical analysis. Prentice Hall, Englewood Cliffs, NJ, 718 p. Zhao, B., J. C. McGovern, and P. J. Harris. 1997. Age, growth, and temporal change in size-at-age of the vermilion snapper from the South Atlantic Bight. Trans. Am. Fish. Soc. 126:181-193. 849 A polyphasic growth function for the endangered Kemp's ridley sea turtle, Lepidochelys kempii Milani Chaloupka Queensland Department of Environment PO Box 155, Brisbane Albert Street, Queensland 4002, Australia E-mail address: m.chaloupka@mailbox.uq edu.au George R. Zug Department of Vertebrate Zoology National Museum of Natural History Washington, D.C. 20560 The Kemp’s ridley, Lepidochelys kempii, is the smallest of the seven extant species of sea turtle ( Marquez, 1994) and is endemic to the Gulf of Mexico and Atlantic coast of the United States (Pritchard, 1989). It has been subject to extensive human exploitation and is the most endan- gered sea turtle species in the world (Marquez, 1994). Seasonal trawl and pound-net fisheries are major hazards, posing a serious risk to the long-term population viability of the Kemp’s ridley (Epperly et al., 1995; Caillouet et al., 1996). Al- though the Kemp’s ridley sea turtle is endangered, the somatic growth and population dynamics of this species are not well known (Cha- loupka and Musick, 1997) despite several important growth studies that have been carried out for cap- tive or head-started stocks (Cail- louet et al., 1986; Caillouet et al., 1995b). We propose a new growth model for the endangered Kemp’s ridley sea turtle that is based on a skeletochronological data set de- rived recently by Zug et al. (1997) from wild stock sea turtles stranded along the Atlantic Bight and Gulf coasts of the United States. The growth model presented provides a basis for improving our under- standing of sea turtle growth dy- namics in general and for modeling Kemp’s ridley population viability. Materials and methods Data set The data set used here comprised 70 size-at-age records for Kemp’s ridley sea turtles — 69 records from stranded turtles plus the inclusion of known mean hatchling size (see Marquez, 1994). The data set (n = 70) also comprised growth records spanning the postnatal de- velopment phase (from 4 to 72 cm straight carapace length, SCL) and including the mature adult phase, but the records were not distributed evenly over this size range. The age estimates were derived from a skeletochronological analysis of wild Kemp’s ridley sea turtles stranded along the Atlantic coast of the United States and in the Gulf of Mexico (see Zug et al., 1997). Straight carapace length (SCL) was measured to 0.1 cm and age to 0.1 yr. The original sample of stranded turtles comprised 73 individuals, but age estimates for 4 individuals were not possible because of either 1) a lack of discernible growth rings or 2) uninterpretable irregular growth. Records for these 4 individu- als were discarded, yielding the 69 individual turtles used in this study. The data set also included strand- ing location, with 79% of the sample comprising turtles stranded on the Atlantic coast. Sex was recorded for 37% of the strandings; no propor- tional difference was evident be- tween the Atlantic and Gulf of Mexico subsamples. Further details of the strandings data set and the skeletochronological methods used for age estimation can be found in Zug et al. (1997). The limitations of skeletochron- ological ageing techniques and the need for caution in interpreting such age estimates for sea turtles have been well discussed elsewhere (Zug et al., 1986; Zug, 1990; Zug et al., 1997). Chaloupka and Musick (1997) have also provided a critical review of sea turtle skeletochron- ological studies and have discussed the limitations of such studies in terms of age validation, length back-calculation, growth estima- tion, layer loss adjustment proto- cols, and implications of the specific time-dependent sampling design implicit in the data set. For in- stance, the implicit sampling de- sign in the current study was mixed cross-sectional because only the ter- minal age-size estimate was avail- able for each of the 69 stranded turtles. This sampling design con- founds age and cohort effects and thus only an expected or mean growth function can be estimated (see Chaloupka and Musick, 1997). Statistical modeling approach The functional relation between size (cm SCL) and estimated age for the 70 Kemp’s ridley sea turtles was modeled with a two-stage ap- proach: 1) exploratory data visual- ization including nonparametric smoothing (see Cleveland, 1993) to Manuscript accepted 4 April 1997. Fishery Bulletin 95:849-856 (1997). 850 Fishery Bulletin 95(4), 1997 evaluate the implicit functional form of the growth model without having to specify an explicit and per- haps invalid nonlinear function; and 2) a polyphasic parametric growth function fitted to the size-at-age data on the basis of the functional form implied by the nonparametric smooth. Polyphasic growth means that there is more than one growth phase or cycle in postnatal development, suggesting ontogenetic shifts in growth rates manifested by at least two growth spurts between birth and the onset of adult matu- rity (see Gasser et al., 1984). The polyphasic growth function used in our study was the Peil and Helwin (1981) parameterization comprising a summation of logistic functions as follows: yt =X{a'[1 + tanh(^a_^))]} + e<' (1) i= 1 assumed that the polyphasic form (Eq. 1) used here also has sound statistical properties. In principle, Equation 1 was fitted by heteroscedasticity-robust nonlinear least-squares (HRNLS) with a hetero- scedasticity-consistent covariance matrix estimator (HCCME) to account for growth variability and mea- surement error (see Davidson and MacKinnon, 1993). In practice, Equation 1 was fitted with RATS (Doan, 1992), which implements HRNLS with White’s HCCME. Otherwise, the generalized method of mo- ments (GMM) approach can be used for robust non- linear regression estimation (Davidson and Mac- Kinnon, 1993). The age-specific growth-rate function for the Kemp’s ridley sea turtle was derived analyti- cally by taking the first derivative of the fitted Equa- tion 1 with the software program MATHEMATICA (Wolfram Research, 1993). where yt = tanh(z) = j = mean length at age t ; (asymptotic mean)/; length in phase i\ growth coefficient in phase i; age at the inflection point of phase i; ( ez - e~z)/(ez + e~z ) and 2 = ((!.(/ - 8;)); number of growth phases; and an appropriate random error structure. Parameters of the standard logistic function (mono- phasic with skewed symmetric inflexion and suggest- ing one growth spurt) are well known to have excel- lent statistical properties (Ratkowsky, 1990). It was Results The size and estimated age data for the 70 Kemp’s ridley sea turtles presented in Zug et al. (in press) are shown in Figure 1A with a locally weighted re- gression smoothing known as LOWESS (see Cleve- land, 1993) superimposed to reveal the implicit func- tional form. The LOWESS procedure can be imple- mented by using S software (Becker et al., 1988). The nonparametric smooth (Fig. 1A) implies a polyphasic function with two sequential growth phases, with the first decelerating around 30 cm SCL and the second Age estimate (years) Figure 1 (A) Scatterplot of size-at-age estimates for the 69 Kemp’s ridley sea turtles stranded along the Atlantic Bight and Gulf coasts of the United States, with an additional estimate of mean hatchling size (age=0 yr). Open circles and solid dot are the original data estimates (n= 70) from Zug et al. (1997). Solid dot is the outlier discounted in the parametric model (Table 1). The curve in (A) is a LOWESS (locally weighted robust regression) smooth superimposed to highlight the underlying size-at-age function without presuming the functional form. (B) Scatterplot of the Atlantic Bight subsample estimates (n= 55), with hatchling size included and a LOWESS smooth superimposed. (C) Scatterplot of the Gulf of Mexico subsample estimates (n=14), with hatchling size included and a LOWESS smooth superimposed. NOTE Chaloupka and Zug: A polyphasic growth function for Lepidochelys kempii 851 at >60 cm SCL. It is proposed that a polyphasic growth model might be a better mathematical de- scription of growth than the monophasic (monotonic) von Bertalanffy (Caillouet et al., 1995b; Schmid, 1995; Zug et al., 1997) or the monophasic (non- monotonic) Gompertz functions (Caillouet et al., 1986) proposed for this species. A similar polyphasic growth function comprising two phases is also evi- dent for the Atlantic Bight subsample (Fig. IB) and is suggested for the Gulf subsample (Fig. 1C) despite a very sparse data field in the latter case. Figure 1 also highlights the considerable variabil- ity (heterogeneity) inherent in sea turtle growth and why heteroscedasticity-robust estimation procedures (e.g. HCCME, GMM) should be used to derive re- gression parameter estimates for growth model fits. There is also a major outlier in Figure 1A indicated by a solid dot — this value was discounted in the ex- plicit parametric model fit because no parametric model could be as robust in respect to this outlier as the nonparametric smooth displayed in Figure 1A. Growth variability in sea turtle studies is a complex function of demographic (sex, maturity status) and geographic factors as well as a function of the time- dependent nature of the implicit sampling design (confounding year and cohorts effects) and instru- mental measurement error. For instance, Caillouet et al. (1986) have shown conclusive evidence of so- matic growth variability due to cohort (year-class) effects for captive reared Kemp’s ridley sea turtles. The small sample size, mixed cross-sectional sam- pling design, and insufficient data on demographic and geographic covariates precluded any reliable estimate of these additional sources of growth record variability in the current study. The parametric growth curve proposed here to match the nonparametric smooth (Fig. 1A) for the Kemp’s ridley data comprises separate logistic growth functions for each of the two inferred growth phases integrated into a single explicit polyphasic function — Equation 1. The statistical fit of this func- tion to the growth data (Fig. 1A) is shown in Table 1. The growth model with robust estimation and with elimination of the extreme outlier (see Fig. 1A) fit- ted the data well with significant parameter esti- mates even allowing for family-wise error-rate ad- justment, small parameter estimate standard errors, and no aberrant residual behavior (see Judge et al., 1985, or Ratkowsky, 1990, for a discussion of nonlin- ear regression fitting and goodness-of-fit criteria). Despite the good fit, significant growth variability, probably due to instrumental measurement error and confounding of year and cohort sampling effects, was not accounted for by the model (residual variance: g2=29.1). Table 1 Parameter estimates for the polyphasic logistic growth function (Eq. 1) fitted to the Kemp’s ridley sea turtle size- at-age growth data in Zug et al. (1997). See Equation 1 for definitions of parameters. Asymptotic Parameter Estimate standard error t-ratio Inference «i 13.6467 2.7463 4.97 P < 0.001 Pi 0.7901 0.2989 2.64 P < 0.008 »i 1.1169 0.4303 2.59 P < 0.009 a2 17.6595 3.9288 4.49 P< 0.001 P2 0.3059 0.1274 2.40 P < 0.016 S2 7.6361 0.5407 14.12 P< 0.001 The expected polyphasic size-at-age function is shown in Figure 2 A (age=skeletochronological age estimate) and presented numerically in Table 2 for comparative purposes. The explicit size-at-age growth function (Fig. 2A) was then differentiated with respect to estimated age by an analytical solu- tion to Equation 1 to derive the age-specific growth rate function (Fig. 2B). The expected age-specific growth rate function (Fig. 2B) displays an initial posthatchling growth rate >5 cm SCL/year, increas- ing to 11 cm SCL/year >1 year of age (A1) or 13 cm SCL, slowing to 2 cm SCL/year by 3-4 years of age (ca. 27 cm SCL), marking the end of the first growth phase (i.e. 20^=27.3 in Table 1; mid-curve asymptote in Figs. 1A and 2A). The growth rate then rises to 6 cm SCL/year near 8 years of age (52) or to 46 cm SCL before declining slowly to negligible growth ap- proaching adulthood >15 years of age at a size >62 cm SCL, marking the end of the second growth phase (i.e. 2(a1+a2)=62.6 in Table 1; upper asymptote in Figs. 1A and 2A). Discussion Monophasic von Bertalanffy growth functions have been proposed for the Kemp’s ridley sea turtle by Caillouet et al. (1995b), Schmid (1995), Zug et al., (1997), and others (Marquez, 1994, and references therein). With the von Bertalanffy growth function, however, a monotonic decreasing growth-rate func- tion is implied and hence no growth spurt at any age or size. The statistical validity of that function fitted to a limited data span and of the Fabens mark- recapture analogue used by Schmid (1995) and Zug et al. (1997) has been reviewed critically by Chaloupka and Musick (1997). It is questionable whether a monophasic von Bertalanffy function fits the mean growth profile for the complete postnatal 852 Fishery Bulletin 95(4), 1 997 Table 2 Comparison of size-at-age growth functions for three Kemp’s ridley sea turtle growth models. Age = known age for the Caillouet et al. (1995b) model, whereas age = skeletochronological age estimate for the Zug et al. (1997) model and the polyphasic model presented here. SCL = straight carapace length. Size-at-age estimate (cm SCL) Size-at-age estimate (cm SCL) Age (years) Caillouet et al. (1995b) Zug et al. (1997) This study (Fig. 2A) Age (years) Caillouet et al. (1995b) Zug et al. (1997) This study (Fig. 2 A) 0 (hatchling) 2.79 8.86 4.32 13 61.30 59.49 61.33 1 18.95 14.85 12.99 14 61.57 61.63 61.91 2 30.72 20.39 22.96 15 61.76 63.62 62.23 3 39.29 25.50 27.93 16 61.90 65.45 62.40 4 45.53 30.23 30.46 17 62.00 67.14 62.50 5 50.08 34.60 33.11 18 62.07 68.70 62.55 6 53.39 38.63 36.77 19 62.13 70.15 62.58 7 55.80 42.36 41.56 20 62.17 71.48 62.59 8 57.56 45.81 46.91 21 62.19 72.72 62.60 9 58.84 48.99 51.92 22 62.21 73.86 62.61 10 59.77 51.94 55.88 23 62.23 74.91 62.61 11 60.45 54.65 58.61 24 62.24 75.89 62.61 12 60.94 57.17 60.32 25 62.25 76.79 62.61 development phase of any sea turtle species (see Chaloupka and Limpus, 1997; Chaloupka and Musick, 1997). On the other hand, the monophasic form of the Gompertz growth function used by Caillouet et al., (1986) in a single cohort growth study (weight gain) of 10 Kemp’s ridley sea turtles held in captivity clearly fitted the data well at least for the observed range (ca. 2-7 years old). In the Gompertz function, a nonmonotonic growth-rate function is assumed with a growth spurt in early development similar to the first growth spurt in our study (see Fig. 2B). Whether growth in Caillouet et al.’s (1986) study might have been better fitted by using a polyphasic Age (years) o 5 10 15 20 Age (years) Figure 2 (A) Kemp’s ridley sea turtle size (SCL cm) plotted as a function of the correction-factor age estimates derived from Zug et al. (1997). Solid curve shows the polyphasic logistic growth Equation 1 fitted to the growth data shown in Figure 1A, excluding the single outlier. (B) The age-specific growth-rate function for the Kemp’s ridley sea turtle growth function shown in Panel A represented by the first derivative of Equation 1, which isy'w = X (O', P, (sech2 (P, (t - 5; )))) with the same parameters defined for Equation 1 and with sech (hyperbolic secant) = (1 - tanh). NOTE Chaloupka and Zug: A polyphasic growth function for Lepidochelys kempii 853 function is inconclusive because the data span was incomplete, missing not only the first growth cycle (if it occurred at all) but also the onset of adult matu- ration. By the end of the study the remaining 8 turtles were still growing and below estimates of adult size (weight) recorded for wild stocks. Moreover, growth in captivity might well bear little similarity to the growth dynamics of wild stock Kemp’s ridley sea turtles, which seem to grow much slower at a given size (Caillouet et al., 1986). The nonparametric smooth shown here in Figure 1A fitted to a more complete age and size range for wild stock Kemp’s ridley sea turtles implied that growth comprised two consecutive phases and that an explicit polyphasic model (Table 1; Fig. 2) might be a better parametric description of growth than monophasic models proposed previously for this spe- cies. Nonetheless, two major cautions are warranted prior to drawing further conclusions from Figure 2 about Kemp’s ridley growth dynamics. These cau- tions relate to the implications for growth inferences due to 1 ) data sparsity in the early growth years for this data set and 2) the size composition anomaly between stranding subsamples for this data set. The data were sparse in the lower region of the first inferred growth spurt (see Fig. 1). The growth- layer loss protocols used in deriving the skeleto- chronological age estimates provided differing cov- erage of this growth region (see Zug et al., 1997). The age estimates used in our study were based on the correction-factor protocol that was considered more reliable than ranking protocol estimates (Zug et al., 1997) despite providing no coverage of the first growth year except for hatchling size. The model that was fitted (Fig. 2A) interpolated between known hatchling age and the end of the first year on the basis of the explicit form implied by the specified parametric function (Eq. 1). Although the conclusion of the first growth phase completed by ca. 25-30 cm SCL (see Fig. 1A) is firm despite sparse data during early growth, a specific growth spurt >1 year of age is tenuous. Given the lack of data during the first year, maximum growth might just as feasibly occur immediately following hatching, resulting in a mono- tonic decreasing age-specific growth-rate function for the first cycle and not the nonmonotonic function seen in Figure 2B. On the other hand, a growth spurt might occur a little later after hatching, resulting in a nonmonotonic age-specific growth-rate function for the first cycle similar to that proposed in Figure 2B but with the spurt occurring at say 3 or 6 months rather than at 12 months of age. Because of a spar- sity of data during the early growth years, all these growth scenarios for the first growth cycle are fea- sible; therefore data for the first 12 months of life following hatchling dispersal from the nesting beach are essential to resolve this important issue. Nonetheless, other sources of information corrobo- rate the growth profile proposed for the first phase in Figure 2B. First, the polyphasic function described by Equation 1 fitted the growth data well, including an estimated mean adult size (upper asymptote= 2(aj+a2)=62.6 cm SCL from Table 1) consistent with empirical estimates of mean nesting female size of 64-65 cm SCL (Marquez, 1994). Moreover, the polyphasic function also predicted a mean hatchling size of 4.3 cm, which is consistent with the empirical estimate of mean hatchling size of 4.4 cm (Marquez, 1994). No other growth model has come close to pre- dicting both the mean upper and lower size asymp- totes of the postnatal development phase for the Kemp’s ridley sea turtle. It is worth noting here that it is a common misconception in growth studies (par- ticularly sea turtle growth studies) involving more than one animal that the upper asymptote of a para- metric growth function estimates maximum adult size rather than the correct interpretation of mean adult size (see Ricker, 1979). Second, a growth spurt >12 months of age and a growth phase completed by ca. 27 cm SCL (30-36 months of age) is coincident with developmental changes to the blood oxygen system of the Kemp’s ridley sea turtle prior to acquisition of an adult blood system by 28 months of age (see Davis, 1991) — at least this was the case for captive-reared Kemp’s rid- ley sea turtles. Davis ( 1991) also found that the oxy- gen capacity of the blood had increased substantially during the first 12 months of growth. The size range and timing of the first growth cycle is also consis- tent with apparent dietary and habitat shifts around 20 cm SCL (ca. 18 months of age: Fig. 2A; Eq. 1 ) from a presumed epipelagic habit to a coastal benthic habit (see Shaver 1991; Burke et al., 1994; Musick and Limpus, 1997). The second major caution relates to a lack of infor- mative cofactors (sex, geographic subsample) being included in the model because of insufficient records or small subsamples. For instance, sex was recorded for only 37% of the strandings, whereas the Atlantic coast subsample (cf. Gulf coast) accounted for 79% of the strandings data (see Fig. 1, B and C). Moreover, the Atlantic subsample comprised a significantly dif- ferent size composition compared with that of the Gulf of Mexico (see Fig. 3). The apparent size compo- sition anomaly might be due to 1) differential and inadequate spatial sampling of strandings and 2) developmental migration of Kemp’s ridley sea turtles >40 cm SCL from the Atlantic coast to the Gulf of Mexico (see Collard and Ogren, 1990; Morreale et al., 1992; Epperly et al., 1995; Musick and Limpus, 854 Fishery Bulletin 95(4), 1 997 • extreme outlier Atlantic Gulf coast coast Figure 3 Boxplots of the size distribution of the Kemp’s ridley sea turtle from the Atlantic Coast and Gulf of Mexico subsamples. The boxes show the interquartile range (25th to 75th percen- tiles), bars show 10th and 90th percentiles, and the notches show comparison-wise 95% confidence intervals. The notches on the two boxplots do not overlap each other, indicat- ing a significant difference in median size between the two subsamples. 