MBL Volume 89 Number 1 January 1991 U.S. Department of Commerce \ risncry ^, S5; BulletJjJ^ U.S. Department of Commerce Robert Mosbacher Secretary National Oceanic and Atmospheric Administration John A. Knauss Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William W. Fox. Jr. Assistant Administrator for Fisheries / w \ c ') r 2 = 0.589 6-0.367(1+1.025)) r 2 = 0.532, e -0.280(t + 2.657)\2.88 r 2 _ Q.549 -0.219(t+3.061)\2.8E )2.85 r 2 = 0.474, The growth curves suggest rapid growth for sheeps- head to an age of 6-8 years, after which an asymptote is approached. The values of K, the von Bertalanffy growth coefficient, for sheepshead are relatively high when compared with other sparids (Table 1 ) and to non- sparid perciform fish of similar size (see Pauly 1980). This indicates that sheepshead exhibit relatively rapid growth to an asymptotic size. This could be a result of living in the highly productive waters adjacent to the Mississippi Delta. Beckman et al : Age and growth estimation of Archosargus probatocephalus in Louisiana waters o z rx O 250 200 -r FEMALE — MALE 10 12 14 16 18 20 22 AGE (years) Figure 3 Von Bertalanffy growth models by length for male and female sheepshead. M = male; F = female. Due to the large variability in age at a given body size, size does not accurately estimate age for sheeps- head, especially beyond 2-3 years of age. For exam- ple, a given sheepshead greater than 400 mm or 1.5 kg could be of any age from 3 to 20 years. Otoliths ranged in weight from 28.3 mg (for a 229- mm, 312-g, 2-year-old fish) to 323.5 mg (for a 450-mm, 2410-g, 18-year-old fish). Age-otolith weight (OW) re- gressions (Fig. 5) were significantly different for males and females (P< 0.0001 for slopes, P = 0.0006 for inter- cepts); therefore multiple-regression models were fit separately by sex. Dependent variables included in multiple regression models at the 0.10 level of signif- icance were otolith weight and total weight for males, and otolith weight and fork length for females. The ad- dition of any other variables did not significantly im- prove the fit of the models. The model statistics are presented in Table 2. Since otolith weight accounted for more of the variability in age (83-85% vs. <60% for fish length or weight), it was the best estimator of age of all mor- phometric variables considered. However, there was still considerable variability in otolith weight within each age class. Although some of the remaining vari- ability (1-2%) was accounted for by considering fish length or weight in addition to otolith weight, the unex- plained variability was great enough that these models could not be used for precise estimation of sheepshead age, such as is needed to determine population year- class structure. However, they could be used to approx- imate age distribution patterns in the population. In order to obtain precise age estimates of individual fish, 4000 3600 3200 2S00 2400 2000 1600 1200 800 400 • . » • " i> Q 5 s£ if. ;" : s ° r, = 753 10 12 14 AGE (years) Figure 4 Von Bertalanffy growth models by weight for male and female sheepshead. M = male; F = female. 300 FEMALE y * "* / ^^MALE 250 F ' ' S ' - /^" O) t/ *• I/** "♦" E 0/^'*' 200 I ' * "C ^ x^ * ' " " * ' OI3M H , ' / "- M<+ v "-* •"y. -1 100 (4Mk " " O H \im%' o §7 "' 50 8* r ) 2 4 6 8 10 12 14 16 18 20 22 AGE (years) Figure 5 Log-linear growth models by otolith weight for male and female sheepshead. M = male; F = female. otolith annulus counts are necessary. The linear relationship between otolith weight and age indicates that otolith growth is continuous for sheepshead, whereas fish size (length and weight) asymptotes at intermediate ages and, therefore, is not continuous. This suggests that fast-growing (younger) sheepshead have lighter otoliths than equal-sized slow- growing (older) sheepshead, i.e., otolith growth con- tinues with age, independent of fish growth. This may be a general characteristic of fish growth, as similar observations have been made for other fish species (Templeman and Squires 1956, Beamish 1979, Wilson Fishery Bulletin 89(1), 1991 Table 2 Regression coefficients and statistics on multiple-regression models of age for sheepshead. Models were fit in a stepwise manner using independent variables of otolith weight, fish standard length, fish total weight , condition factor, and associated interaction terms. Variables were log-transformed for analyses. Partial Variable Coefficient SE P r 2 Males (n = 330) (1-variable model) Intercept = -4.420 Otolith weight 1.321 0.0305 <0.0001 0.850 (2-variable model) Intercept = -3.207 Otolith weight 1.569 0.0594 < 0.0001 0.850 Total weight -0.331 0.0686 <0.0001 0.010 Females (n = 366) (1-variable model) Intercept = -3.824 Otolith weight 1.176 0.0276 <0.0001 0.832 (2-variable model) Intercept = 1.258 Otolith weight 1.484 0.0502 <0.0001 0.832 Fork length -1.094 0.1531 < 0.0001 0.021 and Dean 1983, Radtke et al. 1985, Reznick et al. 1989, Secor et al. 1989). Since fish length or weight accounts for significant variability in age after considering otolith weight, large fish have larger otoliths than equal-aged small fish, i.e., otolith growth is affected by fish growth. These observations support the proposi- tion by Secor and Dean (1989) that not only do otoliths grow in a continuous manner, independent of somatic growth, but also that otolith growth is coupled in some manner to somatic growth. These growth patterns should be considered when using otoliths for back-calculation of fish size at age. Since continued otolith growth uncoupled from somatic growth would result in slower-growing fish having larger otoliths at a given fish size, the fish-otolith size relationship would be different for fast-growing and slow-growing fish. This bias would be more pronounced at older ages, since otolith growth could continue even if somatic growth has stopped. Due to gear selectivity and sorting of some samples by fishermen, the age distribution of our samples was not considered to be representative of the Louisiana sheepshead population. Future research should include fishery-independent sampling to accurately character- ize the age structure of the sheepshead population, determine variability in recruitment, estimate mor- tality rates, and identify sources of variability in growth. Additional samples of older fish are required to complete validation of age estimates for the oldest individuals. Acknowledgments Sampling efforts were supported by the Louisiana State University's Sea Grant Developmental Program, and the Louisiana Board of Regents. We thank the seafood dealers (including Preston Battistella, Fred and Debbie Black, and Harlon Pearce), tournament organ- izers, and fishermen for providing samples and facil- ities. Thanks to Dean Blanchard, David Nieland, Robert Parker, David Stanley, and Bruce Thompson for assistance in data collection. Citations 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. Beamish, R.J. 1979 New information on the longevity of the Pacific ocean perch (Sebastes alutus). J. Fish. Res. Board Can. 36: 1395-1400. Beamish, R.J., and D.A. Fournier 1981 A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-983. Beckman, D.W., C.A. Wilson, and A.L. Stanley 1989 Age and growth of red drum, Sciaenops ocellatus, from offshore waters of the Northern Gulf of Mexico. Fish. Bull., U.S. 87:17-28. Beckman, D.W., A.L. Stanley, J.H. Render, and C.A. Wilson 1990a Age and growth of black drum in Louisiana waters of the Gulf of Mexico. Trans. Am. Fish. Soc. 119:537-544. Beckman et al.: Age and growth estimation of Archosargus probatocephalus in Louisiana waters Beckman, D.W.. C.A. Wilson, D.L. Nieland, and A.L. Stanley 1990b Age structure, growth rates, and reproductive biology of black drum in the northern Gulf of Mexico. Final report under 1988-1989 U.S. Department of Commerce cooperative agreement NA89WC-H-MF017, Marine Fisheries Initiative (MARFIN) Program. Louisiana State Univ., Coastal Fish. Inst., Baton Rouge, 79 p. Boehlert, G.W. 1985 Losing objective criteria and multiple regression models for age determination in fishes. Fish. Bull., U.S. 83:103-117. Buxton, CD., and J.R. Clarke 1986 Age, growth and feeding of the blue hottentot Pachy- metopon aeneum (Pisces: Sparidae) with notes on reproductive biology. S. Afr. J. Zool. 21:33-38. 1989 The growth of Cymatoceps nasutus (Teleostei: Sparidae), with comments on diet and reproduction. S. Afr. J. Mar. Sci. 8:57-65. Chang. W.B. 1982 A statistical method for evaluating the reproducibil- ity of age determination. Can. J. Fish. Aquat. Sci. 39: 1208-1210. Coetzee, P.S., and D. Baird 1981 Age, growth and food of Cheimerius nufar (Ehrenberg, 1820) (Sparidae), collected off St Croix Island, Algoa Bay. S. Afr. J. Zool. 16:137-143. Finkelstein, S.L. 1969 Age and growth of scup in the waters of eastern Long Island. N.Y. Fish Game J. 16:84-110. Geoghegan, P., and M.E. Chittenden Jr. 1982 Reproduction, movements, and population dynamics of the longspine porgy, Stenotomus caprinus. Fish. Bull., U.S. 80:523-540. Hecht, T., and D. Baird 1977 Contributions to the biology of the pange, Pterogymnus laniarus (Pisces: Sparidae): Age, growth and reproduction. Zool. Afr. 12:363-372. Hildebrand, S.F., and L.E. Cable 1938 Further notes on the development and life history of some teleosts at Beaufort, N.C. Bull. U.S. Bur. Fish. 48:505-642. Horn, P.L. 1986 Distribution and growth of snapper Chrysophrys auratus in the North Taranaki bight, and management implications of these data. N.Z. J. Mar. Freshwater Res. 20:419-430. Johnson, A.G., W.A. Fable Jr., M.L. Williams, and L.E. Barger 1983 Age, growth, and mortality of king mackerel, Scom- beramorus cavalla, from the southeastern United States. Fish. Bull., U.S. 81:97-106. Koranteng, K.A., and T.J. Pitcher 1987 Population parameters, biannual cohorts, and assessment in the Pagellus bellottii (Sparidae) fishery off Ghana. J. Cons. Cons. Int. Explor. Mer 43:129-138. Louisiana Outdoor Writers Association 1987 Official fish records. The Menhaden Advisory Council for the Gulf of Mexico, New Orleans [pamphlet]. Manooch, C.S., and G.R. Huntsman 1977 Age, growth, and mortality of the red porgy, Pagrus pagrus. Trans. Am. Fish. Soc. 106:26-33. NMFS (National Marine Fisheries Service) 1982 Fisheries of the United States, 1981. Current Fish. Stat. 8200, Natl. Mar. Fish. Serv., NOAA, Silver Spnng, MD, 131 p. 1990 Fisheries of the United States, 1989. Current Fish. Stat. 8900, Natl. Mar. Fish. Serv., NOAA, Silver Spring, MD, 111 p. on, L. 1988 An introduction to statistical methods and data analysis, 3d ed. PWS-KENT Publ. Co., Boston, 835 p. Pauly, D. 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. Pulfrich, A., and C.L. Griffiths 1988 Growth, sexual maturity and reproduction in the hotten- tot Pachymetopon blochii (Val.). S. Afr. J. Mar. Sci. 7:25-36. Radtke, R.L., M.L. Fine, and J. Bell 1985 Somatic and otolith growth in the oyster toadfish (Op- sanus tau L.). J. Exp. Mar. Biol. Ecol. 90:259-275. Reznick, D., E. Lindbeck, and H. Bryga 1989 Slower growth results in larger otoliths: An experimen- tal test with guppies (Poecilia reticulata). Can. J. Fish. Aquat. Sci. 46:108-112. Sakamoto, T. 1984 Age and growth of the red sea bream in the outer waters adjacent to the Kii Strait. Bull. Jpn. Soc. Sci. Fish. 50: 1829-1834. Seeor, D.H., and J.M. Dean 1989 Somatic growth effects on the otolith-fish size relation- ship in young pond-reared striped bass, Morcme 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. P.-V. Reun. Cons. Int. Explor. Mer 191: 431-438. Sokal, R.R., and F.J. Rohlf 1981 Biometry, 2d ed. W.H. Freeman, San Francisco, p. 459-460. Springer, V.G., and K.D. Woodburn 1960 An ecological study of the fishes of the Tampa Bay area. Fla. Board Conserv. Mar. Lab. Prof. Pap. Ser. 1, 104 p. Spurr, A.R. 1969 A low-viscosity epoxy resin embedding medium for elec- tron microscopy. J. Ultrastruct. Res. 26:31-34. Statistical Analysis Systems 1985 SAS User's Guide: Statistics, Version 5 ed. SAS Inst. Inc.. Cary, NC, 956 p. Templeman, W., and H.J. Squires 1956 Relationship of otolith lengths and weights in the had- dock Melanogrammus aeglefinus (L.) to the rate of growth of the fish. J. Fish. Res. Board Can. 13:467-487. von Bertalanffy, L. 1938 A quantitative theory of organic growth. II. Inquiries on growth laws. Hum. Biol. 10:181-213. 1957 Quantitative laws in metabolism and growth. Q. Rev. Biol. 32:217-231. Waltz, C.W., W.A. Roumillat, and C.A. Wenner 1982 Biology of the whitebone porgy, Calamus leucosteus, in the south Atlantic Bight. Fish. Bull., U.S. 80:863-874. White, M.L., and M.E. Chittenden Jr. 1987 Age determination, reproduction and population dynam- ics of the Atlantic croaker, Micropogonias undulatus. Fish. Bull, U.S. 75:109-123. Wilson, C.A., and J.M. Dean 1983 The potential use of sagittae for estimating age of Atlan- tic swordfish, Xiphias gladius. In Prince, E.D., and L.M. Pulos (eds.), Proceedings of the international workshop on age determination of oceanic pelagic fishes: Tunas, billfishes, and sharks, p. 151-156. NOAA Tech. Rep. NMFS 8. Wilson, C.A., R.J. Beamish, E.B. Brothers, K.D. Carlander, J.M. Casselman, J.M. Dean, A. Jearld, E.D. Prince, and A. Wild 1987 Glossary. In Summerfelt, R.C., and G.E. Hall (eds.), Age and growth of fish, p. 527-530. Iowa State Univ. Press, Ames. Fishery Bulletin 89(1), 1991 Wilson, C.A., J.H. Render, D.W. Beckman, and A.L. Stanley 1988 The age structure and reproductive biology of sheeps- head (Archosargus probatocephaliis) landed in Louisiana. Final report of funded projects FY 1987-1988, Louisiana Board of Regents' Rockefeller Fund Interest Earnings Grant Program. Louisiana State Univ., Coastal Fish. Inst., Baton Rouge, 49 p. Wilson, C.A., D.W. Beckman, D.L. Nieland, and A.L. Stanley 1989 Age, growth, and the reproductive biology of schooling red drum from the northern Gulf of Mexico. 1988-1989 final report to Louisiana Department of Wildlife and Fisheries and U.S. Department of Commerce, Marine Fisheries Initiative (MARFIN) Program. Louisiana State Univ., Coastal Fish. Inst., Baton Rouge, 18 p. Abstract. - Batch fecundity, weight-specific fecundity (number of eggs per gram somatic weight), size of ripe ovarian eggs, and the somatic and gonadal conditions of adult fe- male queenfish Seriphus politus were estimated for five spawning seasons during an 8-year (1979-86) period. The effects of female somatic weight were evaluated in analyses of covari- ance comparing batch fecundity, egg size, and gonadal condition among years. Batch fecundity was positively (and allometrically) related to female somatic weight. Fecundities were remarkably similar during four of the five years evaluated. After adjust- ment for annual differences in fe- male size, fecundities were still sig- nificantly lower (by about one-fifth) during 1984, a major El Nino year, compared with the preceding (1979- 80) or following (1985-86) pairs of years. Gonadal condition also was uniquely low in 1984. The 1984 de- clines in fecundity and gonadal con- dition co-occurred with low somatic condition during 1984, particularly for larger females. Mean size (diam- eter, dry weight) of eggs was indis- tinguishable among years. There was a positive relation between egg size and female body size, and a general decline in egg size as the spawning season advanced for females of all sizes. Likely links between declines in fecundity, gonadal and somatic con- dition, and the crash in planktonic production during the 1982-84 El Nino are discussed. Annual Variations in Fecundity, Egg Size, and the Gonadal and Somatic Conditions of Queenfish Seriphus politus (Sciaenidae) Edward E. DeMartini Honolulu Laboratory, Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 2570 Dole Street, Honolulu, Hawaii 96822-2396 Manuscript accepted 24 August 1990. Fishery Bulletin, U.S. 89:9-18 (1991). Few data exist on annual variations in reproductive traits (fecundity, egg size, gonadal allocation) of marine fishes. At a minimum, however, such data are necessary if fisheries ecol- ogists are to begin to understand the many processes, including the vagaries of planktonic transport, that influ- ence the large annual and longer- term temporal fluctuations in the recruitment and subsequent year- class abundance of marine fish stocks (Sinclair 1988, Bailey and Almatar 1989). The influences of egg size and quality on the early growth and sur- vivorship of most species are poorly understood (Ware 1975). The queenfish Seriphus -politus is a small croaker abundant in the inner- shelf waters off southern California. It has planktonic egg and larval stages prior to the recruitment of juveniles to epibenthic habitat. In- dividual females are indeterminate serial spawners that produce as many as 20 batches of eggs during a pro- tracted (6-month) spawning season (DeMartini and Fountain 1981). Juve- nile and small adult, including male, queenfish feed on zooplankton (cope- pods and mysids), and large adults, females in particular, specialize on juveniles of the northern anchovy Engraulis mordax (DeMartini et al. 1985). In this paper, I present data on batch and relative (weight-specific) fecundities, egg size, and the gonadal and somatic conditions of adult fe- male queenfish collected during five spawning seasons spanning an 8-year period from 1979 to 1986. Fecundity, egg size, and condition indices are compared among years and related to concurrent variations in female body size. Because data are available prior to, during, and immediately following a major El Nino event, I interpret my observations in terms of known interannual variations in planktonic production and potential food limitation. Methods Sample collection Queenfish were collected during the March- August reproductive seasons (DeMartini and Fountain 1981) of 1979, 1980, 1984, 1985, and 1986. Nighttime (2000-0200 hours), bi- weekly to fortnightly collections with a lampara seine, made at 5-16 m bot- tom depths at three longshore loca- tions in San Onofre-Oceanside waters (DeMartini et al. 1985), were used to index the abundance and to describe the size (length, weight) composition of the nearshore queenfish stock. Sample fish for gonad analyses were provided by daytime lampara seining, otter trawling, and gillnetting at <16-m depths in the same area, and by screenwell samplings of the San Onofre Nuclear Generating Station, Fishery Bulletin 89(1), 1991 near San Clemente, California. Night- time samples were used to characterize abundance and size composition, because net avoidance is less at night (Allen and DeMartini 1983). Fish analyzed for repro- ductive variables and condition were collected during daylight hours, because oocytes destined for imminent spawning are macroscopically recognizable within ovaries only as they hydrate during the half-day preceding dusk spawning (DeMartini and Fountain 1981). Processing of samples Queenfish were refrigerated until pro- cessed within one day of collection. Sex and maturity were determined from macroscopic characteristics of gonads (DeMartini and Fountain 1981). Adult females were measured (stan- dard length, SL, in mm), and their gonadectomized, wet body weights (as a proxy for somatic weight) were determined to 0.1 g. Ovary and egg analyses Both ovaries were removed (fresh) from adult females, weighed damp to 0.01 g, and, for fish in ripe(ning) con- dition, the presence of hydrated-state oocytes noted based on macroscopic criteria ("Stage 3" ovaries: DeMartini and Fountain 1981). Hydrated-state ovaries were fixed and preserved in modified Gilson's Fluid (Bagenal and Braum 1971) for about three months, after which declines in oocyte diameters and dry weights should have stabilized (Wit- thames and Walker 1987). These specimens are here- after referred to as "Gilson's-fixed." Batch fecundity was then determined for a maximum of 10 females per month and year of collection. Fecundity was estimated by gravimetric method (Bagenal and Braum 1971, DeMartini and Fountain 1981). Counts from each sec- tion were standardized to the total weight of both ovaries and then averaged (Hunter et al. 1985). In a subsample of the Gilson's-fixed ovaries, I esti- mated the median diameter (random axis) of hydrated- state eggs: 10 randomly chosen oocytes per ovary pair were measured within ±25 ^m (±1 "eyepiece unit" or "EU") using a compound microscope with ocular micrometer at 40 x magnification. I examined a max- imum of 10 females per month and year. A linear dimension such as diameter might not ac- curately represent egg volume or mass because of vari- ations in chemical composition or density (Blaxter and Hempel 1963). Therefore, I compared the diameter and dry weight of oocytes from Gilson's-fixed ovaries. For Table 1 Size composition and catch-per-net-haul (CPUE) of adult female queenfish Seriphus politus collected during the March-August periods of 1979-86, off northern San Diego County, California. All fish were captured by lampara seines of consistent dimensions, mesh sizes, and method of deployment, during the night at 5-16 m bottom depths (DeMartini and Fountain 1981, DeMartini et al. 1985). Year SL (mm) Mean somatic wt(g) Mean CPUE Number X % >165 mm Fish Net hauls 1979 141 19 45 22 872 40 1980 134 16 40 22 995 45 1984 132 2 35 15 294 20 1985 134 6 38 20 624 32 1986 138 7 40 24 948 40 46 females with hydrated-state ovaries present in March- August 1984 collections, I determined the mean dry weight of hydrated oocytes for one member of each ovary pair. I determined the mean diameter of hy- drated oocytes for the other member of the ovary pair. Oocyte diameters were measured as described above. I determined mean oocyte dry weight by drying lots of 100 eggs to constant weight (1-2 days) in a vacuum jar over anhydrous calcium chloride. Eggs were dried at room temperature to avoid weight loss of volatiles (Hay 1984, Hislop and Bell 1987), and an aggregate weight determined to the nearest mg on an analytical balance. Calculation of condition indices The relative allocation of energy to gonadal tissue was indexed by the RGI of Erickson et al. (1985), as RGI = (G/W b ) x 100, where G = wet weight of ovaries in g, W = wet somatic weight in g, and b = the expo- nent of the power equation, G = aW b . The relative gonadal index (RGI) is equivalent to 100 x a, where a is defined by the linearized (log-transformed) equation, InG = lna + bin W (Erickson et al. 1985). It was neces- sary to adjust the gonadal index for somatic weight because the slope of the logarithmic gonad-to-somatic weight equation was significantly greater than one (i.e., the relation was allometric). I first attempted to index somatic robustness as relative somatic condition, K n = CW/SL' 1 (Le Cren 1951), where W = wet somatic weight in g, SL = stan- dard length in mm, C = a constant (10 5 ), and b = the exponent of the regression, W = aSL b . However, the exponent, evaluated as the slope of the log-transformed weight-to-length equation InW = lna + blnSL, differed among years, thereby invalidating the use of such an index in analysis of covariance (Erickson et al. 1985, DeMartini: Annual variations in reproductive traits of Senphus politus I I Cone 1989). I therefore limited my evalu- ation of robustness to comparisons of estimates of ordinary least-squares re- gression parameters (Cone 1989). The relationship between wet and dry body weights might change throughout the spawning season (Love 1970). There- fore, I evaluated seasonal changes in dry somatic weight using 57 females collected at the start of (April-May, n = 32) and im- mediately following (August-September, n = 25) the 1985 spawning season. After ovarectomy, fish were frozen in air-tight "zip-lock" bags. Fish were then thawed, and each entire fish was macerated and individually oven-dried to a constant weight at 120°C for 24-32 hours. Statistical analyses I used nonparametric analysis of variance (Kruskal-Wallis One-way ANOVA) to compare the body lengths and somatic weights of females among years. Year was evaluated as a fixed-effect class vari- able, because I was interested in evalu- ating potential differences among a pre- established series of years. Analysis of covariance (ANCOVA) was used when- ever possible to evaluate the effects of sampling year on batch fecundity, and on relative gonadal condition, after adjust- ment for year differences in somatic weight. A two- way Model I ANCOVA was used to evaluate sub- seasonal (approximately bimonthly) influences of egg diameter among years for females of differing body lengths. Dry egg weight was related to egg diameter by parametric regression. Dry somatic weight was regressed on wet weight for sample fish collected at the beginning and at the end of the 1985 spawning season; regressions were then compared using ANCOVA with body length as the covariate. Computa- tions were made using the GLM, REGRESS and TTEST software procedures of the Professional Data- base Analysis System (PRODAS; Conceptual Software Inc. 1986). Results Variations in female body size and CPUE The size composition of the nearshore, adult female queenfish stock differed among years. Mean female length and weight were significantly lower in 1985 and especially 1984 (Kruskal-Wallis one-way ANOVA; both Table 2 Summary statistics for adult female queenfish Seriphus politus used in analyses of reproductive variables and condition. Gonadal condition (RGI), batch fecun- dity, and egg diameter variables are least-square means (adjusted for annual differences in the body-size covariate). All variables, including fecundity, refer to sample females only. See Methods for explanations of the covariate used to adjust particular variables. One standard error is given in parentheses below each mean. Somatic Batch Egg SL weight fecundity 1. diameter' Year N (mm) (W,g) RGP (no. eggs) (EU) 1979 44 165 77 3.8 14,752 19.10 (5) (7) (.2) (1) (.18) 1980 126 151 56 4.7 15,019 20.61 (2) (3) (.1) (1) (.18) 1984 71 137 37 3.9 11,784 20.11 (2) (1) (■2) (1) (.23) 1985 77 141 44 5.3 14,550 21.03 (2) (2) (■2) (1) (.18) 1986 75 143 44 4.1 14,199 21.10 (2) (2) (.2) (1) (.18) Estimates ( ± SE) of the exponent "b" in the power equation, gonad weight (G) = aW b , were 1.185 ± 0.092 (in 1979), 1.162 ± 0.053 (1980), 1.232 ± 0.124 (1984), 1.280 ± 0.102 (1985), and 1.011 ± 0.102 (1986). Estimates did not differ among years, and were consistently greater than one (ANCOVA; In somatic W x Yr interaction: F 4383 = 0.87, P = 0.48; pooled slope = 1.1810. Batch fecundity estimates were back-calculated from the means and SE's of In-transformed data, multiplied by the correction factor of Sprugel (1983). Using the largest size-class of oocytes (ripe, hydrated-state) present in Gilson's- fixed ovaries (see Methods). P<0.001). Large females (> 165mm SL, chosen be- cause they were relatively rare during 1984-86) in fact were nearly absent in 1984, when overall mean female abundance was at an estimated low (Table 1). Females used in analyses of reproduction and condi- tion also differed in mean body size (both somatic weight and length) among sample years (Kruskal- Wallis ANOVA, both P< 0.001; Table 2). Dry vs. wet somatic weight The dry somatic weight of female queenfish aver- aged 24% of wet weight for fish collected at the begin- ning and at the end of the 1985 spawning season. Body length obviously influenced dry weight; sub- season, however, had no significant effect on dry weight (ANCOVA of effects of body length and sub- season on dry weight: length effect— F 154 = 346, P<0.0001; subseason effect— F 154 = 0.42, P = 0.52). Wet weight, therefore, could be used as an accurate proxy for dry weight throughout the queenfish spawn- ing season, once adjusted for variations in female body size. 12 Fishery Bulletin 89(1). 1991 Batch fecundity Batch fecundity in queenfish was posi- tively related to female body size in each year (Table 3; Fig. 1). Fecundity was generally better related (based on higher R 2 values) to somatic weight than body length. Batch fecundity was dispropor- tionately large in heavier females, as in- dicated by the value of the slope in the linear double-log plot (Fig. 1). Fecundity also differed among years, even after ad- justment for annual differences in female size, with mean fecundity in 1984 signif- icantly lower (by 20%) than mean fecun- dity in the other four years (Tables 2, 3, 5; Fig.l). Table 3 Summary of ANCOVAs" testing the effects of female somatic weight and year on batch fecundity (F, no. eggs) and relative gonadal index (RGI). Both model R 2 are significant at P<0.001 (In F R 2 = 0.694; RGI, R - = 0.114). Dependent variable Source df SS MS F P In F In W 1 118.1 118.1 693 < 0.0001 Yr 4 3.4 0.8 4.9 < 0.0001 Error 387 65.9 0.2 RGI In W 1 0.06 0.06 0.03 0.87 Yr 4 111 27.7 12.5 < 0.0001 Error 387 860 2.2 "In W x Yrir teraction terms deleted for a nalysis of In F (P = 0.61) and for analysis of the RGI (P = 0.84) tn oi UJ too d z "O 9.5 3 O CO 1979 (.) LNF-1.125LN W*5.26 R-BS4 N-44 P<001 1980 (o) LnF-1.190LnW*503 R-B44 N-126 P<.001 1984 (•) LnF-1212 LnW*4.68 R-.666 N-71 P165»iSO Early Middle Late Subseasonal Period Figure 2 Relationship between mean (±1 SE) egg size (diameter of Gilson's-fixed, hydrated oocytes) and bimonthly period within spawning season for female queenfish of three arbitrary body sizes (small < 13.5 cm, medium 13.6-16.5 cm, and large > 16.5 cm SL), by sample year. See Tables 4 and 5 for results of ANCOVA testing the effects of subseasonal period (within year) on egg size, with female body lengths evaluated as a covariate. Data were pooled by bimonthly period to increase sample sizes. Table 5 Summary of a posteriori Bonferroni t -tests for identity of year effects detected by ANCOVAs summarized in Tables 3 and 4. Means connected by underlines are not significantly dif- ferent at P = 0.05. See Table 2 for values of adjusted means and SEs. Variable Year contrasts In F 1984 < 1985 = 1986 < 1980 < 1979 Egg diameter 1979 = 1984 = 1980 = 1985 = 1986 RGI 1979 = 1984 = 1986 < 1980 = 1985 period x year interaction (Tables 4,5). The latter in- teraction illustrates that the pattern of subseasonal decline in egg size varied among years (Fig. 2). Ad- justed for female length, mean egg diameter appeared to vary 10% among years (Table 2). This difference in egg diameters, expressed in terms of volume (as 4/3 n r 3 , the volume of a sphere), was 35% of the smaller value. Condition indices Somatic condition varied with body size. Larger females usually were more robust (Table 6), but somatic condition also varied among years; the slopes and in- tercepts of length-weight relations were lower in 1984 and 1985 than in the other three years (Table 6). Larger females in particular were less robust in 1984 and 1985 (i.e., there was a highly significant In SL x year in- teraction; Table 6), and this invalidated a straightfor- ward interpretation of the effects of body size and year on a summary index of somatic condition. As indicated by values consistently greater than one for the exponent "b" in the equation, G = aW b , per- centage gonad-to-somatic weight allocation increased for larger females (Table 2; Fig. 3). However, relative 14 Fishery Bulletin 89|1), 1991 gonadal condition, as described by the RGI of Erickson et al. (1985) in which gonad weight has been standardized by female somatic weight, did not vary with female size (Table 3). The RGI did differ among years in concert with size-adjusted variations in fecundity; mean RGI in 1984 was about 14% lower than the RGI aver- aged over the other four years (Tables 2, 5). An identical pattern of annual varia- tion in gonadal condition is observed if" the classical gonadal index (GSI = [G/W] x 100) is used as the dependent variable in ANCOVA instead of the RGI. Potential biases of condition indices Table 6 Least-squares regression parameters for length-weight relationships of adult female queenfish in each a of the study years. Ordinary least-squares regres- sions were calculated for the double-log transformed equation, InW = lna + blnSL, where lna is the intercept and b is the slope. Intercept Slope Year Estimate SE Estimate SE R 2 N 1979 1980 1984 1985 1986 11.864 11.660 10.450 10.433 11.888 0.246 0.178 0.378 0.206 0.616 3.151 3.106 2.853 2.862 3.152 0.048 0.036 0.077 0.042 0.124 0.990 0.984 0.952 0.984 0.898 44 < 0.0001 126 <0.0001 71 <0.0001 77 <0.0001 75 <0.0001 "Estimated slopes of the InW - InSL relations differed among years (ANCOVA; InSL x Yr interaction: F 4393 = 6.10, P<0.0001). These differences in somatic and gonadal conditions were not the result of year variations in condition- selective collection methods. Lampara-seined fish com- prised 84-100% of the specimens examined for condi- tion in each year; among non-lampara fish, a maximum of 14% of the fish examined (in 1984) were collected by otter trawl. The somatic conditions (K = 10 5 W/SL 3 ; Le Cren 1951) of fish collected by lampara seine and otter trawl during April-June 1984, the only period when testable numbers of fish were collected using more than one method, were indistinguishable (seine: mean ± SE condition = 1.40 ±0.016; trawl: mean ±SE = 1.39 ±0.010; Student's £ = 0.28, df = 151, P = 0.78). Discussion Annual variations in body size Interannual variations in the body size of adult female queenfish were marked, with the percentage contribu- tion of large fish varying tenfold and mean somatic weights of the nearshore female stock varying by 25% over the study period. Clearly, potential year effects on fecundity and other size-sensitive variables are con- founded with the effects of annual variations in body size, necessitating the use of size as a covariate in analyses. Fecundity and female body size During each of the queenfish spawning seasons moni- tored, batch fecundity was positively related to female body size, especially somatic weight. The overall mean value of b in the equation, F = aW h was 1.2087, which is significantly greater than unity. I ■o ITS c o O 1979 lal Ln G 1.185 Ln W-3 H 893 N 44 P< 001 1980 IDI Ln G 1.162 Ln W- R .893 N 126 H< 001 1985 IOI Ln G 1.280 Ln R.823 N-77 P<.001 1986 (» w- 1 3 367 Ln G 1.011 Ln R-.758 N-75 P-.001 w -2 592 /3 3.5 4 4.5 5.0 Ln Somatic Weight (g) Figure 3 Relationship between the log of ovary weight (ln G) and log female somatic weight (In W) during each of the five study years. For illustration, mean ovary weight data ( + 1 SE) are plotted for each 10-g weight class. The equation. G = aW (in log-linear form as InG = lna + blnW), and its summary statistics are provided for each fitted regression line. DeMartmi: Annual variations in reproductive traits of Senphus politus 15 In another detailed study, Parrish et al. (1986) detected allometric weight-specific fecundities in a size- (age-) structured stock of the northern anchovy. Em- pirical data for additional species of "weedy" (fast- growing, high-fecundity) fishes suggest that allometric fecundity-weight relations may be a general phenom- enon (Blaxter and Hunter 1982, Clarke 1987). Reiss (1987), in a review and interpretation of relevant data, calculated that larger (older) fish in general have disproportionately large reproductive investments. Williams (1966), Wootton (1979), and Reiss (1987) have argued that disproportionate investments by older in- dividuals should be adaptive for many iteroparous fishes with indeterminate growth. The queenfish data further suggest that allometry in gonadal allocation may be more common than is generally appreciated. Adjustments for allometry are required when calcu- lating egg-based stock size estimates for species like queenfish and northern anchovy (Parrish et al. 1986). Covariates of egg size Egg size was positively related to queenfish body length, but declined for females of all sizes as water temperatures increased between the beginning (March- April) and end (July- August) of the spawning season. Egg size has been observed to increase with female body size, and decreases in egg size have been related to increases in water temperature during spring-summer production cycles, for diverse marine fishes (Williams 1967, Ware 1977, Blaxter and Hunter 1982 and references, Kashiwagi et al. 1985, Knutsen and Tilseth 1985, Daoulas and Economou 1986 and references, Imai and Tanaka 1987, Tanasichuk and Ware 1987). A positive female size-egg size relation and a seasonal decline in egg size with increasing water temperatures, the latter either ecophenotypic or genetic responses to the changing prey or predator (Clarke 1989) spectra confronting larvae, are now recognized as general phenomena in marine pelagic- spawners (Ware 1975; Markle and Frost 1985). It is obvious that estimates of mean egg size (and fecundity) must account for the effects of female body size and subseasonal variation. Annual variations in fecundity and egg size Batch fecundities of queenfish (adjusted for annual variations in female size) varied less than 6% during four out of the five years of this study. A marked change in batch fecundity, after adjustments for varia- tions in female size, occurred only in 1984. Weight- specific fecundities paralleled batch fecundities. Queenfish egg size varied little among the five years studied. Apparent egg volume averaged about 24% smaller in 1979 than during the other four years. Few data exist on annual variations in the egg pro- duction of marine fishes (Bagenal 1957 and references, Antony Raja 1971, Pinhorn 1984, Hunter et al. 1985, Bailey and Almatar 1989). As one might expect, the size-specific fecundity of individuals varies among years, but sometimes fecundities are remarkably similar within a short series of years (Antony Raja 1971, Pinhorn 1984, Hunter et al. 1985). Observational and experimental studies (e.g., Tyler and Dunn 1976, Wootton 1979, Hunter and Leong 1981, Hay and Brett 1988) demonstrate that fluctuations in fecundity can and do result from naturally occurring food limitation. Food rations can also affect egg size (Hislop et al. 1978, Le Clus 1979). The trivial inference is that food can sometimes, although not invariably, limit egg produc- tion. Of greater interest is that, for queenfish, the max- imum observed deviation from long-term average fecundity was only a 20% decline in a single year of unique oceanographic conditions, as described in the following section. El IMirio effects The anomalously low fecundities and somatic condition of queenfish in 1984 occurred at a time when the 1982-84 El Nino was still evident in the Southern California Bight (McGowan 1985). During 1983-84, zooplankton production was at unusually low levels in inner-shelf waters (Petersen et al. 1986), mirroring the nadir in phyto- and zooplankton production in the California Current, farther offshore (McGowan 1985). This decline in planktonic production off southern California lagged the more extreme declines in produc- tion that resulted from the parent El Nino that oc- curred off the western coast of South America during 1982-83 (Barber et al. 1985). During the 1982-84 California El Nino, tropical pelagic fishes migrated northward; many species became abundant off southern California, with some noted as far north as Washington-British Columbia (Smith 1985, Mysak 1986). El Nino effects on sub- tropical and cold-temperate fishes are poorly under- stood. Bailey and Incze (1985) and Mysak (1986) summarized the fragmentary data then available on distributional shifts and fluctuations in stock sizes of temperate fishes. Bailey and Incze (1985) speculated that El Nino effects on water temperature, nutrients, and planktonic production could effect egg and larval physiologies, disrupt the transport of early-life-history stages, and impact the somatic condition and egg pro- duction of adults. For vagile species, major impacts such as these should prompt movements to regions more favorable for growth and reproduction (Bailey and Incze 1985). 16 Fishery Bulletin 89(1), 1991 My data on interannual variations in body size com- position suggest that large female queenfish (those in- dividuals whose somatic condition and reproduction were particularly stressed by reductions in their an- chovy prey) responded to El Nino conditions in part by emigrating out of the study area. The observed 1984 nadir in females >165 mm, followed by the return of fish of this size in 1985-86, demonstrates that emigra- tion had to have occurred, because 165-mm long queen- fish are more than 3 years-old (E. DeMartini, unpubl. data). Large fish might have emigrated to deeper depths, tracked colder water masses upcoast, or done both; unfortunately, lack of data prevent discrimina- tion among these possibilities. The only published evidence thus far for El Nino effects on adult fish condition and egg production off the west coast of North America are for the central stock of the northern anchovy (Fiedler et al. 1986), yellowtail rockfish Sebastes flavidus (Lenarz and Echeverria 1986), and for Pacific herring Clupea harengus pallasi (Tanasichuk and Ware 1987). In an- chovy, individual egg production was lower in 1983-84 than in 1980-82 and 1985; this reflected lower spawn- ing frequencies more than declines in batch fecundity (Fiedler et al. 1986). Specific fecundity (the daily pro- duction of eggs per unit biomass) of anchovy was low in 1983 (although high in 1984, the second El Nino year) compared with other years between 1980 and 1985. Growth rates of juvenile-adult anchovy were low in 1983-84 (Fiedler et al. 1986). For yellowtail rockfish off the central California coast, the visceral fat and gonad volumes of adults were lower in 1983 than in 1980 (Lenarz and Echeverria 1986). Off British Co- lumbia, Pacific herring responded to the El Nino with increased batch fecundities coincident with reductions in mean egg size (Tanasichuk and Ware 1987). Adults of the anchovy and rockfish do not regularly occur in inner-shelf waters: most anchovy frequent the Califor- nia Current, tens to several hundred kilometers off- shore of central and southern California. Adult yellow- tail rockfish are an epibenthic predator of continental borderlands offshore of the coasts of British Columbia- California. It is tempting to speculate that the observed decrease in fecundity and somatic condition of queenfish during 1984 reflects lower production of the planktonic and anchovy prey of adults during a major El Nino year, compared with the 1979-80 and 1985-86 periods. If true, these data are among the first to show annual variations in egg production resulting from differences in size-specific batch fecundity, rather than changes in the duration of the spawning season or changes in the spawning frequency of females (Hunter et al. 1985). Interestingly, queenfish egg mass was not detectably lower in 1984, concurrent with the 20% decline in the number of eggs produced per batch, so the impact on egg production may have involved only the quantity, not the quality of eggs. Total number of spawnings per female season, whether due to changes in the duration of the spawning season or frequency of spawning by individual females, also might have varied for queen- fish during 1979-86, but data are lacking. Length of spawning season and the interval between batches are unlikely to cancel out the batch fecundity pattern, though, since they might be expected to vary in con- cert with fecundity, if they covary at all (Hunter et al. 1985). My observations also provide one of the first sugges- tions of food web impacts of the 1982-84 California El Nino on an inner-shelf fish species. Unfortunately, data on potential El Nino effects on the survivorship of early-life stages and year-class establishment are lack- ing for queenfish, as for offshore fishes. Future studies should concurrently measure survivorship and recruit- ment, together with population fecundity and egg pro- duction of the stock, in addition to individual fecundity and condition. Acknowledgments Many skilled technicians assisted in the collection of sample fish and in dissecting and processing ovaries. I especially thank S. Garner, K. George, G. Greenwald, and D. Opalenik. Special thanks go to R. Fountain, without whose diligent notekeeping this study would not have been possible. Also gratefully acknowledged are the constructive criticisms of T. Present and E. Schultz on drafts of the manuscript. Two anony- mous reviewers contributed substantially to the im- provements of draft manuscripts. This study was done in conjunction with complementary work funded by the Marine Review Committee, Inc., of the California Coastal Commission, and I thank the Committee for their support. Citations Allen. L.G., and E.E. DeMartini 1983 Temporal and spatial patterns of nearshore distribution and abundance of the pelagic fishes off San Onofre-Oceanside, California. Fish. Bull., U.S. 81:569-586. Antony Raja, B.T. 1971 Fecundity fluctuations in the oil sardine Sardinella longiceps Val. Indian J. Sci. 18:84-98. Bagenal, T.B. 1957 Annual variations in fish fecundity. J. Mar. Biol. Assoc. U.K. 36:377-382. Bagenal, T.B., and E. Braum 1971 Eggs and early life history. In Ricker, W.E. (ed.), Methods for assessment of fish production in fresh waters, p. 166-198. DeMartini: Annual variations in reproductive traits of Senphus politus IBP (Int. Biol. Prog.) Handb. 