1997). If the Atlantic and Gulf coast subsamples in the Zug et al., (1997) data set represent two discrete populations with population-specific growth behav- iors, then the growth model here (Table 1; Fig. 2) is applicable to the Atlantic group only although a simi- lar polyphasic model is apparent for both subsamples despite a sparsity of data for the >50 cm SCL group of the Atlantic subsample (Fig. IB) and <40 cm SCL group of the Gulf subsample (Fig. 1C). The inclusion of the Gulf subsample serves to provide sufficient data to derive the upper asymptote for estimating mean adult size (see Fig. 1C). The Atlantic subsample comprised only immature Kemp’s ridley sea turtles consistent with recorded size distributions for popu- lations resident in various habitats along the US At- lantic coast (see Burke et al., 1994, Epperly et al., 1995; Schmid, 1995). If the Zug et al. (1997) data set is representative of a single panmictic interbreeding stock displaying some form of staged developmental migration, then the model presented here is a reasonable approxi- mation of the growth dynamics of the endangered Kemp’s ridley sea turtle. There is compelling sup- port for this view given current knowledge of Kemp’s ridley sea turtle movement patterns (Musick and Limpus, 1997). Further support comes from the dis- covery of a Kemp’s ridley sea turtle (70 cm CCL, 67 cm SCL) nesting at Rancho Nuevo in 1996 (116 eggs laid) that had been tagged as a juvenile (51 cm CCL, 49 cm SCL) seven years earlier in Chesapeake Bay (Musick1 ). Growth for this nesting ridley was con- sistent with the polyphasic growth function (Fig. 2A) although clearly a single record is not sufficient to provide conclusive evidence. Nonetheless, if the At- lantic and Gulf subsamples represent a single pan- mictic interbreeding stock, then a juvenile growth spurt at 46 cm SCL (ca. 8 years old, Fig. 2B) would indicate an ontogenetic shift associated with devel- opmental migration from juvenile foraging habitats in the South Atlantic Bight (Musick and Limpus, 1997) and from within the Gulf of Mexico (Collard and Ogren, 1990) to foraging grounds in habitats along the Gulf coast prior to the onset of sexual maturity. It is also conceivable, given the dispersal scenarios proposed by Collard and Ogren (1990), that the Kemp’s ridley sea turtle is a single panmictic inter- breeding stock that comprises two distinct post- hatchling developmental groups. One group remains within the Gulf of Mexico displaying relatively rapid growth owing to the warmer water (see Caillouet et al., 1995b). The second group represents the posthatchlings swept from the Gulf of Mexico that settle as juveniles (ca. 20 cm SCL) in the inshore developmental habitats of the mid-Atlantic (Morreale et al., 1992; Burke et al., 1994) and South Atlantic Bights (Epperly et al., 1995; Schmid, 1995). In this case the polyphasic growth model presented here (Table 1; Fig. 2) would be applicable to describing the mean stochastic growth dynamics of the cohorts swept each year from the Gulf of Mexico and under- going growth in the Atlantic Bights prior to migrat- ing back to the Gulf of Mexico. A separate growth model would need to be derived for the Gulf of Mexico devel- opmental group although polyphasic growth behavior is also apparent for that subsample in our study (see Fig. 1C). Clearly, a better understanding of the dispersal and developmental dynamics of the Kemp’s ridley sea turtle based on a mark-recapture program with a high recapture likelihood is needed to resolve these complex issues. Although several local tagging pro- grams have been undertaken (e.g. Caillouet et al., 1995a; Schmid, 1995; Burke et al., 1994, and refer- ences therein) a more comprehensive spatial and sampling-intensive program spanning the distribu- tional range of this species is needed. 1 Musick, J. 1997. Virginia Institute of Marine Science, Col- lege of William and Mary, Gloucester Point, VA. Personal commun. NOTE Chaloupka and Zug: A polyphasic growth function for Lepidochelys kempii 855 Meanwhile, it is common practice to use somatic growth functions to estimate mean age at sexual maturity. The difficulty in using growth functions for this purpose is that there are no conclusive growth criteria to indicate onset of sexual maturity. Mini- mum or mean female nesting size, or an arbitrary size set slightly below mean nesting size, is a com- monly used criterion. Using an arbitrary size crite- rion based on reasonable biological considerations, Caillouet et al., (1995b) proposed that head-started Kemp’s ridley sea turtles irrespective of sex, took 10 years to reach sexual maturity at ca. 60 cm SCL. Zug et al. (1997) estimated 11-16 years for age-at- maturity for wild stock Kemp’s ridley sea turtles on the basis of mean female nesting size. The upper asymptote of a parametric growth function is the correct estimate of mean adult size, if the correct growth function was used (see Ricker, 1979). By us- ing the upper asymptote metric, it is then apparent that sexual maturity could be reached at >20 years of age for the current study (see Fig. 2B; Table 2) compared with 30 years of age for the Caillouet.et al. (1995b) growth function (see Table 2). But as Caillouet et al., (1995b) point out, mean adult size (nesting females) is a questionable crite- rion for estimating age at sexual maturity. That is why Caillouet et al. (1995b) defined an arbitrary size criterion to estimate age-at-maturity. However, the correct function for estimating mean age at matu- rity is an age-specific maturity-rate function condi- tioned on time-varying age, year, and cohort effects derived from a mixed longitudinal sampling study (see Chaloupka and Musick, 1997, for time-depen- dent demographic sampling issues). In the absence of such a complex function, a useful growth criterion for estimating age-at-maturity might be negligible growth derived from the age-specific growth-rate function indicating the onset of maturity. It is increas- ingly apparent that growth for sea turtles becomes negligible approaching the onset of sexual maturity (see Chaloupka and Limpus, 1997). Although this is a study-dependent and subjective metric, it is clear that negligible growth in the current study occurs > 15 years of age or < 0.25 cm SCL/year (see Fig. 2B and Table 2). Growth was imperceptible by 21 years of age (Table 2); thus the age range of 15-20 years appears to be a reasonable interval estimate of ex- pected age at sexual maturity for the Kemp’s ridley sea turtle. It is therefore noteworthy that the Kemp’s ridley sea turtle tagged in Chesapeake Bay as a juvenile (ca. 49 cm SCL) and discovered nesting seven years later at Rancho Nuevo (see “Discussion” above) was estimated by reference to Figure 2 to be about 9 years old when tagged and therefore 16 years old at the first recorded nesting. The nesting turtle was ca. 67 cm SCL, which is larger than the estimated upper asymptote of ca. 63 cm SCL (Table 1: 2(a1+a2) cm SCL). Recall, however, that the upper asymptote here represents mean adult size (or mean nesting size, as- suming growth is not sex-specific); therefore 50% of a random sample of adult or nesting Kemp’s ridley sea turtles would be >63 cm SCL , whereas 50% of the sample would be smaller. Despite sampling design constraints, cautions about skeletochronological methods, small sample size, and perhaps nonequivalent geographic sub- samples, the data set presented in Zug et al. (1997) is of considerable importance for helping to improve our understanding of the growth dynamics of the en- dangered Kemp’s ridley sea turtle. The re-analysis of these data with exploratory nonparametric smoothing suggested that expected age-specific growth for the Kemp’s ridley sea turtle was polyphasic and could be modeled with a sequence of parametric curves. A parametric model comprising a summation of logistic functions fitted the data well, implying growth spurts at >1 year of age (mean size=13 cm SCL) and ca. 8 years of age (mean size=46 cm SCL) followed by negligible growth approaching the onset of maturity ca. 15-20 years of age (mean size=63 cm SCL). Polyphasic growth behavior is therefore one of many reasons why a monophasic growth function cannot fit the entire postnatal de- velopmental phase of the Kemp’s ridley sea turtle, let alone for any other sea turtle species (see Chaloupka and Musick, 1997). Acknowledgments We thank members of the USA sea turtle stranding network for collection of skeletal material and Heather Kalb and Stephen Luzar for processing skeletochronological data. We are grateful for the Smithsonian Research Opportunities Fund that pro- vided support of the skeletochronological work be- ing undertaken by George Zug. We thank Charles Caillouet, Col Limpus, Jack Musick, and the anony- mous referees for helpful comments on the manuscript. Literature cited Becker, R. A., J. M. Chambers, and A. R. Wilks. 1988. The new S language. Chapman and Hall, New York, NY, 684 p. Burke, V. J., S. J. Morreale, and E. A. Standora. 1994. Diet of the Kemp’s ridley sea turtle, Lepidochelys kempii, in New York waters. Fish. Bull. 92:26-32. 856 Fishery Bulletin 95(4), 1997 Caillouet, C. W., C. T. Fontaine, S. A. Manzella-Tirpak, and D. J. Shaver. 1995a. Survival of head-started Kemp’s ridley sea turtles ( Lepidochelys kempii ) released into the Gulf of Mexico or adjacent bays. Chelonian Conserv. Biol. 1:285-292. Caillouet, C.W., C. T. Fontaine, S.A. Manzella-Tirpak, and T. D. Williams. 1995b. Growth of head-started Kemp’s ridley sea turtles (. Lepidochelys kempii) following release. Chelonian Conserv. Biol. 1:231-234. Caillouet, C. W., D. B. Koi, C. T. Fontaine, T. D. Williams, W. J. Browning, and R. M. Harris. 1986. Growth and survival of Kemp’s Ridley sea turtle Lepidochelys kempi in captivity. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFC-186, 34 p. Caillouet, C. W., 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. Chaloupka, M. Y., and C. J. Limpus. 1997. Robust statistical modelling of hawksbill sea turtle growth rates (southern Great Barrier Reef). Mar. Ecol. Prog. Ser. 146:1-8. Chaloupka, M. Y., and J. A. Musick. 1997. Age, growth and population dynamics. Chapter 9 in P. L. Lutz and J. A. Musick (eds.), The biology of sea turtles, p. 233-276. CRC Marine Science Series, CRC Press Inc., Boca Raton, FL. Cleveland, W. S. 1993. Visualizing data. Hobart Press, NJ, 360 p. Collard, S. B., and L. H. Ogren. 1990. Dispersal scenarios for pelagic post-hatchling sea turtles. Bull. Mar. Sci. 47:233-243. Davidson, R., and J. G. MacKinnon. 1993. Estimation and inference in econometrics. Oxford Univ. Press, New York, NY, 874 p. Davis, B.J. 1991. Developmental changes in the blood oxygen trans- port system of Kemp’s ridley sea turtle, Lepidochelys kempi. Can. J. Zool. 69:2660-2666. Doan, T. 1992. RATS 4.0 (Regression Analysis of Time Series) user’s manual. Estima, Evanston, IL, 602 p. Epperly, S. P„ J. Braun, A. J. Chester, F. A. Cross, J. V. Merriner, and P. A. Tester. 1995. Winter distribution of sea turtles in the vicinity of Cape Hatteras and their interactions with the summer flounder trawl fishery. Bull. Mar. Sci. 56:547-568. Gasser, T., H-G. Muller, W. Kohler, L. Molinari, and A. Prader. 1984. Nonparametric regression analysis of growth curves. Ann. Statistics 12:210-229. Judge, G. G., W. E. Griffiths, R. C. Hill, H. Lutkepohl, and T-C. Lee. 1985. Theory and practice of econometrics, second ed. Wiley and Sons, New York, NY, 1124 p. Marquez M, R. 1994. Synopsis of biological data on the Kemp’s ridley turtle Lepidochelys kempi (Garman 1880), p. 1-91. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFSC-343. Morreale, S. J., A. B. Meylan, S. S. Sadove, and E. A. Standora. 1992. Annual occurrence and winter mortality of marine turtles in New York waters. J. Herpetol. 26:301-308. Musick, J. A., and C. J. Limpus. 1997. Habitat utilization and migration in juvenile sea turtles. Chapter 6 in P. L. Lutz and J. A. Musick (eds.), The biology of sea turtles, p. 137-163. CRC Marine Sci- ence Series, CRC Press Inc., Boca Raton, FL. Peil, J., and H. Helwin. 1981. A phenomenologic-mathematical model of growth dynamics. Biom. J. 23:41-54. Pritchard, P. C. H. 1989. Evolutionary relationships, osteology, morphology and zoogeography of Kemp’s ridley sea turtle. In C. W. Caillouet and A. M. Landry (eds.). Proceedings of the first international symposium on Kemp’s ridley sea turtle biol- ogy, conservation and management, p. 157-164. Texas A&M Univ., Sea Grant College Program, TAMU-SG-89-105. Ratkowsky, D. A. 1990. Handbook of nonlinear regression models. Marcel Dekker, New York, NY, 241 p. Ricker, W. E. 1979. Growth rates and models. Chapter 11 in W. S. Hoar, D. J. Randall, and J. R. Brett (eds.,) Fish physiology, vol. 8: Bioenergetics and growth. Academic Press, New York, NY. Schmid, J. R. 1995. Marine turtle populations on the east-central coast of Florida: results of tagging studies at Cape Canaveral, Florida, 1986-1991. Fish. Bull. 93:139-151. Shaver, D. 1991. Feeding ecology of wild and head-started Kemp’s ridley sea turtles in south Texas waters. J. Herpetol. 25:327-334. Wolfram Research. 1993. MATHEMATICA 2.2 — a system for doing mathe- matics. Wolfram Research, Inc., Champaign, IL, 961 p. Zug, G. R. 1990. Age determination of long-lived reptiles: some tech- niques for sea turtles. Annales des Sciences Naturelles, Zoologie, Paris 13e, serie 11:219-222. Zug, G. R., H. J. Kalb, and S. J. Luzar. 1997. Age and growth in wild Kemp’s ridley seaturtles, Lepi- dochelys kempii, from skeletochronological data. Biol Conserv. 80:261-268. 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. Smithsonian Contrib. Zool. 427:1-34. 857 Age determination of larval strombid gastropods by means of growth increment counts in statoliths Felix A. Grana-Raffucci* Richard S. Appeldoorn** Department of Marine Sciences University of Puerto Rico Mayaguez, Puerto Rico 00681-5000 E-mail address (for R.S. Appeldoorn): R_Appeldoorn@rumac.upr.clu.edu ‘Present Address: Coastal Zone Division Department of Natural and Environmental Resources PO Box 5887, San Juan, Puerto Rico 00906 The queen conch, Strombus gigas Linnaeus, and the milk conch, S. costatus Gmelin, are important gas- tropods of the Caribbean region (Appeldoorn and Rodriguez, 1994). To age strombid larvae by means of their statoliths would be useful in order to study aspects of their larval life histories and ecology. Statoliths, like fish otoliths, are formed of aragonitic calcium car- bonate deposited on a protein ma- trix and exhibit periodic growth increments (Radtke, 1983). Re- search on statolith microstructure has been limited primarily to age determination of commercially im- portant cephalopods (e.g. Jackson, 1994). D’Asaro (1965) observed sta- tocysts that appeared in four-day- old embryos of S. gigas and that were fully functional by day six; he also noted growth increments or rings on these structures. These increments were confirmed by Salley (1986). Our objective is to validate the use of statolith micro- structure in S. gigas and S. costatus to provide information on age, growth, and length of larval life. Materials and methods Egg masses from eight Strombus costatus and one S. gigas were col- lected from ovidepositing females at a site 7 km south of La Parguera, southwest coast of Puerto Rico (17.92°N, 67.05°W). Egg masses were held in 75-L aquaria subjected to the natural light-dark cycle. Cul- ture methods of Ballantine and Appeldoorn (1983) were used. Aquaria were cleaned daily, after which one liter of Tahitian Iso- chrysis ( 106 cells/mL) was added. A minimum daily sample of ten indi- viduals was removed from each aquarium and preserved in 70% ethanol. We examined the statolith microstructure of larvae from the longest surviving cultures. Since no veligers reared in our laboratory developed through metamorphosis, we obtained preserved (5% buffered formalin, pH 8.0) S. gigas veligers and juveniles of known age from the Trade Wind Industries’ hatch- ery in the Turks and Caicos Islands. Preserved veligers were exam- ined with a dissecting microscope, their larval shell length (apex to siphonal canal) was measured, and their shell removed. A drop of 60% solution of alizarin red in glycerin was added to increase contrast be- tween stained soft tissues and the unstained statolith. Coverslips were added, sealed with Permount, and samples were inspected under a compound scope at l,000x. Sta- tolith diameters were measured with an ocular micrometer. Incre- ments on statoliths were counted by focusing up and down through the statolith. For each day of age, counts were made from one sta- tolith from each of 20 veligers; all counts were made by the same reader in a blind manner. General physical structure was observed and described, with emphasis placed on the periods preceding and following hatching, and, for S. gi- gas juveniles, preceding and follow- ing metamorphosis to determine the presence and nature of any transitional marks associated with these events. Statolith diameter, number of growth increments, and shell length were averaged for each day and ar- ranged with age (in days after hatching). Linear least-squares re- gressions were calculated to deter- mine the relations among these three variables. To determine the precision (or reproducibility) of counts of increments (i.e. verifica- tion, Wilson et al., 1983), repeated counts for both statoliths were made on subsamples. The number of increments in these representa- tive samples were counted three separate times (double-blind), and the results were averaged for indi- vidual veligers. Standard devia- tions (SD) were calculated for each individual; standard error of the means (SE) was calculated as ap- propriate. A nested analysis of vari- ance (ANOVA) was done for each species to determine if variability in incremental counts was due to errors in measurement or to natu- ral variability in increment depo- sition (Sokal and Rohlf, 1981). Data were grouped at four levels: 1) all ** Author to whom correspondence should be addressed. Manuscript accepted 11 April 1997. Fishery Bulletin 95:857-862 (1997) 858 Fishery Bulletin 95(4), 1997 individuals across age groups, 2) between individuals within age groups, 3) between statoliths within indi- viduals, and 4) within statoliths. For one culture of eight-day-old veligers of S. costatus, feeding was suspended for two consecutive days to induce change in statolith structure and to determine if changes in feeding regimen affected in- crement deposition. Statolith-increment numbers and statolith diameters (rc=10) were compared with those of nonstarved larvae (n=10) of equivalent age with two-sample /-tests (Sokal and Rohlf, 1981). Results Statoliths from Strombus costatus and S. gigas ve- ligers were similar in size, shape, and pattern of in- crement formation and were translucent in appear- ance. They have either a circular or elliptical appear- ance from a longitudinal view and a biconvex struc- ture in transverse section. On the basis of changes in size and shape, three regions are apparent (Fig. 1). At the very center is a primordial granule, around which all other increments grow; newly hatched ve- ligers show five increments (including the primor- dial granule). This five-increment region (region 1) is quite distinct because of its lighter color, greater width, and seemingly dome-like nature. Prehatching increments in S. costatus and S. gigas had mean widths of 1.11 pm and 1.22 pm, respectively. Depos- ited around region 1 is a second, slightly darker, re- gion. Thinner and smaller in width, this region (re- gion 2) is composed of increments formed between hatching and completion of metamorphosis. Incre- ment widths corresponding to the first day after hatching averaged 0.33 pm for both S. costatus and S. gigas ; over the first six days after hatching the average increment width was 0.24 pm for both spe- cies. A third region (region 3), observed only in juve- niles of S. gigas, appears immediately after comple- tion of metamorphosis. A darker band visible at the outer edge of region 2 results from the even smaller spacings between five or six increments deposited just before metamorphosis. Increments correspond- ing to the last day before metamorphosis (age 20 days) measured 0.09 pm on average, whereas incre- ments corresponding to the first day after meta- morphosis had a mean of 0.43 pm. Region 3 appears lighter than region 2, because of the wider increments. In both species, most veligers hatched on the sixth day after egg mass deposition, defined as age 0. The mean number of increments on the first day after hatching was 6.00 for S. costatus, 6.70 for S. gigas (Table 1). Statoliths in S. costatus show a deposition pattern of 1.11 increments/day (SE=0.59) over days A 30 jum Figure 1 Generalized structure of Strombus sta- toliths: (A) longitudinal view; (B) inferred transverse view. Numbers identify the three morphological regions: 1 = prehatch- ing, 2 = planktonic larva, and 3 = benthic postmetamorphic juvenile. 0-11; in S. gigas the deposition rate was 1.13 incre- ments/day (SE=0.59) over days 0-9 (Table 1). In S. gigas this pattern continued until metamorphosis. For the day preceding metamorphosis (age 20 days), mean number of increments was 26.05; two days af- ter metamorphosis (age 23 days) mean increment number was 29.40 (Table 1). Significant regressions CP<0.05) were found be- tween age (A, in days) and mean increment number (I) ( S . costatus : I = 6.08 + 0.986A, r2=0.98; S. gigas: I - 6.17 + 0.984A, r2=0.99; data from Table 1). For analysis of S. gigas, we pooled locally-reared larvae with those brought from the Turks and Caicos. Sepa- rate regressions for these two sets of larvae were not significantly different. In the regression equations for both species, the slope did not differ from 1 sig- nificantly and the intercept did not significantly dif- fer from 6 (/-test, P<0.05). Thus, the equations can be generalized to predict strombid veliger age from the number of statolith increments: A = / - 6. NOTE Grana-Raffucci and Appeldoorn: Age determination of larval strombid gastropods 859 Table 1 Mean statolith diameter (pm ± SD), mean number of increments (1 SD) hatching) for Strombus costatus and S. gigas. n = 20 for each day. META = and mean shell length (pm ± SE) by age (days after day of metamorphosis, data not available. Age (d) Strombus costatus Strombus gigas Statolith diameter Number of increments Shell length Statolith diameter Number of increments Shell length 0 13.35 ± 0.48 6.00 1 0.65 345.00 ± 32.60 14.65 ± 0.80 6.70 1 0.73 325.00 ± 35.36 1 14.05 ± 0.22 7.06 ± 0.64 390.00 1 13.69 15.05 ± 0.59 7.30 ± 0.66 375.00 ± 40.82 2 15.30 ± 0.56 8.80 ± 0.70 440.00 ± 13.69 15.70 ± 0.71 7.25 1 0.79 387.501 17.68 3 15.50 ± 0.50 8.75 ±0.50 470.00 ± 18.71 16.45 ± 0.50 8.00 1 1.03 408.25 ± 14.43 4 15.93 ± 0.46 9.57 ± 1.28 625.12 ± 17.83 — — — 5 — — 16.35 1 0.51 10.72 ± 0.46 443.00 ± 18.32 6 16.14 ± 0.64 11.71 ± 0.73 625.80 ± 21.77 17.40 ± 0.97 12.35+ 1.26 458.25 ± 28.87 7 15.65 + 0.91 12.65 ± 1.04 635.00 ± 15.69 17.15 + 0.57 13.45 ± 1.28 461.35 ± 15.22 8 16.50 + 0.67 14.85 1 0.99 641.75 ± 30.28 — — — 9 — — 18.10 ± 0.30 15.55 ± 2.63 466.75 ± 14.43 10 16.35 ± 0.57 15.15 1 2.03 650.00 ± 14.42 17.93 ± 0.46 17.07 ± 1.59 475.80 ± 13.67 12 17.33 ± 1.55 18.33 ± 1.15 700.67 + 24.32 19 27.60 ± 0.82 24.35 1 0.75 1160.001 31.75 20 27.90 ± 0.55 26.05 ± 0.83 1177.501 40.00 21 META META META 22 28.15 ± 1.39 26.75 + 4.28 1340.001 44.50 23 30.45 1 1.10 29.40 ± 1.10 1655.00 1 83.25 Individual statolith increment counts and the magnitude of corresponding standard deviations (Tables 2 and 3) indicate that observed variability may be due to errors in measurement, not to vari- ability in increment deposition. Nested ANOVA of subsamples of each species supported this hypoth- esis (Table 4), showing significant variability only at the level of readings between age groups. Within each species, the relation between age (A) and statolith diameter (D, pm) was more variable than that between age and increment count (S. costatus : D = 14.21 + 0.26A, r2= 0.80; S. gigas: D - 14.93 + 0.34A, r2= 0.93). Similarly, the relation be- tween age and shell length (L, pm) was more vari- able than that between age and increment count but was similar to that between statolith diameter and age ( S . costatus: L = 397.6 + 29. 2A, r2=0.83; S. gigas : L = 356.7 +13.7A, r2=0.89). Significant regressions occurred between shell length and statolith diameter ( S . costatus: L = -1015 + 100.4D, r2=0.85; S. gigas: L = -237 +39.9D, r2=0.90). Data for S. gigas obtained from the Turks and Caicos hatchery did not, however, fit this relation, suggest- ing that 1) environmental or genetic factors influ- ence the relative growth of these structures, or 2) the relation is not linear over the entire larval and postmetamorphic period. Larvae of S. costatus subjected to starvation showed no unusual pattern in region 2; mean num- ber of increments between seven-day-old starved (12.09 ± 0.74 SD) and nonstarved (12.65 ± 1.04 SD) larvae were not significantly different R18=0.262; P>0.05). Mean statolith diameter was also similar in the two groups ( 15.65 ± 0.91 SD pm for nonstarved larvae; 14.83 ±0.75 SD pm for starved ones) (£18=1.414; P>0.05). Discussion Statoliths of strombid larvae have a distinctly rec- ognizable structure at hatching. Region 1 consists of a primordial granule surrounded by four increments. Three regions within the statolith result from changes in density of increments caused by abrupt changes in increment width at times of hatching and metamorphosis. Large transitions in incremental width may be caused by differences in metabolism during normal larval growth and development, in- cluding variable mineral deposition in the larval shell (Maeda-Martinez, 1987). In both S. costatus and S. gigas , the rate of incre- ment formation is constant after hatching. Variabil- ity found in incremental deposition was due largely to errors in measurement, not variation in deposi- tional rate. A two-day period of starvation did not produce any noticeable structural change or precise mark. Starved veligers may have continued growing on stored energy reserves (Rodriguez Gil, 1995), ne- gating differences between treatments. That starved 860 Fishery Bulletin 95(4), 1997 veligers still produced statolith rings without struc- tural change implies that age estimates of veligers from statolith increment counts are robust. Period- icity of statolith growth in larval S. costatus and «S. gigas is sufficiently reliable to be considered a bet- ter tool for age determination than diameter of sta- tolith or measurement of shell length. Counts of in- crements were measurably less variable with age than were measurements of statolith diameter, or shell length. In fishes, otoliths have been used to address ques- tions regarding larval dispersal (Thresher and Broth- ers, 1985); settlement dynamics (Victor, 1983); growth rates (Radtke and Bean, 1982); mortality rates (Essig and Cole, 1986); and larval patch-size estimation (Victor, 1984). Statolith-based age and growth determination, however, has not been used in early life history studies of mollusks; characteris- tics such as size, shell structure, and distance off- shore have been used to infer relative age or length of planktonic life (e.g. Scheltema, 1978; Jablonski and Lutz, 1980, 1983; Pechenik et al., 1984; Pechenik, 1986). Our results here indicate a more highly quan- titative method for ageing larvae. Acknowledgments This project was supported by funds from the Uni- versity of Puerto Rico Sea Grant College Program, Office of Research and External Funding of the Fac- ulty of Arts and Sciences, and Department of Ma- rine Sciences. We thank D. L. Ballantine for supply- ing facilities and algal cultures for rearing larvae, and M. Davis and C. Hess for supplying samples from the Trade Wind Industries Caicos Conch Farm. D. L. Ballantine, P. M. Yoshioka and numerous anony- mous reviewers gave valuable criticism. Table 2 Variability in statolith increment counts within laboratory-reared larval Strombus costatus. Age = days after hatching. Statolith A Statolith B Counts Counts Age (d) 1 2 3 Mean SE 1 2 3 Mean SE 0 6 6 6 6.00 0.00 6 5 5 5.33 0.58 0 6 6 6 6.00 0.00 5 5 5 5.00 0.00 0 6 7 7 6.67 0.58 7 7 6 6.67 0.58 (average) 6.22 0.44 5.67 0.87 1 6 6 6 6.00 0.00 7 7 7 7.00 0.00 1 7 7 8 7.33 0.58 8 6 8 7.33 1.15 (average) 6.67 0.82 7.17 0.75 2 8 8 9 8.33 0.58 9 9 9 9.00 0.00 2 10 8 9 9.00 1.00 9 9 9 9.00 0.00 2 9 9 9 9.00 0.00 10 10 8 9.33 1.15 (average) 8.78 0.67 9.11 0.60 4 10 10 10 10.00 0.00 10 9 9 9.33 0.58 6 11 12 11 11.33 0.58 12 13 12 12.33 0.58 6 11 11 11 11.00 0.00 11 12 11 11.33 0.58 (average) 11.17 0.41 11.83 0.75 7 10 10 11 10.33 0.58 11 11 12 11.33 0.58 7 13 11 13 12.33 1.15 12 10 13 11.67 1.53 7 14 14 12 13.33 1.15 15 13 13 13.67 1.15 7 14 14 15 14.33 0.58 15 13 13 13.67 1.15 (average) 12.58 1.73 12.58 1.51 8 15 14 15 14.67 0.58 14 13 16 14.33 1.53 8 13 15 14 14.00 1.00 12 14 13 13.00 1.00 (average) 14.33 0.82 13.67 1.37 10 15 14 14 14.33 0.58 16 15 15 15.33 0.58 10 14 16 15 15.00 1.00 15 14 16 15.00 1.00 (average) 14.67 0.82 15.17 0.75 NOTE Grana-Raffucci and Appeldoorn: Age determination of larval strombid gastropods 861 Table 3 Variability in statolith increment counts within laboratory-reared larval Strombus gigas. Age = days after hatching. Age (d) Statolith A Statolith B Counts Mean SE Counts Mean SE 1 2 3 1 2 3 0 6 6 6 6.00 0.00 6 6 7 6.33 0.58 0 8 8 7 7.67 0.58 8 8 7 7.67 0.58 0 7 7 7 7.00 0.00 7 6 7 6.67 0.58 (average) 6.89 0.78 6.89 0.78 1 8 7 7 7.33 0.58 8 7 6 7.00 1.00 1 8 7 7 7.33 0.58 9 8 7 8.00 1.00 1 7 8 7 7.33 0.58 8 8 8 8.00 0.00 (average) 7.33 0.50 7.67 0.87 2 8 6 7 7.00 1.00 7 7 7 7.00 0.00 2 7 7 6 6.67 0.58 7 6 7 6.67 0.58 (average) 6.83 0.75 6.83 0.41 3 7 7 6 6.67 0.58 7 7 7 7.00 0.00 5 11 11 11 11.00 0.00 11 11 11 11.00 0.00 5 10 11 10 10.33 0.58 10 10 11 10.33 0.58 (average) 10.67 0.52 10.67 0.52 6 13 15 13 13.67 1.15 12 14 14 13.33 1.15 6 10 11 11 10.67 0.58 11 12 12 11.67 0.58 (average) 12.17 1.83 12.50 1.22 7 14 14 12 13.33 1.15 14 12 11 12.33 1.53 7 12 12 12 12.00 0.00 12 13 13 12.67 0.58 7 14 16 15 15.00 1.00 16 13 14 14.33 1.53 (average) 13.44 1.51 13.11 1.45 9 18 17 15 16.67 1.53 16 20 16 17.33 2.31 9 16 15 17 16.00 1.00 16 18 16 16.67 1.15 9 14 14 14 14.00 0.00 13 15 13 13.67 1.15 (average) 15.55 1.51 15.89 2.20 10 16 18 17 17.00 1.00 20 20 19 19.67 0.58 Table 4 Nested ANOVA of statolith increment counts for subsamples of laboratory-reared larval Strombus costatus and S. gigas. SS = sum of squares, df = degrees of freedom, MS = mean square, P = probability of error, S = P < 0.05, NS = P > 0.05. Source SS df MS F-ratio P Strombus costatus: Among age groups 6,868.3 7 981.2 103.3 P<0.05 S Among individuals 171.5 18 9.5 5.3 P>0.05 NS Between statoliths 68.1 37 1.8 0.2 P>0.05 NS Within statoliths 1,063.4 123 8.7 Total 8,171.3 185 44.2 Strombus gigas: Among age groups 6,732.5 8 841.6 195.7 P<0.05 S Among individuals 81.7 19 4.3 10.8 P>0.05 NS Between statoliths 14.1 39 0.4 0.7 P>0.05 NS Within statoliths 74.8 119 0.6 Total 6,885.2 185 37.2 862 Fishery Bulletin 95(4), 1997 Literature cited Appeldoorn, R. S., and B. Rodriguez Q. (eds.) 1994. Queen conch biology, fisheries and mariculture. Fundacion Cientifica Los Roques, Caracas, Venezuela, 356 p, Ballantine, D. L., and R. S. Appeldoorn. 1983. Queen conch culture and future prospects in Puerto Rico. Proc. Gulf Caribb. Fish. Inst. 35:57-63. D’Asaro, C. N. 1965. Organogenesis, development, and metamorphosis in the queen conch, Strombus gigas , with notes on breeding habits. Bull. Mar. Sci. 15:359-416. Essig, R. J., and C. F. Cole. 1986. Methods of estimating larval fish mortality from daily increments in otoliths. Trans. Am. Fish. Soc. 115:34-40. Jablonski, D., and R. A. Lutz. 1980. Molluscan larval shell morphology: ecological and paleotological applications. In D. C. Rhoads and R. A. Lutz (eds.), Skeletal growth of aquatic organisms, p. 323- 377. Plenum Press, New York, NY. 1983. Larval ecology of marine benthic invertebrates: paleobiological implications. Biol. Rev. 58:21-89. Jackson, G. B. 1994. Application and future potential of statolith incre- ment analysis in squids and sepioids. Can. J. Fish. Aquat. Sci. 51:2612-2625. Maeda-Martinez, A. N. 1987. The rates of calcium deposition in shells of mollus- can larvae. Comp. Biochem. Physiol. 86A:21-28. Pechenik, J. A. 1986. Field evidence for delayed metamorphosis of larval gastropods: Crepidula plana Say, C. fornicata (L. ), and Bittium alternatum (Say). J. Exp. Mar. Biol. Eco. 97:313- 319. Pechenik, J. A., R. S. Scheltema, and L. S., Eyster. 1984. Growth stasis and limited shell calcification in lar- vae of Cymatium parthenopeum during trans-Atlantic transport. Science 224:1097-1099. Radtke, R. L. 1983. Chemical and structural characteristics of statoliths from the short-finned squid Illex illecebrosus. Mar. Biol. (Berl.) 76:47-54. Radtke, R. L., and J. M. Dean. 1982. Increment formation in the otoliths of embryos, lar- vae, and juveniles of the mummichog, Fundulus hetero- clitus. Fish. Bull. 80:201-215. Rodriguez Gil, L.A. 1995. Biochemical composition of larval diets and larvae, temperature, and induction of metamorphosis related to the early life history of the milk conch, Strombus costatus Gmelin. Ph.D. diss., Univ. Puerto Rico. Mayaguez, Puerto Rico, 159 p. Salley, S. 1986. Development of the statocyst of the queen conch lar- vae, Strombus gigas L. (Gastropoda: Prosobranchia). M.S. thesis, McGill Univ., Montreal, Canada, 116 p. Scheltema, R.S. 1978. On the relationship between dispersal of pelagic ve- liger larvae and the evolution of marine prosobranch gastropods. In B. Battaglia and J.A. Beardmore (eds.), Marine organisms: genetics, ecology and evolution, p. 303- 322. Plenum Press, New York, NY. Sokal, R. R, and F. J. Rohlf. 1981. Biometry, second ed. W.H. Freeman and Co., San Francisco, CA, 859 p. Thresher, R. E., and E. B. Brothers. 1985. Reproductive ecology and biogeography of Indo-West Pacific angelfishes (Pisces: Pomacanthidae). Evolution 39:878-887. Victor, B. C. 1983. Recruitment and population dynamics of a coral reef fish. Science 219:419-420. 1984. Coral reef fish larvae: Patch size estimation and mix- ing of the plankton. Limnol. Oceanogr. 29:1116-1119. Wilson, C. A., E. B. Brothers, J. L. Casselman, C. L. Smith, and A. Wild. 1983. Glossary. In E. D. Prince and L. M. Pulos (eds.), Proceedings of the international workshop on age deter- mination of oceanic pelagic fishes: tunas, billfishes and sharks, p. 207-208. U.S. Dep. Commer., NOAATech. Rep. NMFS 8. 