3, Blackwell Sci. Publ., Oxford. Bailey, K.M., and L.S. Incze 1985 El Nino and the early life history and recruitment of fishes in temperate marine waters. In Wooster, W.S., and D.L. Fluharty (eds.), El Nino North: Nino effects in the eastern subarctic Pacific Ocean, p. 143-165. Wash. Sea Grant Prog., Univ. Wash., Seattle. Bailey, R.S., and S.M. Almatar 1989 Variation in the fecundity and egg weight of herring (Clupea harengus L.). Part II. Implications for hypotheses on the stability of marine fish populations. J. Cons. Cons. Int. Explor. Mer 45:125-130. Barber, R.T., J.E. Kogelschatz, and F.P. Chavez 1985 Origin of productivity anomalies during the 1982-83 El Nino. Calif. Coop. Oceanic Fish. Invest. Rep. 26:65-71. Blaxter, J.H.S., and G. Hempel 1963 The influence of egg size on herring larvae (Clupea harengus L.). J. Cons. Cons. Int. Explor. Mer 28:211-240. Blaxter, J.H.S., and J.R. Hunter 1982 The biology of the clupeoid fishes. Adv. Mar. Biol. 20: 1-223. Clarke, T.A. 1987 Fecundity and spawning frequency of the Hawaiian an- chovy or nehu, Encrasicholina purpurea. Fish. Bull., U.S. 85:127-138. 1989 Seasonal differences in spawning, egg size, and early developmental time of the Hawaiian anchovy or nehu, En- crasicholina purpurea. Fish. Bull., U.S. 87:593-600. Conceptual Software, Inc. 1986 PRODAS database, statistics, graphics: Manual and user's guide. Conceptual Software, Inc., Houston, Texas. Cone, R.S. 1989 The need to reconsider the use of condition indices in fishery science. Trans. Am. Fish. Soc. 118:510-514. Daoulas, C, and A.N. Economou 1986 Seasonal variation of egg size in the sardine, Sardina pilchardus Walb., of the Saronikos Gulf: Causes and probable explanation. J. Fish Biol. 28:449-457. DeMartini, E.E., and R.K. Fountain 1981 Ovarian cycling frequency and batch fecundity in the queenfish, Seriphus politus: Attributes representative of serial spawning fishes. Fish. Bull., U.S. 79:547-560. DeMartini, E.E., L.G. Allen, R.K. Fountain, and D. Roberts 1985 Diel and depth variations in the sex-specific abundance, size composition, and food habits of queenfish, Seriphus politus (Sciaenidae). Fish. Bull., U.S. 83:171-185. Erickson, D.L., J.E. Hightower, and G.D. Grossman 1985 The relative gonadal index: An alternative index for quan- tification of reproductive condition. Comp. Biochem. Physiol. 81A:117-120. Fiedler, P.C., R.D. Methot, and R.P. Hewitt 1986 Effects of the California El Nino 1982-1984 on the north- ern anchovy. J. Mar. Res. 44:317-338. Hay, D.E. 1984 Weight loss and change of condition factor during fixa- tion of Pacific herring, Clupea harengus pallasi, eggs and larvae. J. Fish Biol. 25:421-433. Hay, D.E., and J.R. Brett 1988 Maturation and fecundity of Pacific herring (Clupea harengus pallasi): An experimental study with comparisons to natural populations. Can. J. Fish. Aquat. Sci. 5:399-406. Hislop, J.R.G., and M.A. Bell 1987 Observations on the size, dry weight and energy content of the eggs of some demersal fish species from British marine waters. J. Fish Biol. 31:1-20. Hislop, J.R.G., A. P. Robb, and J. A. Gauld 1978 Observations on effects of feeding level on growth and reproduction in haddock, Metanogrammus aeglcfiinis (L.) in captivity. J. Fish Biol. 13:85-98. Hunter, J.R., and R. Leong 1981 The spawning energetics of female northern anchovy, Engraulis mordax. Fish. Bull., U.S. 79:215-230. Hunter, J.R., N.C.H. Lo, and R.J.H. Leong 1985 Batch fecundity in multiple spawning fishes. In Lasker, R. (ed.). An egg production method for estimating spawning biomass of pelagic fish: Application to the northern anchovy, Engraulis mordax, p. 67-77. NOAA Tech. Rep. NMFS 36. Imai, C, and S. Tanaka 1987 Effect of sea water temperature on egg size of Japanese anchovy. Bull. Jpn. Soc. Sci. Fish. 53:2169-2178. Kashiwagi, M., F. Nakamura, T. Okada, and N. Yamada 1985 A periodic variation of egg size of Japanese whiting Sillago japonica during the spawning season. Aquiculture 33:134-138. Knutsen, G.M., and S. Tilseth 1985 Growth, development, and feeding success of Atlantic cod larvae Gadus morhua related to egg size. Trans. Am. Fish. Soc. 114:507-511. Le Clus, F. 1979 Dry mass of yolked oocytes of the South West African pilchard Sardinops ocellata in relation to maturity stages and spawning cycles, 1972-1974. S. Afr. Sea Fish. Branch Invest. Rep. 119, 29 p. Le Cren, E.D. 1951 The length-weight relationship and seasonal cycle in gonad weight and condition in the perch (Percafluviatilis). J. Anim. Ecol. 20:201-219. Lenarz, W.H., and T.W. Echeverria 1986 Comparison of visceral fat and gonadal fat volumes of yellowtail rockfish, Sebastesflavidus, during a normal year and a year of El Nino conditions. Fish. Bull., U.S. 84:743-745. Love, R.M. 1970 The chemical biology of fishes. Academic Press, London. Markle, D.F., and L.-A. Frost 1985 Comparative morphology, seasonality, and a key to planktonic fish eggs from the Nova Scotian Shelf. Can. J. Zool. 63:246-257. McGowan, J. A. 1985 El Nino 1983 in the Southern California Bight. In Wooster, W.S., and D.S. Fluharty (eds.), El Nino north: Nino effects in the eastern subarctic Pacific Ocean, p. 166-187. Wash. Sea Grant Prog., Univ. Wash., Seattle. Mysak, L.A. 1986 El Nino, interannual variability and fisheries in the north- east Pacific Ocean. Can. J. Fish. Aquat. Sci. 43:464-497. Parrish, R.H., D.L. Mallicoate, and R.A. Klingbeil 1986 Age dependent fecundity, number of spawnings per year, sex ratio, and maturation stages in northern anchovy, Engraulis mordax. Fish. Bull., U.S. 84:503-518. Petersen, J.H., A.E. Jahn, R.J. Lavenberg, G.E. McGowen, and R.S. Grove 1986 Physical-chemical characteristics and zooplankton bio- mass of the continental shelf off southern California. Calif. Coop. Oceanic Fish. Invest. Rep. 27:36-52. Pinhorn, A.T. 1984 Temporal and spatial variation in fecundity of Atlantic cod (Gadus morhua) in Newfoundland waters. J. Northwest Atl. Fish. Sci. 5:161-170. Reiss, M.J. 1987 The intraspecific relationship of parental investment to female body weight. Functional Ecol. 1:105-107. 18 Fishery Bulletin 89(1), 1991 Sinclair, M. 1988 Marine populations: An essay on population regulation and speciation. Wash. Sea Grant Prog., Univ. Wash. Press, Seattle, 252 p. Smith, P.E. 1985 A case history of an anti-El Nino to El Nino transition on plankton and nekton distribution and abundances. In Wooster, W.S., and D.L. Fluharty (eds.), El Nino North: Nino effects in the eastern subarctic Pacific Ocean, p. 121-142. Wash. Sea Grant Prog., Univ. Wash., Seattle. Sprugel, D.G. 1983 Correction for bias in log-transformed allometric equa- tions. Ecology 64:209-210. Tanasichuk, R.W., and D.M. Ware 1987 Influence of interannual variations in winter sea tem- perature on fecundity and egg size in Pacific herring (Clupea harengus pallasi). Can. J. Fish. Aquat. Sci. 44:1485-1495. Tyler, A.V., and R.S. Dunn 1976 Ration, growth, and measures of somatic and organ condition in relation to meal frequency in winter flounder, Pseudopleuronectes americanus, with hypotheses regarding homeostasis. J. Fish. Res. Board Can. 33:63-75. Ware, D.M. 1975 Relation between egg size, growth, and natural mortal- ity of larval fish. J. Fish. Res. Board Can. 32:2503-2512. 1977 Spawning time and egg size of Atlantic mackerel, Scomber scombrus, in relation to the plankton. J. Fish. Res. Board Can. 34:2308-2315. Williams, G.C. 1966 Adaptation and natural selection. Princeton Univ. Press, Princeton, 307 p. 1967 Identification and seasonal size changes of eggs of the labrid fishes, Tautogolabrus adspersus and Tautoga onitis of Long Island Sound. Copeia 1967:452-453. Witthames, P.R., and M.G. Walker 1987 An automated method for counting and sizing fish eggs. J. Fish Biol. 30:225-235. Wootton, R.J. 1979 Energy costs of egg production and environmental deter- minants of fecundity in teleost fishes. Symp. Zool. Soc. Lond. 44:133-159. Abstract.- Age and growth of the school shark Galeorhinus galeus was studied from rings in the verte- bra and length-frequency data. Sam- ples were collected by trawling off the southern Brazilian coast from June 1980 to September 1986. Histo- logical studies were also conducted on the characteristics of the verte- bra. Standard histological techniques and microradiography were used to determine the pattern of vertebral calcification. The vertebrae of G. ga- leus are composed of calcified carti- lage. Chondrocytes in calcified zones remain alive, probably nourished through vascular channels extending from the perichondrium into the car- tilage matrix. A narrow zone of un- calcified matrix at the outer edge of the centrum indicates that calcifica- tion is preceded by initial develop- ment of hyaline cartilage. The verte- bra presents a pattern of alternating heavily and less heavily mineralized zones, narrow and wide, respective- ly. The narrow zones were named rings, which are translucent under transmitted light and white to the microradiograph. These rings are probably laid down yearly in a slow- growing phase extending throughout the four winter months of June to September. Lengths at age were back-calculated and the von Ber- talanffy growth parameters are: males-K = 0.092, L„ = 152 cm, and to -2.69; females-K = 0.075, L„ = 163 cm, and tn = -3.00. ELEFAN software was used to determine the growth curve best fitted to length- frequency data, but results overesti- mated the growth rate due to the slow growth and modal overlap. Age, Growth, and Structure of Vertebra \n the School Shark Galeorhinus galeus (Linnaeus, 1 758) from Southern Brazil Beatrice Padovani Ferreira Fundacao Universidade do Rio Grande. Rio Grande, RS. 96200, Brazil Present address: Marine Biology Department James Cook University of North Queensland, Townsville Q481 I, Australia Carolus Maria Vooren Departamento de Oceanografia, Fundacao Universidade do Rio Grande Rio Grande, RS, 96200, Brazil Manuscript accepted 12 October. Fishery Bulletin, U.S. 89:19-31 (1991). The elasmobranch skeleton consists of calcified cartilage. In most elasmo- branchs, vertebrae are the only avail- able structure that display periodic rings, which are useful for determin- ing age. These rings result from dif- ferent ratios of organic matrix/min- eral, making the zones optically distinct (Casselman 1974). Since the description by Ridewood (1921), sev- eral techniques have been developed for enhancing the visibility of these rings thus making the counts easier and more accurate. However, infor- mation about the histology of elasmo- branch vertebral rings is limited, and statements about their chemical com- position and optical properties are contradictory (Casselman 1983). The school shark Galeorhinus ga- leus is one of the principal species in the shark fishery in southern Brazil (Vooren and Betito In press). With the purpose of providing information for management decisions, age and growth of the Brazilian school shark were determined from rings in the vertebrae and length-frequency data. A histological study was conducted to investigate the structure of the verte- brae and to determine variations in their composition. Materials and methods The study area was the continental shelf and upper slope off southern Brazil, between latitudes 34° and 30° S, at depths between 10 and 500 m. Data used for size-frequency anal- ysis were obtained during trawl fish- eries of this area from June 1980 to September 1986 (Table 1). All fishes were sexed and their total length (TL, cm) was measured from the tip of the snout to the extremity of the upper lobe of the tail, which was stretched back to be aligned with the body axis. Vertebrae were collected during the cruises listed in Table 1, and samples were chosen to include both sexes and the full size-range available. Sexual maturity and repro- ductive stage (Table 2) were deter- mined according to size criteria es- tablished for this species by Peres (1989). The data were processed on the Statistical Package for the Social Sciences, SPSS (Nie 1975). Length- frequency distributions for both sexes were plotted by cruise, by month, and also by a selected combination of the two largest samples. The ELEFAN (Electronic Length Frequency Anal- ysis) software (Pauly and David 1981) was used to determine growth param- eters (k and L m ) from length-fre- quency data. The goodness of fit was given by the index Rn. Vertebrae for age determination were dissected in the field from the 19 20 Fishery Bulletin 89(1), 1991 region under the first dorsal fin. The vertebrae were cleaned, frozen, and pre- served in 70% ethanol. Several tech- niques were tried to aid enhancement and interpretation of the vertebral rings. Vertebrae of 10 sharks were sectioned into two halves, through transverse or sagittal planes, and the exposed faces were polished on wet 600-grit sandpaper. They were decalcified in 1% formic acid for 1 hour, rinsed in running water for - 24 hours, dried and stained with graphite powder. The material was observed with a dissection microscope at 10 x magnifi- cation using reflected light. Vertebrae from 82 individuals (35 males, 45 females, and 2 embryos) were dehydrated in ethanol, cleared in styrene, embedded in polyester resin, sectioned with a jeweller's saw in transverse and sagittal planes, and polished on wet 400- grit sandpaper to obtain slices of 50-250 microns. Radiographs of all sections were taken with Soft X-ray equipment by Moureuil (France), at settings of 10-30 kV, 5-15 mA, and exposure times of 3-5 minutes. Kodak Industrex M radiography film was used. The radiographs were mounted on glass slides as were vertebra sections, either directly or after staining with Harris's haematoxylin or basic fuch- sin. Vertebrae from five sharks were prepared and sectioned by standard histo- logical techniques for calcified material. Sections were stained with Sudan 4 for observation of the lipid con- tents of cells. A dissecting microscope at 10 x magnification was used for measuring and counting growth rings, and a compound microscope at 100 x magnification for obser- vation of cell structure and details of the margin of the vertebra. The criteria to define a ring requires that it must occupy a distinct translucent zone relative to ad- jacent opaque zones and that the ring traverse the cor- pus calcareum and intermedialia (modified from Casey et al. 1985). Measurements of growth increments were made with an ocular micrometer positioned to measure distances from the focus (notochordal remnant) to successive growth bands. The radius of each centrum was mea- sured from the focus to the distal margin of the corpus calcareum (Fig. 1). The widths of individual translucent and opaque zones were measured in the microradio- graphs of 10 individuals. The periodicity of the forma- tion of the rings was studied by examining the margins of the vertebrae collected from June to September. The Table 1 Catch data of Galeorhiniis galeus during cruises conducted by the NOc A tlan- tico Sul. Asterisks indicate vertebrae collected. Cruise code/year (month) Sites Latitudes (min.-max.) Depth (m) Male Female -(n) 3/80 (June) 1 - 120 19 370 4/80 (July) 1 — — — 65 94 7/80 (Sep.) 7 30°44'-35°52'S 120 108 114 *7/81 (Sep.) 5 — — — 55 161 9/81 (Sep.) 8 30°36'-34°24'S 120 184 276 3/82 (July) 10 — — — 491 160 10/83 (Aug.) 7 30°42'-34°22'S 100 — 30 3/83 (Nov.) 14 30°44'-34°16'S 160 43 36 *4/84 (June) 15 — — — 301 230 2/85 (June) 7 — — — 72 104 *3/85 (July) 23 — — — 76 101 *4/86 (July) 12 32°00'-34°00'S 500 65 414 Table 2 Sexual maturity of Galeorhinus galeus determined by size class. Data from Peres (1989). Total length (cm) Males Females Juveniles Adolescents Adults 43-88 43-103 93-108 108-123 113-143 128-153 marginal zone was classified into three types: pre-ring, consisting of a wide, less calcified zone; ring, consisting of a more calcified zone; and post-ring, consisting of a narrow, less calcified zone. Measurements of mar- ginal increments were considered to be ineffective because the widths of zones varied greatly between size classes. A linear regression of TL as a function of centrum radius was fitted to the data for juveniles and adults (sexes combined and separate) and the equality of Ferreira and Vooren: Age, growth, and structure of vertebra in Galeorhinus galeus 21 h.a. B Figure 1 Drawings of Galeorhinus galeus vertebrae showing patterns of calcification in centra. (A) View from end of centrum toward focus, showing growth rings; (B) view at center of centrum, showing Maltese Cross of calcified radialia; (C) longitudinal view showing double cones and growth rings, c = centrum, h.a. = haemal arch, i = intermedialia, n.a. = neural arch. slopes of the regression lines tested (Sokal and Rohlf 1981). As they were found to be different, a power rela- tionship was fitted to TL x centrum radius for each sex. Each ring was measured and these values inserted into the power equation to back-calculate the length of each shark at formation of the respective ring. The von Bertalanffy growth model was fitted to the mean lengths-at-age, using the methods of Ford and Bever- ton as cited by Gulland (1977). Results Size composition School sharks were present in the study area from April to November. During April and May all captured individuals were adolescent or adult gravid females (Fig. 2). From June to September the largest captures were registered (Table 1) and individuals from both sexes and all sizes (43-148 cm) were present (Figs. 2, 3). Larger juveniles were still captured during Sep- tember, but individuals smaller than 70cm TL were caught only from June to August. A size variation within the sample size composition was also observed when analyzing the length frequen- cies plotted by cruise (Figs. 4, 5). Some size classes were better sampled during certain cruises. For in- stance, the sample was mainly composed of adults during cruises 3/80, 7/80, and 4/86, and of juveniles 40- APR 2W- j^-L^ .. 40- MAY 20- n^ N 7 20- JUN 10- , 1 *^ N 704 20- JUL 10- r ^^JT^_ T S^-^ "'" 20- r~l aug 10- r^ 1 P-r-L. 20- SEP 10- ^^^ J "H, 50- OCT 19- 50" n n N 4 -1 NOV 25- n r 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 43 53 03 '3 83 93 103 113 133 133 143 153 163 JL (cm) Figure 2 Length-frequency distribution of female Galeorhinus galeus by month. 22 Fishery Bulletin 89(1), 1991 JUN N391 JUL N697 83 93 103 113 123 133 143 153 163 TL (cm I Figure 3 Length-frequency distribution of male Galeorhinus galeus by month. during cruises 4/80, 9/81, and 3/82. Smaller length classes were more frequently caught during cruises 4/84 and 3/85, but did not form a marked mode in the length-frequency distribution. Techniques for enhancing vertebral rings Rings were visible in all the attempted methods, but because of the large number of rings and marginal crowding in the centra of older sharks, only detailed microscopic observations were successful in providing comprehensive readings and measurements. The graphite method was an easy and simple technique that provided good results in enhancing rings of vertebrae of younger sharks. However, the large number of rings near the margin of vertebrae of older sharks was dif- ficult to determine using this technique, and the number of rings was always underestimated compared with results by other techniques. Stained sections of vertebrae provided good results in enhancing rings both in calcified and decalcified material. In the latter, no measurements were taken because of shrinkage and distortions observed after the decalcification procedure. The matrix shrank in the space formerly occupied by the mineral, and this zone became narrower than in the calcified state. 380 JUN 480 JUL 780 SEP 781 SEP 981 SEP 382 JUL 1083 AUG 383 NOV 484 JUN 285 JUN 385 JUL 480 JUL 33 133 143 153 163 Figure 4 Length-frequency distribution of female Galeorhinus galeus by cruise. The most satisfactory technique for enhancing and counting rings was microradiography. With this method, the more calcified zones appeared white while the less calcified appeared dark (Figs. 6, 8). Direct observations of the sections showed that the former more mineralized zones were translucent and the latter less-mineralized zones were opaque when observed under transmitted light (Figs. 7, 8). Sections of various thickness were examined. The best contrast between zones was obtained from sections between 250 and Ferreira and Vooren: Age, growth, and structure of vertebra in Galeorhinus galeus 23 380 JUN 480 JUL 780 SEP 7(1 SEP 382 JUL 1083 AUG 383 NOV 484 JUN 285 JUN 385 JUL Figure 5 Length-frequency distribution of male Galeorhinus galeus by 750 ^ for readings of interior rings and from 50 to 100 jy. thick for readings on the margins. Pattern of vertebral calcification The vertebra of Galeorhinus galeus is composed of calcified cartilage. Its centrum consists of a corpus calcareum of two obtuse, hollow cones with their apices joined and opposed. The space around the two cones is organized into four oblique basalia and four calcified Figure 6 Microradiograph of transverse section of Gateorhinus galeus vertebra. Figure 7 Sagittal section of Galeorhinus galeus vertebra under reflected light with dark background. Translucent zones appear dark and the opaque zones appear white. intermedialia: dorsal, ventral and lateral (Fig. 1) (Good- rich 1958, Ridewood 1921). In transverse section through the center of the centrum (focus), these radial 24 Fishery Bulletin 89(1), 1991 calcifications within the basalia have the shape of a Maltese Cross (White 1937) (Figs. IB, 6). This "car- charinoid" pattern is characteristic of the families Triakidae, Sphyrnidae, and Carcharhinidae (Applegate 1967). Above the centrum is a neural arch. In the center of the centrum is a hole, marking the position of the Figure 8 Microradiograph of sagittal section of Galeorhinus galeus vertebra from Figure 7. The more calcified zones (rings and cone in general) appear white, and the less calcified zones (opaque zones and intermedialia in general) appear dark. primitive notochord, which we adopted as the focus of the vertebra. Within each cone, the focus is surrounded by a series of concentric rings which are read through techniques using the whole centra. The inside of each cone is lined by a perichondrium, which consists of a fibrous layer covering a germinative layer of chondroblast cells (Fig. 9). During growth phases, the chondroblasts differentiate into chon- drocytes to form the mature cartilage, a densely cellular tissue consisting of rounded cells embedded in their secreted organic matrix. The body of the vertebra forming the intermedialia is also invested by a peri- chondrium. In sagittal section the differences between the two regions can be observed: the cells of the cone are smaller and embedded in a more abundant matrix than those of the intermedialia (Fig. 10). Mineraliza- tion occurs throughout the matrix and both regions present an alternate pattern of more and less mineral- ized zones, corresponding to the concentric rings that can be seen inside the cone. The properties of the narrow and wide zones, which occur in an alternating sequence, were defined by com- paring microradiographs and direct observations with transmitted and reflected light of sections of the same vertebra. In this species the narrow zone, which we define as a ring, is optically translucent and appears white on the radiograph, being opaque to the X-ray beam, and therefore more calcified. The wide zone, defined here as a growth zone, is optically opaque and appears dark on the radiograph, being semi-trans- parent to the X-ray beam, and therefore less calcified (Figs. 7, 8). Figure 9 Sagittal section of Galeorhinus galeus vertebra showing part of the cone and intermedialia. Starting from external side: a = perichondrium; b = ring crossing cone; c = ring crossing inter- medialia. (Harris's haematoxylin, 100 x) Ferreira and Vooren: Age, growth, and structure of vertebra in Galeorhmus galeus 25 Figure 10 Sagittal section of Gnleorhinus galeus vertebra showing contrasting tissue of cone and intermedialia. (Harris's haematoxylin, lOOx) Figure 1 1 Section of vertebral tissue of Gale- orhinus galeus showing the thin chan- nels linking the cells. (Harris's haematoxylin, 400 x) Thin vascular channels connecting the cartilage cells were observed in haematoxylin-stained sections of resin-embedded vertebrae. These canaliculi form a net- work in the matrix between cells providing an oppor- tunity for fluids and nutrients to reach the interior from the external medium (Fig. 11). Absence of lipid inclu- sions in the chondrocytes is interpreted as evidence of a healthy and active cellular metabolism. The presence of isogenic groups of cells suggests that cells divide in- terstitially and thus effect interstitial growth. The widths of translucent and opaque zones varied in individuals. Rings were usually narrower than adja- cent opaque zones, but with increased body size (TL), both attained the same average size (Figs. 12, 13). Width of a male's opaque zones decreases gradually, and after about the 15th ring the two zones are equal 26 Fishery Bulletin 89(1), 1991 Q opaque translucent IujmW tfUnlftm, Figure 12 Widths (m.u. = 10~ 3 cm) of translucent and opaque zones distributed by ring for female Galeorhinus galeus (134 cm TL, 29 yrs-old). e so — n MALE o Q opaque * J H translucent 30— | 10— pi r '..LI. llJ.l.l.1. >iH^p|pinipi.yjbc^ ring Figure 13 Widths (m.u. = 10~ 3 cm) of translucent and opaque zones distributed by ring for male Galeorhinus galeus (141 cm, 26 yTS-old). in width. Females have the same general pattern, but some variation in width still can be detected after the 15th ring has been formed (Fig. 14). From June to September, 91% of the vertebrae ex- amined showed more calcified zones forming at the margins (Fig. 15). During June this percentage was 80%, rising to 100% during July and falling to 75% in September. The vertebrae observed with less calcified a f w.o.z. (m.u.) Figure 14 Frequency distribution of widths of opaque zones after the 15th ring for female and male Galeorhinus galeus. 120 a>100 c o ~ 80 is c o> 60 40 g. 2 I pre-ring O ring ^ post-ring m& JUN JUL AUG months SEP Figure 15 Percentage of type of marginal zone in vertebrae of Galeo- rhinus galeus observed June-September (n = 82). zones at margins during June (20%) were of the pre- ring type (wide less-calcified zone), and the ones ob- served during September (25%) were of the post-ring type (narrow less-calcified zone). These results indi- cated that the ring formation probably occurred during the winter months (June to September). One mark per year seems to be the most likely case for the school shark, and results here reflect this assumption. In vertebrae of embryos at 8 months of intrauterine age, only the cartilage of cones was mineralized and no rings were visible (Fig. 16). Their intermedialia was Ferreira and Vooren Age. growth, and structure of vertebra in Galeorhmus galeus 27 formed by a hyalin, unmineralized cartilage (Fig. 17). A 45cm TL juvenile shark's vertebra showed some mineralization of the intermedialia, but unmineralized areas were still present. Two rings were observed (Fig. 18). The first ring was probably formed during the first winter after the birth in November (Peres 1989). The most recent one was just forming on the margin. In adult individuals the largest number of rings was 41, observed in a female of 155cm TL. Histological observations revealed the presence of a thin uncalcified layer at the very edge of the vertebrae, peripheral to the outermost narrow or wide calcified Figure 16 Microradiograph of a sagitally sectioned vertebral column from a Galeorhinus galeus embryo, showing calcification (white zones) of double cone in centrum (50 x). zones at the margin (Fig. 18). This indicates that the marginal growth of the vertebra starts with the for- mation of an unmineralized layer of cartilage which is subsequently mineralized. This layer is probably pres- ent throughout the year and does not by itself indicate periodicity of ring formation. Back-calculation All linear regressions of total length on vertebral radius were significant (p<0.05). The regression slopes were significantly different between juveniles and adults (j0<0.05) and between adult males and females (p< 0.05). Therefore, a power relationship was found to be more adequate to fit the pooled data for juveniles and adults of each sex. The following equations were ob- tained: Females (AT = 26), TL = 32.59 x R om and males (AT = 33), TL = 25.07 x R - 897 , where N indi- cates sample size and R radius of vertebra, with TL in cm and R in micrometric units (1 m.u. = 10~ 3 cm) (Fig. 19). Lengths were back-calculated by age class and did not reveal the occurrence of Rosa Lee phenomenon (Gulland 1977), so the lengths at age were averaged (Table 3). The von Bertalanffy growth parameters were: females, k = 0.075, L M = 163cm, and t = -3.00; males, k = 0.092, L 00 = 152cm, and t =-2.69. The growth curves calculated from these parameters are shown in Fig. 20. Using the ELEFAN software, several attempts were made to find the growth curve best fitted to the length- frequency data. In order to obtain a data set contain- Figure 17 Sagittal section of vertebra from a Galeorhinus galeus embryo, showing (a) perichondrium, (b) hyaline cartilage in intermedialia, and (c) calcified car- tilage of cones. (Haematoxylin. 100 x) 28 Fishery Bulletin 89(1), 1991 Figure 18 Portion of a transverse section of a vertebral centrum from a 2d-year Galeorhinus galeus juvenile, showing (a) focus, (b) hyaline cartilage, and (c) calcified cartilage of intermedialia. (1) 1st ring, (2) 2d ring, and (3) un- calcified margin. (Haematoxylin, 50 x ) E 15C — SO — females ct ; 3J 59 . r 0827 N 26 / 100— / 50— males ct= 350 7. r 0897 N 33 10 70 » 40 50 60 70 80 90 100110 10 20 30 40 50 60 TO 80 90 100 110 RADIUS (m u.) ] i Figure 19 Relationship between total length and vertebra radius (R) for emale and male Galeorhinus galeus (m.u. = 10" 3 cm). ing the widest possible range of TL at a given time of the year, the data for June and July 1985 were pooled. Von Bertalanffy growth parameters were obtained as follows: females, k = 0.2 and L^ = 157cm (Rn = 612); and males, k = 0.2 and L. = 157 cm (Rn = 429). Table 3 Back-calculated lengths Galeorhinus galeus. (TL, mm) at age for male ; ind female Age (years) Ring Male female (») TL (n) TL 1 18 343 18 323 1 2 17 464 18 424 2 3 16 545 17 496 3 4 16 616 15 579 4 5 12 693 13 668 5 6 12 762 11 747 6 7 8 818 10 818 7 8 5 893 9 890 8 9 4 947 9 946 9 10 4 998 9 992 10 11 4 1027 8 1037 11 12 4 1070 6 1048 12 13 3 1137 4 1091 13 14 4 1123 14 15 4 1152 15 16 2 1196 16 17 2 1215 17 18 1259 18 19 1309 19 20 1334 20 21 1347 21 22 1371 Discussion The area of greatest uncertainty for interpreting and measuring growth zones in shark centra is near the margin (Casey et al. 1985). Distinguishing the presence of the last ring is a common problem (Casey et al. 1985, Stevens 1975, Walker 1986). Using microradiography, the margin can be easily observed in school shark Ferreira and Vooren: Age, growth, and structure of vertebra in Galeorhmus galeus 29 E t50- too- 50- S" ? <5 5 10 15 30 25 30 35 40 45 AGE (years) Figure 20 Von Bertalanffy growth curves for male and female Galeo- rhinus galeus and observed lengths-at-age. vertebrae and the ring formation identified. This tech- nique is not subject to problems derived from the presence of marginal connective tissue (Stevens 1975), and a large number of thin rings near the margin can be counted. From observation of margins of vertebrae, we can conclude that ring formation probably occurred dur- ing the period June-September. The high percentages of ring formation observed in the sample indicated that the process probably extends over a larger period than observed in the current study, covering perhaps half of the year. The growth band would be formed during the remaining period, and the rings would thus be an- nual marks. However, as vertebrae were examined only for four-month periods (June- September), the need for a proper validation remains, as emphasized by Beamish and McFarlane (1983). The hypothesis that two rings could be formed during a year was disre- garded, due mainly to the results of Grant et al. (1979). These authors estimated a longevity of 40 years for Galeorhinus australis ( = Galeorhiyius galeus, Com- pagno 1984) in South Australia, from mark-recapture data. Several individuals were recaptured after periods of more than 25 years at large, and 10% of the recap- tures occurred 15 years after release. This evidence supported the hypothesis of one ring per year, which also led to the conclusion that individuals can attain up to 40 years of age. Casey et al. (1985), observing a large number of rings near the margin, suggested that if these could be in- terpreted as annual marks, the ages determined for several species of sharks may have been underesti- mated. The present results agree with this hypothesis, and provide evidence that the school shark is a long- lived, slow-growing species. The conclusion that the rings are translucent and strongly calcified, while the growth bands are optical- ly opaque and less calcified is in agreement with the fact that the rings are stained dark by the silver nitrate technique (Stevens 1975, Cailliet et al. 1983). Cassel- man (1974, 1983) concluded that the translucent zone in calcified tissue of fish is more heavily mineralized than adjacent opaque zones and that calcium content is directly related to translucency. In reference to the mode of calcification of cartilage, Moss (1977) wrote that "something radically different occurs in shark cartilage, making it certain that there can be no unitary description of vertebrate cartilagi- nous calcification." Hoenig and Walsh (1982) described the occurrence of vascularized cartilage canals in calcified vertebrae of several species of sharks. Many canals were found to contain blood cells, and they sug- gested a nutritive role for these canals. However, dif- fusion through the matrix is impossible after mineral- ization and in mammals; for instance, calcification of endochondral bone causes cellular degeneration and death of chondrocytes, because the isolated cells can- not maintain a normal metabolism (Robbins 1975). In the school shark, however, chondrocytes remain alive after calcification, as is evident from their normal, lipid- free appearance in calcified zones. Interchange be- tween cells and vascular elements is probably sustained after mineralization by the canaliculi that were ob- served permeating the intercellular matrix. The growth of crystals within a preformed organic structure is the basic mode of skeletal formation (Weiner 1984). Narrow uncalcified matrix areas ob- served at the edge of school shark vertebrae show that development of hyaline cartilage precedes the process of calcification. In addition, evidence of interstitial growth indicates that the tissue still remains uncalcified for a while after the appositional growth. In the verte- brae of embryos only the cone area was calcified. Among juveniles some uncalcified zones occurred also in the intermedialia. During the adult phase, the cone and intermedialia display the same mineralization pat- tern, with corresponding opaque and translucent alter- nate zones. This is evidence that the calcification, even when occurring at different times, follows the same pre-established rules. Weiner (1984) suggests that the organic matrix performs active, specific roles in this process. Growth of hydroxy apatite crystallites occurs in the space between collagen fibrils and perhaps within them (Glimcher in Kemp 1984). Therefore, regions of organic matrix formed during slow-growth phases would offer more space for crystal growth when ex- posed to the calcium and phosphate ions which form the mineral crystallites. Such regions, e.g., the rings, 30 Fishery Bulletin 89(1), 1991 would be characterized by relatively heavier calcifica- tion compared with the more rapidly growing, less calcified zones. In addition, the formation of the ring takes at least 4 months, but during the first 10 years of life the width of a ring is much less, in comparison with adjacent growth bands, than it would be if the growth rate were constant throughout the year. It is concluded that in G. galeus the ring represents a period of slow growth, and that mineralization is more intense during this period. Vooren and Betito (In press) have shown that indi- viduals of Galeorhinus galeus migrate northwards into the study area starting in April, until their peak abun- dance in September. At this time the school shark is abundant on the shelf south of lat. 32 °S and scarce or absent further north. After September, the fish mi- grate southward from the study area, are scarce there in November, and absent in January and February. At the onset of winter, a marked change in temperature preference occurs, with most fishes occurring at 18-20°C from April to June and at 11-15°C in August and September, although water masses of higher temperatures are available during the latter months. The present data show that gravid females arrive first, making up all the April and May catches. From June onwards, other groups (adult males, non-gravid adult females and juveniles of both sexes) migrate into the area. The whole population experiences the decrease in temperature from June to August (Vooren and Betito In press). Thus, the period of slow growth and ring formation in vertebral centra coincides with this change in temperature preference of the population at the onset of winter. Similar results have been reported for other species (Stevens 1975, Jones and Geen 1977), confirming the general view that ring formation is associated with low temperature and slow vertebral growth (Longhurst and Pauly 1987). As for the process involved, Casselman (1974) has suggested that during slow-growth phases, the amount of protein available for appositional growth might be reduced although minerals would still be available. In the shark vertebra, slow growth is evidently associated with a reduced rate of deposition of matrix compo- nents, which include both collagen fibrils and poly- saccharides. Digestion and depletion of the carbo- hydrate moiety of the matrix during the slow-growth phase would facilitate interaction between collagen fibrils and mineral ions, thus promoting calcification. Mugiya (1987) has shown that in fish otoliths both calcium uptake and protein synthesis vary in an en- dogenous process controlled by hormones. In the school shark vertebrae, a matrix less dense in protein formed during slow-growth phases could be the cue to a higher calcium uptake to fill the space available for mineraliza- tion. Jones and Geen (1974) related wider rings with warmer years in spines oiSqualus acanthias, and this relationship may explain the variation in the width of the rings which is observed in vertebrae of adult female school sharks (Fig. 15). Gravid females which migrate into the study area during March to May show prefer- ence for higher temperatures, and it is possible that they remain in the warmer part of the species' tem- perature range during the winter. If so, their rings might grow faster than in fishes that remain at lower temperatures. The distribution of the different com- ponents of the population in the study area during the winter should be investigated in detail to test this hypothesis. Reading vertebrae is the most important tool for age determination in many elasmobranchs. Length-fre- quency analysis is most suitable for fast-growing spe- cies because of the assumption that all fishes in the sample have the same age at the same length (Long- hurst and Pauly 1987). In slow-growing, long-lived species, however, a given size class contains several different age groups (Gruber and Stout 1983). The Von Bertalanffy growth parameters estimated by ELEFAN software (Pauly and David 1981), using length-frequency data, overestimated the growth rate determined by vertebral readings. These values were approximately the same as those found by Olsen (1954) when analyzing length-frequency data for the Aus- tralian school shark. Later, Grant et al. (1979), using tag-recapture methods, estimated lower values of growth rates and concluded that length-frequency analysis was impracticable in this species. Acknowledgments Special thanks are due to Dr. Daoiz Mendoza for all his help and assistance with the histological aspects of this work. We thank Monica B. Peres for the valuable information and Mauro Maida, Dr. Garry R. Russ, Marcus V.S. Ferreira, and Annadel Cabanban for their constructive criticisms of the manuscript. We also thank the people from Projeto Talude and RV Atlan- tico Sul who helped with the field work. The research was sponsored by the Brazilian Ministry of Science and Technology (CNPq). Citations Applegate, S.P. 1967 A survey of shark hard parts. In Gilbert, P.W., R.F. Mathewson, and D.P. Rail (eds.). Sharks, skates and rays, p. 37-67. John Hopkins Press, Baltimore. Beamish, R.J., and G.A. McFarlane 1983 The forgotten requirement for age validation in fisheries biology. Trans. Am. Fish. Soc. 112:735-743. Ferreira and Vooren: Age, growth, and structure of vertebra in Galeorhinus galeus 3! Cailliet. G.M., L.K. Martin, I). Kusher, P. Wolf, and B.A. Welden 1983 Techniques for enhancing vertebra] bands in age estima- tion of California elasmobranchs. In Proceedings of the in- ternational workshop on age determination of oceanic pelagic fishes: Tunas, billfishes and sharks, p. 157-165. NOAA Tech. Rep. NMFS 8. Casey, J.G., H.L. Pratt Jr., and C.E. Stillwell 1985 Age and growth of the sandbar shark (Careharhinus plumbeus) from the Western North Atlantic. Can. J. Fish. Aquat. Sci. 42:963-975. Casselman, J.M. 1974 Analysis of hard tissue of pike Esox lucius L. with special reference to age and growth. In Bagenal, T.B. (ed.), Ageing of fish, p. 13-27. Unwin Bros., Surrey, England. 