863 Bias in Chapman-Robson and least- squares estimators of mortality rates for steady-state populations Michael D. Murphy Florida Marine Research Institute Florida Department of Environmental Protection 100 Eighth Avenue SE, St. Petersburg, Florida 33701-5095 E-mail address: murphy_m@harpo. dep.state.fi. us When age-frequency data are insuf- ficient for fisheries scientists to es- timate year- or age-specific mortal- ity, they are often pooled to provide a single estimate for all fully re- cruited age groups. The accuracy of a pooled estimate depends largely on whether or not the sampled population is in a steady state, i.e. a state in which the rates of recruit- ment and mortality are relatively constant with respect to time and age. Departures from this condition introduce known biases to the esti- mate of mortality (Ricker, 1975; Jensen, 1984). These departures may be difficult to detect because trends in time-specific recruitment or time- and age-specific mortality can result in a population age struc- ture that is quite similar to that for a steady-state population. For in- stance, a long-term increasing trend in recruitment could result in a stable age frequency that would indicate a higher mortality rate than was actually occurring (for various scenarios see Ricker, 1975). Pooled-data estimation tech- niques that have been applied to age-frequency data for fish popula- tions include “catch curve” least- squares regression analysis (Seber, 1973; Ricker, 1975) and nonre- gression-based methods developed by Heincke ( 1913), Jackson ( 1939), and Chapman and Robson (1960). Of the nonregression-based estima- tors, the Chapman and Robson es- timator is preferred because it is the least sensitive to sampling er- ror (Robson and Chapman, 1961). Despite the restrictive steady-state requirements, these techniques have been applied to a wide vari- ety of marine animals; recent ex- amples include Atlantic croaker, Micropogonias undulatus (Barbieri et al., 1994); blue rockfish, Sebastes mystinus (Adams and Howard, 1996); red drum, Sciaenops ocel- latus (Ross et al., 1995); red porgy, Pagrus pagrus ( Pajuelo and Lorenzo, 1996); and deep-water shrimp, Aristeus antennatus (Ragonese and Bianchini, 1996). The Chapman-Robson (CR) esti- mator is based on the probability density function of the geometric distribution and provides a unique minimum-variance, unbiased esti- mate of survival (S), N N + ^X, ~ 1 i=l where x( = the number of years the ith fish is older than the age at full re- cruitment; and N = the total number of fully recruited fish. The underlying assumption is that the age of each fish sampled repre- sents a random, independent age observation from a steady-state population. For age-frequency data that have been truncated to elimi- nate some older age groups, a slightly biased maximum likelihood estimator of survival (CRt) that can be solved by iteration is S 1-S ~(K + 1) g(K+l) 1 -SiK+1) where K + 1 = the number of fully recruited age groups used (Chap- man and Robson 1960). The least-squares regression (LS) estimator provides an unbiased es- timate of -log S (denoted as Z, the instantaneous total mortality rate) and is based on a linear fit to a log- transformed exponential decay model E (log Nj ) = log ( pNq ) - Zj, where AT = the number of age j fish in the sample; N = the original number of fish in the population; and p = the probability that a fish in the population is included in the sample (Seber, 1973). As required for linear regression, log-abundance data are assumed to be independent and normally dis- tributed with constant variance along the regression line. Concern about violating these assumptions led Chapman and Robson ( 1960) to recommend that when the LS method is used, the age-frequency data should be truncated to exclude less abundant age groups. Although both the CR and LS estimator returned very accurate estimates of Z for a suite of exact steady-state age frequencies (Jensen, 1985), the effect of random variation within the sample age frequencies Manuscript accepted 30 May 1997. Fishery Bulletin 95:863-868 ( 1997). 864 Fishery Bulletin 95(4), 1997 has not been investigated. Jensen (1996) found that the CR method was less biased and more precise than the LS method when used to estimate mortality from the age structure of pooled, simulated net-hauls of lake whitefish, Coregonus clupeaformis . A random sample drawn from a known geometric distribution of ages will have an age distribution that varies sto- chastically from the true distribution. In this study, I evaluate the effect of sample size, mortality rate, and an age-frequency truncation scheme on the accuracy and precision of the CR (and CRt) and LS estimators when the sample age frequency is drawn randomly from a population of geometrically distributed ages. calculated with these data, and the mean CRt esti- mates of S were converted to Z. Each truncated age frequency was a subset of a simulation from the com- plete age-frequency simulations in which all fish that were older than the oldest age group meeting or ex- ceeding a threshold abundance of 5 fish were re- moved. Although this truncation scheme reduced the effective sample size within each simulation, it ac- curately reflected the application of a truncation scheme to a real sample. Results and discussion Materials and methods I used a stochastic model that allowed for random departures from the exact age distribution of the population to generate the simulated sample age fre- quencies. Under a known, constant survival rate, a geometric distribution function defines the cumula- tive probability of a fish from a fully recruited cohort being less than age j as P(age < j) = 0 ■ y (i -s)sm-1 m- 1 j< 1 j* 1, where S = the annual survival rate. For this simulation, age-0 fish are defined as those in their first year of full vulnerability to capture. I sampled individual aged fish from this distribution by choosing a random, uniform number (probabil- ity) within the interval from 0 to 1 and determining the age corresponding to this value of the cumula- tive distribution function. By repeating this process, I was able to draw randomly a specified number of aged fish from a known geometric distribution de- fined by S. Each generated sample consisted of 100- 1,000 individuals drawn independently from geomet- ric distributions defined by Z values between 0.20 and 2.00. One thousand simulations were run for each combination of sample size and Z. For each simulation, a CR estimate of S and an LS estimate of Z were calculated from the sample age frequency. Means of the Chapman-Robson estimates of S were converted to Z so that they could be compared to the means of the LS estimates of Z. The effect of constraining the right-hand limb of the sample age frequency was investigated by trun- cating each age-frequency distribution and recalcu- lating mortality. The CRt and the LS estimates were Simulations indicated that mean CR estimates of mortality for the complete age frequencies were es- sentially unbiased. At all Z’s and sample sizes ex- amined, the mean CR estimator agreed closely with the true value of Z. All differences between estimated mean Z’s and true Z’s (relative to the true Z) were <1% (Table 1). The maximum likelihood estimator developed for use with truncated age frequencies (CRt) showed a negative bias that was greatest when sample size was low. With a 5-fish threshold rule, the mean CRt estimate of instantaneous total mortality was biased -12% at Z = 0.2 for a random sample of 100 fish (Fig. 1) . At sample sizes of 300 fish or more, bias was re- duced to less than about -4% for all Z’s (Fig. 1). The mean LS estimates of Z for complete age fre- quencies were consistently less than the true instan- taneous total mortality rate. This bias was greatest at low levels of Z when sample sizes were small (Table 2) . At Z = 0.2, the difference between the mean esti- mated Z and true Z ranged from -16% for samples of 1,000 individuals to -37% for samples of 100. De- viations were much less, -4% to -8%, for all sample sizes when the true Z was 2.0. Bias in the LS esti- mator was reduced by truncating the sample age fre- quency. When I used a minimum threshold abundance of five, the negative bias was reduced to less than about 5% at sample sizes of at least 200 fish (Fig. 1). Precision of the CR and CRt estimators was gen- erally better than that of the LS estimator, especially at low Z’s. Although precision improved for all esti- mators as sample sizes became larger, the coefficient of variation (CV) for the CR and CRt estimators ap- proached 1% for large samples at Z = 0.2, whereas the CV for the LS estimator approached only 6-9% (Fig. 2). For all given sample sizes, the precision of the CR and CRt estimators deteriorated as Z in- creased. The precision of the LS estimator changed little as Z increased, except when the estimator was based on samples of only 100 fish. In general, the CWs for the CR or CRt estimators were less than the NOTE Murphy: Bias in Chapman-Robson and least-squares estimators of mortality rates 865 Table 1 Percent deviation from true instantaneous total mortality rate (Z) for the mean of 1,000 Chapman-Robson estimates of Z for each of the given combinations of sample size and true Z. Frequencies for all age groups were used in calculating mortality rates. Sample size True Z 100 200 300 400 500 600 700 800 900 1,000 0.20 -0.2 -0.1 -0.2 0.1 -0.0 -0.0 0.1 0.2 0.1 -0.0 0.40 -0.1 0.0 0.2 -0.1 0.3 -0.1 0.1 0.2 -0.2 0.2 0.60 -0.3 -0.2 0.0 0.0 -0.1 -0.0 -0.2 0.0 -0.2 0.1 0.80 -0.1 -0.2 0.2 -0.1 0.0 0.2 0.1 -0.0 -0.1 0.0 1.00 0.4 0.2 0.4 -0.1 -0.0 0.1 0.1 0.1 0.0 -0.2 1.20 -0.5 0.2 0.1 -0.0 0.1 0.0 0.2 -0.0 -0.2 -0.1 1.40 -0.3 0.3 -0.1 0.1 0.0 -0.0 -0.1 -0.1 0.0 -0.1 1.60 0.6 -0.1 0.2 0.2 0.2 -0.2 0.1 -0.1 -0.1 -0.1 1.80 -0.2 0.1 0.1 -0.1 0.1 -0.2 -0.1 0.1 -0.2 0.2 2.00 -0.1 -0.0 0.3 -0.1 0.0 -0.2 -0.2 -0.1 0.2 0.2 CR and CRt, Z = 0.2 LS, Z = 0.2 Figure 1 Percent deviation from true instantaneous total mortality (Z) for the mean of 1,000 Chapman-Robson (CR) or Chapman-Robson- for-truncated-data (CRt) estimates of survival or least-squares regression (LS) estimates of equivalent Z for different sample sizes when true Z is 0.2 or 2.0. The threshold levels, representing the minimum acceptable abundance for the oldest age used in the calculation of mortality were one fish ( — • — ) and five fish ( — O — ). 866 Fishery Bulletin 95(4), 1997 Table 2 Percent deviation from true instantaneous total mortality rate (Z) for the mean of 1,000 least-squares regression estimates of Z for each of the given combinations of sample size and true Z. Frequencies for all age groups were used in calculating mortality rates. Sample size True Z 100 200 300 400 500 600 700 800 900 1,000 0.20 -36.8 -28.5 -24.6 -22.1 -20.5 -19.9 -18.2 -17.4 -17.3 -15.6 0.40 -25.2 -20.4 -18.3 -16.9 -14.7 -14.3 -14.3 -13.2 -13.2 -12.3 0.60 -19.6 -17.1 -14.0 -12.9 -12.9 -11.9 -11.5 -11.3 -10.7 -10.6 0.80 -15.8 -14.1 -11.3 -11.0 -10.3 -9.8 -9.6 -9.3 -9.2 -9.7 1.00 -13.6 -10.7 -9.5 -9.3 -8.7 -9.1 -8.3 -7.8 -7.9 -7.9 1.20 -13.0 -9.1 -9.2 -8.4 -7.3 -8.2 -7.5 -7.2 -7.3 -6.8 1.40 -10.8 -9.1 -7.5 -7.9 -6.6 -6.8 -6.6 -6.5 -5.9 -7.1 1.60 -8.8 -7.6 -6.9 -6.2 -6.0 -7.0 -5.9 -5.6 -5.7 -6.1 1.80 -8.3 -6.7 -7.2 -6.4 -5.5 -5.1 -4.4 -5.6 -4.8 -4.9 2.00 -7.7 -5.9 -6.1 -6.1 -5.7 -5.3 -5.0 -5.1 -4.6 -4.4 NOTE Murphy: Bias in Chapman-Robson and least-squares estimators of mortality rates 867 CV’s for the LS estimator when Z was low but were similar or higher when Z was high. When all fully recruited fish are equally available to a sampling gear, the CR estimator can provide a more accurate estimate of mortality than the LS es- timator can. Applying the least-squares estimator to these data clearly violates the linear-regression as- sumption of equal variances among age groups. When a population is subjected to a low Z, the frequency distribution of log-abundances for older age groups in a sample becomes skewed to the right because log- abundance reaches a lower limit at zero (log of 1; Fig. 3). The frequency distribution of log-abundance then becomes truncated (undefined) past some dis- tance to the left of its mean when zero abundances occur in the untransformed frequencies. The vari- ances of the log-abundances appear to be positively related to age until the log-abundance frequencies become truncated when zero abundances appear in the samples for older age groups. Empirical evidence led Chapman and Robson (1960) to conclude that haul data (catch rates for each age group) had an approximately constant variance when log transformed. However, the results from my simulations indicate that variances for the log-abun- dances are likely to differ among age groups. The assumption of constant variance is likely to be met only when the sampling gear operates on a few abun- dant age groups, in which there is no chance of only periodically encountering an older age group. This led Chapman and Robson ( 1960) to suggest that these data should be truncated to eliminate the age fre- quencies beyond the oldest age with a minimum abundance of five fish. Although my findings concur with those of Chapman and Robson, the use of this threshold rule to eliminate older age groups does not completely eliminate all bias in the LS estimator — bias that can be attributed to violations of the as- sumptions on which the linear regression is based. For truncated age-frequency data, both estimators gave biased results when small samples were drawn from a population of many age groups (Z=0.2). In these cases, truncation generally resulted in smaller samples that had far fewer age groups than were in the original complete age frequency. At high Z’s, age- frequency truncation reduced bias in the LS estima- tor to less than 5% at all sample sizes and reduced bias in the CRt estimator to less than 2%. Violations of steady-state assumptions probably impart the most serious biases to pooled estimators of mortality. By simply inspecting a plot of log-abun- dance versus age for evidence of concavity or for a trend in the linear regression residuals, one can de- tect gross violations to these assumptions. Subtle biases inherent when the assumptions required by linear regression are not met are more difficult to de- tect. Both the CR and LS methods can provide very accurate and precise estimates of Z for age frequencies that follow an exact geometric distri- bution (Jensen, 1985). However, the LS estimator is biased when sample ages are drawn randomly from a steady- state, geometrically distributed popu- lation, whereas the CR estimator is not. The LS estimator may be more robust when age samples are not taken ran- domly (Chapman and Robson, 1960). The CRt and LS estimators generally showed similar levels of bias when the sample age structure is truncated with a minimum frequency criterion of 5 fish. In summary, the CR estimator will provide a more accurate and at least as precise an estimate of mortality as the LS estimator will when a random and complete age-frequency sample can be obtained from a population in steady-state. Acknowledgments Mike Armstrong, Gil McRae, Bob Muller, Judy Leiby, as wells as Lynn 868 Fishery Bulletin 95(4), 1997 French of the Florida Marine Research Institute, Grant Thompson, and an anonymous reviewer, helped substantially improve the manuscript. Literature cited Adams, P. B., and D. F. Howard. 1996. Natural mortality of blue rockfish, Sebastes mystinus, during their first year in nearshore benthic habitats. Fish. Bull. 94:156-162. Barbieri, L. R., M. E. Chittenden Jr., and C. M. Jones. 1994. Age, growth, and mortality of Atlantic croaker, Micropogonias undulatus , in the Chesapeake region, with a discussion of apparent geographic changes in population dynamics. Fish. Bull. 92:1-12. Chapman, D. G., and D. S. Robson. 1960. The analysis of a catch curve. Biometrics 16:354- 368. Heincke, F. 1913. Investigations on the plaice. General report 1: the plaice fishery and protective measures. Preliminary brief summary of the most important points of the report. Rapp. P. Reun. Cons. Int. Explor. Mer 16. Jackson, C. H. N. 1939. The analysis of an animal population. J. Anim. Ecol. 8: 238-246. Jensen, A. L. 1984. Non-linear catch curves resulting from variation in mortality among subpopulations. J. Cons. Cons. Int. Explor. Mer 41:121-124. 1985. Comparison of catch curve methods for estimation of mortality. Trans. Am. Fish. Soc. 114:743-747. 1996. Ratio estimation of mortality using catch curves. Fish. Res. 27:61-67. Pajuelo, J. G., and J. M. Lorenzo. 1996. Life history of the red porgy Pagrus pagrus (Teleostei: Sparidae) off the Canary Islands, central east Atlantic. Fish. Res. 28:163-177. Ragonese, S., and M. L. Bianchini. 1 996. Growth, mortality and yield-per-recruit of the deep-water shrimp Aristeus antennatus (Crustacea-Aristeidae) of the Strait of Sicily (Mediterranean Sea). Fish. Res. 26:125-137. Ricker, W. E. 1975. Computation and interpretation of biological statistics of fish populations. Bull. Fish. Res. Board Can. 191:1-382. Robson, D. S., and D. G. Chapman. 1961. Catch curves and mortality rates. Trans. Am. Fish. Soc. 90:181-189. Ross, J. L., T. M. Stevens, and D. S. Vaughan. 1995. Age, growth, mortality, and reproductive biology of red drums in North Carolina waters. Trans. Am. Fish. Soc. 124:37-54. Seber, G. A. F. 1973. The estimation of animal abundance and related parameters. Hafner Press, New York, NY, 506 p. 869 The effects of formalin and freezing on ovaries of albacore, Thunnus alalunga Darlene Ramon Norm Bartoo Southwest Fisheries Science Center National Marine Fisheries Service, NOAA La Jolla, California 92038 E-mail address (for D. Ramon): Darlene.Ramon@noaa.gov In almost every biological sampling program, tissue samples are col- lected and preserved for further examination. In this study, the ef- fects of freezing and 10% buffered formalin on albacore, Thunnus ala- lunga, ovaries are compared to ex- amine how each method of preser- vation affects ovarian weight and oocyte diameter. Formalin is fre- quently used to preserve ovaries for histological studies but, due to its toxicity, it may not be a good choice in all cases, and alternative meth- ods, such as freezing, should be in- vestigated. There is limited infor- mation on the effects of preserva- tion on tuna gonads and, specifi- cally, on albacore gonads. We inves- tigated the effects caused by freez- ing and 10% buffered formalin on weight and oocyte diameter, be- cause weight and other measure- ments (such as oocyte diameter) from fresh samples are not consid- ered interchangeable with those from preserved samples (Lagler, 1968). The purpose of our investigation was to compare fresh ovarian weights with weights of frozen ova- ries and formalin-preserved ovaries from albacore as well as to deter- mine how the diameters from oo- cytes are affected by differing meth- ods of preservation. Materials and methods Albacore are seasonal spawners, spawning mainly during summer months (Otsu and Uchida, 1959; Ramon and Bailey, 1996). Ovaries were collected from albacore caught from 1990 to 1993 (Table 1) with longline gear in the South Pacific (group A) and in waters off Hawaii (group B) and with trolling gear in the North Pacific (group C). In group B, samples were labeled as one of three subgroups (Bl, B2, or B3) on the basis of the method of preservation used (Table 1). In the South Pacific, ovaries from each collected fish were dissected and preserved; a total of 150 pairs were collected. In the North Pacific, sampling took place at two sites: 1) between latitude 42°33'N and 52°02'N and from longitude 129°00'W to 145°52'W and 2) in the waters off Hawaii. Table 1 lists the methods of preservation used for all samples. To investigate the effects that freezing and 10% buffered forma- lin have on oocyte diameter, we ex- amined oocyte diameters from 58 immature North Pacific albacore from 70 cm to 89 cm fork length (FL) in group C. Right-side ovaries were preserved in 10% buffered for- malin; left-side ovaries were frozen until processed two months later. The diameters of the most devel- oped oocyte for each side of the ovary were measured to the near- est 0.01 mm and compared statis- tically with Student’s £-test. To compare differences in oocyte diameter and weight between the two methods of preservation, we used ovaries from subgroup B, which consisted of large, mature albacore (>95 cm FL). The mean diameter of oocytes in the most de- veloped mode was measured from each side of the ovary and com- pared statistically with Student’s t- test. The most developed mode of oocytes was determined with the criteria and method described by Schaefer (1987). The effect of preservation on ova- rian weight was examined for all samples collected in Hawaii (group B). Ovaries in group B were weighed fresh to the nearest 0.1 g and were then either preserved in 10% buff- ered formalin or were frozen. The samples were then reweighed two to three months later; samples pre- served in 10% buffered formalin were placed on a paper towel, and excess moisture was patted off be- fore they were weighed to the near- est 0. 1 g on a Mettler PM3000 elec- tronic balance. Frozen samples were thawed before being placed on a paper towel, and excess moisture was patted off before they were weighed. The preserved weights of ovaries were compared with fresh weights by means of Student’s t-test. Results and discussion Effect of preservation on oocyte diameter Because measurements of oocyte diameter were not made on fresh oocytes, we assumed that left and right ovaries develop at the same rate. This assumption was tested with preserved specimens. Mean oocyte diameters of oocytes in the most developed mode in the right and left ovaries in group A were compared with mean oocyte diam- eters of oocytes in subgroups B2 and B3. The results indicated no Manuscript accepted 4 April 1997. Fishery Bulletin 95:869-872 (1997). 870 Fishery Bulletin 95(4), 1 997 Table 1 Summary of albacore ovaries collected in the Pacific Ocean and preservation treatment used. MFL = Mean fork length. Group Location Date Number of samples Collection Preservation purpose method A — South Pacific New Caledonia lat. 21-23°S May 1990-Feb 1992 105 Control group for oocyte diameter Formalin long. 164-166°E MFL = 90 cm (78-103 cm) Tonga lat. 16-29°S Jan 1990-Feb 1992 45 Control group for oocyte diameter Formalin long. 171-177°W MFL = 88 cm (82-102 cm) B — North Pacific Hawaii Jun 1991-Aug 1992 95 Preservation effects on weight and oocyte diameter within 200-mi EEZ MFL = 103 cm ( 96-116 cm) Bl 16 Left and right ovary of each pair Formalin or treated differently Frozen B2 64 Formalin B3 15 Frozen C — North Pacific U.S. jigboat fishery lat. 42-53°N Aug 1990-Sep 1990 60 Preservation effects on oocyte diameter Formalin or long. 129-146°W MFL = 82 cm (78-87 cm) Frozen Left and right ovary of each pair treated differently Table 2 Mean oocyte diameter (mm) data by ovary weight (g) and preservation method for albacore collected in North Pacific, n = number of fish in sample. Mean oocyte diameter (mm) Ovary weight (g) n Formalin Frozen Percent difference (formalin vs. frozen) * +/-SE Range if +/- SE Range 10-19 19 0.10 ± 0.004 0.07-0.14 0.09 ± 0.003 0.07-0.11 10.0 20-29 27 0.11 ± 0.002 0.09-0.13 0.10 ± 0.002 0.08-0.13 9.1 30-39 11 0.11 ± 0.004 0.08-0.13 0.09 ± 0.032 0.08-0.12 18.1 40-49 2 0.11 ± 0.000 0.11-0.11 0.11 ± 0.002 0.10-0.11 0.0 >50 1 0.11 0.11 0.0 Total 60 0.11 ± 0.002 0.07-0.13 0.10 ± 0.002 0.07-0.13 9.1 significant difference in oocyte diameter ((=0.601, df=225, P>0.05) within method of preservation, and we feel the effects of preservation are largely the source of any differences in measurements. Ovaries from immature albacore (<88 cm FL) in group C, which had been preserved in 10% formalin, were found to have oocytes in the most developed mode with a significantly larger mean diameter in compari- son with those from frozen samples (7=4.614, df=59, P< 0.05). The mean diameter of oocytes preserved in 10% buffered formalin was, on average, 9.1% larger than the mean diameter of frozen oocytes (Table 2). In mature albacore >95 cm FL, the mean diam- eter of oocytes in the most developed mode within an ovary was measured for each side of the ovary and compared statistically within group Bl. Oocytes preserved in 10% buffered formalin were found to have a significantly larger diameter (7=3.581, df=14, P<0.05) — 7.1% greater, on average, than the oocyte diameter in frozen ovaries (Table 3). Thus the mean diameter of oocytes in the most developed mode suggested that mean oocyte diam- eters of frozen ovaries shrank more than those pre- served with 10% formalin. Differential preservation effects were reported by Joseph (1963), who looked at the effects of Gilson’s fluid (Simpson, 1951) ver- sus 4% formalin on oocyte diameter in yellowfin and skipjack tuna ovaries. He found that the diameters NOTE Ramon and Bartoo: The effects of formalin and freezing on albacore ovaries 871 ♦ ♦ ♦ ♦ 1 a • ♦ % weight change (formalin) □ % weight change (frozen) ♦ ♦♦ ♦♦ ♦ <%♦♦♦♦ ♦♦ «* Jo □ * V* s □ * □ ♦ ♦ ♦ * ♦♦ QnD ^ ♦ ♦ crv 0^] « □ ♦ ♦ ♦ □ □ □ □ □ ♦ ♦ □ ♦ 0 100 200 300 400 500 600 700 800 900 1000 Fresh ovary weight (g) Figure 1 Percent change in ovary weight (fresh versus preserved) for albacore ovaries versus fresh weight in grams for samples collected in Hawaii. Table 3 Mean oocyte diameter (mm) data by ovary weight (g) and preservation method for albacore collected near Hawaii, n = number of fish in sample. Mean oocyte diameter (mm) Ovary weight (g) n Formalin Frozen Percent difference (formalin vs. frozen) ic +/-SE Range x +/- SE Range <200 ii 0.19 ± 0.031 0.12-0.48 0.17 ±0.027 0.11-0.43 5.3 200-299 i 0.37 0.33 10.8 >300 3 0.59 ± 0.049 0.50-0.67 0.55 ± 0.042 0.48-0.62 6.8 Total 15 0.28 ± 0.049 0.12-0.67 0.26 ± 0.045 0.11-0.62 7.1 of oocytes preserved with Gilson’s fluid shrank an average of 24% in comparison with those preserved in 4% formalin. This finding is in contrast with that of Schaefer and Orange (1956) who also examined the effects of Gilson’s fluid versus formalin on oocyte diameter. They found that size frequencies for oo- cytes were similar with both methods for oocytes in the 5-63 p size range. The strength of formalin used by Schaefer and Orange was greater than that used by Joseph and may be the cause of the different re- sults. Itano1 examined the effects of brine, refrig- 1 Itano, D. 1994. Progress report on a large-scale investigation on the reproductive biology of yellowfin tuna in the central and western Pacific region. Fourth meeting of the western Pacific yellowfin tuna research group; Koror, Palau, 9-11 August 1994. 872 Fishery Bulletin 95(4), 1997 eration, freezing, and 10% formalin on the histologi- cal quality of samples of ovary from yellowfm tuna. He reported that histological quality was best for those samples preserved fresh in 10% buffered for- malin, whereas samples that had been merely re- frigerated could not be used for histology (Itano1). Effect of preservation on weight The preserved weight of the ovaries from mature fish in group B caught in Hawaii was compared with the fresh weight of ovaries from mature fish in group B (t= 2.565, df=94 and P<0.05) and found to have sig- nificantly different means at the 95% confidence level. Preserved ovaries averaged a loss of 2% of their fresh weight. Our observations ranged from a loss of 33% to a gain of 11%. Samples that were frozen (rc=31) lost, on average, 6% of their weight and ranged from a loss of 26% to a gain of 6% (Fig. 1). In comparison, those that were preserved in 10% buffered formalin lost, on aver- age, 1% of their weight and ranged from a loss of 33% to a gain of 11%. The greatest difference in al- bacore ovary weight was observed for ovaries weigh- ing less than 200 g and which had been frozen, as well as for those greater than 600 g and which had been frozen (Fig. 1). The greater percent weight change observed in the smaller ovaries may be a re- sult of processing method. The percentage of weight change for samples greater than 200 g, but less than 600 g, was within 5%. Of the limited number of samples (n= 5) greater than 600 g, three of the ova- ries that displayed a large loss of weight had oocytes that were hydrating and had mean oocyte diameters near 1 mm. Conclusion Freezing was found to have a greater effect on ova- rian weight and oocyte diameter than 10% buffered formalin. Consequently, the different methods of preservation used in comparative studies should be evaluated cautiously. Furthermore, a single method of preservation or protocol should be used in a given study. Literature cited Joseph, J. 1963. Fecundity ofyellowfin tuna (Thunnus albacares) and skipjack ( Katsuwonus pelamis) from the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm., Bull. 7(41:255-292. Lagler, K. F. 1968. Capture, sampling and examination of fishes. In W. E. Ricker (ed.), Methods for assessment of fish production in fresh waters, p. 13-14. International Biological Programme, Blackwell Scientific Publications, Oxford and Edinburgh, Otsu, T., and R. N. Uchida. 1959. Sexual maturity and spawning of albacore in the Pacific Ocean. Fish. Bull. 59:287-304. Ramon, D., and K. Bailey. 1996. Spawning seasonality of albacore, Thunnus alalunga, in the South Pacific Ocean. Fish. Bull. 94:725-733. Schaefer, K. M. 1987. Reproductive biology of black skipjack, Euthynnus Lineatus, an eastern Pacific tuna. Inter-Am. Trop. Tuna Comm. 19(21:169—214. Schaefer, M. B., and C. J. Orange. 1956. Studies of the sexual development and spawning of yellowfin tuna ( Neothunnus macropterus) and skipjack {Katsuwonus pelamis) in the areas of the eastern Pacific Ocean, by examination of gonads. Inter-American Trop. Tuna Comm. Bull. 1(6): 283-320. Simpson, A. C. 1951. The fecundity of the plaice. Fish Invest., ser. II, Mar. Fish. G.B. Minist. Agric. Fish. Food 17(5), 27 p. 873 Short-term retention of coded wire and internal anchor tags in juvenile common snook, Centropomus undecimalis Julie E. Wallin John M. Ransier Sondra Fox Robert H. McMichael Jr. Florida Marine Research Institute, Florida Department of Environmental Protection 100 Eighth Avenue SE, St. Petersburg, Florida 33701-5095 E-mail address (forJ. E Wallin): Wallin_J@sellers. dep.state.fi. us Common snook, Centropomus un- decimalis, are popular gamefish in southern Florida (Bruger and Haddad, 1986) and throughout their range in the southeastern United States and Central America. Snook populations in Florida de- clined during the 1970’s and 1980’s despite increasingly strict limits on the fishery (Taylor1). In an effort to enhance snook populations, the Florida Department of Environ- mental Protection has considered developing a stocking program to supplement depleted wild snook stocks. To evaluate a stocking pro- gram’s success or impact on native populations, hatchery-released fish must be marked to distinguish them from wild fish. Mark reten- tion rates must be known to cor- rectly interpret catch records and to make inferences about the stock- ing program or the wild population (Wallin and Van Den Avyle, 1994). Coded wire tags (CWT’s) are com- mon in fish stocking programs be- cause they can be applied quickly and economically to small, juvenile fish (<100 mm standard length, SL). Although retention rates of CWT’s vary considerably among different species because of mor- phological differences (Heidinger and Cook, 1988; Szedlmayer and Howe, 1995), retention rates are generally very good (>95%) if ap- propriate tagging tissues are iden- tified and used (Dunning et al., 1990; Pitman and Isaac, 1995). A disadvantage with using CWT’s is that they cannot be detected by recreational or commercial fisher- men; therefore, recovery of CWT data must be obtained indepen- dently. Internal anchor-external streamer tags with imprinted infor- mation allow recapture information to be obtained from recreational or commercial fisheries. However, these tags are generally restricted for use on fish >110 mm SL. As with CWT’s, retention rates of these tags may vary because of morphological differences (Vogelbein and Over- street, 1987; Waldman et al. 1990). In this study, we evaluated reten- tion of CWT’s and two types of in- ternal anchor-external streamer tags in age-0 common snook. The fish sizes in this study were simi- lar to sizes that would be produced in a large-scale stocking program. We also evaluated the effects of tag- ging on growth and mortality. Re- tention of coded wire tags inserted in the cheek musculature was evaluated in common snook 60-115 mm SL for approximately 60 days after tagging. Retention of internal anchor-external streamer tags with disk- or T-style anchors was evalu- ated in snook 110-180 mm SL for 30 days after tagging. The major differences between the two types of internal anchor tags were the shape of the anchor and the size of incision needed to apply the tags, either of which could affect reten- tion or mortality. This study pro- vides baseline information on tag retention rates in common snook that can be used to interpret catch records correctly and to evaluate common snook stocking programs. Methods Coded wire tag retention We evaluated retention of CWT’s and the effects of tagging on growth and mortality in 60-115 mm SL common snook. Equal numbers of fish were randomly assigned to one of three treatment groups: 1) tagged fish, 2) untagged fish that were handled in the same manner as tagged fish but were not tagged, and 3) control fish that were not tagged or handled. The cheek muscle was chosen for tag implan- tation because of the high CWT re- tention rates for this location in red drum and striped bass. Tagged and untagged snook were anesthetized (100 ppm MS-222) just prior to treatment. One person applied all tags. Tags (1-mm-long, 0.25-mm-diameter) were injected vertically, 2 mm deep into the left adductor mandibularis muscle, pos- terior to the eye with a Northwest Marine Technology (Shaw Island, WA) Mark IV tagging machine. 1 Taylor, R. 1997. Florida Marine Res. Inst., Florida Dep. of Environmental Pro- tection, 100 Eighth Ave. SE, St. Peters- burg, FL 33701-5095. Personal commun. Manuscript accepted 21 March 1997. Fishery Bulletin 95:873-878 (1997). 874 Fishery Bulletin 95(4), 1997 Fish were passed through a quality control device (QCD) to ensure tag presence. Fish in the untagged treatment were handled and passed through a QCD in the same manner as tagged fish but were not im- paled on a tagging needle. Tagged and untagged fish were placed in separate 1,300-L tanks. Control fish were transferred directly to a 1,300-L tank with mini- mal disturbance. The three tanks were aerated and had sand, di- atomaceous earth, ultra-violet light, and biological filters. To prevent fish from jumping out, half of each tank was covered with 5-mm nylon mesh net and the other half was covered with opaque black plas- tic. Water temperatures averaged 24°C (range: 17- 30°C); salinity averaged 26 ppt (range: 22-30 ppt); dissolved oxygen averaged 6.4 ppm (range: 4. 0-7. 4 ppm), and pH averaged 7.8 (range: 7. 7-7. 9). Fish were fed baitfish or commercially prepared fish chow daily. Fish in the tagged treatment were checked for tag retention at 3, 30, and 60 days after tagging. At each tag check, fish were anesthetized ( 100 ppm MS-222), checked for tag presence with a field sampling de- tector (Northwest Marine Technology), and mea- sured. Fish in the untagged treatment were simi- larly anesthetized and measured at 3 and 30 days after tagging. Dead fish were removed from the tanks and checked for tag presence. We assumed that mortality associated with tag- ging or handling would occur within 30 days of tag- ging (Szedlmayer and Howe, 1995). Therefore, after 30 days, fish in the untagged and control groups were combined and then randomly divided into two groups that received either the tagged or untagged treat- ment. This created a second trial of the experiment to evaluate CWT retention. Fish in the second trial were checked for tag retention at 3 and 30 days after tagging. The entire experiment was repeated three times with fish from different culture facilities, resulting in three trials in which tag retention was evaluated for 30 days and three trials in which tag retention was evaluated for 30 and 60 days. Mean initial fish sizes varied among trials from 62 to 115 mm SL, and numbers of fish per treatment varied among trials from 24 to 76 (Table 1). Table 1 Tag (coded wire) retention rates and fish survival rates at 3, 30, and 60 d after tagging for each treatment (tagged fish, untagged fish, and control fish) and fish sizes at the time of tagging. Starting no. = number of fish at the beginning of the experiment. Survival rates are from the start of the experiment until 3, 30, or 60 d after tagging. Standard length at the time of tagging is given as the mean with standard deviation in parentheses. Tag retention rates (%) Survival rates (%) Starting Standard length Trial Treatment no. (mm) 3-day 30-day 60-day 3-day 30-day 60-day 1 Tagged 26 92 (6.8) 100 100 100 100 92.3 92.3 Untagged 24 95 (5.9) — — — 92.3 92.3 — Control 26 94 (6.4) — — — 100 100 — 2 Tagged 24 115 (9.7) 100 100 — 100 95.8 — Untagged 24 117 (9.6) — — — 100 100 — 3 Tagged 76 62 (8.3) 98.6 92.0 85.4 98.7 67.1 63.1 Untagged 76 59 (8.3) — — — 100 93.1 — Control 75 59 (8.2) — — — 100 81.3 — 4 Tagged 62 63 (9.2) 100 87.1 — 100 100 — Untagged 61 63 (9.2) — — — 100 100 — 5 Tagged 74 74 (7.8) 100 95.8 95.8 98.6 98.6 98.6 Untagged 75 71 (8.4) — — — 98.7 98.7 — Control 75 75 (8.7) — — — 100 100 — 6 Tagged 74 82 (10.1) 100 98.6 — 100 97.3 — Untagged 75 81 (9.4) — — — 100 100 — Overall Tagged 71 (14.5) 99.8 95.6 93.7 99.6 91.9 84.7 Untagged — 73 (16.6) — — — 98.5 97.4 — Control — 72 (17.3) — — — 100 93.8 — NOTE Wallin et a I.: Retention of coded wire and internal anchor tags in juvenile Centropomus undecimalis 875 Retention of interna! anchor tags with external streamers We evaluated retention of internal anchor tags with external streamers (disk and T-style) in 110-180 mm SL common snook. Tags with disk anchors (Model FM-89SL, Floy Mfg., Seattle, WA) consisted of a plas- tic 5 x 15 mm imprinted elliptical disk and 50-mm imprinted streamer. T-anchor tags (IEX tags, Hallprint Ltd., Holden Hill, Australia) consisted of an 18-mm T-shaped anchor and a 42-mm imprinted streamer. Equal numbers of fish were randomly as- signed to one of three treatments: 1 ) fish tagged with disk anchor tags, 2) fish tagged with T-anchor tags, and 3) untagged fish that were handled in the same manner as tagged fish. All fish were anesthetized ( 100 ppm MS-222) and measured just prior to treatment. To apply disk an- chor tags, we used a scalpel to make an approxi- mately 6-mm vertical incision into the body wall and then inserted the disk into the incision. To apply the T-anchor tags, we used a sharpened tag-applicator needle to make an approximately 2-mm diameter puncture and then inserted the T-anchor into the opening. Incisions were made on the left ventral side of the fish, anterior to the vent and posterior to the pectoral fin. Anchors were dipped in betadyne prior to insertion to minimize infection. Fish in the tag treatments were distinguished by fin clips: disk an- chor-tagged fish received a left pectoral fin clip and T-anchor- tagged fish received a right pectoral fin clip. Fish in the untagged treatment were anesthetized, measured, and handled in the same manner as tagged fish but did not receive an incision or finclip. All fish were transferred immediately after treatment to a 13,300-L tank containing approximately 9,000 L of water. The tank was aerated and had sand, diatomateous earth, and biological filters. Water tem- peratures averaged 27°C (range: 24-29°C); salinity averaged 26 ppt (range: 11-30 ppt); dissolved oxy- gen averaged 6.5 ppm (range: 4. 9-8. 4 ppm), and pH averaged 7.8 (range: 7. 7-7. 