1983 Age and growth assessment of fishes from their calcified structures — Techniques and tools. In Proceedings of inter- national workshop on age determination of oceanic pelagic fishes: Tunas, billfishes and sharks, p. 1-17. NOAA Tech. Rep. NMSF 8. Compagno, L.J.V. 1984 Sharks of the world. FAO species catalogue. FAO Fish. Synop. 4 (125), pt. 2:386-389. Goodrich, E.S. 1958 Studies on the structure and development of vertebrates. Vol I. Dover Publ. Inc., NY, 485 p. Grant, C.J., R.L. Sandland, and A.M. Olsen 1979 Estimation of growth, mortality and yield per recruit of the Australian school shark, Galeorhinus australis, from tag recoveries. Aust. J. Mar. Freshwater Res. 30:625-637. Gruber, S.H., and R.G. Stout 1983 Biological materials for the study of age and growth in a tropical marine elasmobranch the lemon shark, Negaprion brevirostris (Poey). In Proceedings of the international work- shop on age determination of oceanic pelagic fishes: Tunas, billfishes and sharks, p. 193-205. NOAA Tech. Rep. NMFS 8. Gulland, J.A. (editor) 1977 Fish population dynamics. John Wiley, NY, 372 p. Hoenig, J.M., and A.H. Walsh 1982 The occurrence of cartilage canals in shark vertebrae. Can. J. Zool. 62:483-485. Jones, B.C.. and G.H. Geen 1977 Age determination of an elasmobranch (Squalus acan- tkias) by X-ray spectometry. J. Fish. Res. Board Can. 34:44-48. Kemp. N.E. 1984. Organic matrices and mineral crystallites in vertebrate scales, teeth and skeletons. Am. Zool. 24:965-976. Longhurst, A.R., and D. Pauly 1987 Ecology of tropical oceans. Acad. Press, San Diego, 407 p. Moss, M.L. 1977 Skeletal tissue in sharks. Am. Zool. 17:335-342. 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., U.S. 85:395-401. Nie, N.H., C.H. Hull, J.C. Jenkins, K. Steinbrenner, and D.H. Bent 1975 Statistical package for the social sciences, 2d ed. MacGraw-Hill, Boston, 675 p. Olsen, A.M. 1954 The biology, migration and growth rate of the school shark, Galeorhinus australis (MacLeay) in south-eastern Australian waters. Aust. J. Mar. Freshwater Res. 5:353-410. Pauly, D., and N. David 1980 An objective method for determining fish growth from length-frequency data. ICLARM (Int. Cent. Living Aquat. Resourc. Manage.) Contrib., Newsl. 3(3):13-15. 1981 ELEFAN 1, a BASIC program for the objective extrac- tion of growth parameters from length frequency data. Meeresforschung/Rep. Mar. Res. 28(4):205-211. Peres, M.B. 1989 Desenvolvimento, ciclo reprodutivo e fecundidade do cacao bico-de-cristal Galeorhinus galeus L. no Rio Grande do Sul. MSc. thesis, Fundacao Universidade do Rio Grande, RS, Brasil, 76 p. [in Portuguese, Engl, abstr.]. Ridewood, W.G. 1921 On the calcification of the vertebral centra in sharks and rays. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 210:311-407. Robbins, S.L. 1975 Patologiaestructuralefuncional. Interamericana, Mex- ico, 1516 p. Sokal, R.R., and F.J. Rohlf 1981 Biometry, 2d ed. W.H. Freeman, NY, 859 p. Stevens, J.D. 1975 Vertebral rings as a means of age determination in the blue shark Prionace glauca. J. Mar. Biol. Assoc. U.K. 55: 657-665. Vooren, CM., and R. Betito In press. Distribution and abundance of demersal elasmo- branch fishes on the continental shelf of Rio Grande do Sul. Brazil. Bull. Mar. Sci. Walker, T.I. 1986 Southern shark assessment project. 2d Review, October 1986. Mar. Sci. Lab. Prog. Rev. 66, Queenscliff, Victoria, Australia, 48 p. Weiner, S. 1984 Organization of organic matrix components in mineralized tissues. Am. Zool. 24:945-951. White, E.G. 1937 Interrelationships of the elasmobranchs with a key to the order Galea. Bull. Am. Mus. Nat. Hist. 74:25-138. Abstract. - Eighteen species of chaetognaths were identified from shelf waters in the Middle Atlantic Bight. Species composition in the water column and the hyponeuston was nearly identical, but the percent frequencies of the more common cold-temperate species were general- ly lower in surface collections. Mean surface salinity, weighted for abun- dance of individual chaetognath spe- cies in the hyponeuston collections, varied from 32.6 and 32.8 "/« for the coastal- and estuarine-inhabiting Sagitta tenuis and Parasagitta ele- gans, to 34.8 and 34.9 for the offshore Pterosagitta draco and Krohnitta pacifica. Weighted mean tempera- tures ranged from below 14°C for Mesosagitta minima, P. elegans, and Serratosagitta tasmanica to over 24 °C forK pacifica. Overall associa- tion among Middle Atlantic Bight chaetognaths, measured for the 15 most frequent species in 716 collec- tions by variance ratio, was signifi- cantly positive. Association between pairs of species was therefore also largely positive, with the important exception of Parasagitta elegans. This species, with a unique regional niche in low salinities and tempera- tures, was negatively associated (p<0.01) with five warm-water spe- cies (Krohnitta pacifica, Ferosagitta hispida, Sagitta tenuis, Sagitta hele- nae, and Flaccisagitta enflata). Most species reached maximum abundance at the surface near midnight. Excep- tions included Sagitta helenae, with daylight maxima, and Krohnitta pa- cifica, Ferosagitta hispida and Ser- ratosagitta serratodentata, showing crepuscular increases in abundance. Chaetognatha from the Central and Southern Middle Atlantic Bight: Species Composition, Temperature -Salinity Relationships, and Interspecific Associations* George C. Grant Virginia Institute of Marine Science and School of Marine Science The College of William and Mary, Gloucester Point, Virginia 23062 Recognition of several chaetognath species along the northeastern coast of the United States is quite recent. Until 1960, only eight species had been identified from shelf waters off the Middle Atlantic states, i.e., the Middle Atlantic Bight from Cape Cod to Cape Hatteras; these were in- cluded in Bigelow and Sears (1939): Eukrohnia hamata, Parasagitta ele- gans, Flaccisagitta enflata, Serrato- sagitta serratodentata (including the then undescribed Serratosagitta tas- manica), Flaccisagitta hexaptera, Pseudosagitta maxima, Krohnitta subtilis, and Pterosagitta draco**. Deevey (I960), in a study of the Dela- ware Bay region, added Ferosagitta hispida, Sagitta helenae, and Meso- sagitta minima to the list of recog- nized species. Since her material had been collected three decades earlier (1929-31), it appears that pre-1960 studies had simply failed to distin- guish between grossly similar spe- cies. Sagitta tenuis, Sagitta bipunc- tata, and Krohnitta pacifica were added by Grant (1963a, b) to the list of shelf species, and Grant (1967) Manuscript accepted 12 October 1990. Fishery Bulletin, U.S. 89:33-40 (1991). 'Contribution No. 1624 from the Virginia Institute of Marine Science and School of Marine Science, The College of William and Mary. * * Taxonomy in this paper generally follows the revisions of Tokioka (1965) and Kassat- kina (1971), but removes Pseudosagitta lyra and P. maxima, from the genus Flacci- sagitta in agreement with species groupings of Alvarino (1963). later confirmed that the endemic shelf population of "S. serratoden- tata'' in this region was actually S. tasmanica. Grice and Hart (1962) found other species in slope waters southeast of Long Island, New York, including Pseudosagitta lyra, Meso- sagitta decipiens, and Solidosagitta planctonis. Thus, at the close of the 1960s, some 15 species were known from the shelf and the presence of others in surface slope waters sug- gested their likely occurrence over the shelf as well. This study of Middle Atlantic Bight chaetognaths is based on an intensive series of collections from the central and southern bight. Presented here are the species composition in hypo- neuston and subsurface collections, temperature-salinity-plankton (T-S-P) diagrams for the more common surface species, measurements of interspe- cific associations among chaetognaths, and a summary of diel abundance in the hyponeuston. Methods and materials Chaetognath collections A transect of six stations (Cl-Jl, Fig. 1) off southern New Jersey was sam- pled quarterly for two years, October 1975- August 1977. Two more north- erly stations (A2, B5) and a southern transect of four stations (L1-L6) were added in the second survey year 33 34 Fishery Bulletin 89(1), 1991 (Grant 1977a, 1979, 1988). Routine collections at each station included paired 60-cm bongo net samplers (202 and 505 ^m mesh nets), towed from just below the sur- face to near, but safely off, the bottom, then back to the surface (so-called double oblique tows; 220 samples), and eight surface layer (upper 10 cm) collections ob- tained at 3-hour intervals over a 24-hour period (496 samples), using a 1-m wide hyponeuston net (505 ^m mesh). Of the 716 collections (Table 1), onjy 1 bongo and 79 hyponeuston (mostly daytime) collections lacked chaetognaths. Laboratory processing Collections were divided into successively smaller ali- quots for the more numerous taxa, using a sample- splitting device of proven design (Burrell et al. 1974). However, chaetognaths were generally obtained from whole or half samples, unless very abundant. Data analysis Collection data were sorted by species, stations, and collection methods. Analysis of the relationship of species abundance to hydrography was limited to hyponeuston collections because subsurface tows were oblique, often traversing multiple layers of different water types. Mean temperatures and salinities of cap- ture in surface-layer collections were calculated for each common chaetognath species, weighting each observed temperature and salinity by the size of catch (log N + 1), where N = total catch in a standard 20-min- ute hyponeuston net tow at 2.5 knots. Thus, tiflog n t + 1) + t 2 (log n 2 + 1) + . . . + tn(log n n + 1) (log n x + 1) + (log n 2 + 1) + . . . + (log n n + 1) Figure 1 Location of Middle Atlantic Bight plankton stations ( • ) sam- pled quarterly, October 1975-August 1977. Those in transect off New Jersey (Cl-Jl, o) were sampled for 2 years. Other stations were added for the second year. Depth contours in meters. s = s^log n t + 1) + s 2 (log n 2 + 1) + . . . + s n (log n n + 1) (log n, + 1) + (log n 2 + 1) + . . . + (log n n + 1) where tj, s i; and n ; are the surface temperature, sur- face salinity, and total catch in each positive collection, respectively. Presence, absence, and joint occurrences of the 15 most frequent species from both bongo and hyponeus- ton collections were used in an analysis of association between and among species. As recommended by Lud- wig and Reynolds (1988), the significance of associa- tion among all 15 species and 716 collections was first tested simultaneously using a variance ratio (VR) derived from a null association model (Schluter 1984). The expected value of VR under the null hypothesis of independence is 1. When VR>1, a positive associa- tion of species is indicated; VR< 1 indicates a negative association. A test statistic W = (N)(VR) provides a test of significance for deviations from 1. If species are not associated there is a 90% probability that W lies be- tween the chi-square limits: X 2 0.05.N < W < X": ii.'.ir,,N - Because of the large degrees of freedom in this study (where N = 716), critical values of x 2 were approx- imated (see Zar 1984, p. 482). Grant: Chaetognatha from central and southern Middle Atlantic Bight 35 Relative strength of the asso- ciation between pairs of species was measured using 2x2 con- tingency tables and Hurlbert's (1969) coefficient of interspecific association (C 8 ), as corrected by Ratliff (1982) for errors resulting from lack of absolute association (Pielou 1977). Yates' correction of chi-squared calculations was applied for low expected frequen- cies (Bailey 1981). Results Species composition Eighteen species (11 genera) of chaetognaths were identified, 17 in both surface hyponeuston and subsurface bongo net collections (Tables 2 and 3). Compositional differences in the two lists were limited to the rarest species: Sagitta bipunctata was found only in hyponeuston collections, while Pseudosagitta maxima was restricted to subsurface collec- tions. After adjustment for the 80 collections devoid of chaeto- gnaths (79 surface and 1 subsur- face collections), the percent fre- quencies of the most common cold-temperate species {Serrato- sagitta tasmanica, Parasagitta elegans, and Mesosagitta minima) were found to be much lower in surface collections. Warm-water chaetognaths (Flaccisagitta enflata, Sagitta helenae, Ptero- sagitta draco, S. tenuis, Serrato- sagitta serratodentata, Fero- sagitta hispida, and Krohnitta pacifica) were either equally frequent in the two types of collections or more frequent in the hyponeuston. Temperature-salinity-abundance relationships The chaetognaths were collected in a wide range of temperatures (2.2-26.6°C) and salinities 27.7-36.0%o (Table 4). Weighted means of surface salinities measured at the time of hyponeuston collections were very similar among all common surface species, rang- ing from 32.6 and 32.8%o for Sagitta tenuis and Parasagitta elegans, respectively, to 34.9%o for Kroh- Table 1 Number of zooplankton collectii >ns obtai ned during eight seasonal cruises in the Middle Atlantic Bight, 1975-77. Subsurface Surface 60-cm oongo nets 1-m hyponeuston net Cruise Dates 202 nm 505 fjm 505 /jm Total 1 23-29 Oct 75 6 6 48 60 2 5-16 Feb 76 6 6 48 60 3 8-16 Jun 76 6 8 52 66 4 31 Aug-9 Sep 76 6 7 48 61 5 5-28 Nov 76 22 21 75 118 6 20 Feb-6 Mar 77 21 21 75 117 7 17-28 May 77 21 21 75 117 8 19-29 Aug 77 21 21 75 117 Total 109 111 496 716 Table 2 Percent frequency of chaetognath species occurrence in hyponeuston collections from eight quarterly cruises, 1975- 77. Cruise: 1 2 3 4 5 6 7 8 Total Number of collections: 48 48 52 48 75 75 75 75 496 Serratosagitta tasmanica 50.0 43.8 48.1 33.3 66.7 61.3 17.3 21.3 42.5 Flaccisagitta enflata 18.8 - 1.9 81.2 77.3 34.7 24.0 73.3 41.9 Parasagitta elegans 35.4 85.4 92.3 14.6 6.7 16.0 58.7 14.7 37.3 Mesosagitta minima 2.1 22.9 — — 50.7 30.7 2.7 9.3 16.5 Sagitta helenae 4.2 — — 8.3 24.0 18.7 20.0 8.0 11.9 Ferosagitta hispida — - - 35.4 21.3 — 1.3 14.7 9.1 Pterosagitta draco 2.1 8.3 — 16.7 8.0 17.3 6.7 — 7.5 Sagitta tenuis — — — 31.2 20.0 — 1.3 2.7 6.7 Krohnitta pacifica 9.3 24.0 5.0 Serratosagitta serratodentata — — — 4.2 4.0 6.7 2.7 8.0 3.6 Flaccisagitta hexaptera - 10.4 - 10.4 2.7 6.7 - — 3.4 Krohnitta subtilis — — — 4.2 1.3 2.7 1.3 1.3 1.2 Mesosagitta decipiens — - - 2.1 - 6.7 - - 1.2 Pseudosagitta lyra — 2.1 - - - 4.0 - - 0.8 Eukrohnia hamata — 2.1 — — — 2.7 — — 0.6 Sagitta bipunctata 2.1 1.3 - 0.4 Solidosagitta planctonis 1.3 — 0.2 nitta pacifica (Table 4). Weighted mean temperatures, on the other hand, varied widely among species, from 12.9°C for Mesosagitta minima to 24.2°C for K. paci- fica. Species occurring infrequently at the surface and excluded from Table 4 were those residing at greater depths, with surface occurrences mostly restricted to low winter temperatures. Relationships between sur- face abundance of common species and the two physical factors were further examined by plotting average daily catch (8 collections at 3-hour intervals) in stan- dard hyponeuston collections, defined previously (Grant 1988) as 1-m neuston nets towed at 2.5 knots for 20 minutes; the sampler used typically fished the upper 36 Fishery Bulletin 89(1), 1991 10 cm of the sea surface so would more accurately be called a 'hypo- neuston' net. Temperatures and salinities measured during collec- tions containing chaetognaths were averaged at each positive station (Figs. 2 and 3). Flaccisagitta enflata, Sagitta tenuis, Ferosagitta hispida, and, to a lesser extent, S. helenae and Pterosagitta draco were absent or uncommon in temperatures less than about 9°C (Figs. 2, 3), while few Parasagitta elegans, Mesosagitta minima, and Ptero- sagitta draco were found in warmest temperatures. Serrato- sagitta tasmanica was present in a very wide range of both tem- perature and salinity. Numbers of Pterosagitta draco were re- duced in lower salinities as well as low temperatures; occurrences were in salinities higher than 34 %o except in the summer of 1976. Sagitta helenae and Meso- sagitta minima were more fre- quent and numerous in higher salinities, while S. tenuis was ab- sent from salinities greater than 35 %o. Other species in Figures 2 and 3 were captured in various salinities. Interspecific association Before examining association be- tween pairs of species, the data were first tested for overall asso- ciation using a matrix of the pres- ence or absence of the 15 most common chaetognaths in 716 col- lections. Schluter's (1984) index of overall association (VR here, as in Ludwig and Reynolds 1988) for these 15 species was 2.108, indicative of a net positive association among the species. The null hypothesis of no association was rejected (W = 1509; 90% probability 654.9 o ° m 1-10 O100-1000 26- 24- S. tasmanica S. helenae o22- ° ° n o " - lu20- p & o o -20 §18- <16 : o ° m>° 8 0° o o a i4- ®> o 5 12 : LU H10- o ° ao • -10 8- _ ° 6- o O • 4 : o 313233343536 313233343536 SALINITY %o SALINITY %o Figure 2 Temperature-salinity-abundance relationships of Flaceisagitta enflata, Parasagitta elegans, Serratosagitta tasmanica, and Sagitta helrnae in the surface layer of the central and southern Bight. Each circle or dot represents the average catch for eight standard hyponeuston collections obtained over a 24-hour period at a single sampling location. 32 34 36 32 34 36 26- S. tenuis M . minima 24 • o O 0° : o -20 §18- '. £16- O ' Hl 14 O |12- °°°QP ty 10- 8- °.% C -10 MEAN SURFACE CATCH 6- • <1 c 10100 1000 4- 26- oMO 100 1000 F. hispida P. draco 24- o o22- VP • • uj20- o • -20 §18- °6 <16- o i,4- o 9 r 5 ,!^ o. . "D |-10- o ° * . -10 8- • • 6- 4- L 3132 333435 36 313233343536 SALINITY %o SALINITY %o Figure 3 Temperature-salinity-abundance relationships of Sag' tta tenuis, Mesosagitta minima, Ferosagitta hispida, and Pti ro- sagitta draco in the surface layer of the central and southern Bight. See explanation in Figure 2 legend. Discussion Diel distribution at the surface Most of the chaetognaths commonly occurring in the surface layer reached peak densities near midnight (Table 6). Flaccisagitta enflata, Krohnitta pacifica, Sagitta tenuis, Mesosagitta minima, Ferosagitta hispida, and Pterosagitta draco all were caught in maximum numbers at that hour. Peak numbers of Parasagitta elegans and Serratosagitta tasmanica occurred somewhat earlier, but they were abundant throughout hours of darkness. There are also sugges- tions of dusk or dawn (or both) increases in abundance for K. pacifica, F. hispida, and Serratosagitta serra- todentata. Unlike any of the other chaetognaths, Sagitta hele- nae was decidedly more abundant at the surface in daylight hours, with 31.0 and 32.8% of total catches occurring around 1500 and 0900 hours, respectively. Eighteen species of chaetognaths are found in con- tinental shelf waters of the Middle Atlantic Bight, in- cluding the 15 species on record after the 1960s plus the present records of three species previously known only from surface slope waters (Grice and Hart 1962). More recent studies of chaetognaths in this region have been restricted to estuaries (Grant 1977b, Sweatt 1980, Canino and Grant 1985) or to oceanic waters (Cheney 1985a, b). All of the western North Atlantic species labeled epipelagic or mesopelagic by Cheney (1985b) have been collected in shelf waters, so the present list appears reasonably complete. The composition of Mid- dle Atlantic Bight chaetognath collections from hypo- neuston and subsurface plankton tows was nearly iden- tical. There were no frequent or abundant species unique to the hyponeuston, and all but the rarest species from subsurface shelf waters were taken at least occasionally from the surface layer. However, 38 Fishery Bulletin 89(1). 1991 Table 5 Number of occurrences (integers a ong the diagonal , coefficients of association (Hurlbert's C f , right side of diagonal), and their statistical significance (** P<0.01 and * <0.05, respectively ; blank = not significant, P>0.05, left side of diagonal) for chaetognaths collected during eight seasonal cruises in the Middle Atlantic Bight. 1975-77. Both subsurface and surface layer collections are in- eluded (N = 716); the three rarest species are excluded. pac ser his ten hel dra enf min tas ele ham hex lyr dec sub Krohnitta pacifica 31 0.26 0.09 0.13 0.05 0.05 -0.09 -0.73 -1.00 0.04 Serratosagitta serratodentata + + 29 0.13 0.11 0.06 0.05 0.04 0.01 -0.19 0.04 0.05 0.05 Ferosagitta hispida * *■ + 53 0.21 0.12 0.08 0.02 0.01 -0.64 -0.54 0.02 -1.00 0.02 0.03 Sagitta tenuis + + 46 0.09 0.07 0.08 -0.48 -0.18 -0.59 -1.00 0.02 Sagitta helenae + + + + + + 88 0.21 0.15 0.12 0.03 -0.40 0.06 -0.34 0.03 0.03 0.05 Pterosagitta draco + + *♦ * + 54 0.07 0.15 0.06 -0.22 0.13 0.09 0.05 0.08 0.09 Flaccisagitta enflata + + * + + * + + 322 0.22 0.18 -0.46 0.03 0.02 0.01 0.02 0.02 Mesosagitta minima + + + + + + + + + 182 0.27 -0.15 0.08 0.08 0.05 0.03 0.04 Serratosagitta tasmanica + + ** 381 0.08 0.03 0.04 0.02 0.01 0.02 Parasagitta elegans + + + + + + ** + + + + 341 0.01 -0.27 Eukrohnia hamata ** + + * + + + + * 30 0.30 0.25 0.12 0.11 Flaccisagitta hexaptera * + ** + + 39 0.11 0.09 0.03 Pseudosagitta lyra + + + ** + * ** 14 0.27 0.18 Mesosagitta decipiens + + + + + + + + + + + + + + 12 0.18 Krohnitta subtilis 15 there were apparent tem- perature-related differences in the percent frequency of chaetognath species in the two habitats. Cold-temper- ate species were less fre- quent at the surface than in the underlying water col- umn, while warm-temper- ate or subtropical species were either equally frequent in the two habitats or more frequent in the hyponeuston. The idea for T-S-P dia- grams apparently origin- ated with Pickford's (1946) study of the cephalopod Vampyroteuthis, and was first applied to chaetognaths by Bary (1959, 1963). T-S-P diagrams have since been used to relate abun- dance of chaetognaths to hydrography by numerous authors, including Sund (1961, 1964), Aurich (1971), Kotori (1976), Michel and Foyo (1976), O'Brien (1977), Nagasawa and Marumo (1982), and Andreu (1984). Flaccisagitta enflata, abundant in the Middle Atlantic Bight in warmer temperatures (averaging 18.6°C) and in various salinities (<32 to 36%o), occurred through- out the temperature and salinity ranges sampled by Nagasawa and Marumo (1982) and was apparently limited only by depth in the Caribbean (Michel and Foyo 1976). Sund (1961) also recorded F. enflata from 13-28°C and 32.6-35°/.*. T-S-P diagrams for Para- Table 6 Diel distribution of the more common chaetognaths in hyponeuston collections. Data from 24-hour stations only, combined from eight seasonal cruises, 197E -1977. Hours (EST) Total 1200 1500 1800 2100 2400 0300 0600 0900 (% of total) - N Flaccisagitta enflata 10.6 8.9 7.9 11.1 25.7 10.4 13.8 11.6 89,807 Parasagitta elegans 2.3 1.7 1.2 37.4 26.8 19.5 7.8 3.3 87,093 Serratosagitta tasmanica 1.3 1.3 6.7 30.3 27.9 23.5 6.2 2.8 16.283 Sagitta helenae 18.9 31.0 2.5 1.6 10.5 1.8 0.8 32.8 15,073 Krohnitta pacifica 3.5 10.3 11.8 2.5 34.0 1.9 23.0 13.1 2,573 Sagitta tenuis 0.5 0.9 10.3 13.7 45.2 18.2 7.9 3.3 2,283 Mesosagitta minima 0.9 1.4 2.4 8.9 60.4 13.6 7.5 4.9 1,383 Ferosagitta hispida 1.0 4.6 23.4 14.6 26.6 12.9 16.9 628 Serratosagitta serratodentata 0.2 2.9 28.2 8.3 0.7 3.4 56.2 411 Pterosagitta draco 1.7 11.9 32.2 27.1 26.3 0.8 118 sagitta elegans have been plotted by Bary (1963), who included the species in his "coastal (neritic) group," by O'Brien (1977) for populations to the west of Ireland, and by Kotori (1976) with ranges of temperature and salinity close to those in the present study. All agree in showing higher occurrence and abundance in colder water and an apparent tolerance for reduced salinity. Although the ranges of surface salinity in which Mid- dle Atlantic Bight species were found were very similar, it is noteworthy that the five species with the lowest weighted means were the same five species recorded from within Chesapeake Bay (Grant 1977b) and in approximate inverse order of their estuarine abundance (Sagitta tenuis, x S = 32.6"A». and the most Grant: Chaetognatha from central and southern Middle Atlantic Bight 39 abundant, to Serratosagitta tasmanica, x S = 33.7°/oo, the rarest in Chesapeake Bay). Aurich (1971) includes the only other known T-S-P diagram of S. tasmanica, used for a display of habitat differences with 5. serra- todentata. Michel and Foyo (1976) and Nagasawa and Marumo (1982) provided T-S-P diagrams for both Meso- sagitta minima and Pterosagitta draco. Middle Atlan- tic Bight T-S-P diagrams agree with their depiction of greater abundance at higher salinities for these two species. Finally, Michel and Foyo's (1976) T-S-P dia- gram of Ferosagitta hispida shows a few occurrences at lower temperatures (13-17°C), but most at 27-28°C and in a broad range of salinities. Association among the Middle Atlantic Bight species was generally positive, both in the multispecies case and between pairs of species. Parasagitta elegans pro- vided an important and consistent exception, evidenced by its highly significant (P<0.01) negative associations with five warm-water species: Krohnitta pacifica, Ferosagitta hispida, Sagitta tenuis, Flaccisagitta en- flata, and Sagitta helmae. Although both P. elegans and S. tenuis occur abundantly in coastal and estuarine waters in the Chesapeake region, they do so in opposite seasons, hence their negative association. The sole significant (P<0.05) positive association of P. elegans was with Serratosagitta tasmanica, and it appears to occupy a niche of low temperatures and salinities in this region, not unlike that of Aidanosagitta crassa^ in the East China Sea (Matsuzaki 1975). Highly significant positive associations were shared by (1) the endemic and abundant shelf species Serratosagitta tasmanica, Mesosagitta minima, and Flaccisagitta enflata, (2) the warm-water species Krohnitta pacifica, Serratosagitta serratodentata, Ferosagitta hispida, and Sagitta tenuis, and (3) the offshore, shelf-edge species Flacci- sagitta hexaptera, Eukrohnia hamata, Pseudosagitta lyra, Mesosagitta decipiens, and Krohnitta subtilis. Three of the latter species comprised Matsuzaki's (1975) "Kuroshio water" species group. Pearre (1973) determined that extensive diurnal migration takes place with Parasagitta elegans, per- haps related to feeding. Data in the present study on the diurnal distribution of the species in the hypo- neuston also indicates a strong upward migration at night; numbers caught were about an order of magni- tude higher at night than in daylight. Several other species showed similar sharp increases in abundance during darkness, including Serratosagitta tasmanica, Krohnitta pacifica, Sagitta tenuis, Mesosagitta mini- ma, and Pterosagitta draco. Some were more abundant at dawn or dusk, or, in the case of Flaccisagitta enflata and Sagitta helenae, were present in considerable num- bers in daylight. Among these species, Nagasawa and Marumo (1982) found evidence for diurnal migration in F. enflata, P. draco, and K. subtilis, none for K. pacifica, and mixed evidence for M. minima and F. hex- aptera. However, in the present comparison of day and night hyponeuston collections, only a short migration would be required to populate and depopulate the sur- face layer diurnally. Such fine-scale diurnal movements likely do occur within the surface mixed layer, but are most difficult to measure. Acknowledgments Collections and identifications serving as the basis for this report were supported by the U.S. Dept. of the In- terior, Bureau of Land Management Contracts No. 08550-CT-5-42 and AA550-CT6-62. Much appreciated were early and helpful reviews of this manuscript by Romuald N. Lipcius, Kenneth L. Webb, and Daniel W. Sved, and the suggestions by two anonymous reviewers for final alterations and adjustments. Citations Alvarino, A. 1963 Quetognatos epiplanctonocos del Mar de Cortes. Rev. Soc. Mex. Hist. Nat. 24:97-203 [in Spanish]. Andreu, P. 1984 Sagitta decipiens (Chaetognatha) en el Mediterraneo oc- cidentale: Diagramas T-S plancton. Result. Exped. Cient. 12:23-30 [in Spanish, Engl, abstr.]. Aurich, H.J. 1971 Die Verbreitung der Chaetognathen im Gebiet des Nor- datlantischen Strom-Systems. Ber. Dtsch. Wiss. Komm. Meeresforsch. 22:1-30 [in German, Engl, abstr.]. Bailey, N.T.J. 1981 Statistical methods in biology, 2d ed. John Wiley, NY. 216 p. Bary, B. McK. 1959 Species of zooplankton as a means of identifying different surface waters and demonstrating their movements and mix- ing. Pac. Sci. 13:14-54. 1963 Temperature, salinity and plankton in the eastern North Atlantic and coastal waters of Britain, 1957. II. The relation- ships between species and water bodies. J. Fish. Res. Board Can. 20:1031-1066. Bigelow, H.B., and M. Sears 1939 Studies of the waters on the continental shelf, Cape Cod to Chesapeake Bay. III. A volumetric study of the zooplank- ton. Mem. Mus. Comp. Zool. Harv. Univ. 54:179-378. Burrell, V.G. Jr., W.A. van Engel, and S.G. Hummel 1974 A new device for subsampling plankton samples. J. Cons. Perm. Int. Explor. Mer 35:364-367. Canino, M.F., and G.C. Grant 1985 The feeding and diet of Sagitta tenuis (Chaetognatha) in the lower Chesapeake Bay. J. Plankton Res. 7:175-188. Cheney, J. 1985a Spatial and temporal patterns of oceanic chaetognaths in the western North Atlantic — I. Hydrographic and seasonal abundance patterns. Deep-Sea Res. 32:1041-1059. 1985b Spatial and temporal patterns of oceanic chaetognaths in the western North Atlantic — II. Vertical distribution and migrations. Deep-Sea Res. 32:1061-1075. 40 Fishery Bulletin 89(1), 1991 Deevey, G.B. 1960 The zooplankton of the surface waters of the Delaware Bay region. Bull. Bingham Oceanogr. Collect. Yale Univ. 17:5-53. Grant, G.C. 1963a Chaetognatha from inshore coastal waters off Delaware, and a northward extension of the known range of Sagitta tenuis. Chesapeake Sci. 4:36-42. 1963b Investigations of inner continental shelf waters off lower Chesapeake Bay. Part IV. Description of the Chaetognatha and a key to their identification. Chesapeake Sci, 4:107-119. 1967 The geographic distribution and taxonomic variation of Sagitta serratodentata Krohn 1853 and Sagitta tasmanica Thompson 1947 in the North Atlantic Ocean. Ph.D. diss., Univ. Rhode Island, Kingston, 116 p. 1977a Middle Atlantic Bight zooplankton: Seasonal bongo and neuston collections along a transect off southern New Jersey. Spec. Rep. ppl. Mar. Sci. Ocean Eng. (SRAMSOE) 173, Va. Inst. Mar. Sci., Gloucester Pt., 138 p. 1977b Seasonal distribution and abundance of the Chaeto- gnatha in the lower Chesapeake Bay. Estuarine Coastal Mar. Sci. 5:809-824. 1979 Middle Atlantic Bight zooplankton: Second year results and a discussion of the two-year VIMS-BLM Survey. Spec. Rep. ppl. Mar. Sci. Ocean Eng. (SRAMSOE) 192, Va. Inst. Mar. Sci., Gloucester Pt., 236 p. 1988 Seasonal occurrence and dominance of Centropages con- geners in the Middle Atlantic Bight, U.S.A. Hydrobiologia 167/168:227-237. Grice, G.D., and A.D. Hart 1962 The abundance, seasonal occurrence and distribution of the epizooplankton between New York and Bermuda. Ecol. Monogr. 32:287-309. Hurlbert, S.H. 1969 A coefficient of interspecific association. Ecology 50: 1-9. Kassatkina, A. P. 1971 New neritic species of chaetognaths from Possjet Bay in the Sea of Japan. Res. Mar. Fauna, Fauna and Flora of Possjet Bay, vol. 8, p. 265-294. Nauka Press, Leningrad [in Russian], Kotori, M. 1976 The biology of Chaetognatha in the Bering Sea and the northern Pacific Ocean, with emphasis on Sagitta elegans. Mem. Fac. Fish., Hokkaido Univ. 23:95-183. Ludwig, J. A., and J.F. Reynolds 1988 Statistical ecology. John Wiley, NY, 337 p. Matsuzaki, M. 1975 On the distribution of chaetognaths in the East China Sea. The Oceanographical Mag. 26:57-62. Michel, H.B., and M. Foyo 1976 Caribbean zooplankton. Part I. Siphonophora, Hetero- poda, Copepoda, Euphausiacea, Chaetognatha and Salpidae. U.S. Dep. Navy, Office Naval Res., Wash., DC, 549 p. Nagasawa, S., and R. Marumo 1982 Vertical distribution of epipelagic chaetognaths in Suruga Bay, Japan. Bull. Plankton Soc. Jpn. 29:9-23. O'Brien, F.I. 1977 The relationship between temperature, salinity and Chaetognatha in the Galway Bay area of the west coast of Ireland. Proc. R. Ir. Acad. Sect. B Biol. Geol. Chem. Sci. 77:245-252. Pearre, S. 1973 Vertical migration and feeding in Sagitta elegans Ver- rill. Ecology 54:300-314. Pickford, G.E. 1946 Vampyroteuthis infernalis Chun. An archaic dibranchiate cephalopod. I. Natural history and distribution. Dana- Rep. Carlsburg Found. 26:197-210. Pielou, E.C. 1977 Mathematical ecology. John Wiley, NY, 385 p. Ratliff, R.D. 1982 A correction of Cole's C 7 and Hurlbert's C 8 coefficients of interspecific association. Ecology 63:1605-1606. Schluter, D. 1984 A variance test for detecting species associations, with some example applications. Ecology 65:998-1005. Sund, P.N. 1961 Some features of the autecology and distributions of Chaetognatha in the eastern tropical Pacific. Bull. Inter-Am. Trop. Tuna Comm. 5:305-340. 1964 The chaetognaths of the waters of the Peru region. Bull. Inter-Am. Trop. Tuna Comm. 9:113-216. Sweatt, A.J. 1980 Chaetognaths in lower Narragansett Bay. Estuaries 3:106-110. Tokioka, T. 1965 The taxonomical outline of Chaetognatha. Publ. Seto Mar. Biol. Lab. 12:335-357. Zar, J.H. 1984 Biostatistical analysis, 2d ed. Prentice-Hall, Englewood Cliffs, NJ, 718 p. Abstract.- The trawl fishery for ocean pink shrimp (Pandalus jordani Rathbun) has increased dramatical- ly since the early 1970s. Catch and effort statistics and catch sampling data from 1968-88 were analyzed to evaluate changes in the shrimp popu- lation structure. Carapace length at age one and two have increased sig- nificantly since 1978, concurrent with a reduction in fishery catch per unit effort, strongly indicating den- sity-dependent growth. The seasonal pattern of growth provides further evidence for density-dependent growth. The number of age three shrimp in the catch has declined markedly since 1978, while age one shrimp have increased from 30.6% of the catch to 69.2%. The percentage of age one shrimp maturing as fe- males has increased to 30-50% in some years, while the overall per- centage of males shows no trend. The changes in growth, and age and sex composition of the catch are at- tributed in part to the impact of the trawl fishery, which is currently con- tinuing to intensify. Density-depen- dent growth, and the ability to accel- erate the sex change process make pink shrimp resistant to over-har- vest. However, at some exploitation level the reduction of the age 1 spawning stock should begin to re- duce subsequent recruitment. Re- cent strong year classes indicate that the fishery probably has not reached that level of exploitation. Fishery-induced Changes \n the Population Structure of Pink Shrimp Pandalus jordani Robert W. Hannah Stephen A. Jones Oregon Department of Fish and Wildlife, Marine Region Marine Science Drive, Bldg. 3, Newport, Oregon 97365 Manuscript accepted 24 August 1990. Fishery Bulletin, U.S. 89:41-51 (1991). The Pacific trawl fishery for pink shrimp Pandalus jordani Rathbun has developed from a fishery with landings of around 220 mt in the early 1960s to a fishery regularly landing in excess of 18,000 mt. In six of the thirteen years since 1975, combined landings for the states of California, Washington, and Oregon have ex- ceeded 24,000 mt. Pink shrimp range from San Diego, California to Un- alaska, Alaska (Butler 1964); how- ever, the majority of the catch is taken between Cape Mendocino, California and Destruction Island, Washington. The development of the fishery has been well summarized by Dahlstrom (1970), Fox (1972), Zirges and Robinson (1980), and others. Saelens and Zirges (1985) de- scribed the 1984 fishery for pink shrimp and suggested that there was some evidence of changes in the shrimp population structure that were possibly the result of fishing. They noted improved growth and higher levels of age-1 shrimp in the catch, relative to earlier years. Fish- ing effort and shrimp catch have con- tinued to increase since 1984. Given the continued development of this fishery, we felt that some fishery- induced changes in the population structure would be evident in a tho- rough review of the fishery sampling data. We examined 23 years of infor- mation from the pink shrimp fishery off the coasts of Oregon and northern California to search for the classic population responses to increased fishing. Specifically, we looked for persistent shifts in age and sex com- position, changes in the age of female maturity, evidence of reduced stock biomass, and improved shrimp growth as a response to lower bio- mass. As a final step, we attempted to relate the observed changes to the development of the fishery and alter- natively to environmental factors. Methods We examined monthly sample data from the landed catch of pink shrimp for the years 1966-88. The data is comprised of several samples from each statistical area (Fig. 1) and month of the fishing season. The season currently runs from April through October, but has been longer in the past. Individual samples for each area and month (area-month) were combined for analysis of age and sex composition. Sample sum- maries provide individual carapace lengths and average weight ex- pressed as the number of whole shrimp per pound. Shrimp are classi- fied as male, female, or transitional based upon close examination of the inner ramus of the first pleopod (Tegelberg and Smith 1957). Shrimp age is determined by modes in the combined length-frequency histogram (Zirges et al. 1981). Nadirs in the histograms define the range of cara- pace lengths corresponding to each age group, then ages are assigned to individual shrimp. Zirges et al. (1982) concluded that pink shrimp from statistical areas 41 42 Fishery Bulletin 89(1). 1991 STUDY AREA Washington 46" 22 Oregon 21 2 P) 19 y / California Cape Mendocino 48 44 42" - 4 Figure 1 Location of commercial concentrations of pink shrimp Pan- dalus jordani along the U.S. Pacific coast (shaded areas) and statistical areas 18-32. 18-28 constituted a single stock, based upon an analysis of growth, maturation rates, and coastal oceanographic conditions. We used the same stock unit to allow us to draw upon the summarized sample data from Zirges et al. (1982) for the years 1966-81. To evaluate the effect of shrimp density on growth, we compared shrimp carapace length at age for two time-periods representing different levels of population biomass. We use the term "density" in the sense of biomass per unit area rather than the number of in- dividuals per unit area. We used catch-per-unit-effort (CPUE) as our index of shrimp density. During the years 1975-78 major improvements in trawl design were implemented by the shrimp industry. Prior to 1976, the predominant shrimp net was a 57-foot (headrope) Gulf of Mexico style, Marinovitch trawl, with a 4-foot vertical opening (Zirges and Robin- son 1980). During the years 1975-78 the majority of T 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 66 68 70 72 74 76 78 80 82 84 86 YEAR Figure 2 Catch-per-unit-effort (kg/single-rig equivalent hour) for the pink shrimp fishery in statistical areas 18-28 for the years 1966-88. the fleet switched to locally produced 70-90 foot (headrope) box trawls. The new trawls have proven to be much more efficient for pink shrimp. Besides being generally bigger, they open to a height of 12-18 feet, improving fishing for pink shrimp, which come up off the bottom under reduced light conditions. Increased cloud cover and time of day were shown to bring con- centrations of shrimp up off the bottom at least 8 feet (Beardsley 1973). The Oregon box trawls are also con- structed differently from the Marinovitch trawls. They are much longer, employ a slower taper, and are hung with considerable "slack" webbing, all contributing to a much more efficient trawl for pink shrimp. The effort data series is not corrected for these gear improve- ments. Therefore, CPUE data understate the magni- tude of biomass reduction since 1978 (Fig. 2). Accord- ingly, we tested growth for the two time-periods, 1966-78 and 1979-88. We considered these periods to be representative of the virgin stock biomass and the reduced biomass, respectively. We used .F-tests to iden- tify significant differences in length at age. Four area-months were selected as indices for anal- ysis of age-1 and -2 shrimp growth, based upon the com- pleteness of the time-series data. For age-1 shrimp, the months of April- June were excluded because in some years age-1 shrimp are not fully recruited to the trawl gear in those months. For age-2 shrimp, the months of September and October were excluded because in recent years age-2 and older shrimp comprise a small percentage of late-season catches. Given these criteria, the index area- months selected for age-1 shrimp were Hannah and Jones: Fishery-induced population changes in Pandalus jordani 43 Area 22- August, Area 21- July, Area 26- August, and Area 19- August. The most complete time-series for age-2 shrimp were found to be Area 19- August, Area 22- April, Area 21-June, and Area 26-May. Age-2 length com- parisons are not independent of the age-1 results. The age-2 analysis was employed to help rule out any appar- ent changes in growth of age-1 shrimp caused by changes in fishery or gear selectivity over time. To further investigate factors influ- encing age-1 shrimp growth, we con- ducted some exploratory correlation analysis. Since the time-series for the most complete index areas still con- tained some missing samples, and since many of the environmental fac- tors we wished to test are not area- specific, it was desirable to combine our four age-1 growth indices into one, more complete, time-series. Carapace length at age in pink shrimp exhibits a gradient effect increasing from north to south along the coast and also in- creasing through the season. To re- move these effects and yet preserve the interannual variations in size, one of the four indices was chosen as a standard and the other three were adjusted by an additive factor equal to the difference between the mean of the chosen standard and that of the individual index area-month. Subsequently, the adjusted index area-months were averaged into one time-series. Area 22 was chosen as the standard, and the resultant time- series for age-1 was without gaps and most points were based on two or more adjusted means. Linear regression was used to examine factors in- fluencing variation in this age-1 shrimp growth in- dex. The independent variables tested included sea sur- face temperature at Charleston, Oregon (Oreg. Dep. Fish. Wildl., unpubl. data) upwelling at 45°N, 125°W (Bakun 1973; NMFS Pacific Environ. Group, Mon- terey, CA, unpubl. data), inverse-barometer corrected sea level at Newport, Oregon, (Pittock et al. 1982; Pittock, unpubl. data), and catch per unit effort in the fishery as an index of shrimp density. Kruse (1981) found inverse barometer-corrected sea level at Newport and Neah Bay to be highly correlated with sea-bottom shelf temperatures near Newport. We tested each variable with no time lag (year t) and a 1-year time lag (t - 1) to match growth in an earlier life stage. Adjusted sea level at Newport was tested both with and without the 1983 data point, a year of abnormally high sea level caused by a strong Table 1 Comparison of mean carapace length for age-1 and -2 pink shrimp for the years 1966-78 and 1979-88 (single classification ANOVA with unequal sample sizes). Area-Month Years Mean carapace length (mm) N (years) F value P>F Age 1 Area 22-August 1966-78 1979-88 16.04 17.32 13 7 23.40 0.0001 Area 21-July 1966-78 1979-88 15.65 16.44 9 9 6.11 0.0251 Area 26-August 1966-78 1979-88 15.56 16.67 9 7 14.09 0.0021 Area 19-August 1966-78 1979-88 16.44 17.79 8 7 27.07 0.0002 Age 2 Area 19-August 1966-78 1979-88 19.98 21.84 8 7 76.43 0.0001 Area 22-April 1966-78 1979-88 18.03 19.59 13 9 26.70 0.0001 Area 21-June 1966-78 1979-88 19.01 20.60 8 9 37.57 0.0001 Area 26-May 1966-78 1979-88 18.75 19.69 11 9 7.98 0.0112 El Nino event. Finally, we tested the average CPUE for the years t and t - 1 combined, to represent the average density encountered over the life of an age-1 shrimp. Second-order polynomial regression was also used to test each variable for a significantly curvi- linear relationship with the age-1 growth index (Ricker 1975). Correlation analysis with time-series data of short duration is often of limited value, but does help to generate initial hypotheses to be tested with the ac- cumulation of future data (Ricker 1975). Short time- series often exhibit unidirectional time trends causing spurious correlations. For these reasons, we felt that correlation analysis would be a relatively poor tool for differentiating the relative importance of the various factors to shrimp growth, but would help to identify the factors which deserve future analysis. Consequent- ly, correlation analysis was not pursued further in this study. For analysis of trends in the sex composition data, we once again relied upon four index-area months with the most complete time-series. The four index-areas employed were Areas 19 and 28 in October and Areas 21 and 22 in September. Pink shrimp are protandrous hermaphrodites and mate primarily in September and October each year (Pacific Fisheries Management Council 1981). They usually mature first as males in the fall at about IV2 years of age, and after spawning 44 Fishery Bulletin 89( 1991 18.0 -r— — — ——^—^—— ^ ^^^^~ E 19.0 - E E E . □ **■"" □ D i X i- 18.0- i- C3 17.0- ° z ID z D a LU □ LU _l _l LU 17.0- LU 16.0- D B o D o El < D □ Q < D. □ «i o. ' < □ < D q Q DC < 16.0- Q Q AREA 19 LX < o 15.0- Q AREA 21 z AUGUST z ei JULY < < LU LU 2 65 70 75 80 85 90 5 65 70 75 80 85 90 YEAR YEAR "E E E, • DEI D E 17.5- X □ G> X D D 17.0 - □ □ i- Q Q z G Q r. □ z 16.5- LU Q Q a LU n D _l D _l m a D o LU 16.0- LU □ D o < G n D < 15.5- D G D. □ Q " Q. < < " LX cc < o z 15.0- B D AREA 22 < 14.5- AREA 26 . AUGUST o H AUGUST < < LU 2 LU 2 13.5 - 14.0 - 6 5 70 75 80 85 9 6 5 70 75 80 85 9 YEAR YEAR Figure 3 Mean carapace length of age-1 pink shrimp from 1966-88 fo r the selected index area-months. go through a transitional phase, usually maturing as a female the following year at age 2%. Age-1 shrimp that mature directly into females, bypassing the male phase, are called primary females. We examined the data for trends in the percentage of primary females and the overall percentage of male shrimp. Results We found significantly improved growth of pink shrimp for the 1979-88 period as compared with the 1966-78 period. Mean carapace lengths for all four age-1 index area-months were significantly greater in 1979-88 (Table 1, Fig. 3) based on two-tailed F -tests. Mean carapace lengths were also greater for the four age-2 index area-months (Table 1, Fig. 4). Since age-2 shrimp are fully recruited to the trawl gear, this result rules out any apparent increase in mean length due to fishery or gear selectivity or accelerated sex change of age-1 shrimp. The time-series of catch per unit effort (Fig. 2) indicates that the population biomass has been reduced since 1978. The increase in growth demonstrated for the 1979-88 catch years represents a substantial increase in average weight. Using the age-1 composite growth index, the mean carapace length of age-1 shrimp has increased from 16.1 to 17.4mm. From the length- weight relationship developed by Zirges et al. (1982); In W = -7.94746 + 3.20971n L, Hannah and Jones: Fishery-induced population changes in Pandalus jordani 45 — . 99fl ^* « E ^ "i" c\.\j- E, X AREA 21 D n E, x AREA 26 MAY DD a D a 1- 21.0- JUNE D i- 20.0- D z □ Z D LU LU □ _l D ° D H H _l UJ 20.0 - < D LU o < 19.0 - Dl n a D D < cc □ < cc D D D a < 19.0- < 18.0- B o o z D ° D = z < < LiJ 5 18.0- LU 17.0- . | . . . 1 65 70 75 80 85 90 " 65 70 75 80 85 90 YEAR YEAR •=• 21.0- E E E_ I 20.0 - 1- o | 19.0- D Q El B Dd X 1- e> z LU _l 22.0- AREA 19 a AUGUST D Q a LU D LU 21.0- o D o □ < 18.0- Q. D < a. D < O 17.0- B ° D n AREA 22 EI D D APRIL < cc < o 20.0- □ D D z < z < □ "j 16.0- 6 111 5 19.0- i i i 1 5 70 75 80 85 90 65 70 75 80 85 90 YEAR YEAR Figure 4 Mean carapace length of age-2 pink shrimp from 1966- 88 for the four index area-months. where W = Weight (g), and L = Length (mm), this represents a 28.4% increase in mean weight at age. This is an approximate figure scaled for Area 22 data; however, other areas yield similar results. The exploratory correlation analysis (Table 2) shows that sea surface temperature at Charleston, Oregon, and sea level at Newport, Oregon (Fig. 5), are positively correlated with the age-1 growth index. However, CPUE (Fig. 6) is negatively correlated with the age-1 growth index suggesting density-dependent growth. Adjusted sea level at Newport and sea surface tem- perature at Charleston displayed evidence of a curvi- linear relationship, of decreasing slope, with the age-1 growth index. However, the regression coefficients of the second-order polynomial regressions were only significantly different from zero (t-test, P>0.05) for adjusted sea level in year t - 1 . The curvature of this relationship is strongly influenced by a single outlier, the 1983 sea level, data point. With this point deleted, the coefficients of the polynomial regression are not significantly different from zero, while the adjusted r-squared value for the simple linear regression in- creases to 0.528 (Table 2). This suggests that bottom temperature is also influencing shrimp growth. Graphs of mean length at age for areas 19, 21, 22 and 26 (Fig. 7) show that much of the difference in growth between the 1965-76 and 1977-86 broods is already apparent at age-13 months when the fishery first catches age-1 shrimp. This is not surprising, since density-dependent growth has been demonstrated more frequently for species during the immature phase, with density-dependent changes in fecundity more prevalent in the adult phase (Bailey and Almatar 1989). With the exception of area 19, the curves for the 46 Fishery Bulletin 89(1), 1991 1965-76 broods show some tendency for the rate of increase in length to slow in the fall, near the end of the fishing season, at ages 17-19 and 27- 30 months. The curves for the 1977-86 broods differ somewhat in that this slowing of growth in the fall is less pro- nounced. These data suggest that den- sity-dependent growth in pink shrimp may continue into the adult stage. The impact that the trawl fishery has had upon the shrimp population is very evident from an examination of age composition of the catch (Fig. 8). In catch years 1966-78, age-3 shrimp comprised an average 20.4% of the catch by number, falling to an average 4.9% for the years 1979-88. Converse- ly, age-1 shrimp have risen from an average 30.6% in the early period to 69.2% of the catch in recent years. The increases in relative abundance of age-1 shrimp may be explained in part by the fact that they are recruited to the trawl gear earlier in the season due to increased size at age in recent years. The decline in absolute numbers of age-3 shrimp (Fig. 8) cannot be explained by changes in gear selection. The observed changes in age structure of the catch must be at least partly due to the impact of the trawl fishery. The change in age composition of the catch is also reflected in the mean size of shrimp in the landed catch. Table 2 Results of exploratory regression analysis of potential factors influencing carapace length of age-1 pink shrimp. The dependent variable is an index of age-1 shrimp growth based upon mean carapace length from four selected statistical area-months. Model Independent variables* Intercept Coefficient Adjusted r 2 P 1 u, — — NS 2 u t ., — — — NS 3 SL, — — — NS 4 SST, 12.474 0.426 0.191 0.0307 5 SST,., 12.587 0.419 0.198 0.0323 6 SL,., -20.096 0.125 0.348 0.0075 8 CPUE, 17.709 -0.005 0.349 0.0018 7 CPUE,_, 17.649 -0.005 0.287 0.006 9 Mean CPUE, , , 17.877 -0.006 0.391 0.0009 10 SLi-x" -46.171 0.214 0.528 0.0009 upwelling from 45°N lat. 125°W long, verse-barometer corrected sea level at Newport, OR ary mean sea surface temperature at Charleston, OR effort for the pink shrimp fishery in statistical areas of growth index. * U = April- October SL = Mean annual in SST = January-Febn, CPUE = Catch-per-unit- 18-28. t = Calendar year **1983 sea level deleted. The average number of shrimp per pound increased from 109.4 during the 1966-78 period to 118.9 for the 1979-88 period. The decline in numbers of older shrimp has been accompanied by an increase in the percentage of shrimp maturing directly into females at age-1 (Fig. 9). In recent years, levels of primary females as high as 30-50% are common. This effect has compensated for the higher cumulative harvest rates on age-2 and o — AGE ONE GROWTH INDEX t f AGE ONE GROWTH INDEX |mn EAN ANNUAL SEA LEVELAT NEWPO 196S 1970 1975 1980 1985 1990 YEAR Figure 5 Age-1 growth index (mm) and mean inverse-barometer cor- rected sea level (cm) at Newport, Oregon for the years 1966-88 and 1971-88, respectively. AGE ONE GROWTH INDEX CPUE Figure 6 Age-1 growth index (mm) and pink shrimp catch-per-unit- effort (kg/single-rig equivalent hour) for 1966-88. Hannah and Jones: Fishery-induced population changes in Pandalus jordani 47 — 24.0 E E 22 a < Q. < < O < 20 0- 18.0- 16 120 10 24.0 AREA 19 1965-76 BROODS 1977-86 BROODS 1 I 15 20 25 AGE IN MONTHS 30 35 AREA 26 20 25 AGE IN MONTHS 35 35 AGE IN MONTHS _ 24.0 E E 22.0 X t- O Z 20.0 ui 16.0 < a < K < O z 14.0 < ui 5 12.0 AREA 21 -G — 1 965-76 BROODS 1 977-86 BROODS -i— i— i— i— i— i— i— r- 10 20 25 AGE IN MONTHS 30 35 Figure 7 Mean length at age for the 1965-76 and 1977-86 broods of pink shrimp from Oregon statistical areas 19, 21, 22, and 26. _ 8000 YEAR OF CATCH Figure 8 Age composition by number of the catch of pink shrimp from statistical areas 18-28 for the years 1966-88. older shrimp, and a sexually balanced breeding popula- tion has been maintained (Fig. 10). Discussion Our analysis supports a conclusion that pink shrimp are exhibiting density-dependent growth. The transition to larger mean size at age between 1978 and 1979 coin- cides nearly exactly with the large drop in fishery CPUE, and the shift in age composition of the catch toward younger ages (Figs. 6, 8). This is probably due to the intensive development of the fishery during the mid 1970s (Fig. 11). The fishery continued to intensify after 1978, with subsequent effort levels falling to pre-1977 levels only in the years 1983-1985 (Fig. 12). The persistence of reduced CPUE and reduced mean age at capture are classic results of an intense size- selective harvest causing reduced levels of population biomass. 48 Fishery Bulletin 89(1), 1991 AREA 19 -OCTOBER UJ AREA 21 -SEPTEMBER < 80 - a AREA 22- SEPTEMBER PRIMARY FEf • AREA 28 -OCTOBER • • PERCENTAGE OF loo . ' - • : D " " a _ • y o • 1965 1970 1975 1980 1985 1990 YEAR Figure 9 Percentage of age-1 pink shrimp maturing as females for four index area-months for the years 1966-88. CALIFORNIA WASHINGTON OREGON JlullJUl 58 60 62 64 70 72 74 76 78 80 82 84 YEAR Figure 1 1 Combined landings of pink shrimp for the states of Califor- nia, Oregon, and Washington for the years 1957-88. • AREA 19-OCTOBER 90 - AREA21-SEPTEMBER 80 - B • AREA 22-SEPTEM8ER AREA 28-OCTOBER /0 - 60- 9 n Q " a * • D * • • so - • D □ n a 40- • • • o s a • D a n G 30 - 9 9 20- 9 10- -\ • i 1 ' Figure 10 Percentage of male pink shrimp for four index area-months for the years 1966-88. o u. UL 100000 fl 66 68 70 72 74 Figure 12 Fishing effort (single-rig equivalent hours) for pink shrimp in statistical areas 18-28 for the years 1966-88. The pattern of growth in the index areas (Fig. 7) shows a slight tendency toward improved late-season growth in recent years, coincident with the season of minimum shrimp density on the grounds (Fig. 13). Dur- ing the fall, age-0 shrimp begin to appear in the trawl catch in small numbers (Zirges et al. 1982). These data suggest that the improved growth occurring prior to age-13 months may be a result of decreased shrimp densities created by the trawl fishery. The time-series of CPUE probably understates the true drop in shrimp biomass since 1978 due to the gear improvements mentioned previously. The difference in mean size at age demonstrated after 1979 may also be understated by the data shown in Figure 9. Prior to 1969 the minimum codend mesh allowed in the Oregon pink shrimp fishery was 38 mm (including one knot), while from 1969 onward codend mesh size has been unregulated (J.T. Golden, ODFW, Newport, 1981 draft). The lower curves in Figure 7, therefore, prob- ably overestimate the mean size at age-1. Since 1979, with larger mean size, age-1 shrimp have been more completely sampled by the gear. D.R. Bernard (Oregon State Univ., Corvallis, 1983 draft) estimated that pink shrimp were fully recruited to the 38 mm mesh trawl Hannah and Jones: Fishery-induced population changes in Pandatus jordani 49 d 1966-78 MEAN SRE hour) • 1979-88 MEAN D D B D a Q. E V) /♦ • / LU ID 10- 0, o *^-^* z _j 2 4 6 8 10 12 MONTH Figure 13 Log of mean catch-per-unit-effort (shrimp/single-rig equivalent hour) by month for 1966-78 and 1979-88. at 16.6mm carapace length. Since 1979, age-1 shrimp are fully recruited to a 38mm trawl in the later months of the season and to smaller gear even sooner. In earlier years, samples were biased toward only larger age-1 shrimp, and thus, by comparison, understate the in- crease in growth observed since 1979. Our exploratory correlation analysis is inconclusive in differentiating between density-dependent and en- vironmental factors as influences on shrimp growth. While the underlying relationship between the age-1 growth index and the environmental variables tested is most likely curvilinear (Ricker 1975), the relatively narrow range of environmental variability being tested in this case warranted the simple linear approximation. The combined age-1 growth index was closely cor- related with our index of shrimp density, CPUE, despite the fact that CPUE is a relatively poor index of density. We showed CPUE to be negatively cor- related with mean size at age over rather large changes in CPUE. Of course, smaller changes in CPUE, not associated with major changes in population density, should be positively correlated with growth, causing CPUE to be a poor index of shrimp density. We also found adjusted sea level at Newport, Oregon in year t- 1 to be closely correlated with the age-1 growth index, indicating that warmer bottom temperatures may have caused improved shrimp growth after 1978. Rothlisberg (1975) showed shrimp growth to be posi- tively correlated with temperature under laboratory conditions. It is possible that elevated sea levels im : prove growth over the normal range observed, but at extreme levels such as occurred in 1983, the virtual complete shutdown of coastal upwelling has the reverse effect (Pearcy et al. 1985, Miller et al. 1985). The hypothesis that sea surface or bottom tempera- tures (as inferred from sea level data) are controlling shrimp growth will most likely be tested over the next decade or two. In time, the relatively warm ocean con- ditions experienced off the Oregon coast since 1978 will probably be replaced by a colder, upwelling-dominated regime, similar to the early 1970s. The mean increase in length we have measured is equivalent to a 28% in- crease in average weight at age. If sea-bottom tem- peratures are controlling growth and return to lower levels, the drop in fishery yield will be profound. Con- versely, it is unlikely that the shrimp fishery will be substantially reduced, allowing standing stocks of shrimp to rebuild to near virgin levels of the early 1970s. Thus, as we see the subsequent trend in mean carapace length at age of pink shrimp, our hypothesis of density-dependent growth will be tested further. Charnov et al. (1981) showed that reductions in the population of age-2 and older shrimp (predominantly female) should result in increases in primary females. If Charnov is correct, the trawl fishery, through selec- tive removal of older shrimp, should be causing this effect in the population. The result of accelerated sex change should be higher levels of primary females and a roughly stable sex ratio. Jensen (1965) and Charnov (1980) noted increased levels of young females in Pandalus borealis populations after intensive fishery development. We question what these changes in population struc- ture imply for the future productivity of the pink shrimp resource. The evidence for density-dependent growth argues for a harvest-resistant shrimp stock. Our data also support the hypothesis of Charnov et al. (1978) that the population age structure determines the age of sex change in shrimp. As a consequence of in- tensive harvest, the age structure has shifted toward younger shrimp. The percentage of primary females has increased, however, resulting in the maintenance of a sexually balanced breeding population. The capa- city to accelerate sex change in pink shrimp also in- creases the stock's ability to withstand harvest pres- sure, by decreasing the potential for declines in larval production. Both catch and effort levels in the pink shrimp fishery are continuing to increase. The preliminary total catch for the states of California, Oregon, and Washington is nearly 36,000 mt in 1989. The large harvests in 1987-89 (Fig. 11) appear to be the result of a combina- tion of factors. Improved CPUE in 1987 and 1988 (Fig. 2; 1989 data unavailable) indicate some strong year classes of shrimp moving through the fishery. The total harvest levels of age-1 shrimp in these years is 50 Fishery Bulletin 89(1 1991 unprecedented in the history of the fishery. The in- creased size of age-1 shrimp since 1979 has made them more vulnerable to the gear and should have increased the harvest rate on age-1 shrimp relative to 1966-78. Fishing effort in the study area in 1987 and 1988 reached the two highest totals ever recorded, with 1989 likely to be as high or higher. The only other year in which effort approached levels of 1987-88 was in 1980 (Fig. 12). Postulating some improvement in yessel and gear efficiency in the years since 1980, the strong land- ings of 1987-89 must have been partly a result of record levels of effective fishing effort. In combination, the strong landings, high effort, and dominance of age 1 shrimp in the catch for 1987 and 1988 argue strong- ly for increased exploitation rates in those years. The large landings in 1989 are probably caused by the same factors. This raises the question of what impact the increas- ing harvest of age-1 shrimp may have on the spawn- ing population and subsequent recruitments. A spawner-recruit relationship has not been demon- strated for pink shrimp (Gotshall 1972). However, the Pacific Fishery Management Council (1981) identified some potential indicators of over-harvest of shrimp stocks. These included increases in the percentage of age-1 shrimp in the catch and in the percentage of primary females. In the past, reductions in age-2 and older shrimp were balanced by accelerated sex change in age-1 shrimp, and possibly by increased fecundity at age due to density-dependent growth. Levels of primary females have reached nearly 50% in some years. In such years, pink shrimp are virtually a single- age spawning stock. At some level of exploitation, ac- celerated sex change and density-dependent growth will not prevent declines in larval release and subse- quent recruitment. The strong year-classes passing through the fishery since 1986 indicate that we've prob- ably not reached that level of exploitation as yet. Acknowledgments This project was financed in part with Federal Inter- jurisdictional Fisheries Act funds through the U.S. Na- tional Marine Fisheries Service. Numerous individuals provided assistance in the completion of this project. P. Collier of the California Department of Fish and Game and M. Gross of the Washington Department of Fisheries provided unpublished data used in our anal- ysis. J. Robinson, J. Golden, R. Starr, and M. Saelens provided historical background information on the shrimp fishery. The draft manuscript was reviewed by R. Starr and also by D. Hankin of Humboldt State University. Assistance on statistical questions was pro- vided by P. Lawson. Citations Bakun A. 1973 Coast upwelling indices, west coast of North America 1946-71. NOAA Tech. Rep. NMFS SSRF-671, 103 p. Beardsley A. 1973 Design and evaluation of a sampler for measuring the near-bottom vertical distribution of pink shrimp (Pandalus jordani). Fish. Bull., U.S. 71:243-253. Bailey, R.S., and S.M. Almatar 1989 Variation in the fecundity and egg weight of herring (Clupea hareng-us L.). Part II. Implications for hypotheses on the stability of marine fish populations. J. Cons. Cons. Int. Explor. Mer 45:125-130. Butler, T.H. 1964 Growth, reproduction, and distribution of pandalid shrimps in British Columbia. J. Fish. Res. Board Can. 21: 1403-1452. Charnov, E.L. 1981 Sex reversal in Pandalus borealis: Effect of a shrimp fishery? Mar. Biol. Lett. 2 (1981):53-57. Charnov, E.L., D.W. Gotshall, and J.G. Robinson 1978 Sex ratio: Adaptive response to population fluctuations in pandalid shrimp. Science (Wash. DC) 200:204-206. Dahlstrom, W.A. 1970 Synopsis of biological data on the ocean shrimp Pandalus prdayii, Rathbun, 1902. FAO Fish. Rep. 57(4):1377-1416. Fox, W.W. 1972 Dynamics of exploited pandalid shrimps and an evalua- tion of management models. Ph.D. diss., Univ. Wash., Seattle, 148 p. Gotshall, D.W. 1972 Population size, mortality rates, and growth rates of northern California ocean shrimp Pandalus jordani, 1965 through 1968. Calif. Dep. Fish Game, Fish Bull. 155:1-47. Jensen, A.J. 1965 Pandalus borealis in the Skagerak (length, growth and changes in the stock and fishery yield). Rapp. P.-V. Reun. Cons. Int. Explor. Mer 156:109-111. Kruse, G.H. 1981 Relationship between shelf temperatures, coastal sea level, the coastal upwelling index, and English sole (Parophrys vetulus) spawning activity off Oregon. M.S. thesis, Oregon State Univ.. Corvallis, 68 p. Miller, C.B., H.P. Batchelder, R.D. Brodeur, and W.G. Pearcy 1985 Responses of the zooplankton and ichthyoplankton off Oregon to the El Nino event of 1983. In Wooster, W.S., and D.L. Fluharty (eds.), El Nino north: Nino effects in the eastern subarctic Pacific ocean, p. 184-187. Wash. Sea Grant Prog., Univ. Wash., Seattle. Pacific Fishery Management Council 1981 Discussion draft fishery management plan for the pink shrimp fishery off Washington, Oregon, and California. Pac. Fish. Manage. Counc, Portland. 169 p. Pearcy, W.G., J. Fisher, R. Brodeur, and S. Johnson 1985 Effects of the 1983 El Nino on coastal nekton off Oregon and Washington. In Wooster. W.S., and D.L. Fluharty (eds.), El Nino north: Nino effects in the eastern subarctic Pacific ocean, p. 188-204. Wash. Sea Grant Prog., Univ. Wash., Seattle. Pittock, H.L., W.E. Gilbert, A. Hyer, and R.L. Smith 1982 Observations of sea level, wind, and atmospheric pressure at Newport, Oregon, 1967-1980. Natl. Sci. Found. Data Rep. 98 Ref. 82-12, Oreg. State Univ., Corvallis, 158 p. Hannah and Jones: Fishery-induced population changes in Pandalusjordani 51 Ricker, W.E. ]975 Computation and interpretation of biological statistics of fish populations. Fish. Res. Board Can. Bull. 191, 322 p. Rothlisberg. P.C. 1975 Larval ecology of Pandalus jordani Rathbun. Ph.D. diss., Oreg. State Univ., Corvallis, 117 p. Saelens, M.R., and M.H. Zirges 1985 The 1984 Oregon shrimp fishery. Inf. Rep. 85-6, Oregon Dep. Fish Wildl., Newport, 29 p. Tegelberg, H.C., and J.M. Smith 1957 Observations on the distribution and biology of the pink shrimp (Pandalus jordani) off the Washington coast. Wash. Dep. Fish. Res. Pap. 2(l):25-34. Zirges, M.H., and J.G. Robinson 1980 The Oregon pink shrimp fishery, management history and research activities. Inf. Rep. Ser. Fish. 80-1, Oregon Dep. Fish Wildl., Newport, 15 p. Zirges, M.H.. M.R. Saelens, and J.E. McCrae 1981 Length-frequency, size, sex, and age composition data by month and area for pink shrimp landed in Oregon 1966 to 1980. Inf. Rep. Ser. Fish 81-2, Oregon Dep. Fish Wildl., Newport, 348 p. 1982 Total catch and effort tables, and summary biological statistics for pink shrimp caught in Oregon statistical areas 18-28 by month and area 1966-1981; catch numbers expanded by age and sex, effort in hours by vessel type. Inf. Rep. Ser. Fish. 82-4, Oregon Dep. Fish Wildl., Newport, 145 p. Abstract.- At the Alaska Fish- eries Science Center, one in five age readings produced for routine stock assessments are re-aged indepen- dently by a second age-reader. The Center now has a large database of repeated age readings that covers a variety of groundfish species and years. The purpose of this paper is to point out the problems and utility of interpreting such a database. The main problem of interpretation is fundamental, and relates to the fact that the true age of a fish is seldom known. Nevertheless, from a prag- matic point of view, these data can still provide useful insights into the age-determination process. Data from six marine fish species are used to show the overall levels of between- reader bias, agreement, and variabil- ity that have occurred on production age readings. Other uses for these data include objectively ranking the relative difficulty in ageing different species, maintaining quality control, examining between-reader differ- ences in ageing criteria, and evalu- ating the possible importance of between-reader bias and variability in later analysis and modeling ap- plications. Assuming reader bias is negligible, modeling results pre- sented here indicate that estimated percentage agreements are consis- tent with the hypothesis that age determinations are normally distrib- uted with a constant coefficient of variation over relatively wide age ranges. This result supports use of the coefficient of variation for mea- suring variability in age precision studies. Between-Reader Bias and Variability \n the Age -Determination Process Daniel K. Kimura Julaine J. Lyons Alaska Fisheries Science Center. National Marine Fisheries Service, NOAA 7600 Sand Point Way NE, Seattle. Washington 981 15-0070 Manuscript accepted 6 August 1990. Fishery Bulletin, U.S. 89:53-60 (1991). In this paper we evaluate a unique database developed for all marine fish species being routinely aged at the Alaska Fisheries Science Center. Here, large subsamples (one in five age readings produced for routine stock assessments) have been re-aged independently by a second experi- enced age-reader, mainly for the pur- pose of maintaining quality control. However, it became apparent that this database could be used to provide additional insights into the age-deter- mination process. Most everyone familiar with the ageing of fish knows this process is fraught with difficulties. At the very least, there must be random variabil- ity about some true age. Most likely there is also bias in the ageing meth- odology at some ages, as well as between-reader differences. Because reader bias is probably affected by the individual reader, the true age of the fish being aged, and perhaps even individual fish, the analysis of re- peated age readings made by differ- ent readers does not easily fall under the purview of classical statistical theory. The types of analysis that can be performed on age-determination data are dependent on the kind of data col- lected and the assumptions the data analyst is willing to make. For exam- ple, if replicated readings are made by each reader, it is possible to per- form a variance components analysis, assuming that reader effects are ran- dom and unbiased (Kimura et al. 1979). Comparative calibration is the area of statistical analysis that com- pares different methods of measure- ment (e.g., different readers) where all methods of measurement are as- sumed to contain error, and perhaps bias (Theobald and Mallinson 1978). Recently Kimura (unpubl.) examined the limits of possible inference for the functional comparative calibration model. In the analyses presented here we examine between-reader bias and variability based on subsamples aged independently by two age-readers. For these types of data, we define between-reader bias as the average difference (a^ - a 2 ) in ages assigned by these readers when ageing the same specimens of the same nominal age. Thus between-reader bias pre- sumably arises from the two readers using different ageing criteria. If the average difference between age- readers is negligible at some nominal age, then between-reader bias at that age is defined to be negligible, re- gardless of what the unknown abso- lute bias of the readers might be. Estimates of between-reader age- ing variability from these types of data can be computed by averaging the sample variances calculated from the two age readings (df = 1) from each age structure over some nom- inal age. These sample variances (between-reader variances) probably overestimate measurement error, because they include a component of variability that might be thought of as between-reader bias. Age determination is a statistical process that has a characteristic level of variability. This variability is spe- cies-dependent, and provides a basis for comparing the ageing of different 53 54 Fishery Bulletin 89(1), 199! species. For example, a species that can be aged with a larger percentage agreement (percentage of speci- mens aged the same on two occasions by the same or different reader), or smaller coefficient of variation, provides a statistical confirmation of the statement that species "x" is easier to age than species "y." Between-reader bias provides a measure of the ade- quacy of criteria for distinguishing ages in a particular species of some nominal age. Presumably, if there is no between-reader bias, ageing criteria are being ap- plied similarly by both readers, and the data only con- tain random measurement error. If between-reader bias and measurement error are independent, at this point between-reader variance would also be mini- mized. Significant between-reader bias may indicate a lack of resolving power in the criteria, insufficient training, or even peculiarities in the structures being aged. Between-reader variance is generally an indi- cator of overall "ageability," but is not as effective as between-reader bias measurements for pointing out between-reader differences in criteria. Species often have a characteristic age above which between-reader biases become larger. This age may be interpreted as a line distinguishing which ages are more reliable. For age-readers themselves, between-reader bias is usually of more interest than variability. This is because while measurement error is inherent in the age-determination process, between-reader bias can be controlled to a greater extent. In age-determination studies the term "precision" is used to describe "agreement," or variability between readings of the same specimen by the same or different age-reader. The term "accuracy" is reserved to de- scribe a comparison of ages generated by readers with the "true" age for specimens of known age. By emphasizing the importance of between-reader bias and variability, we do not mean to denigrate the obvious importance accuracy and age validation play in the age-determination process (Beamish and McFar- lane 1983, 1987). Validation (the comparison of ages determined by counting rings on hard parts with known ages) can be carried out in a variety of ways, all of which are difficult. These include combining an ex- ternal tag with an oxytetracycline (OTC) injection that labels calcium rings with a mark visible under ultra- violet light, following unusually strong year-classes through time, ageing young fish of known ages, and, most recently, measuring the activity of naturally occurring radioisotopes. Scientists at the Pacific Bio- logical Station (Beamish et al. 1983, Cass and Beamish 1983, Leaman and Nagtegaal 1987, McFarlane and Beamish 1987) have made wide use of the OTC mark. And, recently, two studies appear to have succeeded in validating longevity in rockfish using radioisotopes (Bennett et al. 1982, Campana et al. 1990). Typically, validation can be carried out on only a very few fish. Often, doubts remain concerning criteria for certain age groups, or structures that look different. Never- theless, the validation process is a critical one, and age- readers must constantly strive to improve the accuracy of their age determinations. Because we seldom knew the true age of a fish, ab- solute bias and total mean-square error in the ageing process were not known. Therefore, our discussions here will be limited to between-reader bias and vari- ability. These quantities are defined by the between- reader bias and coefficient-of-variation formulas described in the following section. Materials and methods The Ageing Unit at the Alaska Fisheries Science Center has the broad responsibility of ageing commer- cially important fish species and fish stocks in U.S. waters from California to the eastern Bering Sea. Historically, data have been accumulated from three principal sources: scientific surveys using various fishing gear, and foreign and domestic vessels fishing in U.S. waters. The present data consist of ages read from the otoliths (ear bones) of various groundfish species collected using assorted gear. Since 1981, the preferred method of reading ages from these structures has been to either break or saw the otolith cross-wise, burn the exposed surface, and read the cross-section under a microscope (Chilton and Beamish 1982). Only young or unusually clear speci- mens of select species can be read from the intact surface. In 1983 a quality-control program was initiated wherein 20% of all routine age readings would be in- dependently re-aged by an age-reader (i.e., the tester) particularly experienced in a species. Statistics were calculated on these reader/tester data (one reading per otolith from each age-reader) in the following manner: 1 mean (x) = (tester + reader)/2. 2 standard deviation (SD) = v^[(tester - i) 2 + (reader - x) 2 ] 3 nominal age (age): x (rounded), or tester age 4 n (count): sample size (number of specimens aged) 5 percentage agreement: (n agree/n) x 100 6 coefficient of variation (CV) = (SD/i) x 100 7 between-reader bias: reader age - tester age 8 percentage bias: [(reader age - tester age)/x] x 100 Elements 5-8 were averaged over the "n" specimens of the same nominal age, and over all ages (weighted by n) for overall statistics. Kimura and Lyons: Between-reader bias and variability in age-determinations 55 Table 1 Statistics comparing reader/tester data for Pacific whiting in 1986. Bias is oetween-reader bias; nominal age is mean. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias 2 258 100.0 0.0 0.00 0.0 3 58 50.0 15.1 -0.14 -5.3 4 8 62.5 7.6 -0.38 -10.7 5 13 92.3 3.6 0.23 5.1 6 651 95.5 0.6 -0.00 -0.2 7 118 20.3 9.9 0.23 3.4 8 66 37.9 11.9 -0.23 -2.9 9 153 77.8 2.4 -0.02 -0.2 10 28 10.7 12.1 0.25 2.4 11 23 26.1 12.0 0.00 -0.2 12 13 0.0 12.0 -0.92 -7.9 13 29 75.9 1.4 0.03 0.3 14 11 0.0 6.1 0.45 3.3 15 4 25.0 13.1 0.25 1.5 16 2 100.0 0.0 0.00 0.0 Average 78.6 3.2 When the main purpose of an analysis is to compare criteria of the age readers, the nominal age for classi- fication should be the tester age. When the overall char- acteristics of ageing a species is of interest, perhaps x is the better nominal age. The estimated statis- tics by age look different depending on which nominal age is being used. For example, if x is used, the per- centage agreement for younger ages will appear larger. Both the average percent error (Beamish and Four- nier 1981) and the coefficient of variation have been proposed as an "age independent" method of estimat- ing precision for age-determination studies. Assuming normality, Chang (1982) favored the coefficient of variation on the basis of efficiency, and we favor the coefficient of variation on the basis of common usage. Under differing distributional assumptions, the aver- age percent error may actually be superior. If age determinations are independently and normal- ly distributed about some true age, then the percent- age agreement at each age can be predicted from the area under the normal curve. Suppose the age of an "a"-y ear-old fish can be determined with a certain coef- ficient of variation. The difference between two in- dependent age determinations (b and c, say), would be distributed as z = b - c ~A7[0,2CV 2 a 2 ]. The predicted percentage agreement (ppa) at age "a" is then ppa = [(z 2 ) - cp( Zl )] x 100, Table 2 Statistics comparing reader/tester data for yellowfin sole in 1986. Bias is between-reader bias; nominal age is mean. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias 3 1 100.0 0.0 0.00 0.0 4 4 50.0 10.1 0.50 14.3 5 11 45.5 9.7 0.64 13.7 6 15 86.7 1.7 0.13 2.4 7 64 79.7 2.2 0.14 2.2 8 32 68.8 2.9 0.06 0.8 9 23 78.3 1.8 -0.13 -1.5 10 39 66.7 2.7 0.21 2.1 11 19 57.9 3.2 -0.16 -1.5 12 32 62.5 3.0 0.25 2.1 13 19 52.6 3.3 -0.05 -0.4 14 14 42.9 6.9 0.79 5.7 15 22 68.2 2.2 0.45 3.1 16 26 34.6 3.6 0.04 0.2 17 12 25.0 5.3 0.08 0.5 18 16 43.8 3.0 0.38 2.2 19 12 58.3 1.9 0.33 1.8 20 5 0.0 7.1 0.00 0.0 23 3 0.0 5.1 -1.00 -4.3 25 1 0.0 2.9 1.00 4.1 26 1 0.0 2.8 1.00 3.9 Average 60.9 3.2 where = the cumulative distribution function of the unit normal distribution, Zl = -0.5/[(1.4142)(CV)(a)] and, z 2 = +0.5/[(1.4142)(CV)(a)]. Results and discussion The above statistics were calculated for several species sampled in 1986 and subsequently aged (Tables 1-6). These are overall statistics calculated using data from all readers and testers, and therefore represent group rather than individual performance. A summary of percentage agreements and coefficients of variation averaged over all ages is given below: Percentage Species agreement CV Pacific whiting 78.6 0.032 Merluccius productus yellowfin sole Limanda aspera 60.9 0.032 Pacific ocean perch 40.8 0.049 Sebastes alutus walleye pollock 63.8 0.050 Theragra chalcogramma Atka mackerel 66.8 0.068 Pleurogrammus monopterygius sablefish Anoplopoma fimbria 43.7 0.129 56 Fishery Bulletin 89(1), 1991 Table 3 Statistics comparing reader/tester data for Pacific ocean perch in 1986. Bias is between-reader bias; nominal age is mean. Age Percentage cv Bias Percentage Age Percentage cv Bias Percentage (yr) Count agreement (%) (yr) bias (yr) Count agreement (%) (yr) bias 2 1 100.0 0.0 0.00 0.0 34 2 0.0 12.7 1.00 3.0 3 7 71.4 8.1 0.29 11.4 36 1 0.0 3.9 -2.00 -5.6 4 1 0.0 20.2 1.00 28.6 37 2 0.0 1.9 1.00 2.7 5 34 73.5 4.2 -0.03 -0.7 38 3 66.7 0.6 0.33 0.9 6 29 51.7 6.2 -0.28 -■-5.0 39 5 0.0 8.0 0.80 2.0 7 9 33.3 10.7 -0.78 -11.7 40 2 0.0 2.7 -1.50 -3.8 8 62 71.0 2.9 0.02 0.2 41 8 12.5 2.2 -0.50 -1.2 9 42 45.2 4.7 0.05 0.5 42 6 50.0 1.7 0.00 0.0 10 41 36.6 5.8 0.20 2.1 43 4 25.0 3.7 0.75 1.7 11 12 25.0 7.7 0.33 3.1 44 1 0.0 4.9 3.00 6.9 12 21 38.1 5.2 0.19 1.6 45 3 66.7 1.0 0.67 1.5 13 6 16.7 7.3 -0.67 -5.2 46 3 0.0 2.1 -0.67 -1.4 14 8 0.0 6.5 -1.00 -7.3 47 1 0.0 3.0 2.00 4.3 15 9 33.3 5.3 0.44 3.0 48 1 100.0 0.0 0.00 0.0 16 14 21.4 4.8 0.50 3.2 49 1 0.0 1.5 1.00 2.1 17 5 20.0 5.0 0.40 2.4 50 2 0.0 1.4 0.00 0.0 18 7 14.3 6.8 0.00 -0.0 52 5 0.0 7.4 4.20 8.1 19 8 37.5 7.0 1.13 5.9 54 2 0.0 2.6 -1.00 -1.9 20 7 28.6 4.6 -0.71 -3.7 55 2 50.0 3.2 -2.50 -4.6 21 1 0.0 10.3 3.00 14.6 56 1 0.0 3.8 3.00 5.4 22 1 100.0 0.0 0.00 0.0 57 2 0.0 7.4 -6.00 -10.5 23 2 0.0 6.1 0.00 0.0 59 3 0.0 6.0 -3.00 -5.1 24 2 50.0 1.5 -0.50 -2.1 60 1 0.0 2.4 -2.00 -3.3 25 4 25.0 4.3 -1.50 -6.1 61 5 0.0 2.8 -1.60 -2.6 26 2 0.0 2.8 0.00 0.0 65 2 50.0 1.6 -1.50 -2.3 27 3 33.3 3.6 1.33 5.0 66 1 0.0 1.1 -1.00 -1.5 28 2 100.0 0.0 0.00 0.0 68 1 0.0 3.1 3.00 4.4 29 2 0.0 17.1 3.00 10.3 72 1 0.0 1.0 1.00 1.4 30 1 0.0 4.7 2.00 6.7 73 1 0.0 8.8 9.00 12.4 31 2 0.0 4.6 0.00 0.0 75 1 100.0 0.0 0.00 0.0 32 2 0.0 4.4 0.00 0.0 78 1 0.0 3.6 -4.00 -5.1 33 1 100.0 0.0 0.00 0.0 Average 40.8 4.9 Table 4 Statistics comparing reader/tester data for pollock in 1986. Bias is between-reader bias; nominal age is mean. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias 1 18 100.0 0.0 0.00 0.0 2 64 93.8 2.9 0.03 2.1 3 132 92.4 2.1 0.00 0.0 4 159 74.8 5.3 0.04 1.0 5 136 66.2 5.6 0.07 1.5 6 119 64.7 5.2 0.06 1.1 7 113 49.6 6.5 -0.19 -2.7 8 181 56.9 4.7 -0.19 -2.5 9 85 21.2 7.1 0.11 1.2 10 26 15.4 10.7 -0.08 -0.8 11 17 41.2 7.4 0.41 3.8 12 8 25.0 5.3 -0.38 -3.2 13 2 0.0 8.3 -0.50 -3.7 Average 63.8 5.0 Table 5 Statistics comparing reader/tester data for Atka mackerel in 1986. Bias is between-reader bias; nominal age is mean. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias 1 15 100.0 0.0 0.00 0.0 2 76 97.4 1.2 0.00 0.0 3 47 55.3 14.2 0.02 0.9 4 28 64.3 10.3 -0.11 -2.8 5 29 48.3 10.1 0.03 1.1 6 28 50.0 6.8 0.11 1.8 7 20 65.0 5.4 -0.40 -6.0 8 35 40.0 7.2 -0.26 -3.4 9 4 25.0 8.1 0.00 0.3 10 4 50.0 3.7 -0.50 -5.3 Average 66.8 6.8 Kimura and Lyons: Between-reader bias and variability in age-determinations 57 Table 6 Statist ics comparing reader/tester data for sablefish in 1986. Bias is between-reader bias; nominal age is mean. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias 1 13 100.0 0.0 0.00 0.0 2 43 88.4 5.5 0.02 1.6 3 43 58.1 13.4 -0.21 -7.8 4 24 58.3 10.1 -0.25 -7.1 5 49 53.1 10.2 0.51 10.8 6 50 34.0 12.7 0.82 14.4 7 29 37.9 12.1 0.79 11.7 8 17 11.8 15.0 1.18 15.2 9 21 0.0 20.6 2.33 26.9 10 11 0.0 24.2 3.00 30.3 11 11 9.1 17.4 1.73 16.0 12 12 8.3 24.1 3.83 32.7 13 5 0.0 29.9 5.40 42.3 14 4 0.0 36.7 4.50 33.3 15 1 0.0 4.9 1.00 6.9 16 1 0.0 31.9 7.00 45.2 18 2 0.0 8.1 1.00 5.7 19 2 0.0 35.5 9.50 50.2 23 1 0.0 15.7 -5.00 -22.2 29 2 50.0 1.2 0.50 1.8 Average 43.7 12.9 Because percentage agreement decreases with the age of fish (Tables 1-6), and age distributions vary greatly among different species and among samples of the same species, percentage agreement lends itself only to age-specific comparisons. This is illustrated above by Pacific ocean perch and sablefish which show a similar percentage agreement; however, Pacific ocean perch is much more "ageable" than sablefish, as re- flected in the corresponding coefficients of variation. Although percentage agreement and coefficient of variation both reflect the relative difficulty of ageing each species, only the coefficient of variation adjusts for the absolute age of the fish. Therefore, one might conclude that the easiest species to age are Pacific whiting (Table 1) and yellowfin sole (Table 2); and the medium-difficult species are Pacific ocean perch (Table 3) and walleye pollock (Table 4). The most difficult species to age— species with un- resolved criteria, or species for which readers needed further training— were Atka mackerel (Table 5) and sablefish (Table 6). In fact, in this study the age-reader for Atka mackerel was inexperienced with the species, and there were unresolved criteria for sablefish. The most important usage of reader/tester data is in maintaining quality control. Unlike the data pre- sented in Tables 1-6, for quality-control purposes we need to compare only one tester with one reader, with Table 7 Statistics comparing reader/tester data by individual readers for pollock in 1986. Reader A is less experienced than reader B. Bias is between-reader bias ; nominal age is tester age. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias Results for Reader A 1 21 85.7 6.7 0.14 9.5 2 46 87.0 4.1 0.09 2.9 3 95 80.0 4.2 0.16 4.3 4 94 72.3 4.9 0.02 -0.3 5 76 69.7 4.2 0.09 1.2 6 59 62.7 4.8 0.07 0.5 7 63 38.1 7.5 -0.17 -3.5 8 122 44.3 6.0 -0.18 -3.0 9 30 26.7 10.0 -0.30 -5.0 10 5 0.0 7.4 -1.00 -10.5 11 3 0.0 6.7 -1.00 -9.5 12 5 20.0 12.5 -1.80 -17.6 13 1 0.0 11.8 -2.00 -16.7 14 1 0.0 10.9 -2.00 -15.4 Average 61.0 5.5 Results for Reader B 2 11 100.0 0.0 0.00 0.0 3 30 86.7 3.2 0.00 -0.8 4 57 80.7 3.3 0.05 0.7 5 34 76.5 4.0 0.06 0.4 6 44 79.5 2.8 0.00 -0.4 7 35 80.0 2.0 0.09 1.0 8 54 74.1 2.5 -0.02 -0.5 9 29 24.1 7.1 -0.28 -3.8 10 8 50.0 5.3 0.00 -0.6 11 14 50.0 6.0 -0.57 -6.0 12 3 33.3 3.9 0.00 -0.2 Average 72.4 3.4 the nominal age being the tester age. For pollock (Table 7) there were significant between-reader biases at older ages in the case of inexperienced reader A. There were no such between-reader biases for the experienced reader B. To ensure data quality, these types of be- tween-reader biases are constantly reviewed and the samples partially re-aged, before the data are released for use. Sablefish is an especially difficult species to age (Table 8). There were so many problems with age deter- mination that we suspended ageing, reviewed criteria with other ageing labs, and re-aged several large sam- ples. It is probable that between-reader bias for this species can be reduced, but it is doubtful that the coef- ficient of variation for this species can be substantial- ly reduced. Nevertheless, the availability of reader/ tester data was useful in revealing problems. Also, data users deserve a quantitative presentation of variabil- ity in age determinations and may be able to use these 58 Fishery Bulletin 89(1), 1991 Table 8 Statistics comparing reader/tester data for individual Reader A ageing sablefish in 1986. Bias is between-reader bias; nominal age is tester age. Age Percentage CV Bias Percentage (yr) Count agreement (%) (yr) bias 1 16 81.3 8.8 0.19 12.5 2 47 80.9 7.3 0.19 4.6 3 40 62.5 10.9 -0.08 ~-6.6 4 41 34.1 14.7 0.68 12.5 5 62 41.9 14.1 0.82 10.0 6 40 42.5 12.2 1.05 12.7 7 29 37.9 11.9 1.28 14.6 8 23 8.7 21.6 2.83 24.2 9 15 0.0 21.4 2.87 23.7 10 6 0.0 14.4 1.67 13.3 11 8 12.5 15.6 2.88 17.2 12 6 16.7 13.3 1.17 6.4 14 1 0.0 4.9 1.00 6.9 16 2 0.0 19.2 -1.00 -9.9 17 1 0.0 11.5 3.00 16.2 18 1 0.0 4.0 -1.00 -5.7 25 1 0.0 15.7 -5.00 -22.2 28 1 0.0 2.5 1.00 3.5 29 1 100.0 0.0 0.00 0.0 Average 43.7 12.9 data for making decisions on aspects of their data analysis and modeling. We examined the question of consistency between the percentage agreement and coefficient of variation measures of variability when analyzing between-reader data. Earlier we showed that assuming a constant coef- ficient of variation, and the normal error model, the percentage agreement can be predicted for all nominal ages. By comparing these theoretical curves with estimated percentage agreements calculated from data (Tables 1-6), some confidence in the consistency of the two measures can be derived. However, this compari- son is crude due to the probable existence of between- reader biases that are not factored into the analysis. Four different values for the coefficients of variation were used to calculate theoretical percentage agree- ment curves (Fig. 1A). These curves were then com- pared with estimated percentage agreement values for yellowfin sole (CV = 0.032, Fig. IB), walleye pollock (CV = 0.050, Fig. 1C), and sablefish (CV = 0.129, Fig. ID). The percentage agreements for all three species ap- pear consistent with the hypothesis that the coefficient of variation is constant over a wide age range, although the percentage agreements for pollock are biased low. These results support averaging the coefficient of variation across age ranges, and generally support using the coefficient of variation for interpreting preci- sion data from age-determination studies. However, there is considerable variation in these data which makes our results somewhat tentative. An important factor that also affects the ageing process is the presence of a strong year-class. For example, if two adjacent year-classes have absolute strengths of 10 and 100 fish, a 10% imprecision of + 1 year will add 5 fish from the strong cohort to the weak one (a 50% change) but only one-half a fish from the weak year-class to the strong (a 0.5% change). The data on Pacific whiting (Table 1) show how this phe- nomenon can lead to poor percentage agreements for weaker year-classes. Users of age data are often concerned that after some age, say 9 years, for example, the dominant year- classes become spread over several ages. Since percent- age agreement by such an age has often decreased to 50% or less, it is expected that age distributions will be smoothed. The only reason the ages would not be smoothed is if the dominant year-class is being antici- pated by the age-reader. For example, if samples are 90% 10-year-olds, it would be difficult for the age- reader not to anticipate that age and between-reader agreement would be high. However, if say 9-, 10-, and 11-year- olds occur in equal numbers, the agreement would not be nearly as good. Two possible ways of handling this problem are evi- dent. A controversial method would be to assure that all age-readers are reasonably coached as to the prob- able occurrence of a strong year-class; the other is to group the older ages in any model analyzing these data (e.g., Deriso et al. 1989). Both approaches avoid ask- ing the age-reader to perform the impossible. Finally, interpretation and analysis of repeated read- ings given here assume that the repeated readings were statistically independent. In the present context, this simply means that each reader did not have infor- mation regarding the other reader's results. When repeated readings are not made on an independent basis, or are of inadequate sample size, the data will be difficult or impossible to interpret statistically. From such a database, it is impossible to make assertions regarding precision. Acknowledgments Mr. George Hirschhorn had the foresight to initiate the reader/tester quality control system. We thank the age readers of the Ageing Unit at the Alaska Fisheries Science Center who provided the data for this study. We also thank the Scientific Editor and three anon- ymous referees whose comments contributed significantly to this paper. Kimura and Lyons: Between-reader bias and variability in age-determinations 59 *-* YELLOWFIN SOLE CV = 0.032 10 15 20 AGE (YR) 10 15 20 AGE (YR) SABLEFISH CV = 0.129 15 20 AGE (YR) 10 15 20 AGE (YR) 25 ^0 Figure 1 Predicted percentage agreement curves assuming different coefficients of variation and normally distributed error (upper left); overlayed with estimated percentage agreements for yellowfin sole (upper right); overlayed with estimated percentage agreements for pollock (lower left); and overlayed with estimated percentage agreements for sablefish (lower right). Citations Beamish, R.J., and D.A. Fournier 1981 A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-983. Beamish, R.J., and G.A. McFarlane 1983 The forgotten requirement for age validation in fisheries biology. Trans. Am. Fish. Soc. 112:735-743. 1987 Current trends in age determination methodology. In Summerfelt, R.C., and G.E. Hall (eds.), Age and growth of fish, p. 15-42. Iowa State Univ. Press, Ames. Beamish, R.J., G.A. McFarlane, and D.E. Chilton 1983 Use of oxytetracycline and other methods to validate a method of age determination for sablefish (Anoplopoma fim- bria). In Proceedings of the international sablefish sym- posium, p. 95-116. Alaska Sea Grant Rep. 83-8, Univ. Alaska, Fairbanks. Bennett, J.T., G.W. Boehlert, and K.K. Turekian 1982 Confirmation of longevity in Sebastes diploproa (Pisces: Scorpaenidae) from Ra-226/Pb-210 measurements in otoliths. Mar. Biol. (Berl.) 71:209-215. Campana, S.E., K.C.T. Zwanenburg, and J.N. Smith 1990 210 Pb/~ 6 Ra determination of longevity in redfish. Can. J. Fish. Aquat. Sci. 47:163-165. Cass, A.J., and R.J. Beamish 1983 First evidence of validity of the fin-ray method of age 60 Fishery Bulletin 89(1), 1991 determination for marine fishes. N. Am. J. Fish. Manage. 3:182-188. Chang. W.Y.B. 1982 A statistical method for evaluating the reproducibility of age determination. Can. J. Fish. Aquat. Sci. 39:1208-1210. 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. Deriso, R.B., T.J. Quinn II, and P.R. Neal 1989 Further aspects of catch-age analysis with auxiliary information. In Beamish, R.J., and G.A. McFarlane (eds.), Effects of ocean variability on recruitment and an evaluation of parameters used in stock assessment, p. 127-135. Can. Spec. Publ. Fish. Aquat. Sci. 108. Kimura, D.K., R.R. Mandapat, and S.L. Oxford 1979 Method, validity, and variability in the age determina- tion of yellowtail rockfish (Sebast.es flaindus), using otoliths. J. Fish. Res. Board Can. 36:377-383. Leaman. B.M., and D.A. Nagtegaal 1987 Age validation and revised natural mortality rate for yellowtail rockfish. Trans. Am. Fish. Soc. 116:171-175. McFarlane, G.A., and R.J. Beamish 1987 Validation of the dorsal spine method of age determina- tion for spiny dogfish. In Summerfelt, R.C., and G.E. Hall (eds.), Age and growth of fish, p. 287-300. Iowa State Univ. Press, Ames. Theobald, CM., and J.R. Mallinson 1978 Comparative calibration, linear structural relationships and congeneric measurements. Biometrics 34:39-45. Abstract.- Early juvenile (Stages V-IX) American lobsters Homarus arnericanus were fed diets of meso- plankton in filtered seawater, meso/ microplankton combination in fil- tered seawater, and frozen brine shrimp in both filtered and unfiltered seawater to determine if mesoplank- ton diets could sustain survival and growth throughout most of the first year of molts and if smaller zooplank- ters and phytoplankton in the meso/ microplankton diet could be utilized as food and could sustain survival in periods of low food supply. At the beginning of the experiment, there were no significant differences in either carapace length or weight be- tween the groups of sibling lobsters. Lobsters fed mesoplankton had high survival (80%) and significant in- creases in both carapace length and weight, although they weighed less at Stage IX than those fed frozen brine shrimp in unfiltered seawater. Lobsters fed frozen brine shrimp in filtered seawater had low survival (15%), but did not differ significant- ly at Stage IX from those fed meso- plankton in terms of both carapace length and weight. Lobsters fed brine shrimp in unfiltered seawater had high survival rates (95%) and weighed nearly twice as much at Stage IX than both the brine shrimp- fed lobsters in filtered seawater and the mesoplankton-fed lobsters; how- ever, none of these three surviving groups differed significantly in cara- pace length at Stage IX. Intermolt periods for the three surviving groups were not significantly different until the molt between Stage VIII and IX when the mesoplankton-fed lobsters took nearly twice as long to molt as either of the brine shrimp-fed groups. Lobsters fed meso/microplankton did not molt out of Stage V and died within 36 days of the 107-day experi- ment. These results indicate that mesoplankton diets promote growth and survival of lobsters throughout most of their first season of molting and that larger planktonic organisms may contain essential nutritional re- quirements not met by brine shrimp alone. However, the meso/microplank- ton diet, consisting mostly of dia- toms, does not provide sufficient nutrition for survival during periods of starvation. Manuscript accepted 6 August 1990. Fishery Bulletin, U.S. 89:61-68 (1991). Survival and Growth of Early-Juvenile American Lobsters Homarus americanus Through Their First Season While Fed Diets of Mesoplankton, Microplankton, and Frozen Brine Shrimp Kari L. Lavalli Boston University Marine Program, Marine Biological Laboratory U/oods Hole, Massachusetts 02543 Little is known of the natural forag- ing activities of the settled postlarvae (Stage IV) and early-juvenile (<1 year-old) stages of the American lob- ster Homarus americanus, presum- ably due to the inability of past inves- tigators to locate them in the benthic environment. Recently, Barshaw and Bryant-Rich (1988) examined the be- havior of the early-juvenile American lobster in naturalistic settings in the laboratory and found that they spent a considerable amount of time pleo- pod fanning (15% of the time) and antennule flicking (15-40% of the time) at the entrance of their bur- rows. During their 8-month investi- gation, Barshaw and Bryant-Rich never observed an early-juvenile lob- ster leave its burrow; of the several instances where lobsters were seen feeding, they captured amphipods near the entrance of the burrow twice while other observations in- dicated that the lobsters were cap- turing planktonic organisms via self- generated currents which drew the organisms toward the burrow en- trance. Their observations are cor- roborated by field cage studies of Gregory Roach (Nova Scotia Dep. Fish., Halifax, N.S., Canada B3J 3C4, pers. commun., Nov. 1989) where he, too, never observed early- juvenile American lobsters leave their burrows during one year of observations. While little is known about the nat- ural diet of recently settled American lobsters, Cobb et al. (1983) observed presettlement Stage-IV American lobsters capturing crab megalopae and insects in the field. Stomach con- tent analyses indicate that the Stage- IV diet is similar to that of the larvae, consisting of copepods, decapod lar- vae, amphipods, algae, and diatoms (Williams 1907, Herrick 1911, Tem- pleman and Tibbo 1945). Although most laboratory investigations have used artificial feeds which wild early- juvenile lobsters would never en- counter, some studies have provided information on naturalistic diets. Emmel (1908) found that Stage-IV American lobsters were capable of surviving on planktonic organisms obtained from the water alone. The intermolt period for this group of lobsters was significantly longer than that for groups fed on beef, soft- shelled clam, lobster muscle, or shredded fish, but this result was probably due to differences in the overall amount of food available to the groups, as unequal weights of food were used. More recently, An- drea (1975), D'Agostino (1980), and Good et al. (1982) found that when amphipods were used as a food source, growth rates of larval, post- larval, and early-juvenile American lobsters improved significantly over brine shrimp diets (both live and 61 62 Fishery Bulletin 89(1), 1991 frozen) and artificially prepared compound diets. Daniel et al. (1985) demonstrated that Stage-IV and early-juvenile American lobsters were capable of surviving and growing on a frozen filtrate diet consisting of 99% barnacle larvae and 1% calanoid copepods; however, these filtrate-fed lobsters were significantly smaller (by 17%) than lobsters fed on frozen adult brine _. shrimp. Similarly, Barshaw (1989) found that Stage-IV American lobsters were also capable of surviving and growing through two molts on a diet of live, uniden- tified plankton (size 152-1000/um), al- though the plankton-fed lobsters were smaller and had a greater intermolt period from Stage V to VI than those fed on frozen brine shrimp. In all of the above studies, there were no differences in mor- tality between the different groups of fed lobsters. This study examined the survival and growth of early-juvenile (Stages V-IX) American lobsters fed on diets of meso- plankton (95-lOOO^m) and a meso/micro- plankton combination (25-95 ^m) while using frozen brine shrimp diets for reference. Studies with other crustaceans indicate that phytoplankton may be used as a supplement when zooplankton abun- dance is low and its presence may extend the period of survival over that observed for starved animals (McConaugha 1985). However, the nutritional value of phytoplankton is highly dependent on its content of essential fatty acids which can vary in response to temperature, dissolved nutrients, light, and age (Castell and Kean 1986). While American lobsters have not been classified as algal feeders (Lebour 1922), stomach content analyses of the larvae and postlarvae indicate that diatoms and other algae form part of their diet (Herrick 1895, Williams 1907, Herrick 1911). Recently, Lavalli and Barshaw (1989) have shown that Stage-IV and -V American lobsters are capable of removing particles from the water column to at least a size of 70^m, indicating that early- juvenile lobsters may be able to utilize small organ- isms in the mesoplankton and microplankton. This study was designed, in particular, to determine two things: (1) Whether early juveniles could utilize an already-proven diet (mesoplankton) for Stage-IV and -V lobsters throughout much of their first season of molting activity, and (2) whether early juveniles could extend survival by utilizing the organisms found in the smaller range of mesoplankton and in the microplankton. Brine Shnmp/Unlillered Seawater Brine Shrimp/Filtered Seawater Mesoplankton D " Meso/microplankton Figure 1 Mean number of original 20 lobsters surviving through the 107 days of the experiment (molt stages V-IX) on each of the diet regimes: brine shrimp in unfiltered seawater, brine shrimp in filtered seawater, mesoplankton (95-1000 /jm) in filtered seawater, meso/microplankton combination (25-95 fim) in filtered seawater, and starved. Materials and methods Prior to the beginning of the experiment, Stage-IV American lobster siblings Homarus americanus were held collectively in a seawater table supplied with am- bient, unfiltered seawater and were fed ad libitum on frozen adult Artemia (San Francisco Bay type). Sib- lings were used for the experiment, since genetic dif- ferences between females can produce significant dif- ferences in weight among similarly raised juvenile lobsters (Conklin et al. 1975, Hedgecock and Nelson 1978). The lobsters were then randomly assigned to one of four groups of 20 animals: a mesoplankton-fed group (95-1000^m), a meso/microplankton combination-fed group (25-95 ^m), a frozen brine shrimp-fed group, and a starved group. Upon assignment, individual lobsters were placed into plastic trays (Rubbermaid Drawer Organizers, No. 2915) with dimensions 224 mm long x 75 mm wide x 50mm deep, and volume of ~750 mL. Each tray was modified to include a sidewall screen for water flow and a dark -grey PVC tube (10 mm diameter) glued to the bottom which could act as a shelter. The trays were provided with ambient seawater which was filtered with a dual-cartridge filtering system (a 50-^m honeycomb filter followed by a 5-^m nominal filter). They were arranged in a Latin square design to inter- sperse the treatments and were kept in darkness, ex- cept during cleaning and feeding periods, as previous investigations demonstrated that juvenile lobsters Lavalli: Survival and growth of Homarus amencanus fed plankton diets 63 grew more quickly and were more active in a nearly constant dark regime (Bordner and Conklin 1981). The water flow to the trays was turned off for 1 hour after the introduction of food to allow the lobsters to more easily capture the food. Filters were replaced during these feeding periods if they were clogged. Lobsters were fed according to group; excess food and other debris were removed daily with a kitchen baster. All trays were thoroughly scrubbed each week to remove algal growth from the sidewalls and bottom. During cleaning the lobsters were held in a moist, small-mesh fish net. Every attempt was made to feed equal wet weights of food, and representative portions of each diet were weighed each week. For the plankton diets, representative portions were also photographed using the technique of silhouette photography (Edger- ton 1977, Ortner et al. 1979) so that identification of the planktonic organisms could be made without the aid of a microscope. Plankton was collected three to four times per week by towing with a #10 plankton net (152 /mi) and a phy- toplankton net (25^m) in the Waquoit Bay/Nantucket Sound areas. After collection it was sieved to remove objects > lOOO^m and to divide the plankton into each size group. Half of the plankton was used immediately while the other half was refrigerated overnight and used the following day. Carapace lengths of Stage-V lobsters were measured to the nearest 0.1 mm using calipers, and their weights were recorded on a Brainweigh B300D scale to the nearest 0.001 g. The lobsters were blotted with absor- bent paper to remove excess water prior to weigh- ing. The experiment ran until all surviving lobsters at- tained Stage IX. During this time, the dates for all molts (for the determination of intermolt periods) and deaths were recorded. Although no post-mortems were performed, it was noted whether lobsters died in the process of molting or of unknown causes. Coloration of the lobsters was also noted. After achieving Stage IX, the lobsters were once again measured and weighed. During the time of this experiment, a fifth group of lobsters (also siblings of the other four groups of lobsters) was raised in seawater tables for another ex- periment. The lobsters in this fifth group were placed individually into separate circular containers (85 mm diameter; 200 mm high) consisting of a black plastic bottom glued to a cylinder made of screening (1-mm mesh). These lobsters were fed the same amount of brine shrimp as the brine shrimp group of lobsters above, but lived in unfiltered, ambient seawater and were subject to ambient daylight plus overhead fluores- cent lighting. Organic debris was cleaned out of the seawater table and containers at least once per month. While data on the initial (Stage V) weights and cara- pace lengths are unavailable for this fifth group of lobsters, their final (Stage IX) weight and carapace length were recorded. Intermolt periods were recorded except for the period between Stages V and VI, since this group was held communally until after they had molted into Stage V. Data for each of the measurements taken (intermolt period, initial (Stage V) and final (Stage IX) carapace lengths and weights) were analyzed using the Student's /-test when comparisons between two groups or two measurements within a group (i.e., initial and final weights or carapace lengths) were made, and by 1-way ANOVA tests when more than two groups were com- pared. Where ANOVA tests indicated significant dif- ferences were present, the groups were compared to determine which groups were different by using the Tukey test with unequal sample sizes. Differences in survival rates were tested with a 2 x 2 chi-square con- tingency table. This experiment was conducted at the Marine Biological Laboratory in Woods Hole, MA from 13 July to 27 October 1987. The ambient seawater temperature ranged from 23 to 14.5°C and averaged 19.6°C. Results Survival was high in the groups fed brine shrimp in unfiltered seawater (95% survival), brine shrimp in filtered seawater (95% survival), and mesoplankton (90% survival) for the molt between Stage V and VI. During the subsequent molts, however, the group fed brine shrimp in filtered seawater had significantly higher mortality Or, P<0.001; Fig. 1), with only 15% survival by the end of the experiment. The survival of the brine shrimp-fed group in unfiltered seawater re- mained unchanged, while that of the mesoplankton-fed group fell to 80% by the end of the experiment. How- ever, there was no significant difference in survival between these two groups. Of the deaths noted for each of the groups, one lobster fed brine shrimp in unfiltered seawater and one fed mesoplankton died during its molt; of the 17 lobsters which died on the brine shrimp diet in filtered seawater, 14 died while in the process of molting. Coloration of the surviving groups differed, with the brine shrimp-fed group in filtered seawater being pale blue, typical of brine shrimp-fed lobsters, and the mesoplankton-fed group and brine shrimp-fed group in unfiltered seawater being the wild-type coloration. None of the starved or meso/microplankton com- bination-fed lobsters molted beyond Stage V. All of the lobsters in these two groups died within 36 days of the 107-day experiment, and although the lobsters fed the meso/microplankton combination diet took slightly 64 Fishery Bulletin 89(1), 1991 longer to die (23.842 ± 10.035 (SD) days vs. 21.75 ± 7.063 days), this difference was not significant. Intermolt duration data (Fig. 2) showed that the brine shrimp-fed group in fil- tered seawater took significantly longer to molt (by 1 day) into Stage VI than the mesoplankton-fed group (10.412 ± 1.502 days vs. 11.389 ± 1.243 days; Student's r-test, P< 0.025). Data are not available on the intermolt period between Stages V and VI for the brine shrimp-fed group in unfiltered seawater. There was no significant difference between the groups brine shrimp-fed in filtered seawater, brine shrimp-fed in unfiltered seawater, and mesoplankton-fed for the intermolt periods between Stages VI and VII (14.857 ±2.035 vs. 13.444 ±1.653 vs. 14.059 ± 1.853 days) and Stages VII and VIII (22.0 ± 7.810 vs. 20.556 ± 4.681 vs. 20.529 ± 2.528 days). However, the in- termolt periods of both brine shrimp-fed groups were significantly different (18.0 ± 1 and 16.842 ± 2.292 days; 1-way ANOVA, P<0.001; Tukey test, P< 0.001) from those of the mesoplankton- fed group (36 ± 5.057 days) for the molt between Stages VIII and IX, with the two brine shrimp-fed groups taking nearly half the time of the mesoplank- ton-fed group to molt into Stage IX. There was no significant difference between any of the groups brine shrimp- fed in filtered seawater, mesoplankton- fed, meso/microplankton combination-fed, and starved lobsters at the beginning of the experiment in either weight (0.06 ± 0.011 vs. 0.066 ± 0.011 vs. 0.059 ± 0.009 vs. 0.061 ±0.011g, respectively; Fig. 3) or carapace length (4.66 ± 0.214 vs. 4.761 ± 0.214 vs. 4.739 ± 0.236 vs. 4.716 ± 0.236mm respectively; Fig. 4). Although measurements are not available for the brine shrimp- fed group in unfiltered seawater, they probably did not differ significantly from the other groups since they were maintained in conditions identical to their siblings until immediately before the molt to Stage V. Each of the surviving groups of lobsters fed brine shrimp in fil- tered seawater, brine shrimp in unfiltered seawater, and mesoplankton showed significant growth (Student's t- test, P<0.001) in terms of both increased weight and carapace length (Figs. 3 and 4). However, final (Stage IX) weights did differ between groups (1-way ANOVA, P< 0.001). The weight of the brine shrimp-fed group in unfiltered seawater (0.837 ±0.117 g) was significantly greater (Tukey test, P<0.001) than that of both the Brine Shrtmp/Unflltered Seawater Brine Shrimp/Filtered Seawater Mesoplankton CH Stage V to VI Stage VI to VII Stage VII to vni Stage VIII to LX MOLT PERIOD Figure 2 Mean intermolt durations for lobsters on each of three diet regimes: brine shrimp in unfiltered seawater, brine shrimp in filtered seawater, and mesoplankton (95-1000/jm) in filtered seawater. Bars indicate standard deviation values. ( a ) Stage VI molt date missed for one lobster, so intermolt period could not be determined for Stages V-VI and VI-VII for that lobster. ( b ) Stage VII molt date missed for one lobster, so intermolt period could not be determined for Stages VI-VII and VII-VIII for that lobster. ( c ) Stage IX molt date missed for one lobster, so intermolt period could not be determined for Stages VIII-IX for that lobster. brine shrimp-fed group in filtered seawater (0.484 ± 0.183g) and the mesoplankton-fed group (0.484 ± 0.037 g). However, there was no significant difference between the latter two groups. Final (Stage IX) carapace lengths did not differ between the three surviving groups (brine shrimp-fed in filtered seawater, 9.9 ± 1.353 mm; brine shrimp-fed in unfiltered seawater, 10.459 ± 0.564mm; mesoplankton-fed, 9.907 ± 0.732mm). There was no significant difference in the wet weights of each diet fed the lobsters. The average wet weights of the diets were 0.408 ± 0.095g for the meso- plankton; 0.364 ± 0.108g for the meso/microplankton combination diet; and 0.391 ± 0.072g for the brine shrimp diets. The mesoplankton diet consisted pre- dominantly of Acartia copepods, barnacle nauplii, pagurid shrimp zoea, invertebrate eggs, brachyuran crab zoea, foraminifera, centric and pennate diatoms, and marine algae, with occasional instances of ascidian tadpoles, barnacle exoskeletons, fish eggs and young, amphipods, hydroids, brachyuran crab prezoea, Lavalli: Survival and growth of Homarus amencanus fed plankton diets 65 Brine Shrlmp/Unfiltered Seawater Brine Shrimp/Filtered Seawater Mesoplankton Meso/microplankton Starved ~ * No Data Available **/•*• Died Before Molting 18 18 19 i * m — i i Initial Weights Final Weights WEIGHT MEASUREMENTS TAKEN Figure 3 Initial (Stage V) and final (Stage IX) mean wet weight measurements (g) for lobsters on the five diets: brine shrimp in unfiltered seawater, brine shrimp in filtered seawater, mesoplankton (95-1000 ^m) in filtered seawater, meso/microplankton combination (25-95jjm), and starved. Bars indicate standard deviation values. ( a ) Two lobsters missed in weighing schedule. ( h ) One lobster missed in weighing schedule. caridean shrimp zoea, Centropages and Calanns copepods, dinoflagellates, and juve- nile nemertea. The meso/microplankton combination diet typically consisted of cen- tric and pennate diatoms with occasional in- stances of fragments of marine algae and crustaceans. Discussion The results clearly indicate that early juve- nile American lobsters are not capable of extending survival on a diet consisting most- ly of diatoms, despite their common pres- ence in stomach content analyses (Herrick 1895, Williams 1907, Herrick 1911). Larger planktonic organisms are required for sur- vival and growth. This result is not entirely surprising even though Lavalli and Barshaw (1989) showed that post-larval and early juvenile (Stage V) American lobsters could remove particles from the water down to a size of at least 70^m. Other crustaceans fed on phytoplankton can gain some nutrients and extend their survival in periods of low Brine Shrimp/Unaltered Seawater Brine Shrimp/Filtered Seawater Mesoplankton Meso /microplankton Starved • No Data Available ••/»*• D,ed Before Molting 15b CARAPACE MEASUREMENTS TAKEN Figure 4 Initial (Stage V) and final (Stage IX) mean carapace length measurements (mm) for lobsters on the five diets: brine shrimp in unfiltered seawater, brine shrimp in fil- tered seawater, mesoplank- ton (95-lOOO^m) in filtered seawater, meso/microplank- ton combination, and starved. Bars indicate standard devia- tion values. ( a ) Two lobsters missed in carapace-length measuring schedule. ( b ) One lobster missed in carapace- length measuring schedule. 66 Fishery Bulletin 89|1), 1991 food abundance, but this type of diet does not support molting or growth (McConaugha 1985). Post-larval lobsters are known to contain diatoms and other algae in their guts (Herrick 1895, Williams 1907, Herrick 1911) which suggests some nutritional role for these items, but one not fully understood nor clarified by this experiment. The smaller planktonic organisms in the meso/microplankton combination diet may not have been present in sufficient numbers to make up for the small amount of nutrients derived. Because the meso/ microplankton diet consisted mostly of diatoms which have a high content of silicon-based ash, it is likely that this diet had a greater percentage of non-digestible fiber or bulk than that in the mesoplankton or brine shrimp diets (John Castell, Dep. Fish. & Oceans, Hali- fax, N.S., Canada B3J 2S7, pers. commun., May 1990). Furthermore, these smaller organisms may have been more easily flushed out of the containers when the water flow resumed. The results presented here also clearly support those of Barshaw (1989) and Daniel et al. (1985) in terms of postlarval and early-juvenile lobsters being capable of surviving on mesoplankton, and in demonstrating high survival among the brine shrimp-fed (in filtered sea- water) and mesoplankton-fed groups through Stage VI. These studies differ, however, in that Barshaw (1989) found molt delays in her plankton-fed group between Stages V and VI, whereas no molt delays were found in this study until Stage VIII. Barshaw's lobsters also took longer to molt into Stage VI (34 days for the plankton-fed lobsters and 23 days for the brine shrimp- fed lobsters) than did the lobsters in this experiment (10 and 11 days for the same groups), indicating that they were not receiving enough food and thus took longer to build up the reserves to molt. In addition, both Daniel et al. (1985) and Barshaw (1989) found that lobsters fed on frozen brine shrimp in filtered seawater were significantly larger than the filtrate-fed or plankton-fed lobsters. This study found no such dif- ference between the similarly treated groups. The differences between the two groups of lobsters fed on brine shrimp diets were striking. Lobsters fed brine shrimp in the filtered seawater had pale blue col- oration and poor survival, with the majority of deaths occurring during molting. However, this difference in survival was not present until after Stage VI where Barshaw's (1989) experiment ended. Similar drops in survival of brine shrimp-fed lobsters in filtered sea- water after Stage VI have been observed by Colleen Boggs (Edgerton Res. Lab. [in collaboration with the Kravitz Lab., Harvard Medical School], New England Aquarium, Boston 02110, pers. commun., summer 1990). Certain strains of brine shrimp promote better growth than others (McConaugha 1985), and the suc- cess of one strain versus another is linked to its fatty acid content (Fujita et al. 1980), the presence of which is extremely important for the survival of postlarval and early-juvenile American lobsters (D'Abramo et al. 1981). The San Francisco Bay brand used in this ex- periment is intermediate in lipid content (McConaugha 1985), but even different lots of the same strain of brine shrimp are known to be highly variable in quality (Eagles et al. 1984, 1986). Thus, whatever nutritional component was lacking in the lot of the brine shrimp used in this experiment was compensated by the planktonic organisms entering through the ambient water supply, since the brine shrimp-fed group of lobsters in unfiltered seawater showed high survival, a greater weight increase compared with those in filtered seawater, and wild-type coloration. What is particularly interesting, though, is that while the lob- sters fed brine shrimp in unfiltered seawater were nearly twice as heavy at Stage IX as both those fed brine shrimp in filtered seawater and mesoplankton, there was no significant difference at Stage IX between any of these groups in terms of carapace lengths. Weight, therefore, might be a more important index of growth in early-juvenile lobsters. The carapace lengths achieved by the three surviving groups of lobsters at Stage IX were shorter than those predicted by calcula- tions of Hudon (1987) from early juveniles captured in the field. This contradiction may have resulted from the lobsters used in this experiment being hatchery- and laboratory-reared and thus being typically smaller than wild lobsters at Stage V (pers. observ.). The difference in weights at Stage IX between lob- sters fed brine shrimp in unfiltered seawater and those fed mesoplankton indicates that growth (as well as sur- vival) might be significantly enhanced if the lobsters have access to both a planktonic diet and a diet of small benthic organisms. Andrea (1975) demonstrated that lobster larvae (Stages I-IV) fed frozen copepods or frozen amphipods had significantly higher survival rates than those fed frozen brine shrimp. Furthermore, those larvae fed live copepods had higher survival than those fed both live and frozen adult brine shrimp when held under the same rearing conditions. Andrea's data also showed that the increase in carapace length and the gain in weight by lobsters fed diets of live copepods were comparable to the increases found in lobsters fed live brine shrimp. Evidence to date indicates that early juveniles are found in shallow subtidal areas (Cooper and Uzmann 1980, Hudon 1987, Able et al. 1988, Wahle 1990) where they would have access to suprabenthic plankton and epiplankton (Wieser 1960, Cornet et al. 1983) as well as surface plankton that vertically migrate in response to light/dark conditions (Hardy 1970). They would also have access to the many benthic organisms found in subtidal areas (Orth 1973, Reise 1977). In support of Lavalli: Survival and growth of Homarus amencanus fed plankton diets 67 this hypothesis, postlarvae and early juveniles in labor- atory settings have been observed to lunge out of their burrows to grab at food (amphipods) passing by (Ber- rill 1974 with//, gammarus; Barshaw and Bryant-Rich 1988) or to stalk swimming amphipods (Good et al. 1982). Also, Crnkovic (1968) suggested that the crea- tion of new openings in existing Nephrons norvegicus burrows may be linked to searching for food within the sediment. The intermolt periods, with the exception of that for the mesoplankton-fed group between Stages VIII and IX, were consistent with or shorter than previous studies at the same average temperature (19°C) (Tem- pleman 1948, as reported in Wilder 1953) and were close to the values predicted by Hudon (1987) for the same stages. These results show that early-juvenile lobsters fed on mesoplankton are able to capture it ef- fectively enough to keep pace with the brine shrimp- fed lobsters in terms of intermolt periods until Stage VIII. At that time, the mesoplankton-fed lobsters spend nearly twice as much time in intermolt than either of the brine shrimp-fed groups. This result could be in- dicative of one of three conditions or some combina- tion of all of them: (1) Either the lobsters became less efficient at capturing the plankton, (2) the planktonic organisms were not present in sufficient numbers in this study to compete with a brine shrimp diet at later stages, or (3) dietary requirements change with later molt stages. In support of the first hypothesis is the fact that the claws of the postlarvae are small and symmetrical prior to Stage VIII. The claws slowly develop into the crusher and seizer claws during the early-juvenile stages; concomitant with this gradual development is a change in the posture of the lobster from one that is completely defensive (withdrawing or tail-flipping) to one that is more aggressive (Lang et al. 1977), and a change in the muscle fiber pattern and innervation of the two types of claws (Govind 1984). At Stage VIII the claw asymmetry is well established and the fiber composition and innervation are nearly the same as that found in the adult (Govind and Pearce 1986). These changes may indicate a shift in the feeding strategies used by the lobster, where capture of small benthic organisms becomes more important than the capture of planktonic organisms at or near Stage VIII. As for the second hypothesis, Bordner and Conklin (1981) determined that older juvenile lobsters could consume up to 10% of their body weight per day. Dur- ing this entire experiment, each group of lobsters was fed more than 10% of their body weight per day. There- fore, it seems unlikely that the later stages of lobsters were underfed on the mesoplankton diet. Finally, diet- ary requirements might indeed change as the lobster becomes more able to defend itself and thereby forage, and as the claws develop the ability to crush small molluscs; however, this experiment was not designed to answer such a question. In conclusion, the results from this experiment con- tradict those of Barshaw (1989) and Daniel et al. (1985) in that they show no difference in growth and survival of early-juvenile lobsters (Stages V and VI) fed on a diet of mesoplankton versus a diet of frozen brine shrimp in filtered seawater. Stage VI-VIII lobsters are able to survive and grow on planktonic diets, but after Stage VIII they experience molt delays when compared with lobsters fed frozen brine shrimp diets. Despite this delay, the mesoplankton diet allows the early juveniles the opportunity to reach the predicted (Hudon 1987) winter stages of Stage VI (for late-fall settlers) to IX or X (for August settlers) without the need for other benthic food. Diets composed of smaller members of the mesoplankton plus microplankton do not provide sufficient nutrition to support survival in periods of low food abundance. Acknowledgments I thank Mike Syslo and Kevin Johnson of the Massa- chusetts State Lobster Hatchery for supplying me with Stage IV lobsters, and Dr. Harold Edgerton for in- struction in the technique of silhouette photography. I also would like to thank Dr. Joe Costa (formerly of BUMP) for getting the boat up and running, and Paula Dolan (formerly of University of Tampa), Leslie Sam- mon (formerly of Mt. Holyoke College), and Lee Kefauver for their help in running the experiments and their willingness to go on plankton tows during the early hours of the morning in rain or shine. Drs. Jelle Atema, Stanley Cobb, Diana Barshaw, John Castell, and three anonymous reviewers provided helpful com- ments on the manuscript for which I am grateful. Citations Able, K.W., K.L. Heck, M.P. Fahay, and C.T. Roman 1988 Use of salt-marsh peat reefs by small juvenile lobsters on Cape Cod, Massachusetts. Estuaries 11:83-86. Andrea, J.J. 1975 The nutritional adequacy and acceptability of several natural prey species for larval Homarus americanus (Milne- Edwards) in culture. M.S. thesis, Mar. Sci. Res. Cent., State Univ. New York, Stony Brook, 53 p. Barshaw, D.E. 1989 Growth and survival of post-larval lobsters, Homarus americanus, on a diet of plankton. Fish. Bull., U.S. 87: 366-370. Barshaw, D.E., and D.R. Bryant-Rich 1988 Long-term survival and behavior of early juvenile lob- sters, Homarus americanus, in the three naturalistic sub- strates: Mud, rock, and eelgrass. Fish. Bull., U.S. 86:789-796. 