9 ppm). Fish were fed live baitfish daily. Fish were checked for tag retention at 14 and 30 days after tagging. At each tag check, fish were anes- thetized ( 100 ppm MS-222), checked for tag and fin- clip presence, measured, and weighed. The entire experiment was repeated twice; numbers of fish per treatment varied between trials from 61 to 78 fish per treatment (Table 1 ). During the first trial, necrop- sies were performed on three to ten fish from each treatment after 14 days to evaluate a bacterial in- fection. Six of ten T-anchor- tagged fish examined had anchors that were inserted into the swim bladder, whereas all disk-anchor-tagged fish examined (n=3) had anchors that were correctly placed in the perito- neal cavity. Because of the potential for misapplica- tion with the T-anchor tag, during the second trial we applied T-anchor tags by making a 3-mm scalpel incision rather than by making a puncture with a tag applicator needle. At the end of the second trial, ten randomly selected fish from each tag treatment were dissected and examined for tag implant loca- tions and tissue healing. Analyses Estimates of the rates of tag retention and survival were arcsine-transformed prior to analysis (Sokal and Rohlf, 1981). Fish size at the time of tagging dif- fered significantly among trials in both the CWT experiments ( ANOVA, df=5, P=517.6, P<0.01 ) and in the internal-anchor-tag experiments (ANOVA, df=l, F= 6.7, P<0.01). Therefore, mean size at tagging was used as a covariate in analysis of covariance (Sokal and Rohlf, 1981; SAS, 1989) to compare rates of tag retention, survival, and growth among time periods and treatment groups. Post hoc comparisons were conducted with the method of least-square means (SAS Institute, Inc., 1989). Results Overall, retention rates of coded wire tags averaged 99.8%, 95.6%, and 93.7% at 3, 30, and 60 days after tagging, respectively (Table 1). At 3 days, tag reten- tion rates for all sizes of snook ranged from 98.6 to 100% (Fig. 1). Fish size at the time of tagging sig- nificantly affected tag retention rates (F=12.5, P<0.01); larger fish had higher retention rates, par- ticularly during the 30- to 60-d period after tagging (Fig. 1). Tag retention of common snook >70 mm SL at the time of tagging was >95% at 30 and 60 days after tagging, whereas tag retention in snook <65 mm SL was 92.0% and 85.4% at 30-60 days after tagging, respectively (Table 1; Fig. 1). Tag retention rates decreased between 3 and 30 days after tagging (LSMEANS, P=0.02) but did not change significantly between 30 and 60 days (LSMEANS, P=0.89; Fig. 2). Survival rates of CWT fish were not significantly different from those of the untagged or control fish (df=2, F= 0.5, P=0.61). Survival rates of fish in all treatments averaged 99.6%, 91.9%, and 84.7% at 3, 30, and 60 days after tagging, respectively (Table 1). Poor survival of fish in the third trial (63%, Table 1) is attributed to parasitic infection. Growth rates of CWT fish were also not significantly different from those of untagged or control fish (df=2, F- 0.87, P=0.42). 876 Fishery Bulletin 95(4), 1997 Overall, retention rates of T-anchor and disk-an- chor tags averaged 100% and 99.2%, respectively, after 30 days (Table 2). Fish size at the time of tag- ging did not significantly affect internal anchor tag retention rates (df=l, F=0.3, P=0.64). Retention rates did not differ significantly between disk- and T-an- 102 n 100 - O o 98 - O 5? 96 - 0 c o 94 - c CL) 0 92 - V CD 03 1- 90 - 88 - V 86 - □ o o o V O Tag retention at 3 d S7 Tag retention at 30 d □ Tag retention at 60 d chor-tagged fish or among time periods (for both com- parisons, df=l, F= 0.8, P=0.43). There were also no significant differences in survival (df=2, F=0.7, P-0.55) or growth (df=2, F=0.03, P=0.10) among treatments . Survival rates of T-anchor- and disk-an- chor-tagged fish averaged 97.0 and 91.9% at 14 and 30 days after tagging, respectively (Table 2). Nearly all (80%) fish with internal anchor tags examined at 30 days had inflammation or a proliferation of epithelial cells at the tag insertion site. Of the ten disk-tagged fish ex- amined after the second trial, six fish had an- chors encapsulated in the peritoneum, two fish had anchors inserted in the swim bladder or gastrointestinal tract, and two fish had an- chors freely moving in the peritoneal cavity. Of the ten T-anchor-tagged fish examined, two fish had anchors encapsulated in the perito- neum, one had an anchor inserted in the swim bladder, and seven had anchors that were freely moving in the peritoneal cavity. 84 -| 1 1 1 1 1 1 50 60 70 80 90 100 110 120 Discussion Size at tagging (mm) Figure 1 Retention rates of coded wire tags and average size (standard length) of juvenile snook at the time of tagging. The high retention rates of CWT’s reported in this study (85-100%) indicate that CWT’s are effective for marking age-0 common snook and that the cheek musculature is appropriate for tagging. CWT’s did not affect snook survival or growth during the first 30-60 days after tagging, indicating that CWT’s have negligible adverse effects during this time period. CWT retention rates in- creased with initial fish size (Fig. 1) from 87-92% for snook with an initial mean size of 62 mm SL to 95-100% for snook with initial mean sizes of 71- 117 mm SL. Several au- thors have noted the impor- tance of fish size or target tissue size in the retention of CWT’s (Heidinger and Cook, 1988; Szedlmayer and Howe, 1995). Our study also shows a relation be- tween initial tagging size and CWT retention. Most tag loss occurred between 3 and 30 days (Fig. 2). Other studies have noted Trial 1, SL=92 2 mm Trial 2, SL=1 15.3 mm Trial 3, SL=61 .5 mm Trial 4, SL=62.9 mm Trial 5, SL=81.5 mm Trial 6, SL=73.6 mm Time (days) Figure 2 Retention rates of coded wire tags in fish 3, 30, and 60 days after tagging. The key shows mean initial size (SL) of fish in each trial. NOTE Wallin et al.: Retention of coded wire and internal anchor tags in juvenile Centropomus undecimalis 877 Table 2 Internal anchor tag retention rates and fish survival rates at 14 and 30 d after tagging for each treatment (t-bar-tagged fish, disk-tagged fish, and control fish) and fish sizes at the time of tagging. Starting no. = number of fish at the beginning of the experiment. Survival rates are from the start of the experiment until 14 or 30 d after tagging. Standard length at the time of tagging is given as the mean with standard deviation in parentheses. Trial Treatment Starting no. Standard length (mm) Tag retention rates (%) Survival rates (%) 14-day 30-day 14-day 30-day 1 T-bar-tagged 78 139(13.1) 100 100 93.6 93.6 Disk-tagged 78 138 (13.0) 100 100 91.0 75.6 Control 78 137 (13.7) — — 97.4 82.1 2 T-bar-tagged 61 135 (11.3) 100 100 100 100 Disk-tagged 61 136(10.4) 100 98.4 100 100 Control 61 134(10.7) — — 100 100 Overall T-bar-tagged 136 (12.5) 100 100 96.8 96.8 Disk-tagged — 137 (11.9) 100 99.2 95.5 87.8 Control — 137 (12.6) — — 98.7 91.1 that most CWT loss in juvenile fish occurs within two to four weeks after tagging, the time period correspond- ing with the time required for the tagging puncture wound to heal (Dunning et al., 1990). Although there was no significant change in retention rates after 30 days, tag retention rates in the smallest fish (mean SL 62 mm) continued to decline between 30 and 60 days (Fig. 2), providing further evidence of the impor- tance of fish size for tag retention. Thirty-day retention rates of both types of inter- nal anchor-external streamer tags were very high (>99%), indicating that these tag types are effective tools for marking >110 mm common snook. However, other studies have suggested that retention of disk- and T-anchor tags may decrease over months or years and should therefore be evaluated for longer periods of time (Collins et al., 1994). Although the incisions required to apply disk tags are larger than incisions required for T-anchor tags, there was no difference in survival or growth rates of fish with either tag type. Both tag types caused similar incidences of ir- ritation at the tag insertion site. Similar irritation has been noted in other fish species and can eventu- ally result in tag loss (Mattson et al., 1990; Collins et al., 1994). After 30 days, disk anchors were more frequently encapsulated and attached to the inside of the body wall than were T-anchors. Both tag types were equally likely to be incorrectly inserted in the swim bladder or gastrointestinal tract when the in- cision was made with a scalpel; overall, incorrect in- sertions were noted in 10-20% of the fish examined. We recommend using scalpels to insert the T-anchor tags; when a tag applicator needle was used, inci- dences of incorrect insertion increased to 60%, prob- ably because of difficulty in controlling puncture depth. In conclusion, because CWT retention rates were significantly improved for fish >70 mm SL, we rec- ommend tagging snook at this size whenever pos- sible. We also recommend that CWT retention in snook be evaluated for 30 days after tagging so that accurate estimates of tag-loss rates can be calculated. For snook >110 mm SL, both the T-anchor and disk internal anchor-external streamer tags were effec- tive marking techniques. For all types of tags and all sizes of fish in our study, tagging did not signifi- cantly affect snook survival or growth. Acknowledgments We thank Greg Vermeer for advice in fish health and review of a draft manuscript. We also thank Buck Dennis and the staff of FDEP’s Stock Enhancement Research Facility and the Fisheries- Independent Moni- toring Program for their technical assistance. Harry Grier, Lynn French, Judy Leiby, and Jim Quinn also provided helpful comments on the manuscript. This project was funded in part by the Department of the Interior, U.S. Fish and Wildlife Service Federal Aid for Sportfish Restoration (Project F-43) and by funds from Florida Saltwater Fishing License sales. Literature cited Bruger, G. E. , and K. D. Haddad. 1986. Management of tarpon, bonefish, and snook in 878 Fishery Bulletin 95(4), 1997 Florida. In R. H. Stroud (ed.), Multi-jurisdictional man- agement of marine fisheries. Mar. Rec. Fish. 11:53-57. Collins, M. R., T. I. J. Smith, and L. D. Heyward. 1994. Effectiveness of six methods for marking juvenile shortnose sturgeon. Prog. Fish-Cult. 56:250-254. Dunning, D. J., Q. E. Ross, B. R. Friedman, and K. L. Marcellus. 1990. Coded wire tag retention by, and tagging mortality of, striped bass reared at the Hudson River Hatchery. Am. Fish. Soc. Symp. 7:262-266. Heidinger, R. C., and S. B. Cook. 1988. Use of coded wire tags for marking fingerling fishes. N. Am. J. Fish. Manage. 8:268-272. Mattson, M. T., J. R. Waldman, D. J. Dunning, and Q. E. Ross. 1990. Abrasion and protrusion of internal anchor tags in Hudson River striped bass. Am. Fish. Soc. Symp. 7:121-126. Pitman, V. M., and J. Isaac Jr. 1995. Coded wire tag retention by paddlefish at three im- plant sites. N. Am. J. Fish. Manage. 15:878-880. SAS Institute, Inc. 1989. SAS/STAT users guide, version 6, fourth ed. SAS Institute Inc., Cary, NC, 943 p. Sokal, R. R., and F. J. Rohlf. 1981. Biometry. Freeman, New York, NY, 859 p. Szedlmayer, S. T., and J. C. Howe. 1995. An evaluation of six marking methods for age-0 red drum, Sciaenops ocellatus. Fish. Bull. 93:191-195. Waldman, J. R., D. J. Dunning, and M. T. Mattson. 1990. A morphological explanation for size-dependent an- chor tag loss from striped bass. Trans. Am. Fish. Soc. 119:920-923. Wallin, J. E., and M. J. Van Den Avyle. 1994. Retention of coded wire tags by juvenile striped bass. Proc. Annu. Conf. Southeast. Assoc. Fish Wildl. Agencies 48:550-554. Vogelbein, W. K., and R. M. Overstreet. 1987. Histopathology of the internal anchor tag in spot and spotted seatrout. Trans. Am. Fish. Soc. 116:745-756. 879 Announcement The editorial staff of Fishery Bulletin would like to thank the following referees for their time and efforts in providing reviews of the manuscripts published in 1997. Their contributions have helped ensure the publication of quality science. Dr. Larry G. Allen Mr. Frank P. Almeida Dr. William D. Anderson Dr. Vaughn C. Anthony Dr. Alexander I. Arkhipkin Dr. Michael P. Armstrong Dr. Freddy Arocha Dr. Carin J. Ashjian Dr. Marinelle Basson Dr. Robert E. Bayley Dr. William H. Bayliff Dr. Richard J. Beamish Dr. Lynnath E. Beckley Dr. Daniel W. Beckman Dr. Allen J. Bejda Dr. David G. Bembo Dr. Douglas C. Biggs Dr. Karen A. Bjorndal Dr. Brian W. Bowen Dr. Michael J. Bradford Dr. Steven B. Brandt Dr. Steven Branstetter Dr. Mary T. Bremigan Dr. Richard D. Brodeur Dr. Gerald Brothers Dr. Lawrence J. Buckley Dr. Rik C. Buckworth Mr. Britt W. Bumguardner Dr. John S. Burke Dr. John L. Butler Dr. C. D. Buxton Dr. Gregor M. Cailliet Dr. Dan Campbell Mr. Albert E. Caton Dr. Christopher R. Chambers Dr. Patricia M. Clay Dr. V. G. Cockcroft Dr. Mark R. Collins Dr. Steven H. Coombs Dr. James H. Cowan Dr. J. Cramer Mr. Edwin P. Creaser Dr. Matthew A. Cronin Dr. Mary Carla Curran Dr. Michael J. Dagg Mr. Andrew W. David Dr. Cabel S. Davis Dr. Edward E. DeMartini Dr. M. Demestre Dr. Andrew E. Dizon Dr. Dennis J. Dunning Dr. Maxwell B. Eldridge Dr. John M. Emlen Dr. Sheryan P. Epperly Dr. Karim Erzini Dr. Bo Fernholm Dr. Joseph P. Fisher Dr. Gary R. Fitzhugh Dr. Terry J. Foreman Dr. Kevin D. Friedland Dr. Alejandro Gallego Mr. Damon P. Gannon Dr. R. W. Gauldie Dr. Stratis Gavaris Dr. Audrey J. Geffen Dr. C. A. Gray Dr. George D. Grice Dr. Churchill B. Grimes Dr. Jill J. Grover Ms. Ruth Haas-Castro Dr. Wayne R. Haight Dr. Louis S. Hales Jr. Dr. Harlyn O. Halvorson Dr. John Hampton Dr. Douglas E. Harper Dr. Paul J. B. Hart Dr. R. H. Hesslein Dr. F. Hester Dr. Michael H. Horn Dr. Gene R. Huntsman Mr. S. E. Hutchinson Dr. James N. Ianelli Mr. Josef S. Idoine Dr. Merton C. Ingham Dr. Bjoernar Isaksen Mr. David G. Itano Dr. George D. Jackson Dr. Stephen C. Jewett Dr. Malcolm Jobling Dr. Allyn G. Johnson Dr. James H. Johnson Dr. Lyndal L. Johnson Dr. Robert D. Kenney Mr. Frederick G. Kern Dr. D. E. Kime Dr. Pierra A. Kleiber Dr. Grace Klein -Macphee Dr. Elizabeth A. Koch Dr. Robert G. Kope Dr. J. Anthony Koslow Dr. Jay S. Krouse Dr. Yves de Lafontaine Lt. John T. Lamkin Ms. Kathy L. Lang Dr. Roger B. Larsen Dr. T. N. Lee Dr. W. Y. Lee Dr. C. D. Levings Dr. Nancy C. H. Lo Dr. Alec D. MacCall Dr. Pamela M. Mace Dr. Michael J. Maceina Dr. Koji Maekawa Dr. Gary L. Maillet Mr. John P. Manderson Dr. Daniel Margulies Dr. Jeffrey B. Marliave Dr. Guy Martel Dr. John C. McGovern Dr. Alasdair D. McIntyre Dr. Mark E. McMaster Dr. Robert H. McMichael Dr. Charmion B. McMillan Dr. Bernard A. Megrey Dr. Carol Meise Dr. Richard D. Methot Dr. Tom J. Miller Mr. Raymond Mojica Dr. M. Joanne Morgan Dr. Wallace W. Morse Dr. David Mountain Dr. Robert G. Muller Dr. Thomas A. Munroe Mr. Michael D. Murphy Dr. Ransom A. Myers Dr. Hideki Nakano Dr. Stephen J. Newman Dr. Jorgen Nielsen Dr. Muneo Okiyama Dr. Kathryn A. Ono Dr. J. R. Ovenden Dr. Richard O. Parker Jr Dr. C. John Parmenter. Dr. Glenn R. Parsons Dr. Gilbert B. Pauley Dr. William G. Pearcy Dr. G. Pederson Dr. R. Ian Perry Dr. James P. Peterson Dr. Edward J. Pfeiler Dr. Susan Picquelle Dr. Hans Polet Dr. A. E. Punt Dr. Richard L. Radtke Dr. Kirsten Ramsay Dr. Joseph S. Ramus 880 Fishery Bulletin 95(4), 1997 Dr. R. Anne Richards Dr. Adriaan D. Rijnsdorp Ms. Cecilia M. Riley Dr. D. Ross Robertson Dr. D. A. Roff Dr. Andrew A. Rosenberg Dr. Steven W. Ross Dr. Rodney A. Rountree Dr. Roger A. Rulifson Dr. Thomas L. Rutecki Dr. Susan E. Safford Dr. Saul B. Saila Dr. David B. Sampson Dr. Warwick H. Sauer Dr. Mark W. Saunders Dr. David R. Schiel Dr. Daniel R. Scoles Dr. Stephen G. Scott Mr. Gary R. Shepherd Dr. Vaughn Silva Dr. I. Philip Smith Dr. Adam J. Smolenski Dr. Roxanne Smolowitz Dr. Gunnar Stefansson Dr. Peter Stewart Dr. Robert R. Stickney Dr. Allan W. Stoner Dr. Patrick J. Sullivan Dr. Fred C. Sutter Dr. Gordon L. Swartzman Dr. W. Mark Swingle Dr. Stephen T. Szedlmayer Dr. Bruce A. Thompson Dr. Grant G. Thompson Dr. Andrew W. Trites Dr. Albert V. Tyler Dr. Sam C. Wainwright Dr. John R. Waldman Dr. William A. Walker Dr. Julie E. Wallin Dr. Stanley M. Warlen Dr. James R. Weinberg Mr. Kenneth L. Weinberg Dr. D. W. Welch Dr. Vidar G. Wespestad Mr. Grant West Dr. H. von Westernhagen Dr. Jerry A. Wetherall Ms. Susan E. Wigley Dr. Stuart J. Wilk Dr. John G. Williams Dr. George H. Winters Dr. Peter R. Witthames Dr. Oleg G. Zolotov 881 Fishery Bulletin Index Volume 95 (1-4), 1997 List of titles 95(1) 1 Individual variation in growth and development during the early life stages of winter flounder, Pleuronectes ameri- canus, by Douglas F. Bertram, Thomas J. Miller, and Wil- liam C. Leggett 11 Archeological evidence of large northern bluefin tuna, Thunnus thynnus , in coastal waters of British Columbia and northern Washington, by Susan J. Crockford 25 Dietary overlap of juvenile fall- and spring-run chinook salmon, Oncorhynchus tshawytscha, in Coos Bay, Oregon, by Joseph P. Fisher and William G. Pearcy 39 Fish age determined from length: an evaluation of three methods using simulated red snapper data, by Phillip C. Goodyear 47 The life history and stock separation of silver kob, Argyrosomus inodorus, in South African waters, by Marc H. Griffiths 68 Estimates of tag-reporting and tag-shedding rates in a large-scale tuna tagging experiment in the western tropi- cal Pacific Ocean, by John Hampton 80 Use of gonad indices to estimate the status of reproductive activity of female swordfish, Xiphias gladius: a validated classification method, by Michael G. Hinton, Ronald G. Tay- lor, and Michael D. Murphy 85 Persistent spatial and temporal abundance patterns for late-stage copepodites of Centropages hamatus (Copepoda: Calanoida) in the U.S. northeast continental shell ecoysystem, by Joseph Kane 99 Estimating the annual proportion of nonspawning adults in New Zealand hoki, Macruronus novaezelandiae, by Mary E. Livingston, Marianne Vignaux, and Kathy A. Schofield 114 Structure and dynamics of the fishery harvest in Broward County, Florida, during 1989, by James E. McKenna Jr. 126 Year-class twinning in sympatric seasonal spawning popu- lations of Atlantic herring, Clupea harengus, by Ian H. McQuinn 137 Nursery habitat in relation to production of juvenile pink snapper, Pristipomoides filamentosus, in the Hawaiian Archipelago, by Frank A. Parrish, Edward E. DeMartini, and Denise M. Ellis 149 Application of high-breakdown robust regression to tuned stock assessment models, by Victor R. Restrepo and Jo- seph E. Powers 161 Growth and survival of 0+ English sole, Pleuronectes vetulus, in estuaries and adjacent nearshore waters off Washington, by Yunbing Shi, Donald R. Gunderson, and Patrick J. Sullivan 174 Mitochondrial DNA diversity in and population structure of red grouper, Epinephelus morio, from the Gulf of Mexico, by Linda R. Richardson and John R. Gold 180 Entanglement of California sea lions, Zalophus call- fornianus californianus, in fishing gear in the central- northern part of the Gulf of California, Mexico, by Alfredo Zavala-Gonzalez and Eric Mellink 95(2) 187 Birthdate analysis and its application to the study of re- cruitment of the Atlanto-lberian sardine, Sardina pilchardus, by Federico Alvarez and Francisco Alemany 195 Demersal fish assemblages of the northeastern Chukchi Sea, Alaska, by Willard E. Barber, Ronald L. Smith, Mark Vallarino, and Robert M. Meyer 209 Evaluations of the Nordmpre grid and secondary bycatch- reducing devices IBRD’s) in the Hunter River prawn-trawl fishery, Australia, by Matt K. Broadhurst, Steven J. Kennelly, John W. Watson, and Ian K. Workman 219 Spatial patterns in the dynamics of slope rockfish stocks and their implications for management, by Donald R. Gunderson 231 Fecundity and egg weight in English sole, Pleuronectes vetulus, from Puget Sound, Washington: influence of nu- tritional status and chemical contaminants, by Lyndal L. Johnson, Sean Y. Sol, Daniel P. Lomax, Gregory M. Nelson, Catherine A. Sloan, and Edmundo Casillas 250 The Loop Current and the abundance of larval Cubiceps pauciradiatus (Pisces: Nomeidae) in the Gulf of Mexico: evi- dence for physical and biological interaction, by John Lamkin 267 Influence of release season on size-dependent survival of cultured striped mullet, Mugil cephalus, in a Hawaiian estuary, by Kenneth M. Leber, H. Lee Blankenship, Steve M. Arce, and Nathan P. Brennan 280 Correcting annual catches from seasonal fisheries for use in virtual population anlysis, by Christopher M. Legault, and Nelson M. Ehrhardt 293 Estimates of tag loss from double-tagged sablefish, Anoplopoma fimbria, by William H. Lenarz and Franklin R. Shaw 300 Aerial survey of giant bluefin tuna, Thunnus thynnus, in the Great Bahama Bank, Straits of Florida, 1995, by Molly Lutcavage, Scott Kraus, and Wayne Hoggard 311 A population profile for Atlantic hagfish, Myxine glutinosa ( L. ), in the Gulf of Maine. Part I: Morphometries and re- productive state, by Frederic Martini, John B. Heiser, and Michael P. Lesser 882 Fishery Bulletin 95(4), 1997 321 The reproductive cycle and sexual maturity of Atka mack- erel, Pleurogrammus monopterygius, in Alaska Waters, by Susanne F. McDermott, and Sandra A. Lowe 334 Prey occurrence in pantropical spotted dolphins, Stenella attenuata, from the eastern tropical Pacific, by Kelly M. Robertson and Susan J. Chivers 349 Age, growth, and sexual maturity of shovelnose guitarfish, Rhinobatos productus (Ayres), by Maryellen Timmons and Richard N. Bray 360 Food habits and energy values of prey of striped marlin, Tetrapturus audax, off the coast of Mexico, by Leonardo A. Abitia-Cardenas, Felipe Galvan-Magaha, and Jesus Rodriguez-Romero 369 Note on plankton and cold-core rings in the Gulf of Mexico, by Douglas C. Biggs, Robert A. Zimmerman, Rebeca Gasca, Eduardo Suarez-Morales, Ivan Castellanos, and Robert R. Leben 376 Physical environment and recruitment variability of At- lantic herring, Clupea harengus, in the Gulf of Maine, by Mark A. Lazzari, David K. Stevenson, and Stephen M. Ezzy 386 In vitro digestibility of some prey species of dolphins, by Keiko Sekiguchi and Peter B. Best 394 Development of laboratory-reared sheepshead, Archo- sargus probatocephalus (Pisces: Sparidae), by John W. Tucker Jr. and Sabine R. Alshuth 95(3) 403 Optimal sampling for estimating the size structure and mean size of abalone caught in a New South Wales fishery, by Neil L. Andrew and Yong Chen 414 Seasonal and ontogenetic changes in distribution and abun- dance of smooth flounder, Pleuronectes putnami, and win- ter flounder, Pleuronectes americanus, along estuarine depth and salinity gradients, by Michael P. Armstrong 431 Energetics of larval red drum, Sciaenops ocellatus. Part II: Growth and biochemical indicators, by Ross I. Brightman, Joseph J. Torres, Joseph Donnelly, and M. Elizabeth Clarke 445 Retrospective analysis of virtual population estimates for Atlantic menhaden stock assessment, by Steven X. Cadrin and Douglas S. Vaughan 456 Maturation and reproductive seasonality in bonefish, Albula vulpes, from the waters of the Florida Keys, by Roy E. Crabtree, Derke Snodgrass, and Christopher W. Harnden 466 Seasonal movement of offshore American lobster, Homarus americanus, tagged along the eastern shore of Cape Cod, Massachusetts, by Bruce T. Estrella and Thomas D. Morrissey 477 Daily variability in abundance of larval fishes inside Beau- fort Inlet, by William F. Hettler Jr., David S. Peters, David R. Colby, and Elisabeth H. Laban 494 Effects of geography and bathymetry on growth and ma- turity of yellowfin sole, Pleuronectes asper, in the eastern Bering Sea, by Daniel G. Nichol 504 Habitat models for juvenile pleuronectids around Kodiak Island, Alaska, by Brenda L. Norcross, Franz-Josef Muter, and Brenda A. Holladay 521 Daily growth increments in otoliths of juvenile weakfish, Cy noscion regalis : experimental assessment of changes in increment width with changes in feeding rate, growth rate, and condition factor, by Richard Paperno, Timothy E. Targett, and Paul A. Grecay 530 Daily age and growth of larval and early juvenile Spanish mackerel, Scomberomorus maculatus, from the South At- lantic Bight, by John S. Peters and David J. Schmidt 540 Use of stable isotopes to assess groups of king mackerel, Scomberomorus cavalla, in the Gulf of Mexico and south- eastern Florida, by Lynn A. Roelke and Luis A. Cifuentes 552 Effectiveness of four industry-developed bycatch reduction devices in Louisiana’s inshore waters, by Donna R. Rogers, Barton D. Rogers, Janaka A. de Silva, and Vernon L. Wright 566 Global population structure of yellowfin tuna, Thunnus albacares, inferred from allozyme and mitochondrial DNA variation, by Robert D. Ward, Nicholas G. Elliott, Bronwyn H. Innes, Adam J. Smolenski, and Peter M. Grewe 576 An estimate of orange roughy, Hoplostethus atlanticus, bio- mass using the daily fecundity reduction method, by John R. Zeldis, R. I. Chris Francis, Malcolm R. Clark, Jonathan K. V. Ingerson, Paul J. Grimes, and Marianne Vignaux 598 Maturity and fecundity of arrowtooth flounder, Atheresthes stomias, from the Gulf of Alaska, by Mark Zimmermann 612 The reproductive biology and early life stages of Podothecus sachi (Pisces: Agonidae), by Hiroyuki Munehara 620 Age and growth of totoaba, Totoaba macdonaldi ( Sciaenidae ), in the upper Gulf of California, by Martha J. Roman- Rodriguez, and M. Gregory Hammann 629 Mark retention and growth of jet-injected juvenile marine fish, by John F. Thedinga, Adam Moles, and Jeffrey T. Fujioka 95(4) 637 Yield-per-recruit analysis and management strategies for Atlantic croaker, Micropogonias undulatus, in the Middle Atlantic Bight, by Luiz R. Barbieri, Mark E. Chittenden Jr., and Cynthia M. Jones 646 Daily and seasonal activity patterns of female cunner, Tautogolabrus adspersus (Labridae), in Newfoundland, by Carole Bradbury, John M. Green, and Michael Bruce- Lockhart 653 The composite square-mesh panel: a modification to codends for reducing unwanted bycatch and increasing 883 catches of prawns throughout the New South Wales oce- anic prawn-trawl fishery, by Matt K. Broadhurst and Steven J. Kennelly 665 Movements, feeding periods, and daily ration of piscivo- rous young-of-the-year bluefish, Pomatomus saltatrix, in the Hudson River estuary, by Jeffrey A. Buckel and David O. Conover 680 Population biology and harvest of the coral reef surgeonfish Acanthurus lineatus in American Samoa, by Peter C. Craig, J. Howard Choat, Lynda M. Axe, and Suesan Saucerman 694 Spatial and temporal variation in age and growth of king mackerel, Scomberomorus cavalla, 1977-1992, by Douglas A. DeVries and Churchill B. Grimes 709 Age, growth, distribution, and food habits of recently settled white seabass, Atractoscion nobilis, off San Diego County, California, by Christopher J. Donohoe 722 Analysis of diet and feeding strategies within an assem- blage of estuarine larval fish and an objective assessment of dietary niche overlap, by Daniel J. Gaughan and Ian C. Potter 732 Changes in the life history of red porgy, Pagrus pagrus, from the southeastern United States, 1972-1994, by Patrick J. Harris and John C. McGovern 748 Reproductive success of Pacific herring, Clupea pallasi, in Prince William Sound, Alaska, six years after the Exxon Valdez oil spill, by Scott W. Johnson, Mark G. Carls, Rob- ert P. Stone, Christine C. Brodersen, and Stanley D. Rice 762 Maximum population growth rates and recovery times for Atlantic cod, Gadus Morhua , by Ransom A. Myers, Gor- don Mertz, and P. Stacey Fowlow 773 Age and growth of larval and juvenile Atlantic croaker, Micropogonias undulatus, from the Middle Atlantic Bight and estuarine waters of Virginia, by Stephen W. Nixon and Cynthia M. Jones 785 Temporal and spatial spawning patterns of the Atlantic butterfish, Peprilus triacanthus , in the South and Middle Atlantic Bights, by Teresa Rotunno and Robert K. Cowen 800 A comparison of three genetic methods used for stock dis- crimination of orange roughy, Hoplostethus atlanticus : allozymes, mitochondrial DNA, and random amplified poly- morphic DNA, by Peter J. Smith, Peter G. Benson, and S. Margaret McVeagh 812 Survival of chinook salmon, Oncorhynchus tshawytscha, from a spawning tributary of the Rakaia River, New Zealand, in relation to spring and summer mainstem flows, by Martin J. Unwin 826 A shift in distribution of humpback whales, Megaptera novaeangliae, in response to prey in the southern Gulf of Maine, by Mason Weinrich, Malcolm Martin, Rachel Griffiths, Jennifer Bove, and Mark Schilling 837 Temporal variation in sexual maturity and gear-specific sex ratio of the vermilion snapper, Rhomboplites auro- rubens, in the South Atlantic Bight, by Boxian Zhao and John C. McGovern 849 A polyphasic growth function for the endangered Kemp’s ridley sea turtle, Lepidochelys kempii , by Milani Chaloupka and George R. Zug 857 Age determination of larval strombid gastropods by means of growth increment counts in statoliths, by Felix A. Grana- Raffucci and Richard S. Appeldoorn 863 Bias in Chapman-Robson and least-squares estimators of mortality rates for steady-state populations, by Michael D. Murphy 869 The effects of formalin and freezing on ovaries of albacore, Thunnus alalunga, by Darlene Ramon and Norm Bartoo 873 Short-term retention of coded wire and internal anchor tags in juvenile common snook, Centropomus undecimalis, by Julie E. Wallin, John M. Ransier, Sondra Fox, and Robert H. McMichael Jr. 884 Fishery Bulletin 95(4), 1997 Fishery Bulletin Index Volume 95 (1-4), 1997 List of authors Abitia-Cardenas, Leonardo A. 360 Alemany, Francisco 187 Alshuth, Sabine R. 394 Alvarez, Federico 187 Andrew, Neil L. 403 Appeldoorn, Richard S. 857 Arce, Steve M. 267 Armstrong, Michael P. 414 Axe, Lynda M. 680 Barber, Willard E. 195 Barbieri, Luiz R. 637 Bartoo, Norm 869 Benson, Peter G. 800 Bertram, Douglas F. 1 Best, Peter B. 386 Biggs, Douglas C. 369 Blankenship, H. Lee 267 Bove, Jennifer 826 Bradbury, Carole 646 Bray, Richard N. 349 Brennan, Nathan P. 267 Brightman, Ross I. 431 Broadhurst, Matt K. 209, 653 Brodersen, Christine C. 748 Bruce-Lockhart, Michael 646 Buckel, Jeffrey A. 665 Cadrin, Steven X. 445 Carls, Mark G. 748 Casillas, Edmundo 231 Castellanos, Ivan 369 Chaloupka, Milani 849 Chen, Yong 403 Chittenden, Mark E., Jr. 637 Chivers, Susan J. 334 Choat, Howard J. 680 Cifuentes, Luis A. 540 Clark, Malcolm R. 576 Clarke, M. Elizabeth 431 Colby, David R. 477 Conover, David O. 665 Cowen, Robert K. 785 Crabtree, Roy E. 456 Craig, Peter C. 680 Crockford, Susan J. 11 de Silva, Janaka A. 552 DeMartini, Edward E. 137 DeVries, Douglas A. 694 Donnelly, Joseph 431 Donohoe, Christopher J. 709 Ehrhardt, Nelson M. 280 Elliott, Nicholas G. 566 Ellis, Denise M. 137 Estrella, Bruce T. 466 Ezzy, Stephen M. 376 Fisher, Joseph P. 25 Fowlow, P. Stacey 762 Fox, Sondra 873 Francis, R. I. Chris 576 Fujioka, Jeffrey T. 629 Galvan-Magana, Felipe 360 Gasca, Rebeca 369 Gaughan, Daniel J. 722 Gold, John R. 174 Goodyear, C. Phillip 39 Grana-Raffucci, Felix A. 857 Grecay, Paul A. 521 Green, John M. 646 Grewe, Peter M. 566 Griffiths, Marc H. 47 Griffiths, Rachel 826 Grimes, Churchill B. 694 Grimes, Paul J. 576 Gunderson, Donald R. 161, 219 Hammann, M. Gregory 620 Hampton, John 68 Harnden, Christopher W. 456 Harris, Patrick J. 732 Heiser, John B. 311 Hettler, William F., Jr. 477 Hinton, Michael G. 80 Hoggard, Wayne 300 Holladay, Brenda A. 504 Ingerson, Jonathan K. V. 576 Innes, Bronwyn H. 566 Johnson, Lyndal L. 231 Johnson, Scott W. 748 Jones, Cynthia M. 637, 773 Kane, Joseph 85 Kennelly, Steven J. 209, 653 Kraus, Scott 300 Laban, Elisabeth H. 477 Lamkin, John 250 Lazzari, Mark A. 376 Leben, Robert R. 369 Leber, Kenneth M. 267 Legault, Christopher M. 280 Leggett, William C. 1 Lenarz, William H. 293 Lesser, Michael P. 311 Livingston, Mary E. 99 Lomax, Daniel P. 231 Lowe, Sandra A. 321 Lutcavage, Molly 300 Martin, Malcolm 826 Martini, Frederic 311 McDermott, Susanne F. 321 McGovern, John C. 732, 837 McKenna, James E., Jr. 114 McMichael, Robert H. 873 McQuinn, Ian H. 126 McVeagh, S. Margaret 800 Mellink, Eric 180 Mertz, Gordon 762 Meyer, Robert M. 195 Miller, Thomas J. 1 Moles, Adam 629 Morrissey, Thomas D. 466 Munehara, Hiroyuki 612 Murphy, Michael D. 80, 863 Muter, Franz-Josef 504 Myers, Ransom A. 762 Nelson, Gregory M. 231 Nichol, Daniel G. 494 Nixon, Stephen W. 773 Norcross, Brenda L. 504 Paperno, Richard 521 Parrish, Frank A. 137 Pearcy, William G. 25 Peters, David S. 477 Peters, John S. 530 Potter, Ian C. 722 Powers, Joseph E. 149 Ramon, Darlene 869 Ransier, John M. 873 Restrepo, Victor R. 149 Rice, Stanley D. 748 Richardson, Linda R. 174 Robertson, Kelly M. 334 Rodriguez-Romero, Jesus 360 Roelke, Lynn A. 540 Rogers, Barton D. 552 Rogers, Donna R. 552 Roman Rodriguez, Martha J. 620 Rotunno, Teresa 785 Saucerman, Suesan 680 Schilling, Mark 826 Schmidt, David J. 530 Schofield, Kathy A. 99 Sekiguchi, Keiko 386 Shaw, Franklin R. 293 Shi, Yunbing 161 Sloan, Catherine A. 231 Smith, Ronald L. 195 Smith, Peter J. 800 Smolenski, Adam J. 566 Snodgrass, Derke 456 Sol, SeanY. 231 Stevenson, David K. 376 Stone, Robert P. 748 Suarez-Morales, Eduardo 369 Sullivan, Patrick J. 161 Targett, Timothy E. 521 Taylor, Ronald G. 80 Thedinga, John F. 629 Timmons, Maryellen 349 Torres, Joseph J. 431 Tucker, John W., Jr. 394 Unwin, Martin J. 812 Vallarino, Mark 195 Vaughan, Douglas S. 445 Vignaux, Marianne 99, 576 Wallin, Julie E. 873 Ward, Robert D. 566 Watson, John W. 209 Weinrich, Mason 826 Workman, Ian K. 209 Wright, Vernon L. 552 Zavala-Gonzalez, Alfredo 180 Zeldis, John R. 576 Zhao, Boxian 837 Zimmerman, Robert A. 369 Zimmermann, Mark 598 Zug, George R. 849 886 Fishery Bulletin 95(4), 1997 Fishery Bulletin Index Volume 95 (1-4), 1997 List of subjects Abalone, blacklip 403 Abundance cigarfish, bigeye 250 copepod 85 demersal fishes 195 flounder smooth 414 winter 414 larval fishes 477 rockfish 219 seabass, white 709 snapper, pink 137 tuna, bluefin 300 Acanthurus lineatus - see Surgeonfish, bluebanded Activity 646 Afurcagobius suppositus 722 Age at sexual maturity - see also Maturity bonefish 456 cod 762 mackerel, Atka 321 rockfish 219 snapper, vermilion 837 sole, yellowfin 494 Age determination 39 butterfish 785 conch 857 croaker, Atlantic 773 guitarfish, shovelnose 349 hoki, New Zealand 99 mackerel king, 694 Spanish 530 sardine, Atlanto-Iberian 187 seabass, white 709 totoaba 620 Age validation 530 Alaska Aleutian islands 321 Bering Sea 494 Chukchi Sea 195 Kodiak Island 504 Prince William Sound 748 Albula vulpes - see Bonefish Albulidae - see Bonefish Allozyme 566, 800 Annual proportion spawning 99 Anoplopoma fimbria - see Sablefish Anoplopomatidae - see Sablefish Archeological evidence 11 Archosargus probatocephalus - see Sheepshead Argyrosomus inodorus - see Kob, silver Artificial digestion 386 Assessment 267 Atheresthes stomias - see Flounder, arrowtooth Atlantic Ocean 837, 566 Atractoscion nobilis - see Seabass, white Australia 209, 722 New South Wales 403, 653 Bahama Bank 300 Balaenopteridae - see Humpback whale Baleen, 826 Bathymetry 494 Biochemical indicators 432 Biomass 576 Birthdate analysis 187 Blenniidae 722 Bluefish 665 Bonefish 456 Bootstrapping 722 Brevoortia tyrannus - see Menhaden, Atlantic British Columbia 11 Butterfish, Atlantic 785 Bycatch reducing devices, secondary 209 reduction 209, 552, 653 California 293, 349, 709 Catch-curve analysis “ Centropages hamatus 85 Centropagidae - see Copepod Centropomis undecimalis - see Snook, common Cheloniidae - see Turtles, sea Chesapeake Bay 653, 773 Cigarfish, bigeye 250 Clupea harengus - see Herring, Atlantic pallasi - see Herring, Pacific Clupeidae - see Herring, Menhaden, or Sardine Cod, Atlantic 762 Cold-core rings 369 Commercial fishery 114 Competition 722 wild and hatchery-reared salmon 25 winter and smooth flounder 414 Conch milk 857 queen 857 Condition factor 231, 521, 598, 680 Contaminants, chemical 231 Copepod 85 Croaker, Atlantic 637, 773 Cubiceps pauciradiatus - see Cigarfish, bigeye Cunner 646 Cynoscion regalis - see Weakfish Daily growth 187 ration 665 Delphinidae - see Dolphin Demersal fish assemblages 195 Density dependence 161 Depth 504 gradient 414 Development flounder, winter 1 sheepshead, 394 Diel 665 Diet - see Food habits Diet analysis 386 Dietary overlap fall versus spring chinook salmon, juvenile 25 Dietary niche overlap 722 Direct assessment 300 Distribution cigarfish, bigeye 250 copepod, 85 demersal fishes, 195 flounder smooth 414 winter 414 pleuronectids, juvenile 504 seabass, white 709 tuna, bluefin 11 whale, humpback 826 DNA, mitochondrial grouper, red 174 roughy, orange 800 tuna, yellowfin 566 Dolphin, pantropical spotted 334 Double tagging 68, 293 Drift macrophytes 709 Drum, red 431 Eastern tropical Pacific 334 Ecology 311 Egg development 394 production 231, 576 weight 231 ^ Elasmobranch 349 Elopomorpha 456 Energetics 431 Enoploteuthidae 334 Entanglement, incidence of 180 Epinephelinae - see Grouper Epinephelus morio - see Grouper, red Estradiol 17-13 231 Estuarine 722 and Atlantic croaker 773 and juvenile chinook salmon 25 English sole and nursery habitat 161 flounder habitat 414 Extralimital fish records 11 Exxon Valdez oil spill 748 Favonigobius lateralis 722 Fecundity 887 bonefish 456 flounder, arrowtooth 598 roughy, orange 576 sole, English 231 Feeding - see also Food habits ecology 722 chronology 665 rates 521 strategy 722 Fish larvae 722 Fish marking 629 Fishermen 114 Fishery artisanal 680 harvest 114 Louisiana shrimp 532 management abalone, blacklip 403 croaker, Atlantic 637 guitarfish, shovelnose 349 menhaden, Atlantic 445 rockfish stocks 219 prawn-trawl 653 production 137 subsistence 680 Fishing gear 180 Flatfish 231, 504 Florida 456, 540, 873 Broward county 114 Flounder arrowtooth 598 smooth 414 winter 1, 414 Flow variability 812 Food habits bluefish 665 dolphin 386 pantropical spotted 334 marlin, striped 360 salmon, chinook 25 seabass, white 709 Food web 540 Formalin 869 Freezing 869 Frontal areas, larval abundance in 250 Gadidae - see Cod Gadus morhua - see Cod, Atlantic Genetic studies grouper, red 174 polymorphism 566 tuna, yellowfin 566 Geography 494 Gobiidae 722 Gonadal development 456 Gonadosomatic index 80, 456, 598 Grouper, red 174 Growth butte rfish 785 conch 857 croaker, Atlantic 773 drum, red 431 flounder, winter 1 guitarfish, shovelnose 349 halibut, Pacific 629 herring, Atlantic 126 juvenile fish 629 mackerel king 694 Spanish 530 mullet, striped 267 porgy, red 732 rockfish 219 sablefish 629 sheepshead 394 sole English 161, 231 rock 629 yellowfin 494, 629 totoaba 620 turtle, Kemp’s ridley sea 849 variation in 1 Growth rates 521, 530, 709 Growth overfishing 637 Guitarfish, shovelnose 349 Gulf of Alaska 321, 598 of California 180, 360, 620 of Maine 311, 376, 826 of Mexico 174, 250, 369, 540 of St. Lawrence 