68 Fishery Bulletin 89(1). 1991 Berrill, M. 1974 The burrowing behavior of newly-settled lobsters, Homa- rus vulgaris (Crustacea-Decapoda). J. Mar. Biol. Assoc. U.K. 54:797-801. Bordner, C.E., and D.E. Conklin 1981 Food consumption and growth of juvenile lobsters. Aquaculture 24:285-300. Castell, J.D., and J.C. Kean 1986 Evaluation of the role of nutrition in lobster recruit- ment. Can. J. Fish. Aquat. Sci. 43:2320-2327. Cobb, J.S., T. Gulbransen, B.F. Phillips, D. Wang, and M. Syslo 1983 Behavior and distribution of larval and early juvenile Homarus americanus. Can. J. Fish. Aquat. Sci. 40:2184-2188. Conklin, D.E., K. Devers, and R.A. Shleser 1975 Initial development of artificial diets for the lobster, Homarus americanus. Proc. World Maricult. Soc. 6:237-248. Cooper, R.A., and J.R. Uzmann 1980 Ecology of juvenile and adult Homarus. In Cobb, J.S., and B.F. Phillips (eds.), The biology and management of lobsters, vol. II, p. 97-142. Academic Press, NY. Cornet. M., J. Bouchet, J. Lissalde, J. Sorbe, and L. Amoureux 1983 Donnees qualitatives sur le benthos et le suprabenthos d'un transect du plateau continental sud-gascogne. Cah. Biol. Mar. 24:69-84 [in French, Engl, abstr.]. Crnkovic, D. 1968 Some observations regarding the burrows of juvenile Nephrops norvegicus (L.). Rapp. Comm. Int. Mer Medit. 19: 171-172. D'Abramo, L.R., C.E. Bordner, D.E. Conklin, and N.A. Baum 1981 Essentiality of dietary phosphatidycholine for the sur- vival of juvenile lobsters. J. Nutr. 111:425-431. D'Agostino, A. 1980 Growth and color of juvenile lobsters (Homarus ameri- canus) kept on diets of natural and artificial foodstuffs. In Proc, 1980 lobster nutrition workshop, p. 41-45. Maine Sea Grant Tech. Rep. 58, Univ. Maine, Orono. Daniel, P.C., R.C. Bayer, and S. Chapman 1985 Barnacle larvae (Balanus spp.) as a potential diet for juve- nile lobsters (Homarus americanus). Aquaculture 46:67-70. Eagles, M.D., D.E. Aiken, and S.L. Waddy 1984 Effect of food quality and feeding schedule on survival, growth and development of larval American lobsters fed frozen adult brine shrimp. J. World Maricult. Soc. 15:142-143. 1986 Influence of light and food on larval American lobsters, Homarus americanus. Can. J. Fish. Aquat. Sci. 43:2303-2310. Edgerton, H.E. 1977 Silhouette photography of small active subjects. J. Microsc. (Oxf.) 110:79-81. Emmel, V.E. 1908 The problem of feeding methods in lobster culture. Annu. Rep. R.I. Comm. Inland Fish. 38:98-114 [avail. NMFS Woods Hole Lab., Woods Hole, MA 02543]. Fujita, S., T. Watanabe, and C. Kitajima 1980 Nutritional quality of Artemia from different localities as a living feed for marine fish from the viewpoint of essen- tial fatty acids. In Persoone, L.G., P. Sorgeloos, D. Roels, and E. Jaspers (eds.), The brine shrimp Artemia, vol. 3, p. 277-290. Universal Press, Wettsen, Belgium. Good, L.K., R.C. Bayer, M.L. Gallagher, and J.H. Rittenberg 1982 Amphipods as a potential diet for juveniles of the Ameri- can lobster Homarus americanus (Milne Edwards). J. Shell- fish Res. 2:183-187. Govind, C.K. 1984 Development of asymmetry in the neuromuscular system of lobster claws. Biol. Bull. (Woods Hole) 167:94-119. Govind, C.K., and J. Pearce 1986 Differential reflex activity determines claw and closer muscle asymmetry in developing lobsters. Science (Wash. DC) 233:354-356. Hardy, A. 1970 The open sea: Its natural history. Part I: The world of plankton. Houghton Mifflin, Boston, 335 p. Hedgecock, D., and K. Nelson 1978 Components of growth rate variation among laboratory cultured lobsters (Homarus), p. 125-137. In Proc. Annu. Meet. World Maricult. Soc. 9. Herrick, F.H. 1895 The American lobster: A study of its habits and develop- ment. Bull. U.S. Fish. Comm. 15:1-252. 1911 Natural history of the American lobster. Bull. U.S. Fish. Comm. 29:147-408, 20 pis. Hudon, C. 1987 Ecology and growth of postlarval and juvenile lobster, Homarus americanus, off lies de la Madeleine (Quebec). Can J. Fish. Aquat. Sci. 44:1855-1869. Lang, F., C.K. Govind, W.J. Costello, and S.L Greene 1977 Developmental neuroethology: Changes in escape and defensive behavior during growth of the lobster. Science (Wash. DC) 197:682-685. Lavalli, K.L., and D.E. Barshaw 1989 Post-larval American lobsters (Homarus americanus) living in burrows may be suspension feeding. Mar. Behav. Physiol. 15:255-264. Lebour, M.V. 1922 The food of plankton organisms. J. Mar. Biol. Assoc. U.K. 12:644-677. McConaugha, J. 1985 Nutrition and larval growth. In Wenner, A.M. (ed.). Crustacean issues, vol. 2, p. 127-154. A. A. Balkema, Rotter- dam & Boston. Orth, R.J. 1973 Benthic infauna of eelgrass, Zostera marina, beds. Chesapeake Sci. 14:258-269. Ortner, P.B., S.R. Cummings, R.P. Artring, and H.E. Edgerton 1979 Silhouette photography of oceanic zooplankton. Nature (Lond.) 277:50-51. Reise, K. 1977 Predation pressure and community structure of an inter- tidal soft-bottom fauna. In Keegan, B.F., P.O. Ceidigh, and P.J.S. Boaden (eds.), Biology of benthic organisms: 11th Euro- pean symposium on marine biology, Galway, October 1976, p. 513-519. Pergamon Press, NY. Templeman, W. 1948 Growth per molt in the American lobster. Bull. New- foundland Govt. Lab. 18:12-25. Avail. Dep. Fish. Oceans, St. John's, Newfoundland, Canada. Templeman, W., and S.N. Tibbo 1945 Lobster investigations in Newfoundland 1938 to 1941. Dep. Nat. Resour. St. John's Research Bull. 16, Dep. Fish. Oceans, St. John's, Newfoundland, Canada, 98 p. Wahle, R.A. 1990 Recruitment, habitat selection, and the impact of pred- ators on the early benthic phases of the American lobster (Homarus americanus Milne Edwards). Ph.D. diss., Univ. Maine, Orono, 136 p. Wieser, W. 1960 Benthic studies in Buzzards Bay. II. The meiofauna. Limnol. Oceanogr. 5:121-137. Wilder, D.G. 1953 The growth rate of the American lobster (Homarus americanus). J. Fish. Res. Board Can. 10:371-412. Williams, L.W. 1907 The stomach of the lobster and the food of larval lobsters. Annu. Rep. R.I. Comm. Inland Fish. 37:153-180 [avail. NMFS Woods Hole Lab., Woods Hole, MA 02543]. AbStr3Ct. — Beach and purse seine catches at Jones Beach, River kilometer 75, were used to examine diel movement patterns of juvenile chinook salmon Oncorhynchus tsha- wytscha, coho salmon 0. kisutch, and steelhead 0. mykiss as they migrated downstream in the Columbia River estuary. The patterns were moni- tored during five 24-hour periods in 1978, 1979, and 1980, and compared with patterns obtained from exten- sive morning-hour sampling con- ducted during 1979-83. Diel catch patterns were generally consistent among the sampling periods and there was reasonable agreement with morning-hour sampling. How- ever, diel movement was different than that reported for salmonids in other river systems and in other loca- tions in the Columbia River. The times and lateral position of greatest downstream movement which pro- vided the largest catches of salmonid juveniles were as follows: sunrise to early afternoon nearshore for sub- yearling chinook salmon, sunrise to early afternoon midriver for yearling chinook salmon, midmorning to early evening nearshore and sunrise to early afternoon midriver for coho salmon, and noon to early evening midriver for steelhead. Decreased movement during darkness was ap- parent for all salmonids. No relation- ship between tidal cycle and catch was evident from either beach or purse seine sampling. Diel Sampling of Migratory Juvenile Salmonids in the Columbia River Estuary Richard D. Ledgerwood Coastal Zone and Estuarine Studies Division Northwest Fisheries Science Center, National Marine Fisheries Service, NOAA 2725 Montlake Boulevard East. Seattle. Washington 981 12-2097 Frank P. Thrower Auke Bay Laboratory. Alaska Fisheries Science Center National Marine Fisheries Service, NOAA P.O. Box 210155. Auke Bay. Alaska 99821 Earl M. Dawley Coastal Zone and Estuarine Studies Division Northwest Fisheries Science Center, National Marine Fisheries Service, NOAA 2725 Montlake Boulevard East. Seattle, Washington 981 12-2097 Manuscript accepted 24 July 1990. Fishery Bulletin, U.S. 89:69-78 (1991). Successful and cost-effective timing and survival studies for juvenile salm- on and steelhead are dependent on understanding migratory behavior as well as sampling effectiveness. Literature regarding the migratory behavior of juvenile Pacific salmon Oncorhynchus spp. and steelhead O. mykiss indicates a wide variation in diel movement patterns, from great- est movement during daylight hours (Sims et al. 1976) to greatest move- ment at night (Smith et al. 1968) (see also Table 1). Catches used for the reported observations were obtained using an assortment of sampling equipment in large and small rivers and reservoirs during a range of tur- bidity conditions. Juveniles captured varied in life stage from emergent fry to migrating smolt. It was difficult to determine from some of the litera- ture whether the greatest catches represented increased fish movement or times of greatest susceptibility to sampling equipment. Personnel of the National Marine Fisheries Service conducted a sam- pling program at Jones Beach, Co- lumbia River, kilometer (Rkm) 75, to examine diel movement patterns of juvenile chinook salmon 0. tshawy- tscha, coho salmon O. kisutch, and steelhead in the upper Columbia River estuary. The objective was to estab- lish the optimum time of day and lateral location for the most effective sampling of these fish during the peak of the spring migration. Also, this program was to provide data to compare with previous sampling re- sults at Jones Beach which indicated midriver orientation of yearling fish, shore orientation of subyearling fish, and substantially decreased move- ment of shore-oriented migrants at night (Dawley et al. 1986). Methods Diel migration patterns were moni- tored using beach and/or purse seines during five 24-hour periods: 18-19 May 1978, 14-15 June 1978, and 14-15 May 1980 for beach seine; and 10-11 May 1979, 23-24 May 1979, and 14-15 May 1980 for purse seine. Sampling dates were based on peaks of juvenile salmonid migrations recorded in other years (1966-83) at Jones Beach (Dawley et al. 1986). Purse seining was conducted mid- river from the north edge of the ship channel toward Puget Island; beach 69 70 Fishery Bulletin 89(1), 199) Cross section Rock cliff OREGON 1 I I I I Kilometers Figure 1 Jones Beach sampling site. Beach and purse seining areas are denoted by the two asterisks on the cross section. seining was on the south shoreline, lateral to the purse seine site (Fig. 1). Conditions of the Columbia River were different dur- ing each sampling period. River flows ranged from 6800 to 7600 m 3 /second (U.S. Army Corps of Engi- neers 1978-80). Turbidity and water temperatures ranged from 5 to 11 Jackson Turbidity Units and 12° to 17°C, respectively. Tides at Jones Beach are semi- diurnal (~7 hours of ebb and 4.5 hours of flood current); flow reversal occurs during flood tides throughout most of the year. River flows in the range of 5000-12,000 m 3 /second generally occur during the period May through mid-July, and flood tide effects are diminished at these high flows. Captured salmonids were anesthetized, identified to species, and enumerated (Dawley et al. 1985). Subyear- ling and yearling chinook salmon were separated on the basis of fork length; separation points were deter- mined from the bimodal curves of length frequency. Verification of age from marked fish of known age (4.8-6.5% of catch) showed about 4% error in separa- tion. All captured fish were either held in tanks onshore until sampling was complete and then released or transported downstream from the sampling area and released. Beach seining The beach seine was 95 m long by 5 m deep with 1-2 cm (stretch) webbing (Sims and Johnsen 1974). The net was fished downstream regardless of tidal influence. An anchor was used to secure one end of the net on- shore and the opposite end, containing the bunt, was towed upstream at the 1-m depth contour, then arched Ledgerwood et al.: Diel sampling of migratory juvenile salmomds in Columbia River estuary 71 downstream and back to shore. The effective fishing depth of the net was 2-3 m in water up to 6 m deep. The net was pulled onto the beach which crowded the fish into the bunt for capture. Completion of each set required about 25 minutes; sets were made at 45-min- ute intervals. Catch data from the first two or three seine sets on the first morning of each sampling period were not used for analysis of diel movement because some salmonids probably resided in the sampling area over- night, and the initial sets were used to clear the sam- pling site of those residents. Of the beach seine sets in each 24-hour period, 34% (11 of 32) were made dur- ing darkness and 66% (21 of 32) during daylight. Purse seining The purse seine was 206 m long by 11m deep with 1-2 cm (stretch) webbing (Durkin and Park 1967). A depressor weight was used during the pursing opera- tion to increase the effective fishing depth to about 6 m. The vessels used were a 10-m pontoon barge powered by outboard engines and an outboard- motored seine skiff; lights were mounted on the barge for night opera- tion. A depth finder, a compass, and channel markers were used to locate the sampling site. The seine was set near midriver in water 9- 14 m deep, and towed upstream at constant power in a "U" configuration (Dawley et al. 1985). After 5 minutes, the ends of the net were brought together and the net bot- tom was closed (pursed) and hauled aboard the barge with a boom and hydraulic capstan. Then the cork line and webbing were retrieved and the catch was placed in 75-L containers supplied with circulating river water. Completion of each set required about 40 minutes; sets were made at 90-minute intervals. About 31% (5 of 16) of the purse seine sets were made during darkness and 69% (11 of 16) during daylight. Data analysis Each set represented one time interval within the 24-hour sampling period. Twice as many beach seine sets were made in each 24-hour period as purse seine sets; consequently, time intervals are one-half those for the purse seine. The catch per set (CPS) interval was calculated in terms of the percentage of the total 24-hour catch by species and stock. An overall percent CPS was calculated for each seine type by averaging interval values from the three appropriate sampling dates. Diel catch data for each species were compared graphically and with linear regression to corresponding tidal heights at Jones Beach. Results During 14-15 May 1980, the only period we sampled with both beach and purse seines, the beach seine accounted for 79% of the total catch of subyearling chinook salmon (predominately fall race) (Van Hyning 1973), while the purse seine produced the largest catches of yearling fish: 92% of the yearling chinook salmon (predominately spring race); 82% of the coho salmon; 100% of the sockeye salmon O. nerka; and 97% of the steelhead. Daylight sampling in previous years produced similar beach seine to purse seine catch ratios (Dawley et al. 1986). Examination of catch data indicated there was no ap- parent relationship to tidal variations for any species during any sampling period; correlation coefficients ranged from -0.51 to 0.14. Catch/tidal data are avail- able upon request. Dawley et al. (1986) also observed a lack of correlation between tidal cycles and beach seine catches of subyearling chinook salmon from the Columbia River estuary. Subyearling chinook salmon Beach seine catches of subyearling chinook salmon (13,513 fish) peaked during the interval about 1.5 hours after sunrise (6.9% CPS) followed by steady catches during the daylight intervals, each near 4.0% CPS. About 1.5 hours before sunset, a second, smaller peak was observed in two intervals (CPS of 5.2% each), followed by a sharp and continued decrease with dark- ness through the night intervals (average CPS = 0.9%). The night catch was 10.2% (3.8 SD) of the total catch for a 24-hour period. Catches increased again about 45 minutes before sunrise (Fig. 2A). Purse seine catches of subyearling chinook salmon (1461 fish) increased just before sunrise and decreased throughout the day (Fig. 2B). Again, only 10% (1.7 SD) of the total purse seine catch was at night. Coho salmon About 21% of the yearling coho salmon captured were from beach seining (1092 by beach seine and 3990 by purse seine). The June 1978 sampling period produced only 17 fish and was not included in the assessment of movement behavior. Beach seine catches in daylight were low until about 1000 hours then generally in- creased, with large fluctuations between intervals, to a peak at about 1430 hours (10.7% CPS) (Fig. 2C). In the late afternoon and evening, catches generally decreased with large fluctuations between intervals. The CPS dropped at dusk to 2.5% followed by lower catches during darkness. The night catch averaged 72 Fishery Bulletin 89(1), 1991 lOr BEACH SEINE Subyearling Chinook Salmon n = 13,513 0545 1200 1800 2045 2400 0500 Sunrise Sunset Time PURSE SEINE » Subyearling Chinook Salmon D n = 1461 15 - 10 - 5 - Coho Salmon n = 3990 D Daylight Darkness / -4 i — t- -+ V — 4~ ,,- 0545 Sunrise 1800 2045 2400 0500 Sunset 20 r- PURSE SEINE Steelhead n = 4673 0545 Sunrise I i ' 1 1200 1800 2045 2400 0500 Sunset Time Figure 2 Diel catch patterns for chinook salmon, coho salmon, and steel- head from beach and purse seine sampling at Jones Beach, 1978-80 (samples combined and averaged). 13.7% of the total catch for the two 24-hour sampling periods. Purse seine catches peaked during the interval about 1.5 hours after daylight (15.8% CPS) and remained near 6.5% CPS throughout the daylight intervals the (Fig. 2D). Catches decreased in the night intervals to an average 3.7% CPS; 18.5% (4.6 SD) of the CPS was obtained during darkness. Yearling chinook salmon The majority of yearling chinook salmon migrated mid- river; purse seine catches totaled 2029 fish compared with 113 from the beach seine. The peak catch with the purse seine was during the interval 0946 to 1115 hours (12.3% CPS). Overall, 46% of the total catch was taken in 31% of the sets (1.5 hours after sunrise to about 1330 hours) (Fig. 2E). Purse seine catches were smallest from dusk to midnight (average 3.2% CPS), with larger catches occurring during the remainder of the night intervals (average 5.4% CPS). The night catch was 21.1% (7.0 SD) of the total purse seine catch for a 24-hour period. Sockeye salmon Sockeye salmon juveniles were caught only in the purse seine (222 fish), with 15% (6.3 SD) captured at Ledgerwood et al.: Diel sampling of migratory juvenile salmomds in Columbia River estuary 73 night. An insufficient number of fish were captured to allow a more detailed analysis of the diel migration pattern. Steelhead Over 98% of the juvenile steelhead were caught by purse seine (4673 by purse seine and 74 by beach seine). Purse seine catches were moderate in the four inter- vals after sunrise (average 4% CPS), peaked at the interval from 1416 to 1545 hours (14.6% CPS), decreas- ed at dusk, and remained low throughout the night in- tervals (average 1.7% CPS) (Fig. 2F). The night catch was 8.7% (1.0 SD) of the total catch for a 24-hour period. Discussion Catch data from our beach and purse seines appear to represent movement and position of juvenile salmonids during their migration through the upper Columbia River estuary. Catches of subyearling chinook salmon at both purse and beach seining sites indicate a substan- tially decreased migration during darkness. Beach (nearshore) and purse (midriver) seine catches of coho salmon indicate a fairly uniform migration throughout the daylight period. Data obtained for yearling chinook salmon, sockeye salmon, and steelhead indicate a midriver orientation with decreased migration during darkness. Other researchers have reported different diel movement patterns, but conditions, equipment, and life stages of the fish sampled are so variable that direct comparison between experiments is difficult (Table 1). Table 1 Summary of diel observations of juvenile chinook and coho salmon and steelhead in various Pacific Northwest and Alaska river systems. Environment Life stage Location Method Pertinent observations of movement behavior Source Yearling chinook salmon River Smolt Central Ferry Bridge (Snake R.) Fyke net Largest catches between 0300 and 0600 h, smallest catches between 0600 and 1200h. Largest catches from shoreline areas during low and medium river flow, and midriver areas during high flow. Catches uniform in number from surface to bottom. Mains and Smith 1964 River Smolt Byer's Landing (mid-Columbia R.) Fyke net Largest catches (70%) between 1800 and 0600 h. Largest catches in shore- line areas. Largest catches in surface water (0.8 m). Mains and Smith 1964 Reservoir Smolt John Day (Columbia R.) Purse seine Largest catches during daylight (0700-2100 h). Sims et al. 1976 Reservoir Smolt Lower Monumental (Snake R.) Monofilament gillnet Largest catches at night; 92% of total catch. Catches in the upper 7.3m were 5 during the day and 109 at night, and catches below 7.3 m in- creased from 6 during the day to 19 at night. Largest catches in central portion of reservoir, day and night. Smith 1974 Reservoir Smolt and fry Mayfield (Cowlitz R.) Floating fish trap Largest catches (82% of total obtained between 2000 and 0800 h. Allen 1965 Reservoir Fingerling and fry Upper Mayfield (Cowlitz R.) Trawl and gillnet Largest catches during darkness or periods of high turbidity (trawl). Largest catches (87%) in upper 7.3m. Movement of fish not strongly downstream. Smith et al. 1968 Reservoir Smolt and fingerling North Fork (Clackamas R.) Gillnet Largest catches during darkness near surface (0-5 m) over deep water (15 m). Korn et al. 1967 Reservoir Smolt Rocky Reach (Columbia R.) Sonar Highest movement between dusk and dawn. Leman 1978 74 Fishery Bulletin 89(1), 1991 Table 1 (continued) Pertinent observations of Environment Life stage Location Method movement behavior Source Yearling chinook salmon (continued) Dam Smolt John Day Dip net and Largest catches (92% of total) Sims et al. 1976 (Columbia R.) airlift pump between dusk and dawn (8.5-h period). Turbine intake, 20m below surface. Sims and Ossiander 1981 Dam Smolt The Dalles . Fyke net in Largest catches (94%) at night (1900- Long 1968 (Columbia R.) turbine intake 0700 h). Largest catches (75% of total) from upper third of intakes (top of turbine intake, 6m below surface). Dip net in Largest catches during daylight Sims et al. 1976 gatewell (0700-2100h) with only 11% of total caught in darkness (8.5-h period). Nichols 1979 Dam Smolt The Dalles Fyke net in Largest catches in daylight (0800- Nichols 1979 (Columbia R.) sluiceway 1400 h) with few fish after dark; 3-ft deep surface flow over sluiceway gate. Largest catches near dusk (1700- 2200 h) with few fish at other periods; 2-ft deep surface flow over sluiceway gate. Subyearling chinook salmon Estuary Smolt Puget Island and Beach seine 90% of catch from daylight sets Dawley et al. Jones Beach (0600-2100 h). Largest catches during 1986 (Columbia R. early morning and at dusk. estuary) River Fingerling Sixes R. Traps Both emergence of fry from the Reimers 1973 and fry (Oregon R.) gravel and downstream migration of fry and fingerling primarily during darkness. Reservoir Smolt John Day (Columbia R.) Purse seine Largest catches during daylight (0700-2100 h). Sims et al. 1976 Dam Smolt John Day Dip net and Largest catches (88% of total) Sims et al. 1976 (Columbia R.) airlift pump between dusk and dawn. Turbine Sims and in gatewell intake, 20 m below surface. Ossiander 1981 Dam Smolt The Dalles Fyke net in Largest catches (67%) at night (1900- Long 1968 (Columbia R.) turbine intake 0700 h). Largest catches (49% of total) from upper third of water column (4.4 m) entering the intakes (top of turbine intake, 6m below surface). Dip net in Largest catches during daylight Sims et al. 1976 gatewell (0700-2100h) with 10% of total caught in darkness (8.5-h period). Turbine intake, 6 m below surface. Nichols and Ransom 1980 Dam Smolt The Dalles Fyke net in Largest catches in daylight (0600- Nichols and (Columbia R.) sluiceway 0700 and 1400-2100h) with few fish after dark; 3-ft deep surface flow over sluiceway gate. Ransom 1980 Coho salmon Estuary Smolt Columbia R. Purse seine Largest catches at midday (1000-1400h). Durkin 1982 River Smolt Minter Creek (Puget Sound) Trap Largest catches at dawn and dusk. Salo and Bayliff 1958 River Smolt Taku R. (S.E. Alaska) Scoop trap Largest catches at dawn and dusk. Meehan and Siniff 1962 Reservoir Smolt Brownlee (Snake R.) Fyke net Largest catches at dawn and dusk. Monan et al. 1969 Reservoir Smolt John Day (Columbia R.) Purse seine Largest catches during daylight (0700-2100h). Sims et al. 1976 Ledgerwood et al.: Diel sampling of migratory juvenile salmonids in Columbia River estuary 75 Table 1 (continued) Pertinent observations of Environment Life stage Location Method movement behavior Source Coho salmon (continued) Reservoir Smolt and North Fork Gillnet Largest catches at dawn and dusk; Korn et al. 1967 fingerling (Clackamas R.) near surface during darkness and deeper during daylight. Reservoir Smolt and Round Butte Gillnet Few fish captured during daylight Korn et al. 1967 fingerling (Deschutes R.) during any season. During spring migration period, smolts captured principally near surface (0-3.7 m). Reservoir Fingerling Upper Mayfield Trawl and Largest catches during darkness or Smith et al. and fry (Cowlitz R.) giilnet periods of high turbidity (trawl). Largest catches (87%) near surface (0-7.3 m). Movement of fish not strongly downstream. 1968 Dam Smolt John Day Dip net in Largest catches during darkness Sims et al. 1976 (Columbia R.) gatewell of dam (2100-0700 h). Turbine intake, 20m below surface. Dam Smolt The Dalles Dip net in Largest catches during daylight Sims et al. 1976 (Columbia R.) gatewell of dam (0700-2100h). Turbine intake, 6m below surface Dam Smolt The Dalles Fyke net in Largest catches in daylight Nichols 1979 (Columbia R.) sluiceway (0800-1400h) with few fish after dark; 3-ft deep surface flow over sluiceway gate. Largest catches near dusk (1 700-2200 h) with few fish at other periods; 2-ft deep surface flow over sluiceway gate. Steelhead Reservoir Smolt John Day (Columbia R.) Purse seine Largest catches during daylight (0700-2100h). Sims et al. 1976 Reservoir Smolt Lower Monofilament Largest catches at night; 76% of daily Smith 1974 Monumental gillnet total. Catches in the upper 7.3m in- (Snake R.) creased from 146 during the day to 396 at night and catches below 7.3m from 56 during the day to 420 at night. Uniform distribution across the reservoir day and night. Reservoir Smolt Mayfield Floating Largest catches (82% of total) Allen 1965 and fry (Cowlitz R.) fish trap obtained between 2000 and 0800 h. Reservoir Fingerling Upper Mayfield Trawl and Largest catches during darkness or Smith et al. and fry (Cowlitz R.) gillnet periods of high turbidity (trawl). Largest catches (87%) in upper 7.3m. Movement of fish not strongly downstream. 1968 Reservoir Smolt and North Fork Gillnet Largest catches (few fish) during Korn et al. 1967 fingerling (Clackamas R.) darkness near surface over deep water. Dam Smolt John Day Dip net and Largest catches (77% of total) Sims et al. 1976 (Columbia R.) airlift pump in gatewell of dam between dusk and dawn (8.5 h period). Turbine intake, 20 m below surface. Dam Smolt The Dalles Fyke net in Largest catches (85%) at night (1900- Long 1968 (Columbia R.) turbine intake 0700 h). Largest catches (72% of total) from the upper third of the water column (4.4 m) entering the intakes (top of turbine intake, 6m below surface). 76 Fishery Bulletin 89(1). 1991 Table 1 (continued) Pertinent observations of Environment Life stage Location Method movement behavior Source Steelhead (continued) Dam Smolt The Dalles Fyke net in Largest catches during daylight with Sims et al. 1976 (Columbia R.) turbine intake only 29% of total caught in darkness (8.5 h period). Turbine intake. 6m below surface. Dam Smolt The Dalles _ Fyke net in Largest catches in daylight (0800- Nichols 1979 (Columbian R.) sluiceway 1400 h) with few fish after dark; 3-ft deep surface flow over sluiceway gate. Largest catches near dusk (1700- 2200 h) with few fish at other periods; 2-ft deep surface flow over sluiceway gate. Variability of catch between sets and sampling periods was higher for yearling chinook salmon than for other salmonids. The origin of marked fish varied sub- stantially among the three purse-seine sampling periods (Table 2). The largest portions of the catch originated in the Willamette, mid-Columbia, and Snake Rivers for the first, second, and third sampling periods, respectively. Stock differences and changes in abundance among stocks during the diel sampling periods may have caused the higher variability in the catch. We found reasonable agreement among the diel catch patterns reported here and those from extensive morning sampling (2615 sets) at Jones Beach in May and June 1979-83 (Dawley et al. 1986) (Fig. 3). A noteworthy exception was that beach seine catches near sunrise were lower during the diel study because sets were made before sunrise to remove fish which resided in the area overnight. It is generally agreed that net avoid- ance is probably greatest in daylight; therefore, decreased net catches at night should represent decreased fish abun- dance in the water sampled. It seems unlikely that decreased catches at Jones Beach during darkness were caused by surface- or midwater-oriented juveniles maintaining their position against cur- rent velocities up to 5 km/hour. Data obtained at Jones Beach by Dawley et al. (1986) showed that marked subyearling chinook salmon released into the shoreline sampling area at night were recaptured at a much Table 2 Origin of marked yearling chinook salmon captured by purse seine during diel sampling, 1979 and 1980. Sampling dates 10-11 May 23-24 May 14-15 May 1979 1979 1980 Origin Snake River 7 19 60 mid-Columbia River 13 47 40 Transported and released downstream from Bonneville Dam 33 28 Willamette River 40 Lower Columbia River 7 6 Table 3 Measured movement rates of juvenile salmonids and water velocities in a 155-km reach of the Columbia River between Bonneville Dam and Jones Beach at two volumes of river flow. Water velocity (km/h) b Movement rates (km/hf River Chinook salmon Coho salmon (1000 m 3 /s) Subyearling Yearling Steelhead 8.1 ±0.5 11.3 ±0.5 4.8 5.0 0.9 1.0 0.8 1.4 0.9 3.2 1.7 'Zero to nine marked groups were available for each calculation of average movement rate at these designated river flows (Dawley et al. 1986). 'From Blahm (1974). higher rate than marked fish released during daylight (30.6 vs. 8.0%). Because midriver-oriented yearling fish do not appear in shoreline areas at Jones Beach during darkness, they probably hold near the bottom, Ledgerwood et al.: Diel sampling of migratory juvenile salmomds in Columbia River estuary 77 particularly in deep areas of low current velocity. This premise is supported by studies on water velocity (Blahm 1974) and the movement rates of marked juvenile salmon released below Bonneville Dam and recovered at Jones Beach (Dawley et al. 1986) (Table 3). In all cases, fish migration speeds from release site to capture in the estuary were less than water velocity (Dawley et al. 1986). In conclusion, the most appropriate times and locations for sampling to attain maximum CPSs are as follows: Subyear- ling chinook salmon, sunrise to early after- noon nearshore; yearling chinook salmon, sunrise to early afternoon midriver; year- ling coho salmon, midmorning to early evening nearshore and sunrise to early afternoon midriver; juvenile steelhead, noon to early evening midriver. Acknowledgment The authors acknowledge Dr. Theodore Blahm (deceased) for providing inspiration and encouragement while conducting this study. Citations 20 10 Allen, R. 1965 Juvenile fish collector operation at Lake Mayfield July 1, 1964 to June 30, 1965. Report to Bureau of Commercial Fisheries. Contract 14-17-0001-1357. Wash. Dep. Fish., Olympia, 23 p. Blahm, T.H. 1974 Gas supersaturation research, Prescott Facility-1974. Report to U.S. Army Corps of Engineers, Contract DACW57-74-F-0414. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Seattle, WA 98112, 34 p. Dawley, E.M., R.D. Ledgerwood, and A.L. Jensen 1985 Beach and purse seine sampling of juve- nile salmonids in the Columbia River estuary and ocean plume, 1977-1983; Vol. I. Proce- dures, sampling effort, and catch data. NOAA Tech. Memo. NMFS F/NWC-74, Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Seattle, WA 98112, 260 p. Dawley, E.M., R.D. Ledgerwood, T.H. Blahm, C.W. Sims, J.T. Durkin, R.A. Kirn, A.E. Rankis, G.E. Monan, and F.J. Ossiander 1986 Migrational characteristics, biological observations, and relative survival of juvenile salmonids entering the Columbia River estuary, 1966-1983. Report to Bonneville Power Admin- istration, Portland, Oregon. Contract DE-A179-848BP39652. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, 30 20 10 CUMULATIVE YEARS 30 r SUBYEARLING CHINOOK SALMON o Beach Seine 189 d ■• Purse Seine 145 d * Diel Sampling 3 d I I I I I I I I I I 40 r 30 20 10 SUBYEARLING CHINOOK SALMON 20 10 40 COHO SALMON I I I J I 4 - I 40 r 30 YEARLING CHINOOK SALMON 30 - 20 10 40 COHO SALMON J I I l__l 30 20 10 STEELHEAD * (46) J_ 0600 0900 Hours 1200 0600 0900 Hours 1200 Figure 3 Percent total catch of juvenile salmonids during morning hour sampling at Jones Beach, Columbia River (1979-83) compared with diel sampling results (mean value per time interval and 95% confidence limits); d = total number of sampling days used for each estimate. Seattle, WA 98112, 256 p. Durkin, J.T. 1982 Migration characteristics of coho salmon {Oncorhynehus kisutch) smolts in the Columbia River and its estuary. In Ken- nedy, V. (ed.), Estuarine comparisons, p. 365-375. Academic Press, NY. Durkin, J.T., and D. L. Park 1967 Purse seine for sampling juvenile salmon. Prog. Fish- Cult. 29:56-59. 78 Fishery Bulletin 89(1), 1991 Korn, L.M., L.H. Hreha, R.G. Montagne, W.G. Mullarkey, and E.J. Wagner 1967 The effects of small impoundments on the behavior of juvenile anadromous salmonids. Final Report to U.S. Bureau of Commercial Fisheries, Contract 14-17-0001-597,767,917, 1093, and 1238. Fish Comm. Oreg., Res. Div., Clackamas, 127 p. Leman, B.D. 1978 Coordinated spilling at Columbia River dams during low river flows; response by Public Utility Districts. Unpubl. manuscr., Chelan County, Public Utility District Ml, Wenat- chee, WA, 15 p. Long, C. 1968 Diel movement and vertical distribution of juvenile anadromous fish in turbine intakes. Fish. Bull., U.S. 66: 599-609. Mains, E.M., and J. M. Smith 1964 The distribution, size, time and current preferences of seaward migrant chinook salmon in the Columbia and Snake Rivers. Wash. Dep. Fish., Fish. Res. Pap. 2(3):5-43. Meehan, W.R., and D.B. Siniff 1962 A study of the downstream migrations of anadromous fish in the Taku River, Alaska. Trans. Am. Fish. Soc. 91: 399-407. Monan, G.E., R.J. McConnell, J.R. Pugh, and J.R. Smith 1969 Distribution of debris and downstream-migrating salmon in the Snake River above Brownlee Reservoir. Trans. Am. Fish. Soc. 98:239-244. Nichols, D.W. 1 979 Passage efficiency and mortality studies of downstream migrant salmonids using The Dalles ice-trash sluiceway dur- ing 1978. Report to U.S. Army Corps of Engineers, Contract DACW 57-78-C-0058. Oreg. Dep. Fish Wildl., Portland, 28 p. Nichols, D.W.. and B.H. Ransom 1980 Development of The Dalles Dam ice and trash sluiceway as a downstream migrant bypass system, 1980. Final report to U.S. Army Corps of Engineers, Contract DACW 57-78-C- 0058. Oreg. Dep. Fish Wildl., Portland, 37 p. Reimers, P.E. 1973 The length of residence of juvenile fall chinook salmon in Sixes River, Oregon. Res. Rep. Fish Comm. Oreg. 4(2): 3-43. Salo, E.O., and W.H. Bayliff 1958 Artificial and natural production of silver salmon, Oncor- hynchus kisutch, at Minter Creek, Washington. Wash. Dep. Fish. Res. Bull. 4:1-76. Sims, C.W., and R.C. Johnsen 1974 Variable mesh beach seine for sampling juvenile salmon in the Columbia River estuary. Mar. Fish. Rev. 36(2):23-26. Sims, C.W., and F.J. Ossiander 1981 Migrations of juvenile chinook salmon and steelhead trout in the Snake River from 1973 to 1979, a research summary. Final report to U.S. Army Corps of Engineers, Contract DACW68-78-C-0038. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Seattle, WA 98112, 53 p. Sims, C.W., R.C. Johnsen. and W.W. Bentley 1976 Effects of power peaking operations on juvenile salmon and steelhead trout migrations, 1975. Final report to U.S. Army Corps of Engineers, Contract DACW68-77-C-0025. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Seattle, WA 98112, 61 p. Smith, J.R. 1974 Distribution of seaward migrating chinook salmon and steelhead trout in the Snake River above Lower Monumental Dam. Mar. Fish. Rev. 36(8):42-45. Smith, J.R., J.J. Pugh, and G.E. Monan. 1968 Horizontal and vertical distribution of juvenile salmonids in upper Mayfield Reservoir, Washington. U.S. Fish Wildl. Serv. Spec. Sci. Rep. Fish. 566, 11 p. U.S. Army Corps of Engineers 1978-1980 Annual fish passage reports, Columbia and Snake Rivers, for salmon, steelhead, and shad. U.S. Army Corps of Engineers, North Pac. Div., Portland, OR, 41 p. Van Hyning, J.M. 1973 Factors affecting the abundance of fall chinook salmon in the Columbia River. Res. Rep. Fish Comm. Oreg. 4(l):3-87. Abstract.- In Bahia de la Ascen- si6n, Mexico, the fishery for Panu- lirus argus is based on artificial shel- ters called "casitas." Highest catch- per-unit-effort (kg tails/boat day) in the fishery occurs each year immedi- ately after the opening of the fishing season, and declines sharply over the next months. This trend probably reflects combined effects of natural mortality, fishing mortality, and emi- gration of lobsters from the bay. In 1985, 3470 tagged lobsters were released during the closed season, and 849 (24.5%) were recaptured by fishermen, mainly during the first three months of the following fishing season. In 1986, an additional 1324 tagged lobsters were released, and 407 (30.7%) were subsequently re- captured. Growth of recaptured lob- sters was highly variable, and sexes had different growth rates, that of males being higher. Von Bertalanffy parameters for each sex were calcu- lated using two different techniques; most reasonable estimates were ob- tained by a maximum likelihood ap- proach. Ninety-nine percent of the recaptured lobsters were caught within the bay, but movements gen- erally tended to be toward the reef in front of the bay. Longest straight- line distance moved was 45 km. The population fished in the bay was composed entirely of juveniles, and we hypothesize that an unfished population of adults exists outside the bay. Testing of this hypothesis would benefit future management plans. In addition, the long-term ef- fects of casitas on the lobster popula- tion and on the ecology of the sea- grasses and their associated benthic communities need to be understood. Fishery Characteristics, Growth, and Movements of the Spiny Lobster Panulirus argus in Bahia de la Ascension, Mexico Enrique Lozano-Alvarez Patricia Briones-Fourzan Universidad Nacional Autonoma de Mexico Institute) de Clencias del Mar y Limnoiogia, Estacion "Puerto Morelos" P O Box I 152, Cancun. Q R .. 77500 Mexico Bruce F. Phillips Commonwealth Scientific and Industrial Research Organization (CSIRO) Marine Laboratory. PO Box 20. North Beach, W.A.. 6020 Australia Manuscript accepted 12 September 1990. Fishery Bulletin, U.S. 89:79-89 (1991). Panulirus argus accounts for approx- imately one-third of Mexico's spiny lobster production of about 2400 t (mean for 1978-87), 80% of which is produced in the state of Quintana Roo (Secretaria de Pesca 1987). The fishery for lobsters in Bahia de la Ascension began in 1965. Initially, traps and skin diving were used, but in 1968 "casitas cubanas" were intro- duced (Miller 1982). These "casitas" consist of a frame of about 1.8 x 1.2 m made of the trunks of a local palm, and a "roof" of the same wood, metal, asbestos or, more recently, ferrocement. Casitas are sunk over seagrass-covered bottom. The fisher- men check the casitas by skin diving, and catch the lobsters with a gaff (Lozano et al. 1989). The bottom of the bay suitable for setting casitas has been divided into several parcels of different sizes, alloted to the older fishermen. Miller (1982) suggested that the casitas might increase the fishing pressure on the population and cause overfishing, and Eggleston et al. (1990) propose that casitas pro- vide critical refuge for juvenile lob- sters from their predators. The long- term effects of casitas on the lobster population remain to be determined. Here we report the results of an in- vestigation using tag and recapture methods to study the structure, move- ments, and growth rates of the spiny lobster population in Bahia de la As- cension during 1985-87. Methods Fishing methods in Quintana Roo The coast of Quintana Roo can be divided into three areas on the basis of the lobster fisheries (Fig. la): In the northern area, from Holbox to Tulum and especially around Isla Mujeres, the fishery is well devel- oped. Lobsters are caught mainly by traps in depths of 15-60 m, and by Scuba and "hookah" diving to depths near 40 m. An annual migration of lobsters occurs along the northeast- ern coast of the Yucatan Peninsula in a southerly direction, at the end of autumn or in winter (Kanciruk and Herrnkind 1978). During this migra- tion, fishermen use lobster bottom- nets in areas 2-10 m deep. Twelve cooperatives, involving 65% of the 1084 lobster fishermen of the state, operate in the northern area. In the central area, where Bahia de la Ascension is located, skin diving and "casitas cubanas" are used. In this area, where three cooperatives 79 80 Fishery Bulletin 89(1), 1991 MEXICO Puerto Juarez ^ lsla ^ujeres Puerto Morelos* / j Cozur jmel Punta Allen Ascension Bay Espiritu Santo Bay Punta Herrero 44 mm CL were tagged in order to reduce incidental mortality which might occur on smaller animals (Chittleborough 1974). Tags were inserted in- to the dorsolateral extensor muscle between the cephalothorax and first abdominal segment. After tag- ging, the lobsters were immediately released where they had been caught. Underwater observations re- vealed that after a few minutes, the tagged lobsters returned under the same casita. Tag number, date, release location, sex, reproduc- tive state, and CL (± 0.1mm measured from between the rostral horns to the posterior dorsal edge of the carapace) were recorded. Fishermen were requested to keep the head of a recaptured lobster with its tag so the CL could be measured, and to provide the recap- ture date and location. The tagging program was advertised widely, and a reward was offered in the form of a lottery to encourage tag returns. Analyses of growth data The analysis of growth using capture-recapture data was performed using Fabens' method (1965), and a technique developed by M. Palmer (CSIRO Div. Math. Stat., Floreat Park, W.A. 6014, Australia). This tech- nique assumes an individual lobster grows exponential- ly with time: y = a (1 - e bt ) + E where y = CL (mm), a = asymptotic CL (mm), b = a growth coefficient, t = time, and E = residuals. A mean value of 6mm CL obtained from 50 settling pueruli was introduced as a starting size (zero age) into the model. Parameters for the model, including variability of in- dividual growth, were estimated using a multivariate Gaussian distribution. The residuals around an in- dividual's curve (E) were assumed to be independent Gaussian normal with constant variance. The likelihood estimate, assuming that individual coefficients are known for an individual, was Lj = p (y | a,b) p (a,b) where L ; = initial length, and p(.|.) denotes a probabil- ity distribution. Since the individual animal's coefficients were unknown, we consider them as "nuisance" parameters and integrate them out of the likelihood, giving oo / p (y | a,b) p(a,b) dadb where lj is the likelihood for the i th individual. Then the product of the individual likelihood must be max- imized to find the estimates of the population param- eters. A convenient algorithm to use in this case is the EM algorithm (Dempster et al. 1977). Details of its ap- plication in this context are in Laird et al. (1987), Palmer (1986), and Palmer et al. (1988). Although the time between subsequent captures was known, the age at first capture was unknown. A prob- ability distribution for this unknown parameter was also assumed, but now the initial time is treated as "missing" and is removed from the likelihood by in- tegrating it out. The likelihood for the i th animal is now of the form OO CO ;; p(y | a,b,tx) p(a,b) dadbdti. Maximum likelihood is used to estimate both the growth parameters and the distribution of initial ages. The method of constructing and maximizing the likeli- hood is described in Palmer et al. (1988). Mean weekly growth rates (Hunt and Lyons 1986) of recaptured lobsters were analysed to determine if there were significant changes in growth rate along their size range. 82 Fishery Bulletin 89(1), 1991 50 100 50 100 50 100 50 100 Carapace length (mm) Figure 2 Monthly size-frequency distribution of the commercial catch (July 1985-February 1986) of Panulirus argus in Bahia de la Ascensibn, obtained by converting tail weight (g) to carapace length (mm). 1981 1982 1983 1985 Figure 3 CPUE (kg tails/boat day) of Panulirus argus for the fishing seasons 1981-82 to 1985-86 in Bahia de la Ascensi6n (no data available for January-March, 1986). Results Commercial catch and size composition The total catch of lobster tails in Bahia de la Ascen- sion for 1985-86 was 42.5t, and for 1986-87, 63.0t. The size composition of the commercial catch for July 1985-February 1986 is shown in Figure 2. Data for March were insufficient and not included. There was a mode around the minimum legal size throughout the fishing season, except in July. The CPUE (catch/boat • day) data trends were similar each year (Fig. 3). The highest CPUE occurred during 16 July-15 August, i.e., immediately after the open- ing of the fishing season, followed by a sharp decline over the next few months. This trend probably re- flected both fishing and natural mortality, as well as emigration. The values of F (fishing mortality), between 1.25 and 2.80, derived using a Leslie analysis (Leslie and Davis 1939) were highly influenced by the July data, which were the annual peaks (Fig. 3), implying a different F for that month. Therefore, the results of the analysis were biased and could not be considered a good esti- mate of mortality. Tagging results Of the total 3893 lobsters caught in 1985, 3470 were large enough to be tagged (Fig. 4a). The male:female ratio of the captured population was 1.14:1, and that of the tagged population was 1.13:1. Of the total 1403 lobsters caught during 1986, 1324 were tagged (Fig. 4b). The male:female ratios of both the captured and tagged populations were 1.04:1. The size range of both sexes was similar for both years. A total of 849 tagged lobsters were recaptured dur- ing the 1985-86 fishing season (24.5% of total tagged in 1985). None of the animals tagged in 1985 were recaptured during the 1986 tagging period. However, four lobsters tagged in 1985 were recaptured during the 1986-87 fishing season. The male:female ratio of recaptured animals was 1.14:1. Lobsters were recap- tured in all sampling zones in the bay, as well as at some localities outside of the bay. A total of 407 lobsters were recaptured during the 1986-87 fishing season (30.7% of total tagged in 1986); the male:female ratio was 1.08:1. In this season, no lobsters were recaptured outside the cooperative's fishing grounds. In both fishing seasons, nearly all the recoveries occurred during the first three months of the season (e.g., Fig. 5). Population structure The mean CL (61.4mm, range 10. 2-142. 4mm) of lob- sters caught in the Bay during 1985 was signifi- cantly smaller (z test, Hoel 1976; z= 10.4, P<0.05) than that of lobsters caught during 1986 (65.2 mm, Lozano-Alvarez et al.: Fishery, growth, and movements of Panulirus argus 83 A (a) D Males n = 2074 Females n =18 19 ^ • 10 30 50 70 90 110 130 150 Carapace length (mm) 20 00 80 60 40 20 10 20 30 40 50 60 70 80 90 100 110 120 Carapace length (mm) Figure 4 Size-frequency distribution of Panulirus argus caught in Bahfa de la Ascension for tagging purposes in (a) 1985 and (b) 1986. Tags were applied to lobsters >44mm CL, as indicated by arrow. range 22.0 - 113.1 mm). However, the tagging opera- tion in 1985 occurred one full month earlier than in 1986. The means of CL of lobsters caught in each of the six sampling zones during 1985 and 1986 (Figs. 6, 7) were significantly different (P<0.05, approximate test of equality of means when the variances are heterogeneous; Sokal and Rohlf 1981, Box 13.2). An unplanned comparison among pairs of means (Sokal and Rohlf 1981) shows three groups of mean sizes in each year (Table 1). This result implies that the distri- bution of lobsters by size in the bay is not random. Smaller lobsters occupied the more interior of the six sampling zones (zones II, III, and IV), and larger lobsters occupied zones closer to the reef (zones I, V, and VI). 800 700 - \ "D 1 0> - 600 1 *■! 1 Q. 1 03 1 g 500 - u n= 849 co v 400 _ 17) JD 1 O 1 ~ 300 - o 1 L. 1 n 200 1 E 1 3 Z 1 100 •' ■• — • — m • J ASONDJ FM 1985 1986 Figure 5 Number of Panulirus argus tagged in 1985 that were recaptured each month during the 1985-86 fishing season. Growth and recruitment Of the 849 lobsters recaptured during 1985-86 and the 407 recaptured during 1986-87, only 372 and 268, respectively, were returned with accurate CL informa- tion. All but two showed growth between tagging and recapture. During the first three months of the 1985-86 season, the modal CL of recaptured males increased 10 mm. Growth of individuals was highly variable (Fig. 8) and in some cases indicates more than one molt occurred. For females (data not illustrated), the mode increased 8mm. Fabens' method (1965) for estimating growth param- eters was used for both sexes and both years separately (Table 2a). The estimates of asymptotic length and the growth coefficient k by this method show great variability. On the other hand, the EM algorithm needs some ini- tial estimates before it can begin iterating, and the Fabens estimates were used for this. A set of data for males and females for each year was thus obtained (Table 2b). Combining the data over both years, for each sex, did not lead to a significant increase in the 84 Fishery Bulletin 89(1), 1991 50 40 30 20 ZONE IV n\\ N 426 J . r X 1 -J^ — T \° 1 30 50 70 90 110 50 70 90 110 Carapace length (mm) Figure 6 Size-frequency distribution of Panulirus argus caught in each sampling zone during the 1985 tagging operation in Bahia de la Ascensi6n. log-likelihood ratio, indicating that it was valid to pool the data over the two years, i.e., there was no difference in growth between years. However, combining the male and female data led to a significant increase in the log-likelihood ratio, in- dicating different growth rates for males and females. Males grew faster and larger than females, similar to other panulirid species (Kanciruk 1980). The final set of data is shown in Table 2c. Estimated mean growth curves for males and females, combined for both years, are shown in Figure 9. The algorithm also predicts the age at which each animal was initially caught. The estimated non-parametric density function of age at first capture for the 1985 and 1986 data showed a clear mode around 525 days from settlement. Growth of the lobsters in Bahia de la Ascension was rapid. Males and females enter the fishery at 74 mm CL, 1.65 and 1.7 years, respectively, after settling. Allowing for a six-month larval period (Lewis 1951), males and females enter the fishery at approximately 2.15 and 2.2 years of age, respectively. The analysis of mean weekly growth rates (Hunt and Lyons 1986) combining the 1985 and 1986 data (not illustrated) did not show any points of inflection, sug- gesting that there is no marked decrease in the growth rate of the lobsters in the bay. Movements Most of the recaptured lobsters were caught within the boundaries of the coop- erative's fishing ground. Lobsters that Figure 7 Size-frequency distribution of Panulirus argiis caught in sampling zones II, III, IV, and VI during the 1986 tagging operation in Bahia de la Ascensi6n. Lozano-Alvarez et al.: Fishery, growth, and movements of Panulirus argus 85 Table 1 Mean carapace length (J CL mm) of Panulirus argus caught in each of the six sampling zones. In 1986, no sampling was performed in zone I, and only five individuals were obtained in zone V. Mean CL's followed by corresponding letter in row are not significantly different (P>0.05). Means with different letter in row are significantly different (P<0.05), using test for unplanned comparisons among pairs of means. Zone IV III II VI V I (a) 1985 60.14" 60.89" 62.34" 67.63 b 70.63 b 78.73 1 ' Zone IV II III VI (b) 1986 63.34" 64.85" b 66.41 h 78.14 c 50 60 70 80 90 100 Carapace length (mm) Figure 8 (a) Size at initial capture of 148 male Panulirus argus tagged at Bahia de la Ascension, and (b) size of same males at recapture during first three months of 1985-86 fishing season. Shadings denote individual lobters and show carapace length increment by selected size cohorts, to em- phasize variability in growth. It is possible that more than one molt is involved in some cases. Table 2 Estimates of mean growth parameters for Panulirus argus form. T = dispersion matrix of the coefficients of the growth curve; o 2 = variability around an individual curve. Ses Year L^ k (a) Fabens' (1965) approach Males 1985 101.898 -0.0059 Males 1986 113.815 -0.0049 Females 1985 85.970 -0.0109 Females 1986 148.284 -0.0018 (b) Maximum likelihood approach (Palmer et al. 1988) Males 1985 255.464 -0.00057 Males 1985 261.501 -0.00054 Females 1985 222.807 -0.00065 Females 1986 218.149 -0.00067 p / 10.295 -0.00004 \ 2.16x lO" 10 ] 6 2 = 25.426 (c) Maximum likelihood approach (1985 and 1986 combined) Males 257.204 -0.00056 Females 215.605 -0.00068 p / 10.349 -0.000047\ 2.19 x 10 10 / b- = 25.421 E 125 E 5 100 U) c OJ . 75 o CO S 50 CO ° 25 *?»' :>;>>* 200 400 600 800 1000 1200 1400 Days from settlement Figure 9 Growth curves for male and female Panulirus argus, as estimated by the maximum likelihood approach. Only the continuous lines are based on actual data; broken lines are extrapolations ac- cording to model. dispersed from their zones of initial capture tended to move toward and along the reef in both years (Figs. 10, 11). As an example, of the 79 recaptured lobsters that had been tagged in zone III in 1985 (total number tagged in zone III = 581), 29 remained in zone III 86 Fishery Bulletin 89(1). 1991 (a) (b) \5-^s~\ (I I J Pa jar os £// * ▼ Caught at Punta Herrero (c) (d) Pajaros [Z/^ 2* Caught at Punta Herrero (e) Figure 10 Recorded movements of Panulirus argus tagged in sampling zones II-VI in Bahia de la Ascension, in 1985-86. Shaded areas denote sampling zones; arrows indicate direction of movements and arrowheads the sites of recapture. Numbers indicate lobsters that moved in each direction from denoted sampling zone. and 42 moved to areas nearer or at the outer reef (Fig. 10b). However, because no lobster fishing was con- ducted in the inner half of the bay, no data have been obtained to indicate the possible movement of lobsters to that area. In the 1985-86 fishing season, eleven lobsters were recaptured outside of the bay by fishermen of other cooperatives. Of these, ten had traveled south, while only one had gone north (Fig. 10). The longest straight- line distance traveled by an animal was 45km. All the recaptured lobsters that were tagged in 1986 were caught in the fishing grounds of the cooperative "Pescadores de Vigia Chico." The longest distance traveled by any of these was 23km (Fig. lib). No animals were caught north of Boca Paila or south of Punta Herrero (Fig. 10). However, fishing effort immediately outside the bay was restricted to skin diving on the reef to depths of about 15 m, so any lobsters that moved beyond that depth would not be recaptured. Reproduction During the 1985 tagging program, only four individuals (67.2, 82.5, 83.7, and 116.8mm CL) of 1819 females had spermatophores attached, and only one individual (76.0mm CL) carried eggs. In the 1986 program, five individuals (87.0, 90.2, 94.2, 94.9, and 100.3mm CL) Lozano-Alvarez et al.: Fishery, growth, and movements of Panulirus argus 87 of 689 females were found with re- mains of empty egg capsules and/or eroded spermatophores. In both years, all of the females that showed signs of reproductive activity were caught on the edge of the reef (zones II and V, Fig. lb). No other evidence of reproductive activity was observed. Discussion A decline of CPUE from the begin- ning until the end of the fishing sea- son has been reported for other Panulirus argus fisheries (Warner et al. 1977, Lyons et al. 1981). How- ever, in the fishery of Bahia de la Ascensi6n, the decline from the first month of the season to the second is sharp. Eggleston et al. (1990) sug- gest that casitas enhance survivor- ship of juvenile lobsters by pro- tecting them from their predators, hence increasing lobster production. Thus, during the closed season, in the absence of fishing mortality, it is possible that the aggregation effect of casitas on juvenile lob- sters, in conjunction with recruit- ment by rapid growth, result in the catch from a casita being greater during the first month of the season than during the remainder of the season. The high level of recaptures during both the 1985-86 and 1986-87 seasons suggests a high level of fishing mortality on the population in Bahia de la Ascension. The failure to recapture any of the animals tagged in 1985 during the tagging program in 1986, and the fact that only four were recaptured during the fishing season in 1986-87, may reflect tag loss, high natural mortality, or a strong emigration from the bay. The latter is supported by the movements of the recaptured lobsters. We could not separate fishing mortality from natural mortality because there appeared to be both recruit- ment by growth of small lobsters throughout the season as well as immigration onto the casitas from other areas. This is sustained by the monthly size composi- tion of the catch by the fishery (Fig. 2). Those catches showed a nearly constant size distribution, with a mode near the minimum size limit, indicating recruitment by growth to the fishery throughout the fishing season. (a) ^1 } Punta Pajaros (c) (b) (d) Figure 1 1 Recorded movements of Panulirus argus tagged in sampling zones II, III, IV, and VI in Bahia de la Ascension, 1986-87. Shaded areas denote sampling zones; arrows indicate direction of movements and arrowheads the sites of recapture. Numbers in- dicate lobsters that moved in each direction from denoted sampling zone. The estimates of the growth parameters by the Fabens' method showed great variability (Table 2a), which could be interpreted in two ways: (1) Lobster growth differed greatly interanually, or (2) the pro- cedure yielded unreliable estimates. Palmer et al. (1988) suggested that the Fabens method does not ex- plicitly model individual variability in growth (e.g., Fig. 8), and that it produces inconsistent estimates of the asymptote of growth. Alternatively, the maximum likelihood estimates of the mean curves did not show great variability, so both years could be pooled to obtain a final set of param- eters. The reasonableness of the estimated parameters was further confirmed by the fact that P. argus can attain sizes much larger than the asymptotic sizes estimated from the Fabens method (Sutcliffe 1957, Munro 1974, Olsen and Koblic 1975, Farrugio 1975). However, the tagged lobsters were mostly juveniles and young adults as further demonstrated by the lack 88 Fishery Bulletin 89(1), 1991 of an inflection point in their mean weekly growth rate. Thus, the estimated parameters may reflect growth rates of immature lobsters only, and those of repro- ductive adults could change the last part of the curve (Fig. 9). Lyons et al. (1981), utilizing a method that involved mean growth rates obtained from several authors, estimated an age of slightly more than two years after settlement as postlarvae for P. argus measuring 76 mm CL, allowing for a nine-month larval period. With our maximum likelihood results, and considering the same nine-month larval period, the estimated age for a 76 mm CL lobster would be 2.5 years. Munro (1974) produced a growth curve for P. argus based on data from 156 lobsters tagged and recaptured in Florida and Belize. His estimated age of one year after settling as postlarvae for lobsters measuring 45 mm CL agrees closely with the estimate obtained in the present study by the maximum likelihood approach. Peacock (1974) tentatively estimated an age of one year for 50mm CL P. argus, as did Eldred et al. (1972) and Witham et al. (1968). Therefore, the maximum likelihood approach utilized in this paper seems to have provided a useful set of growth parameters for juvenile and young adult P. argus, with the additional advantage of separating growth data between males and females. The few signs of reproductive activity in female lobsters near the reef, in conjunction with small cara- pace length, indicated that the lobster population in Bahi'a de la Ascension was probably composed mainly of juveniles. Lyons et al. (1981) found little evidence of mating activity of P. argus in the shallow Florida Keys, and they stressed that almost 90% of the spawn- ing occurred at their reef and deep-water stations. Peacock (1974), Davis (1975), and Kanciruk and Herrn- kind (1976) also reported an absence of reproductive activity in shallow bank or lagoon areas. The movements demonstrated by the tagging pro- gram indicate a displacement of lobsters from shallows toward deeper habitats offshore. This was also sup- ported by the analysis of the size composition by zones (Figs. 7, 8; Table 1), which indicated that the lobsters were smaller in the innermost sampling zones com- pared with those caught near or on the reef. Buesa (1970) and Cruz et al. (1986) suggested that juvenile P. argus in Cuba live in protected areas with seagrass beds and move towards the outer reefs as they grow. Other authors that mention similar movements for juvenile P. argus are Peacock (1974) in Barbuda, Olsen and Koblic (1975) in the U.S. Virgin Islands, Warner et al. (1977), Davis (1979), and Lyons et al. (1981) in Florida. Although northern and southern movements were made by lobsters which left the bay, southerly move- ments predominated. In a three-year study of move- ments of P. argus in Biscayne Bay, Florida, Davis (1979) found southerly movements of tagged lobsters during the first year, northerly movements in the sec- ond year, and both northerly and southerly displace- ments during the third. He concluded that juvenile lobsters from Biscayne Bay are recruited into virtual- ly the entire Florida fishery. The extent of the move- ments made through deeper water by the lobsters tagged in Bahi'a de la Ascension— and their final destination— is still unknown, because from Tulum to Mahahual (Fig. la) lobsters are fished only in the bays and on the shallow parts of the reef. A winter migra- tion, similar to that which occurs at the northeastern end of the Yucatan Peninsula (Kanciruk and Herrnkind 1978), may take place in deeper waters outside the cor- al reef that runs across the front of the bay. Small size, rapid growth, movements toward the reef areas, and lack of reproductive activity all serve as evidence that the population of lobsters in Bahi'a de la Ascension is composed of juveniles. We hypothesize the existence of a population composed of reproductive adults off the coral banks of Bahia de la Ascension, an area that is not currently being fished. The existence of adult stocks outside of the bay and the output of lobsters from the bay into offshore deeper areas are issues that need to be assessed for future management plans. In addition, although casitas may provide critical refuge for juvenile lobsters from their natural predators (Eggleston et al. 1990), the long-term effects of the casitas on the lobster populations, as well as on the benthic communities associated with sea- grasses and on the stability and structure of the seagrass beds themselves, remain to be determined through future field studies. Acknowledgments We acknowledge the field assistance of MariCarmen Martinez, Fernando Negrete, Gabriel Carrasco-Zanini, Alfredo Velazquez, Juan Garcia, Edith Zarate, David Ortega, Carlos Rico, Alejandro Torres, and Jorge Simonin. The first three also helped with data process- ing. Mark Palmer, David Gutierrez, Nick Caputi, and Norm Hall gave mathematical advice. William G. Lyons, Alan Campbell, and two anonymous reviewers greatly improved the manuscript. The study was sup- ported by Universidad Nacional Autonoma de Mexico, CSIRO-Australia, and Consejo Nacional de Ciencia y Tecnologia de Mexico (Joint Project No. 140134D202). Scientific fishing permits extended by the Fisheries Secretary of Mexico made this study possible. We also express our thanks to the fishermen of the several cooperatives that assisted the study, particularly the "Pescadores de Vigia Chico." Lozano-Alvarez et al.: Fishery, growth, and movements of Panulirus aigus 89 Citations Buesa, R.J. 1970 Migraciones de la langosta (Panulirus argus). Mar Pesca 60:22-27 [in Spanish]. C'hittleborough, R.G. 1974 Development of a tag for the western rock lobster. CSIRO Div. Fish. Oceanogr. Rep. 56:1-19. Cruz, R., R. Brito, E. Diaz, and R. Lalana 1986 Ecologia de la langosta (Panulirus argus) al SE de la Isla de la Juventud. II. Patrones de Movimiento. Rev. Invest. Mar. 7(3): 19-35 [in Spanish, Engl, abstr.]. Davis, G.E. 1975 Minimum size of mature spiny lobsters, Panulirus argus, at Dry Tortugas, Florida. Trans. Am. Fish. Soc. 104:675-675. 1979 Management recommendations for juvenile spiny lobsters, Panulirus argus, in Biscayne National Monument, Florida. South Fla. Res. Cent. Rep. M-530, Homestead, FL 33030. 32 p. Dempster, A.P., N.M. Laird, and D.B. Rubin 1977 Maximum likelihood from incomplete data via the EM algorithm (with discussion). J. R. Stat. Soc. Ser. 39:1-38. Eggleston, D.B., R.N. Lipcius, D.L. Miller, and L. Coba-Cetina 1990 Shelter scaling regulates survival of juvenile Caribbean spiny lobster Panulirus argus. Mar. Ecol. Prog. Ser. 62: 79-88. Eldred, B., C.R. Futch, and R.M. Ingle 1972 Studies of juvenile spiny lobsters, Panulirus argus. in Biscayne Bay, Florida. Fla. Dep. Nat. Resour. Mar. Res. Lab., Spec. Sci. Rep. 35, 15 p. Fabens, A.J. 1965 Properties and fitting of the von Bertalanffy growth curve. Growth 29:265-289. Farrugio, H. 1975 Observations sur deux langoustes de la Martinique: Panulirus argus et Panulirus guttatus; premieres donnees biometriques et etude comparee de leurs croissances relatives. Sci. Peche, Bull. Inst. Peches Mar. 240:11-20 [in French], Hoel, P.G. 1976 Elementary statistics. John Wiley, NY, 361 p. Hunt, J.H., and W.G. Lyons 1986 Factors affecting growth and maturation of spiny lob- sters, Panulirus argus, in the Florida Keys. Can. J. Fish. Aquat. Sci. 43:2243-2247. Jordan, E. 1988 Estudio regional de los arrecifes coralinos del Mar Caribe mexicano: Su potencial de uso. Informe Final, Proyecto PCCNA-021928, Inst. Cienc. del Mar y Limnol Univ. Nal. Aut6n. Mexico/Consejo Nal. Cienc. y Tecnologia [in Spanish], Kanciruk, P. 1980 Ecology of juvenile and adult Palinuridae (spiny lobsters). In Cobb, J.S., and B.F. Phillips (eds.), The biology and manage- ment of lobsters, vol. II, p. 59-96. Academic Press, NY. Kanciruk, P., and W.F. Herrnkind 1976 Autumnal reproduction in Panulirus argus at Bimim, Bahamas. Bull. Mar. Sci. 26:417-432. 1978 Mass migration of spiny lobsters, Panulirus argus (Crus- tacea: Palinuridae): Behavior and environmental correlates. Bull. Mar. Sci. 28:601-623. Laird, N.M., N. Lange. and D. Stram 1987 Maximum likelihood computations with repeated measures: Application of the EM algorithm. J. Am. Stat. Assoc. 82:97-105. Leslie, P.H., and D.H.S. Davis 1939 An attempt to determine the absolute number of rats on a given area. J. Anim. Ecol. 8:94-113. Lewis. J.B. 1951 The phyllosoma larvae of the spiny lobster Panulirus argus. Bull. Mar. Sci. Gulf Caribb. 1(1):89-103. Lozano, E., P. Briones, and B.F. Phillips 1989 Spiny lobster fishery at Bahia de la Ascensi6n, Q.R. In Chavez, E. (ed.), Proc, Workshop Australia-Mexico on marine sciences, Merida, Mexico, July 1987, p. 379-391. Centro de Invest. Estudios Avanzados, Merida 97310, Mexico. Lyons, W.G., D.G. Barber, S.M. Foster, F.S. Kennedy Jr., and G.R. Milano 1981 The spiny lobster, Panulirus argus, in the middle and upper Florida Keys: Population structure, seasonal dynamics and reproduction. Fla. Mar. Res. Publ. 38, 38 p. Miller, D.L. 1982 Construction of shallow water habitat to increase lobster production in Mexico. Proc. Gulf Caribb. Fish. Inst. 34: 169-179. Munro, J.L. 1974 The biology, ecology, exploitation and management of Caribbean reef fishes. Part V.l.-The biology, ecology and bionomics of Caribbean reef fishes: Crustaceans (Spiny lobsters and crabs). Res. Rep. 3, Zool. Dep., Univ. West Indies, Kingston, Jamaica, 57 p. Olsen, D.A., and I.G. Koblic 1975 Population dynamics, ecology and behavior of spiny lobster, Panulirus argus, of St. John, USVI. II. Growth and mortality. Nat. Hist. Mus. Los Angel. Cty. Sci. Bull. 20:17-22. Palmer, M.J. 1986 Calibration of noisy scrub bird with repeated measure- ments. CSIRO Div. Math. Stat. Consult. Rep. WA(C) 86/3, Floreat Park, W.A. 6014, Australia. Palmer, M.J., B.F. Phillips, and G.T. Smith 1988 Application of random coefficient models to capture- recapture data. Submitted to Biometrics. Peacock, N.A. 1974 A study of the spiny lobster fishery of Antigua and Bar- buda. Proc. Gulf. Caribb. Fish. Inst. 26:117-130. Secretaria de Pesca 1987 Pesquerias Mexicanas: Estrategias para su administra- tion. Dir. Gral. Admin, de Pesquerias, Secretaria de Pesca, Alvaro Obreg6n 269, Mexico, D.F., 1061 p. [in Spanish]. Sokal, R.R., and F.J. Rohlf 1981 Biometry. W.H. Freeman, San Francisco, 859 p. Sutcliffe, W.H. Jr. 1957 Observations on the growth rate of the immature Ber- muda spiny lobster, Panulirus argus. Ecology 38:526-529. Warner, R.E., C.L. Combs, and D.R. Gregory Jr. 1977 Biological studies of the spiny lobster, Panulirus argus (Decapoda: Palinuridae) in south Florida. Proc. Gulf Caribb. Fish. Inst. 29:166-183. Witham, R., R.M. Ingle, and E.A. Joyce Jr. 1968 Physiological and ecological studies of Panulirus argus from the St. Lucie Estuary. Fla. Board Conserv. Mar. Res. Lab. Tech. Ser. 53:1-31. Abstract .- Tilefish Lopholatilus chamaeleonticeps and yellowedge grouper Epinephelus flavolimbatus are deepwater fishes and targets of a relatively recent bottom longline fishery in the Gulf of Mexico. They are long-lived, slow growing, have very limited movements and distri- bution, and are susceptible to long- lines. However, population size and life-history parameter estimates are generally unknown for Gulf fish. This study compared two methods for estimating population sizes to deter- mine the most cost-effective one for use on long-term fishery-indepen- dent stock assessments. Bottom long- lines were used to deplete fish from a small area, and a regression of catch per effort on cumulative catch was used to estimate the area's popu- lation prior to fishing. The popula- tion was also estimated by counting fish burrows from a submersible and expanding the mean number per unit area by the study site's area after correcting for the number of occu- pied burrows. Longlines and submer- sibles provided significantly different estimates of tilefish populations, the only species for which estimates could be compared. Longline esti- mates were probably more accurate because errors in area estimation and double counting were evident in submersible data. Longlines were less expensive to operate ($5000 vs. $8000 per day) and they afforded col- lection of size, age, and sex data on each fish caught. These data were not available from the submersible. Longlines could be used more cost- effectively than submersibles in determining long-term population changes. However, direct observa- tion of fish behavior was not avail- able from longlines, but was from the submersible. Submersibles also pro- vide data on habitat and gear assess- ment, including deployment, effici- ency, bait predation, and potential catch loss during retrieval. Comparison of Two Techniques for Estimating Tilefish, Yellowedge Grouper, and Other Deepwater Fish Populations Gary C. Matlock Fisheries Division, Texas Parks and Wildlife Department 4200 Smith School Road, Austin, Texas 78744 Walter R. Nelson Southeast Fisheries Science Center, National Marine Fisheries Service, NOAA 75 Virginia Beach Drive, Miami, Florida 33149 Robert S. Jones Marine Science Institute. University of Texas at Austin P.O. Box 1267, Port Aransas, Texas 78373 Albert W. Green Environment Assessment. Texas Parks and Wildlife Department 4200 Smith School Road, Austin, Texas 78744 Terry J. Cody Coastal Fisheries Branch, Texas Parks and Wildlife Department 100 Navigational Circle, Rockport, Texas 78382 Elmer Gutherz Mississippi Laboratory, National Marine Fisheries Service, NOAA P.O. Drawer 1207, Pascagoula, Mississippi 39567 Jeff Doerzbacher I 1505 Oak Branch Drive, Austin, Texas 78737 Manuscript accepted 1 August 1990. Fishery Bulletin, U.S. 89:91-99 (1991). Tilefish Lopholatilus chamaeleonti- ceps support an economically impor- tant bottom longline fishery in the Mid- Atlantic Bight (Grimes et al. 1980, Turner 1986), and are the focus of a developing longline fishery in the South Atlantic Bight and the Gulf of Mexico (Katz et al. 1983, Low et al. 1983). Impacts of this development in the Gulf are unknown because popu- lation sizes and life-history param- eter estimates there are generally unknown. However, the potential for recruitment overfishing appears large even at relatively low fishing ef- fort because of the fish's life history (Harris and Grossman 1985). The fish is long-lived, slow growing (Turner et al. 1983, Harris and Grossman 1985), has limited movement (Grimes et al. 1983, 1986), and is restricted to tem- peratures of 9-14°C (Grimes et al. 1986, Freeman and Turner 1977). Tilefish are burrowers, requiring a clay substrate that is soft enough to allow burrowing, but firm enough for maintenance of burrows that may ex- ceed 1 m in diameter and 3 m in depth (Able et al. 1982, Grimes et al. 1986, Twichell et al. 1985). In the Gulf of Mexico this is a narrow geographic area along the outer edge of the con- tinental shelf between depths of 229 and 411m (Nelson and Carpenter 91 92 Fishery Bulletin 89(1), 1991 1968, Wolf and Rathjen 1974). For the above reasons, tilefish are susceptible to capture on longlines (Nelson and Carpenter 1968, Wolf and Rathjen 1974, Grimes et al. 1982, Cody et al. 1981) and overfishing (Harris and Grossman 1985). Yellowedge grouper Epinephelus flavolimbatus are also a target of the developing Gulf longline commer- cial fishery (Prytherch 1983, Graham 1978). However, even less is known about the life history andjtopula- tions of this species than of tilefish. They are apparent- ly present in commercial concentrations off Texas in the 128-274 m depth zone (Nelson and Carpenter 1968). On only 90 trips in the Gulf in 1982 over 65,000kg of yellowedge grouper were landed (Prytherch 1983). However, their frequent distribution around rock and coral formations may preclude sustained commercial catches because of gear loss (Graham 1978). This study was conducted to contrast "fishing out" an area with bottom longlines to direct visual observa- tions from a small research submersible as methods for determining population sizes of tilefish, yellowedge grouper, and other deepwater fishes. The impact of longline fishing on these populations within a limited area was also determined. Materials and methods Preliminary activities In May 1984 the NOAA ship Oregon II spent 6 days surveying an area measuring 95 km east-to-west (94° 10' long, to 95°00'W long.) between 183- and 457-m depths directly south of Galveston, Texas. Bottom con- figuration and acoustic signatures of fish were noted with a color-enhanced fathometer. Eleven bottom longline sets were made during each day to locate areas of high tilefish and yellowedge grouper catch rates (5*0.3 fish/100 hook-hours). Based on these preliminary cruises, specific sites were chosen for the submersible and longline studies described in this paper. Three days (10-13 September 1984) were spent by the Oregon II making detailed charts of each study site prior to the arrival of the submersible. Bathymetric charts of each site were developed using a depth sounder and Loran "C" plotter. These charts represented an area 2.6km 2 (lnmi 2 ) and were contoured by 10-m depth intervals. The trackline interval used in mapping was about one track per 15 m. The charts were duplicated and copies were provided to the Harbor Branch Foundation's RV Johnson prior to the beginning of submersible and fishing activities. This allowed both vessels to track and plot the position of the submersible and location of longline sets precisely. Figure 1 Tilefish study area (center point at 27°40.0'N lat. and 94°22.8'W long.) showing depth contours in meters, submer- sible transect tracks (dashed lines), and distribution of longline sets (solid lines) within the southern part of the study area. Chart represents one square nautical mile. Study area Tilefish A study area (1.9 x 1.1km) was selected off the north Texas coast at 27°40.0'N lat. and 94°22.8W long. (Fig. 1). The area was a broad ridge with a mini- mum depth of 304 m. Approximately 50% of the study area was less than 31 1 m in depth with the bottom drop- ping gradually away to 316-318m at the northern part of the area, and 320-329 m in the southwestern part. The area was almost entirely covered with a soft sand- clay mixture, characteristic of tilefish habitat along the entire U. S. eastern coastline (Freeman and Turner 1977, Able et al. 1982, Twichell et al. 1985, Grimes et al. 1986). However, the substratum was less cohesive than in east coast tilefish grounds. A fin-stabilized metal rod, dropped from a height of 1.2 m by the sub- mersible's manipulator arm, penetrated 80-100 cm in the gulf and 20-30 cm on east coast tilefish grounds (C.B. Grimes et al., NMFS Panama City Lab., unpubl. data for Mid-Atlantic Bight and South Atlantic off Florida). Many of the burrows in the study area were dug at an oblique angle into the substratum or into a sloping face, instead of perpendicular as is characteristic of east coast tilefish on flat bottom (Able et al. 1982, Grimes et al. 1986). It was evident that the low cohesiveness Matlock et al.: Estimating deepwater fish populations 93 of the sediment probably caused the frequently ob- served cave-ins and sloughing of sediment around the mouths of burrows. Extensive secondary bioerosion by galatheid crabs and other burrowers (similar to the mechanism described by Able et al. 1982 and Grimes et al. 1986) further weakens the structure, contributing to cave-ins of the burrow roofs dug at an angle. Con- sequently, large areas of up to 6 x 9m appeared to be plowed or cratered. It is not known if these broad areas were caused by one or several tilefish within each of the "plowed areas," or if one or more generations of tilefish were responsible. Other tilefish burrows found in the study site were more like the typical "vertical" burrows known for these animals (Able et al. 1982, Grimes et al. 1986). Vertical burrows are apparently more stable than oblique ones, requiring less constant re- excavation. Some burrows that had become inactive were filled with sediment at the shaft entrance, but they showed evidence of recent bioerosion around their margins from secondary burrowers. There were also extensive areas that contained numerous depressions, apparently remains of old structures that were 1-2 m across and 0.6-1. 