126 Habitat use 414 Habitat 219 models 504 nursery 137, 709 Hagfish 311 Halibut, Pacific 504 Haliotidae - see Abalone Haliotis rubra - see Abalone, blacklip Harvest 114 Hatchery release 267 Hawaii 267 Hawaiian Archipelago 137 Hepatosomatic index 598 Herring Atlantic 126, 376, 826 Pacific 748 Hexigrammidae - see Mackerel Hippoglossoides elassodon - see Sole, flathead Hippoglossus stenolepis - see Halibut, Pacific Histology bonefish 456 effects of marking on 629 flatfish 629 flounder, arrowtooth 598 hoki, New Zealand 99 mackerel, Atka 321 swordfish 80 Hoki, New Zealand 99 Homarus americanus - see Lobster, American Homing 466 Hoplostethus atlanticus - see Roughy, orange Iberian peninsula 187 Ichthyoplankton 722 Indian Ocean 566 Inshore waters 552 Interannual variability 219 Isotopes stable carbon 540 stable nitrogen 540 Istiophoridae - see Marlin Jeffreys Ledge 826 Jet-injected marking 629 Juvenile studies and jet-injected marking 629 croaker, Atlantic 773 flounder, winter 1 mackerel, Spanish 530 pleuronectids 504 salmon, chinook 25 sardine, Atlanto-Iberian 187 seabass, white 709 sheepshead 394 snapper, pink 137 snook, common 873 sole, English 161 weakfish 521 Kob, silver 47 Labridae - see Cunner Lapillus 773 Larval studies cigarfish, bigeye 250 conch, 857 croaker, Atlantic 773 drum, red 431 mackerel, Spanish 530 flounder, winter 1 sheepshead 394 Lactate dehydrogenase (LDH) activity 431 Least-squares method 149 Length at age 494 Length at sexual maturity 321, 456, 494 Lepidochelys kempii - see Turtle, Kemp’s ridley sea Life history 837 bonefish, 456 kob, silver 47 porgy, red 732 surgeonfish 680 Lipid 43 1 Lobster, American 466 Loop Current 250, 369, 540 Louisiana 552 Low-frequency reproduction 99 Lutjanidae - see Snapper Lutjanus campechanus - see Snapper, red Mackerel Atka 321 king 540, 694 Spanish 530 Macruronus novaezelandiae - see Hoki, New Zealand Marginal increment analysis 530 Mark retention 629 Marlin, striped 360 Massachusetts, Cape Cod 466 Maturity ogive 99 Maturity - see also Age at sexual maturity flounder, arrowtooth 598 888 Fishery Bulletin 95(4), 1997 kob, silver 51 sole, yellowfin 494 surgeonfish 680 Mean size 403 Megaptera nouaeangliae - see Whale, humpback Menhaden, Atlantic 445 Merisitics 785 Merlucciidae - see Hoki Methods age determination from length 39 assessing dietary niche overlap 722 classifying reproductive condition 80 daily fecundity reduction 576 determining sampling size 403 estimation of mortality rates Chapman-Robson 863 least-squares 863 genetic 800 in virtual population analysis 280 skeletochronological 849 Mexico 180, 360 Micropogonias undulatus - see Croaker, Atlantic Middle Atlantic Bight 637, 773, 785 Migration 47, 300 Mississippi River 540 Mitochondrial DNA 174, 566, 800 Model age 39 habitat 504 growth 800 spawner- recruitment 762 stock assessment 149 yield-per-recruit 637 Modeling 637 Morphometries 311, 394, 785 Mortality 161, 637, 863 rates in VPA’s 280 Mouth size 722 Movements 466, 646, 665 Mugil cephalus - see Mullet, striped Mugilidae - see Mullet Mullet, striped 267 Multispecies 114 Myctophidae 334 Myxine glutinosa - see Hagfish Myxinidae - see Hagfish Nephrophidae - see Lobster New Hampshire 414 New York, Hudson River 665 New Zealand 576, 99 Rakaia River 812 Newfoundland 646 Nomeidae 250 Nonlinear estimation 849 Nonparametric smoothing 849 Nordmpre-grid 209 North Carolina 637 Beaufort 477 Nursery areas juvenile pleuronectids 504 kob, silver 47 snapper, pink 137 sole, English 161 Nutrition 231 Oceanic islands 137 Ommastrephidae 334 Oncorhynchus tshawytscha - see Salmon, chinook Oocyte stages 321, 456, 598 Oregon 293 Coos Bay 25 Otariidae - see Sea lion Otolith butterfish 785 croaker, Atlantic 773 mackerel, Spanish 530 seabass, white 709 totoaba 620 weakfish 521 Outliers, detection with robust regression 149 Ovary, effects of preservation 869 Overfishing 219, 732 Pacific Ocean tuna 68, 566 Pacific tuna fisheries 11 Pagrus pagrus - see Porgy, red Parablennius tasmanianus 722 Pelagic fishes 300 Penaeus plebejus - see Prawn, eastern King Peprilus triacanthus - see Butterfish, Atlantic Perch, Pacific ocean 219 Phenotypic variation 1 Physical environment 376 oceanography 250 Pigmentation 394 Piscivore 665 Pleurogrammus monopterygius - see Mackerel, Atka Pleuronectes americanus - see Founder, winter asper - see Sole, yellowfin bilineatus - see Sole, rock putnami - see Flounder, smooth vetulus - see Sole, English Pleuronectidae - see Flounder and Sole Poacher, sail-fin 612 Podothecus sachi - see Poacher, sail-fin Polychlorinated biphenyls (PCB’s) 231 Polycyclic aromatic hydrocarbons (PAH’s) 231 Polymerase chain reaction (PCR) 800 Polyphasic growth 849 Pomatomidae - see Bluefish Pomatomus saltatrix - see Bluefish Population biology of surgeonfish 680 dynamics menhaden, Atlantic 445 rockfish 219 growth rate for cod 762 profile for hagfish 311 structure grouper, red 174 herring, Atlantic 126 tuna, yellowfin 566 Porgy, red 732 Prawn, eastern King 653 Prawn-trawl 209, 653 Preservation 869 Prey digestibility 386 energy values 360 mesopelagic 334 occurrence 334 size 722 Pristipomoides filamentosus - see Snapper, pink Production snapper, juvenile pink 137 Protein 431 Proximate composition 431 Pseudogobius olorum 722 Random amplified polymorphic DNA (RAPD) 800 Ration level 43 1 Ray 349 Recapture rate 267, 466 Recreational fishery 456 Recruitment 250, 773 herring, Atlantic 376 sardine, Atlanto-Iberian 187 Replenishment (marine fish populations) 267 Reproductive biology - see also Spawning bonefish 456 guitarfish, shovelnose 349 hagfish 311 herring, Pacific 748 kob, silver 47 mackerel, Atka 321 porgy, red 732 sole, English 231 swordfish 80 Resource availability 722 Retrospective analysis 445 Rhinobatidae - see Guitarfish Rhinobatos productus - see Guitarfish, shovelnose Rhomboplites aurorubens - see Snapper, vermilion RNA-DNA ratios 431 Robust regression 149 Rockfish darkblotched 219 rougheye 219 splitnose 219 Roughy, orange 576, 800 Sablefish 293 Sagitta 773 Salinity gradient 414 Salmonidae - see Salmon Salmon, chinook 25, 812 Samoa, American 680 Sampling 403, 477 Sanctuaries 219 Sand lance 826 889 Sardina pilchardus - see Sardine, Atlanto-Iberian Sardine, Atlanto-Iberian 187 Schooling 300 Sciaenidae - see Drum, Croaker, Kob, Totoaba, Seabass, and Weakfish Sciaenops ocellatus - see Drum, red Scomberomorus cavalla - see Mackerel, king maculatus - see Mackerel, Spanish Scombridae - see Tuna and Mackerel Scorpaenidae - see Rockfish Sea lion, California 180 Sea ranching 267 Seasonal movements 414 Seasonal timing 267 Sebastes alascanus - see Thornyhead, shortspine aleutianus - see Rockfish, rougheye alutus - see Perch, Pacific ocean crameri - see Rockfish, darkblotched diploproa - see Rockfish, splitnose Sediment 504 Sex ratio 837, 47 Sheepshead 394 Shrimp trawls 552 Size-at-release 267 Size-structure 403 Snapper red 39 pink 137 vermilion 837 Snook, common 873 Sole English 161, 231 flathead 504 rock 504 yellowfin 494, 504 South Africa 47 South Atlantic Bight 530, 732, 785 Spain 187 Sparidae - see Porgy or Sheepshead Spatial patterns 219 Spatial variations 694 Spawner-recruitment 762 Spawning - see also Reproductive biology biomass and hoki 99 bonefish 456 butterfish 785 season and herring 126 stock structure and swordfish 80 surgeonfish 680 Species assemblage Florida fishery harvest 114 rockfish 219 Square-mesh panels 653 Statolith 857 Stellwagen Bank 826 Stenella attenuata - see Dolphin, pantropical spotted Stock assessment 149, 445 discrimination 800 enhancement 267 interchange 466 rebuilding 219 separation 47, 540 structure 566 Stock-recruitment 762 Stromateidae - see Butterfish Stromateoid 250 Strombidae - see Conch Strombus costatus - see Conch, milk Strombus gigas - see Conch, queen Subtropical 114 Surgeonfish, bluebanded 680 Survey, aerial 300 Survival 812 mullet, striped 267 sole, English 161 Swordfish 80, 149 Syngnathidae 722 Tag coded wire 873 double 68, 293 internal anchor 873 loss 293 reporting 68 retention 293, 873 seeding 68 shedding 68 Tag- release-recapture 267, 466 Tagging 466, 646 Tautogolabrus adspersus - see Gunner 646 Temperature 161, 762 effects on red drum growth 43 1 pleuronectids and habitat use 504 Temporal variation 694, 837 Tetrapturus audax - see Marlin, striped Thornyhead, shortspine 219 Thunnus albacares - see Tuna, yellowfin Thunnus thynnus - see Tuna, bluefin Torpor 646 Totoaba macdonaldi - see Totoaba Totoaba 620 Trachichthyidae - see Roughy Tracking 646 Triglycerides 231 Tuna albacore 869 bluefin 11, 149, 300 distributions 11 prehistoric 11 tagging 68 yellowfin 566 Turtle, Kemp’s ridley sea 849 United States northeast 85 southeast 732 Urocampus carinirostris 722 Vancouver Island 11 Variability 477 Variation allozyme and mitochondrial DNA 566 in individual flounder 1 Video camera, baited 137 Virginia 773 Virtual population analysis 280, 445 Washington 161, 293 northern 11, 219 Puget Sound 231 Weakfish 521 Whale, humpback 826 Xiphias gladius - see Swordfish Xiphiidae - see Swordfish Year-class twinning 126 Yield-per-recruit 637 Zalophus californianus californianus - see Sea lion, California Zooplankton 369, 722 890 Fishery Bulletin 95(4), 1997 Superintendent of Documents Order Form *5178 □yes, please send me the following publications: subscriptions to Fishery Bulletin (FB) for $34.00 per year ($42.50 foreign). The total cost of my order is * . Price includes regular shipping and handling and is subject to change. International customers please add 25%. Company or personal name (Please type or print) Additional address/attention line Street address City, State, Zip code Daytime phone including area code Charge your order. It’s easy! P3S Purchase order number (optional) Check method of payment: □ Check payable to Superintendent of Documents □ GPO Deposit Account | f □ VISA □ MasterCard ~^\ (expiration date) To fax your orders (202) 51 2-2250 To phone your orders (202) 512-1800 Thank you for your order! Authorizing signature Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 This statement is required by the Act of August 12, 1970, Section 3686, Title 39, U.S. Code, showing ownership, management, and circulation of the Fishery Bulletin, publication number 366-370, and was filed on 11 Septem- ber 1997. The Bulletin is published quarterly (four issues annually) with an annual subscrip- tion price of $34.00 (sold by the Superinten- dent of Documents, U.S. Government Printing Office, Washington, DC 20402). The complete mailing address of the office of publication is: NMFS Scientific Publications Office, NOAA, 7600 Sand Point Way NE, BIN C16700, Seat- tle, WA 98116. The complete mailing address of the headquarters of the publishing agency is: National Marine Fisheries Service, NOAA, Department of Commerce, 1336 East-West Highway, Silver Spring, MD 20910. The name of the publisher is Willis Hobart and the man- aging editor is Sharyn Matriotti; their mailing address is: NMFS Scientific Publications Of- fice, 7600 Sand Point Way NE, BIN C16700, Seattle, WA 98116. The owner is the U.S. De- partment of Commerce, 14th St., N.W., Wash- ington, DC 20230; there are no bondholders, mortgages, or other security holders. The pur- pose, function, and nonprofit status of the or- ganization (agency) and the exempt status for Federal income tax purposes has not changed during 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 2,033 and the actual number of copies of the single issue published nearest to the filing date was 2,038. Paid circulation (B) is handled by the U.S. Government Printing Office, Wash- ington, DC 20402, and (C) the total number of printed for their sales (mail subscriptions and individual sales) was 600 for both the average number of copies each issue during the preced- ing 12 months and the actual number of copies of the single issue published nearest to the fil- ing date. Free distribution (D) by mail; sam- ples, complimentary, and other free copies (average number of copies each issue during the preceding 12 months) was 1,433 and the actual number of copies of the single issue pub- lished nearest to the filing date was 1,438. Free distribution outside the mail (E) by car- riers or other means was 0 for both average number of copies and actual number of cop- ies. Total free distribution (F) was 1,433 (av- erage number of copies during the preceding 12 months) and 1,438 for actual number of copies of the single issue published nearest the filing date. The toted distribution (G: sum of C and G) (average number of copies each issue during the preceding 12 months) was 2,033 and the actual number of copies of the single issue published nearest to the filing date was 2,038. There were no copies not dis- tributed or returned from news agents (H). The total (I: sum of G and H) is equal to the net press run figures shown in Item A: 2,033 and 2,038 copies, respectively. I certify that the statements made by me above are cor- rect and complete: (Signed) Willis Hobart, Publisher. Fishery Bulletin Guide for Contributors Content Articles published in Fishery Bulletin de- scribe original research in fishery marine science, engineering and economics, and the environmental and ecological sciences, in- cluding modeling. Articles may range from relatively short to extensive; notes are re- ports of 5 to 10 pages without an abstract and describing methods or results not sup- ported by a large body of data. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not upon the editor or the publisher. It is therefore im- portant that the contents of the manuscript are carefully considered by the authors. Submission of an article is understood to imply that the article is original and is not being considered for publication elsewhere. Manuscripts should be written in English. Authors whose native language is not En- glish are strongly advised to have their manuscripts checked by English-speaking colleagues prior to submission. Preparation Title page should include authors’ full names and mailing addresses, the corre- sponding author’s telephone, FAX number, and E-mail address, and a list of key words to describe the contents of the manuscript. Abstract should not exceed one double- spaced typed page. It should state the main scope of the research but emphasize its con- clusions and relevant findings. Because ab- stracts are circulated by abstracting agencies, it is important that they represent the research clearly and concisely. Text must be typed double-spaced through- out. A brief introduction should portray the broad significance of the paper; the remain- der of the paper should be divided into the following sections: Materials and meth- ods, Results, Discussion (or Conclusions), and Acknowledgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e. there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbrevia- tions, and technical terms should be spelled ☆ U.S. Gov. Printing Office 1997 589017/40012 out the first time they are mentioned. The scientific names of species must be written out the first time they are mentioned; sub- sequent mention of scientific names may be abbreviated. Follow the U.S. Government Printing Office Style Manual (1984 ed.) and the CBE Scientific Style and Format (6th ed.) for editorial style, and the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada for fish nomenclature. Dates should be written as follows: 11 November 1991. Measurements should be expressed in metric units, e.g., metric tons as (t); if other units of measure- ment are used, please make this fact explicit to the reader. The numeral one (1) should be typed as a one, not as a lower-case el (1). Use of appendices is discouraged. Text footnotes should be numbered with Arabic numerals. Footnote all personal communications, unpublished data, and un- published manuscripts with full address of the communicator or author, or, as in the case of unpublished data, where the data are on file. Authors are advised to avoid ref- erences to nonstandard (gray) literature, such as internal, project, processed, or ad- ministrative reports. Where these refer- ences., are used, please include whether they are available from NTIS (National Techni- cal Information Service) or from some other public depository. Literature cited comprises published works and those accepted for publication in peer- reviewed literature (in press). Follow the name and year system for citation format. In the text, cite as follows: Smith and Jones (1977) or (Smith and Jones, 1977). If there is a sequence of citations, list by year: Smith, 1932; Smith and Jones, 1986; Smith and Allen, 1986. Abbreviations of serials should conform to abbreviations given in Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and completeness of all citations. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be ex- plained in the table legend. Other inciden- tal comments may be footnoted with italic numerals. Use the asterisk only to indicate probability in statistical data. Because ta- bles are typeset, they need only be submit- ted typed and formatted, with double-spaced legends. Zeros should precede all decimal points for values less than one. Figures must be cited in numerical order in the text. The senior author’s last name and the figure number should be written on the back of each one. Hand-drawn illustrations should be submitted as originals and not as photocopies. Submit photographs as glossy prints or slides that show good contrast, otherwise we cannot guarantee a good final printed copy. Graphic illustrations should be submitted as laser-printed copies, not as photocopies. Label all figures with Helve- tica typeface and capitalize the first letter of the first word in axis labels. Italicize spe- cies name and variables in equations. Use zeros before all decimal points. Use upper- case Times Roman bold typeface to label the parts of a figure, e.g. A, B, C, etc. Send orig- inal figures to the Scientific Editor when the manuscript has been accepted for publi- cation. Each figure legend should explain all symbols and abbreviations in the figure and should be double-spaced and placed at the end of the manuscript. Copyright law does not cover government publications; they fall within the public do- main. If an author reproduces any part of a government publication in his work, refer- ence to source is appreciated. Submission Send printed copies (original and three cop- ies) to the Scientific Editor: Dr. John B. Pearce, Scientific Editor Northeast Fisheries Science Center National Marine Fisheries Service 166 Water Street Woods Hole, MA 02543-1097 Once the manuscript has been accepted for publication, you will be asked to submit a software copy of your manuscript to the Man- aging Editor. The software copy should be submitted in WordPerfect or Microsoft Word text format and placed on a 3.5-inch disk that is double-sided, double or high density, and that is compatible with either DOS or Apple Macintosh systems. A copy of page proofs will be sent to the au- thor for final approval prior to publication. Copies of published articles and notes are available free of charge to the senior author (50 copies) and to his or her laboratory (50 copies). Additional copies may be purchased in lots of 100 when the author receives page proofs. Heckman BINDERY, INC. Bound-To-Please® FEB 00 N. MANCHESTER, INDIANA 46962