5m deep. Concentrations of 15-20 such depres- sions, 6-8 filled-in burrows, and 2-5 active burrows were common in the study area. Yello wedge grouper A study area (1.28 x 1.28km) slightly inshore of the tilefish site was selected at 27°41.3'N lat. and 94°23.6'W long. (Fig. 2). Depth ranged from 267 m along a central ridge to 311m at the outer edge of the study area. The area was char- acterized by isolated boulders and scattered low rock ridges concentrated in depths of less than 283 m. The bottom was comprised of a sand-clay mixture inter- spersed with rubble and shell. Patches of avalanche anemones (Bolocera sp.) and sea pens (Penatula sp.) were frequently attached in the vicinity of rubble and "hard bottom." Bottom temperature fluctuated little at the study site (12.0-12.9°C). Fishing activities were confined to a 640 x 640m (409,600 m 2 ) quadrant of the study area because time available was shortened by bad weather. Study activities Submersible observations Burrow and fish counts were made from the submersible Johnson-Sea-Link during morning. Accordian-type transects with ran- domly selected starting points and alternating 366 (east- west) and 91-m (north-south) legs (up to 2652 m total distance per dive) were run with the submersible within lm of the bottom and traveling at 1.9km/h. At the end of the east-west portions of each transect leg, the submersible would maintain position while the RV Figure 2 Yellowedge grouper study site (center point at 27°41.3'N lat. and 94°23.6'W long.) showing depth contours in meters, submersible transect tracks (dashed lines), and distribution of longline sets (solid lines) within the northeast part of the study area. Chart represents one square nautical mile. Johnson maneuvered directly above and recorded the position on a LORAN plot of the study area. Five transect dives were made on each of the study areas (Figs. 1, 2). Two of these transects were located com- pletely within the portions of the tilefish area fished with longlines; one transect was completely within the fished portion of the grouper area. The number of adult and juvenile burrows that were "filled-in" (denoting previous occupancy) or were "depressions" (characteristic of long-abandoned bur- rows that had been gradually filled in and smoothed over) were counted as inactive burrows. All others were counted as "active" (currently used by fish); only "active" burrow counts were used in estimating pop- ulations. Burrows within 7.3m on either side of the submersible in the tilefish area and 11.0m in the yellowedge grouper area were counted. These viewing distances were based on the range of visibility and viewing angles and were different in the two study areas because grouper and their burrows were general- ly larger than tilefish. So, grouper could be seen far- ther away than tilefish. All tilefish and yellowedge grouper seen were counted. All other fish seen were identified to species, if possible (names follow Robins et al. 1980), and these identifications were verified 94 Fishery Bulletin 89(1), 1991 using photographic and video records made during each dive. Occupancy of "active" burrows was estimated by observing tilefish diving into burrows and by observ- ing "smoking" (sediment plumes extruded) burrows during each of three other dives. These "positive" sightings provided a minimum estimate of occupancy of the total number of active-looking burrows seen on each dive. Occupancy dives were not conducted in the yellowedge grouper site, because grouper were usual- ly seen outside their burrows and only moved out of sight into burrows when the submersible approached. Bottom longline fishing Intensive fishing activities were conducted with bottom longlines in a portion of each study area (Figs. 1, 2). In the tilefish area longline sets consisted of 100 No. 7 circle hooks on 4.6-cm gangions, attached to a 366-m long ground line with halibut line snaps. Weights were used at both ends of the longline to prevent movement of the line along the bottom. Longlines were baited with squid and fished during daylight only. Two lines were fished on a rotating basis, with sets being soaked for approximate- ly 2 hours. A maximum of eight sets (800 hooks) were made per day. In the yellowedge grouper area "Kali" poles were used to reduce gear loss caused by hanging on large boulders. This gear consisted of 40 2.4-m long PVC pipes weighted at one end and buoyed at the other, with 5 hooks 20.3cm apart on each pole separated vertically by about 0.5m. One pole was clip- ped to a floating mainline every 9 m with a halibut line snap. The "Kali" lines had only the lower end of the PVC pipe and anchors touching bottom. The lines were set at randomly selected locations for approximately 2 hours, using squid for bait. Upon retrieval of all longlines, number of hooks returned, number of baits returned, and catch by species were recorded. Catch-per-unit-effort (CPUE) was calculated for each set using number of hooks returned. Data analysis Population densities and sizes were estimated using the Leslie method modified by Braaten (Ricker 1975, p. 151) from longline catches of Cuban dogfish Squalus cubensis, gulf hake Urophycis cirrata, southern hake U. floridana, and tilefish in the longline-fished portion of the tilefish area; and barrelfish Hyperoglyphe per- ciformis, longspine scorpionfish Pontinus longispinis, southern hake, and yellowedge grouper in the longline- fished portion of the grouper area. A regression of CPUE (in number of fish/ 100 hook-hours) for each longline on adjusted (50% of each day's total catch) cumulative daily catch was calculated using least squares regression weighted by the inverse of variance in daily CPUE (SAS 1982). The cumulative catch was adjusted to reduce bias that can result from using the cumulative catch at the start or end of each fishing period (Braaten 1969). The X-axis intercept was the population estimate (N) for the area fished. Associated 95% confidence intervals were calculated following Sokal and Rohlf (1981, p. 498). Catchability coefficients (for species caught on longlines with significant regres- sions) were equal to the slopes of the regressions (Ricker 1975, p. 150). The assumptions of this technique include constant catchability, geographically closed (within the study area) population (i.e. no recruitment or emigration), no natural mortality, and constant fishing effort (Ricker 1975). Data from the submersible were used to estimate tilefish and yellowedge grouper populations (N) within the area fished with longlines (fished) and the remain- ing portion of the study area (unfished). There were two transects made in the fished portion, and three transects in the unfished portion of the tilefish study area. There was one transect made completely in the fished portion, and four transects in the unfished por- tion of the yellowedge grouper study area. These populations were estimated as the product of the mean number of burrows or fish per km 2 , the percent bur- row occupancy (for tilefish only), and the total km 2 in the study area. The mean number of burrows or fish per km 2 (± 1 SE) was estimated for each transect and each area using the mean and variance equations for the delta-distribution (Pennington 1983, 1986) after transforming the density data along each leg of each transect to natural logarithm. Differences in mean den- sities were tested using the t-test (Sokal and Rohlf 1981). Occupancy was estimated as the mean percent (± 1 SE) from three dives using the "smoking" bur- row data discussed previously using the ratio estimator (Cochran 1977). The variance of the population esti- mates (N) based on burrow counts was calculated as the variance of a product (burrows/km 2 x percent oc- cupancy; Goodman 1960). Differences in the population estimates resulting from the longlines and submersibles were tested using the f-test (Cochran 1977, Sokal and Rohlf 1981) and variances associated with the popula- tion estimates. Results Fish data from longlines and submersibles provided significantly different estimates of tilefish populations. Population estimates for yellowedge grouper could not be made from the longline data because it did not yield a significant regression (Table 1). The number of tilefish (with 95% confidence intervals in parentheses) in the Matlock et al.: Estimating deepwater fish populations 95 Table 1 Population estimates and 95% confidence intervals (CI) for fishes caught on longlines and observed during submersibl e dives on the tilefish and grouper study areas off Galveston, Texas during September 1984. Longline estimates for fishes other than tilefish and yellowedge grouper were made only for species with significant relationships between catch-per-unit-effort and cumulative catch (Ricker 1975). Submersible estimates were based on expansions of mear burrows }r fish per unit area seen on one and two transects in the grouper and tilefish study sites, respectively, to the total fishec area (836,067 m 2 ). Submersible Longline N (95% CI) N (95% CI) No. of 7-intercept Slope No. of based on based on Species days ( ± 1 SE) ( ± 1 SE) R 2 F N (95% CI) transects burrows fish seen Tilefish site Tilefish 7 4.026 ± 0.835 -0.050 + 0.012 0.382 18.574** 81 (39-128) 2 446(364-522)' 134(121-147) Cuban dogfish 7 5.410 ± 1.190 -0.084 ±0.020 0.371 17.677** 65 (24-108) Gulf hake 7 0.491 ± 0.330 -0.005 ±0.010 0.007 0.209NS Southern hake 7 1.874 ± 0.501 -0.043 + 0.013 0.260 10.554** 43(- 27-121) Grouper site Yellowedge grouper 3 0.914 ± 0.383 -0.030 ±0.020 0.140 2.277NS 1 150(105-195) 150(118-182) Southern hake 3 3.445 ± 1.154 -0.052 ±0.020 0.320 6.587* 66 (9-170) Longspine scorpionfish 3 0.995 ± 0.287 -0.013 ±0.013 0.066 0.983NS Barrelfish 3 1.713 ±0.705 occupancy of burrows -0.027 ±0.017 0.157 ,vas 35.6 ± 16.8 (95% CI) 2.605NS "Mean percent fished portion of the study area was 81 (39-128) based on longline catches; the estimate from submersible data was 134 (121-147) based on observed fish (Table 1). These estimates were significantly different from each other (t = 4.939, df = 29, P<0.05). The estimated num- ber of tilefish based on burrow counts (446, 364-522) was four to five times higher than either of the fish- based estimates. The estimated number of yellowedge grouper in the fished portion of the grouper area was 150 (118-182) fish based on submersible grouper data and 150 (105-195) based on burrow data (Table 1). Again, no comparable estimate was made from longline data. The estimated number of tilefish seen per unit area was significantly (t = 3.621, df = 51, P<0.05) greater (about 30%) in the fished area than in the unfished por- tion of the study area. There were also significantly (t = 5.899, df = 42, P<0.05) more yellowedge grouper (about 68%) seen in the fished area than in the unfished area (Table 2). This same pattern was apparent in the burrow data for tilefish (t = 3.737, df = 51, P<0.05) and yellowedge grouper (t = 6.381, df = 42, P<0.05). In the unfished portion of the grouper study area, the mean number of yellowedge grouper burrows seen per km 2 (70, 95% CI = 62-78) was less than 50% of the mean number of yellowedge grouper seen (170, 95% CI = 154-186) (Table 2). However, the number of bur- rows seen in the tilefish study area in both the fished and unfished portions exceeded the number of tilefish seen by about 10 to 20 times. Estimates using submersible data could not be made for southern hake, gulf hake, Cuban dogfish, longspine scorpionfish, and barrelfish since they were generally not seen from the submersible. However, longlines caught 322 of these fishes during the 12,000-13,000 hook-hours of fishing on the tilefish and grouper study areas. Longline catch rates declined through time for southern hake in both study areas and Cuban dogfish caught in the tilefish study area, but no significant change in catch rates was detected for gulf hake in the tilefish study area or longspine scorpionfish or bar- relfish in the grouper study area. Therefore, popula- tion estimates and 95% confidence intervals were only made for southern hake and Cuban dogfish; 43 (-27-121) and 65 (24-108) fish, respectively, in the tilefish study area and 66 (9-170) southern hake in the grouper study area (Table 1). More species were seen from the submersible than were caught on longlines (Table 3), but more population estimates could be made from longline data than from submersible data. 96 Fishery Bulletin 89(1). 1991 Table 2 Mean number of tilefish and yellowedge grouper and burrows of each species seen per km 2 on transects by a submersible in areas fished with longlines (study sites) and adjacent areas. Mean number of tilefish burrows seen was multiplied by mean percent occupancy (0.36 + 0.16) to estimate number of tilefish and grouper. Width of each transect was 14.6 and 22.0 m for tilefish and grouper areas, respectively. The number of transect legs (n) is indicated for each transect. Tilefish Yellowedge grouper Area Transect n Fish Burrows Transect n Fish Burrows X J SE X SE X SE x SE Fished with A 11 109 24 1500 80 F 11 370 30 370 50 longlines B Pooled 13 24 150 134 29 13 1400 1500 80 40 Adjacent (unfished) C D 7 11 12 109 5 19 900 900 300 40 G H 11 11 70 300 9 30 60 8 120 20 E 11 150 19 3400 70 I 11 100 20 50 9 Pooled 29 100 7 1700 60 Pooled 33 170 8 70 4 Pooled 53 117 3 1600 30 44 220 7 140 6 Discussion Longlines appear to be a more cost- effective means of monitoring fish population changes than submer- sibles. However, longlines kill all the collected fish whereas submer- sibles do not. More data can be col- lected on each caught fish at a lower cost with longlines than with a sub- mersible. Size, age, and sex data could be collected from longline catches at a cost of about $5000 per day (in 1984 U.S. dollars), while none of these data were available from the submersible even though it cost about $8000 per day to op- erate. Additionally, the tilefish pop- ulation based on burrow data from the submersible may have been overestimated because (1) "active" burrows were overestimated, (2) width of each burrow-count tran- sect was underestimated, and (3) double counting occurred when transects crossed or came close to crossing. The number of tilefish estimated from tilefish seen was about 50% larger than the estimate based on longlines, and was about four times less than the estimate based on burrow counts. The number of burrows may be more a reflec- tion of population size prior to exploitation if this area was heavily fished prior to our study. Table 3 List of species caught on onglines and seen from submersible in the tilefish and grouper study areas. An X indicates presence; blank indicates absence. Common name Scientific name Longline Submersible Tilefish Lopholatilns chamaeleonticeps X X Yellowedge grouper Epinephelus flavolimbatus X X Southern hake Urophycis Jloridana X X Gulf hake Urophycis cirrata X Cuban dogfish Squalus cubensis X X Longspine scorpionfish Pontinus longspinus X X Barrelfish Hyperoglyphe perciformis X X Night shark Carcharhinus signatus X Chain dogfish Scyliorhinus retifer X X Chub mackerel Scomber japonicus X Snowy grouper Epinephelus niveatus X X Beardfish Polymixia lowei X X Moray Gymnothorax kolpos X X Conger eel Conger oceanicus X Argentines Argentine sp. X Shortnose greeneye Chlorophthalmus agassizi X Reticulate goosefish Lophiodes reticulatus X Red hake Urophycis chuss X Buckler dory Zenopsis conehifera X Slimehead Gephroberyx darwini X Deepbody boarfish Antigonia capros X Longspine snipefish Macrorhamphosus scolopax X Longtail bass Hemanthias leptus X Yellowfin bass Anthias nicholsi X Bladefin bass Jeboehlkia gladifer X Polylepion n sp. X Flatheads Bembrops sp. X Scorpionfish Scorpanenodes sp. X Tilefish populations were probably underestimated using longline data. But the amount of bias is unknown. Capture probabilities were not constant, and this usual- ly leads to underestimates (White et al. 1982). Recruit- Matlock et al.: Estimating deepwater fish populations 97 merit and emigration rates are unknown, but were probably low. If they occurred, recruitment must have been less than emigration because the populations were depleted in the study area. As recommended by Ricker (1975), our fishing effort was concentrated into "a rather short period of time" to minimize the effects of violating this assumption and that of no natural mortality. The grouper population based on submersible fish data may have been overestimated because the esti- mated number of fish exceeded the estimated number of burrows and double counting of fish probably oc- curred. On one dive, the same fish (based on a scar on the lower jaw) was seen four different times. Additional research is needed to determine the im- pacts of each of the above factors on fish population estimates based on counts made from submersibles. Future burrow counts should include all burrows, not just apparently active ones. Transect width should be accurately measured by counting burrows only within the range of a fixed physical extension from the sub- mersible. Occupancy rates for tilefish and yellowedge grouper should be determined in randomly selected areas at night when they are most likely to be in their burrows. Although no significant relationship between CPUE and cumulative catch was found for grouper, a more intensive effort should be made before discounting this technique. Additional longline collections over a longer period for yellowedge grouper are needed to determine if using the Leslie method is feasible. Longlines can potentially impact stocks of tilefish. The population estimate of tilefish in the study area (39-128) and the catch made by the intensive fishing effort (79) indicate that from 62 to 100% of the fish were taken out of the area by an effort of approximate- ly 6000 hook-hours, which is a 1.5-2 day effort by a commercial longliner (Prytherch 1983). Catch rates in the northern Gulf of Mexico in 1982 averaged 1-6 fish per 100 hook-hours (Prytherch 1983). Based on the estimated population size within the area, the initial catch rates indicate that the longline effectively catches all fish out of an area that is at least 12 m wide. Some fish are attracted from greater distances (Grimes et al. 1982), and some near the longline are not caught. But the number removed from the population is equal to the length of the longline x a width of 12 m x fish density. Estimates of the total portion of the Gulf of Mexico inhabited by tilefish have not been developed, but the optimal areas are limited by depth, temperature, and bottom type (Grimes et al. 1980, 1986; Grossman et al. 1985). This, combined with slow growth rate, longev- ity, and low natural mortality (Turner et al. 1983, Har- ris and Grossman 1985), indicate that overfishing could easily take place if substantial effort is expended in tilefish habitat. This is especially true in light of the susceptibility to mass mortalities caused by sudden temperature reductions (Hachey 1955). Data from South Carolina tilefish habitat show a substantial decline in catch rate and mean size over a 4-5 year period with low to moderate effort (Low et al. 1983). Further, the number of tilefish burrows per km 2 in the Middle and South Atlantic Bights in the early 1980s was 241 and 125, respectively (Able et al. 1987). These estimates are much lower than the 1600 burrows per km 2 in the Gulf of Mexico estimated in this study. More extensive longline studies of yellowedge grouper catches are required to assess the effect of longlines on these populations. The population estimate of yellowedge grouper in the yellowedge study site from fishing activities was not significant, but the best estimate (26 animals) from the non-significant regres- sion may be realistic. The regression indicated that similar fractions of the yellowedge grouper population (40%) would be caught at similar levels of effort as com- pared with tilefish, and similar impacts from the long- line fishery might be expected. However, the results may not be analagous because different gear were used in the two areas. While the association with hard substrate and high relief was expected for yellowedge and other groupers, the burrowing habits were not expected. A detailed description of grouper habitat and burrow character- istics have been provided by Jones et al. (1989). The finding that this species also inhabits burrows was especially significant. If this were the only habitat, it would limit their distribution and increase their sus- ceptibility to fishing once they are located. However, this species is also associated with rock and reef habitat typical of other grouper species. This diversity of habitat should enhance the survivability of the species overall, but it makes a part of the population more susceptible to longline fishing. The uneven distribution of tilefish and yellowedge grouper between fished and unfished areas was also unexpected. Reasons for the differential distribution are not apparent. But the effects of depth, temper- ature, and bottom type on the fish were probably involved. This study demonstrates the need for additional research to estimate population sizes and life-history parameters for deepwater Gulf fishes to quantify the amount of fishing they can support. Routine monitor- ing of these populations could be accomplished with longlines fished during August through October. Limited data on tilefish and yellowedge grouper have been collected with bottom longlines by the National Marine Fisheries Service since 1968 (Table 4). However, the data are insufficient to identify trends. 98 Fishery Bulletin 89(1), 1991 Table 4 Mean catch (no./lOO hook-hours) ± 95% confidence interval of tilefish and yellowedge groupei month, 1968-84, in the area bounded by 27°37'-27°50' lat. and 93°32'-95°21' long. Numbers in Blanks indicate no data collected. on NMFS bottom longline sets each parentheses indicate numbers of sets. Species Year Jan. Feb. May Aug. Sep. Oct. Nov. Tilefish 1968 9.5 ±12.2 1973 (8) 20.8 ± 6.9 6.8 ± 5.2 1975 j (14) 2.1 ±6.2 1976 7.1 ±4.1 (3) 1977 9.1 ±9.3 (8) 10.0 ±12.9 1983 (6) (4) 6.1 ±3.4 1984 6.9 ±8.0 (9) (7) 3.1 ± 1.3 (57) Yellowedge grouper 1968 1973 5.5 ±8.3 (11) 18.4 0.7 ± 1.2 1975 (1) (8) 0.0 1976 2.5 ±6.0 (1) 1977 0.0 ±0.0 (3) 1981 (3) 0.0 ±0.0 1983 3.7 ±6.7 (2) 1984 1.2 ±2.1 (6) (8) 2.1 ± 1.3 (16) Natural and fishing mortality estimates, growth rates, population structure, distribution throughout all life stages, and weight landed should be determined and used in population simulation models to assist man- agers in protecting these resources from overfishing. A fishery-independent sampling program using long- lines is recommended for monitoring the status of tilefish, hake, barrelfish, longspine scorpionfish, Cuban dogfish, and possibly yellowedge grouper populations. This is a more appropriate source of fish for mortality estimates than are commercial landings (Low et al. 1983, Winters and Wheeler 1985) and can yield reliable population size estimates if catchability coefficients are known. Acknowledgments Funding for submersible activities involved in this project was provided by the NOAA Office of Undersea Research. The Harbor Branch Oceanographic Institute participated on a cost-sharing basis. Support for the fishing activities and the NOAA Ship Oregon II was provided by the National Marine Fisheries Service. We would like to thank Captain Bill Abney and the crew of the RV Johnson for their dedication and long hours, and submersible pilots Tim Askew, Marshall Flake, Don Liberatore, Steve Hall, and the submersible sup- port crew for the Johnson-Sea-Link for doing whatever was necessary to accomplish mission objectives. We would also like to thank Captain Gunnar Gudmundsson and the crew of the NOAA Ship Oregon II for their sup- port and, especially, G. Michael Russell for serving as field party chief to insure accomplishment of planned fishing activities. Thanks are also extended to Art Crowe and Maury Osborn of the Texas Parks and Wildlife Department for their help in baiting hooks and collecting data. Suggestions from two anonymous reviewers were very helpful in improving the content of the manuscript. Matlock et al.: Estimating deepwater fish populations 99 Citations Able, K.W., C.B. Grimes, R.A. Cooper, and J.R. Uzmann 1982 Burrow construction and behavior of tilefish, Lopholatilus chamaeleonticeps in Hudson Submarine Canyon. Environ. Biol. Fishes 7:199-205. Able, K.W., D.C. Twichell, C.B. Grimes, and R.S. Jones 1987 Sidescan Sonar as a tool for detection of demersal fish habitats. Fish. Bull., U.S. 85:725-736. Braaten, D.O. 1969 Robustness of the DeLury population estimator. J. Fish. Res. Board Can. 26:339-355. Cochran, W.G. 1977 Sampling techniques. John Wiley, NY, 428 p. Cody, T.J.. B.E. Fuls, G.C. Matlock, and C.E. Bryan 1981 Assessment of bottom longline fishing off the central Texas coast, a completion report. Tex. Parks Wildl. Dep., Coast. Fish. Branch, Manage. Data Ser. 22, Austin. 51 p. Freeman, B.L., and S.C. Turner 1977 Biological and fisheries data on tilefish, Lopholatilus chamaeleonticeps, Goode and Bean. Tech. Rep. 5, Sandy Hook Lab., Northeast Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Highlands, NJ 07732, 41 p. Goodman, L.A. 1960 On the exact variance of products. J. Am. Stat. Assoc. 55:708-713. Graham, G. 1978 Preliminary progress report: Bottom longline investiga- tions. Unpubl. manuscr., Sea Grant Coll. Prog., Tex. A&M Univ., College Station, 30 p. Grimes, C.B., K.W. Able, and S.C. Turner 1980 A preliminary analysis of the tilefish, Lopholatilus cha- maeleonticeps, fishery in the Mid-Atlantic Bight. Mar. Fish. Rev. 42(11):13-18. 1982 Direct observation from a submersible vessel of commer- cial longlines for tilefish. Trans. Am. Fish. Soc. 111:94-98. Grimes, C.B., S.C. Turner, and K.W. Able 1983 A technique for tagging deepwater fish. Fish. Bull., U.S. 81:663-666. Grimes, C.B., K.W. Able, and R.S. Jones 1986 Tilefish Lopholatilus chamaeleonticeps, habitat, behavior, and community structure in Mid-Atlantic and southern New England waters. Environ. Biol. Fishes 15(4):273-292. Grossman, G.D.. M.J. Harris, and J.E. Hightower 1985 The relationship between tilefish, Lopholatilus chamae- leonticeps, abundance and sediment composition off Georgia. Fish. Bull., U.S. 83:443-447. Hachey, H.B. 1955 Water replacements and their significance to fishery. Pap. Mar. Biol. Oceanogr., Deep-Sea Res. Suppl. to Vol. 3, p. 68-73. Harris, M.J., and G.D. Grossman 1985 Growth, mortality, and age composition of a lightly ex- ploited tilefish substock off Georgia. Trans. Am. Fish. Soc. 114:837-847. Jones, R.S., E.J. Gutherz, W.R. Nelson, and G.C. Matlock 1989 Burrow utilization by yellowedge grouper, Epinephelus flavolimbatus, in the northwestern Gulf of Mexoci. Environ. Biol. Fishes 26:277-284. Katz, S.J., C.B. Grimes, and K.W. Able 1983 Delineation of tilefish, Lopholatilus chamaeleonticeps, stocks along the United States east coast and in the Gulf of Mexico. Fish. Bull., U.S. 81:41-50. Low, R.A., G.F. Ulrich, and F. Blum 1983 Tilefish off South Carolina and Georgia. Mar. Fish. Rev. 45(4-6):16-26. Nelson. W.R., and J.S. Carpenter 1968 Bottom longline explorations in the Gulf of Mexico. Commer. Fish. Rev. 30(10):57-62. Pennington, M. 1983 Efficient estimators of abundance, for fish and plankton surveys. Biometrics 39:281-286. 1986 Some statistical techniques for estimating abundance in- dices from trawl surveys. Fish. Bull., U.S. 84:519-525. Prvtherch, H.F. 1983 A descriptive survey of the bottom longline fishery in the Gulf of Mexico. NOAA Tech. Memo. NMFS-SEFC-122, Southeast Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Miami, FL 33149, 33 p. Ricker, W.E. 1975 Computation and interpretation of biological statistics of fish populations. Fish. Res. Board Can. Bull. 191, 382 p. Robins, C.R., R.M. Bailey, C.E. Bond, J.R. Brooker, E.A. Lachner, R.N. Lea, and W.B. Scott 1980 A list of common and scientific names of fishes from the United States and Canada. Spec. Publ. 12, Am. Fish. Soc, Bethesda, MD, 174 p. SAS Institute, Inc. 1982 SAS user's guide: Statistics, version 5 edition. SAS Inst., Inc., Cary, NC, 965 p. Sokal, R.R., and F.J. Rohlf 1981 Biometry, 2d ed. W.H. Freeman, San Francisco, 776 p. Turner, S.C. 1986 Population dynamics of and the impact of fishing on tilefish, Lopholatilus chamaeleonticeps, in the Middle Atlantic- Southern New England region during the 1970's and early 1980's. Ph.D. diss., Rutgers Univ., New Brunswick, NJ. 289 p. Turner, S.C, C.B. Grimes, and K.W. Able 1983 Growth, mortality, and age/size structure of the fisheries for tilefish, Lopholatilus chamaeleonticeps, in the Middle Atlantic-Southern New England region. Fish. Bull., U.S. 81:751-763. Twichell, D.C, C.B. Grimes, R.S. Jones, and K.W. Able 1985 The role of erosion by fish in shaping topography around Hudson Submarine Canyon. J. Sediment. Petrol. 55:712-719. White, G.C, D.R. Anderson, K.P. Burnham, and D.L. Otis 1982 Capture-recapture and removal methods for sampling closed populations. LA-8787-NERP, Los Alamos Natl. Lab., Los Alamos, NM, 235 p. Winters, G.H., and J. P. Wheeler 1985 Interaction between stock area, stock abundance, and catchability coefficient. Can. J. Fish. Aquat. Sci. 42:989-998. Wolf, R.S., and W.F. Rathjen 1974 Exploratory fishing activities of the UNDP/FAO Carib- bean Fishery Development Project, 1965-1971: A summary. Mar. Fish. Rev. 36(9): 1-8. AbStr3Ct. — Collections were made for gulf butterfish Peprilus burti along a cross-shelf transect at depths of 5-100 m in the Gulf of Mexico off Texas from October 1977 to July 1980. Butterfish mature at 100-160 mm fork length as they approach age I. Spawning occurs primarily from September through May, but length frequencies indicate it concentrates, or is most successful, in distinct "Winter" (late January-mid-May) and "Fall" (early September-late October) periods that coincide with downcoast, alongshore currents (to- ward Mexico). Gonad data and per- sistence of small fish indicate spawn- ing in winter, but at a low level. Spawning probably occurs offshore and upcoast toward the northcentral Gulf. Surface currents of the cyclonic shelf gyre probably transport eggs/ larvae inshore and downcoast to re- cruit to the bottom in water 5-27 m deep, used as nurseries by butterfish when they are 2-5 months old. But- terfish disperse offshore as they ma- ture and congregate in 36-100 m depths when they are 9-12 months old. They average 130-146mm in fork length at age I in the north- western Gulf, but 120-124 mm at age I and about 170 mm at age II in the northcentral Gulf. Estimates for the von Bertalanffy growth parameters L„, K, and t were 164 mm, 1.99/ year, and -0.20 years, respectively, for pooled northwestern Gulf Winter cohorts and 141mm, 2.69/year, and - 0.06 years, respectively, for pooled Fall cohorts. Somatic growth ceases as spawning approaches in the north- western Gulf, but fish from the northcentral Gulf show large annual size increments. Butterfish reach about 200 mm in fork length, the largest ones occurring in the north- central Gulf. Apparent maximum ages are 1-1.5 years in the north- western Gulf, and 2-2.5 years in the northcentral Gulf. Differences in pop- ulation attributes suggest complete mortality at age I in the northwest- ern Gulf or some unknown combina- tion of an offshore and permanent contranatant spawning or postspawn- ing emigration of adults to the north- central Gulf. The genus Peprilus shows zoogeographic differences in population dynamics near Cape Hat- teras, North Carolina. Manuscript accepted 1 October 1990. Fishery Bulletin, U.S. 89:101-116 (1991). Reproduction, Age and Growth, and Movements of the Gulf Butterfish Peprilus burti* Michael D. Murphy Florida Marine Research Institute, Florida Department of Natural Resources 100 Eighth Avenue SE. St. Petersburg. Florida 33701 Mark E. Chittenden Jr. Department of Wildlife and Fisheries Sciences. Texas A&M University College Station, Texas 77843 Present address: Virginia Institute of Marine Science, School of Marine Science College of William and Mary, Gloucester Point, Virginia 23062 The gulf butterfish Peprilus burti ranges in the Gulf of Mexico (Gulf) from the Yucatan Peninsula to Tam- pa Bay, Florida (Horn 1970) and may occur along the U.S. southeast Atlan- tic coast, depending upon its system- atic status and range extensions during cold spells (Caldwell 1961, Collette 1963, Horn 1970, Persch- bacher et al. 1979). This abundant species is important in the industrial fishery and is commonly discarded by the shrimp fishery in the northern Gulf (Roithmayr 1965, Franks et al. 1972, Gutherz et al. 1975, Chittenden and McEachran 1976). Recent ex- ploratory surveys have found large, commercially valuable concentra- tions of P. burti in the northern Gulf (Vecchione 1987). A preliminary bio- mass estimate for this area is 177,000 MT per 10,164 square miles (Gledhill unpubl.). The life history and population dynamics of this species have not been described in detail, only as brief notes in numerous faunal studies in- cluding Gunter (1945), Hildebrand (1954), Miller (1965), Franks et al. (1972), Christmas and Waller (1973), Chittenden and McEachran (1976), and Allen et al. (1986). The paucity 'Contribution No. 1625 of the Virginia Insti- tute of Marine Science, College of William and Mary. of information reflects difficulty in age determination. Allen et al. (1986) found that hard parts such as oto- liths, scales, opercula, and vertebrae were not useful in age determination. In this paper we use an extensive set of length frequencies to infer age of P. burti and to describe size and age at maturity, spawning seasonal- ity and areas, recruitment, seasonal distribution and movements, growth, maximum size and age, and weight- length, girth-length, and total, fork, and standard length relationships. We also discuss hydrographic condi- tions associated with spawning areas and recruitment, and zoogeographic differences in population dynamics in Peprilus near Cape Hatteras, North Carolina. Methods Collections for Peprilus burti were made along a cross-shelf transect in the Gulf off Freeport, Texas (Fig. 1) from October 1977 through July 1980 aboard a chartered shrimp trawler using twin 10.4-m (34-ft) shrimp trawls with a tickler chain and 4.4-cm stretched mesh in the cod end. Initial stations were located at depths of 9, 13, 16, 18, 22, 27, 36, and 47m. Sampling was expanded to include stations at 5 and 24 m after November 1978 and at 55, 64, 73, 82, 101 102 Fishery Bulletin 89(1), 1991 5 10 15 Nautical Miles Figure 1 Location of sampling area off Free- port, Texas. and 100m after May 1979. Collections were made dur- ing the day through September 1978; thereafter, a day and a night cruise usually were made each month. Usually two tows, consisting of 10 minutes of bottom time, were made at each depth. Exceptions were 8 tows made at 16m, and 24 tows made at 22 m, and only one tow made at most depths prior to October 1978. Our spatial sampling design was a single cross-shelf tran- sect from a sampling frame that encompassed much of the northern Gulf. All P. burti were culled from the catch, measured for total length (TL), fixed in 10% formalin for 2-4 days, and then stored in 70% ethanol. For the period De- cember 1978-November 1979 a total of 300 fish, if available, was selected each month after stratifying the catch into cohorts determined by length-frequency analysis (see below). Specimens were then randomly selected within strata to determine total length, fork length (FL), standard length (SL), total weight (TW), gonad weight (GW), and girth (G) measured vertically from the dorsal fin to the preanal pterygiophore. Sagit- tal otoliths were removed from individuals larger than 75 mm, teased free of saccular and labyrinthian tissue, and then stored dry for later immersion in water and viewing with reflected light under a dissecting micro- scope. Female and immature fish were assigned gonad maturity stages (Table 1) modified from Kesteven's system (Bagenal and Braum 1971). Gonad weight and Table 1 Description of gonad maturity stages assigned to immature and female P. burti. Stage Description 1 Immature Gonads barely or not visible; sex indistinguishable to naked eye. 2 Maturing Virgin Gonads small, opaque; usually thin, streamer-like, running along posterioventral wall of body cav- ity; sex indistinguishable to naked eye. 3 Early Developing/ Sex distinguishable; individual Resting eggs not visible to naked eye but ovarian lamellae visible; ovaries occupy <20% of body cavity. 4 Late Developing I Ovaries occupy 20-50% of body cavity; individual eggs visible in close examination; eggs opaque. 5 Late Developing II Ovaries occupy 50-80% of body cavity; individual eggs distin- guishable; eggs opaque. 6 Gravid Ovaries occupy > 50% of body cavity; translucent eggs present, but make up <50%. 7 Ripe Ovaries occupy > 50% of body cavity; >50% of eggs translucent. 8 Spawning/Spent Ovaries flaccid, pink, with few eggs present; fish large enough to have spawned. Murphy and Chittenden Reproduction, age and growth, and movements of Pepnlus burti 103 Table 2 Final quadratic growth regressions, calculated hatching dates, and von Bertalanffy equations for individual Peprilus burti cohorts and pooled Fall and Winter cohorts. Fits regress mean observed fork lengths (mm FL) on age in days (Days) for the quadratic and on age in years (Years) for the von Bertalanffy growth equations. All regressions are significant at a = 0.05. Final calculated Cohort Final quadratic growth equation r 2 hatching date von Bertalanffy growth equation Fall 77 FL = 0.09 + 0.6004(Days)-0.0007(Days) 2 0.93 25 Sep 77 FL = 139(1 - exp( - 2.22(Years + 0.050))) Fall 78 FL = -0.14 + 0.6971(Days)-0.0009(Days) 2 0.87 8 Sep 78 FL = 138(1 -exp(-3.81(Years + 0.122))) Fall 79 FL = -0.16 + 0.7234(Days)-0.0010(Days) 2 0.97 15 Sep 79 Data not adequate to fit alone Fall pooled FL = 4.04 + 0.6386(Days) - 0.0008(Days) 2 0.91 - - - FL= 141(1 -exp(-2.69(Years + 0.063))) Winter 78 FL = -0.10 + 0.1741(Days)-0.0011(Days) 2 0.95 30 Jan 78 FL= 133(l-exp(-3.06(Years + 0.055))) Winter 79 FL = 0.14 + 0.6931(Days)-0.0008(Days) 2 0.94 16 Mar 79 FL = 1 70 (1 - exp( - 1 .98(Years + 0.043))) Winter 80 FL= -0.09 + 0.6975(Days) 0.93 22 Mar 80 Data not adequate to fit alone Winter pooled FL = 8.18 + 0.6327(Days)-0.0007(Days) 2 0.93 - - - FL= 164(l-exp(-1.99(Years + 0.200))) Fall and Winter pooled FL = 6.28 + 0.6362(Days) - 0.0008(Days) 2 0.91 stage were also determined for randomly selected fish in October 1978 and during December 1979-Feb- ruary 1980 to verify spawning seasons. All lengths presented are fork lengths, and all length frequen- cies are moving averages of three, unless specified otherwise. We also made collections in the northcentral Gulf, at depths of 9-91 m, aboard the FRS Oregon II east (89°00'W-89°30'W) and west (89°30'W-91°30'W) of the Mississippi River during 10-19 April 1980 and 21 April- 1 May 1980, respectively (Rohr et al. 1980). Fish were sampled from these catches without randomiza- tion and measured (FL) to compare with size and age compositions in the northwestern Gulf. Age was determined by length-frequency analysis, that is, the modal group progression analysis modifica- tion (Jearld 1983) of the Petersen method (Lagler 1956, Tesch 1971). Cohorts were specified by the season and year when they hatched, for example, Fall 79 . Cohorts hatched during the periods January-May and Septem- ber-October are hereafter referred to as Winter and Fall cohorts, respectively; this abbreviated terminology corresponds with the major spawning each period (see "Maturation and Spawning Periodicity"). Arithmetic means were used to describe central tendencies in cohort length frequencies. Mean hatching dates used to approximate time scales to calculate growth were determined by a one-step iterative process following Standard and Chittenden (1984). A hatching date of 1 March was assigned to start iterations for the Winter 79 and Winter 80 cohorts because, in those groups, fish 30-75 mm, assumed to be 2-4 months old, first appeared mid-May to mid-June. A hatching date of 1 February was assigned to start iterations for the Winter 78 cohort because, in that group, fish 20-50 mm, assumed to be 1-3 months old, became abundant in mid- April. A hatching date of 1 September was assigned to start the approximation for Fall cohorts because fish 30-75 mm, assumed to be 2-4 months old, first appeared November through early December. Quadratic regression of length-on-age in days was then used to estimate initial x-intercepts for each cohort. Final hatching dates were calculated by using x-intercepts to readjust the initial x-variable (time) scale, so that each final quadratic growth curve passed through the origin (Table 2). Descriptions of spawning periodicity using length frequencies assume early size and age combinations (see Results) predicted from quadratic regression of length-on-age pooling all Winter and Fall cohorts (Table 2). Von Bertalanffy growth equations were fit to length and age (in years) data using the nonlinear least-square parameter esti- mation procedure in FSAS (Saila et al. 1988). Data points described a curvilinear regression and evidenced an asymptote, so equations presented met the mini- mum requirements for a von Bertalanffy fit (Knight 1968, Gallucci and Quinn 1979). Maximum age was approximated by the Beverton- Holt model parameter t L (Gulland 1969) following the definition that only 0.5-1.0% of the catch exceeds age t L or its corresponding length (Alverson and Carney 1975, DeVries and Chittenden 1982). Maximum sizes and ages, and sizes-at-age, presented are termed ap- parent, because they may be affected by emigration of fish approaching age I; if so, they underestimate max- imum size and age, and sizes-at-age. 104 Fishery Bulletin 89(1), 1991 > 20 10 Immature n=513 50 30 10 Maturing Virgin n=614 20n to Early Developing/Resting n = 480 7 ^0 Late Developing I LU n=245 \\ Z> 15- \ (J / \ LU j I tr 5- r V U- .a/ V_ 10n Late Developing II n=55 Gravid '1 n=37 I , Ripe 3 1 n = 9 g Spawning/Spent 20 60 100 140 FORK LENGTH (mm) 180 Figure 2 Length frequencies of immature and female Peprilus burti by maturity stage (see Table 1). x a Q < o C3 30 26 2 2 1 8 14 10 06 2 OCT 79 (F„| DEC 79 |F 78 .w 79 | . SEP 79 |F 79 | NOV 79 |F 78 .W 79 | 95 105 115 125 135 145 155 165 175 FEMALE JAN 80 |W 79 .F 78 | NOV 79 |F„| OCT 79 |F 78 | NOV 79 |W 7 ,| SEP 79 |F 78 | JON 79 IE 78 | APR 79 |F 78 | OCT 78 |W 7a | OCT 79 |W, 9 | 95 105 115 125 135 145 155 165 175 FORK LENGTH (mm) Figure 3 Monthly gonad weight-length regressions for Peprilus burti by cohort group (W = Winter, F = Fall). Line lengths show observed size ranges. Figure 4 (facing page and overleaf) Monthly length frequencies of Peprilus burti off Freeport, Texas. Results Maturation and spawning periodicity Peprilus burti mature at 100-160 mm in length. Few- fish larger than 100 mm were immature (Fig. 2). Males were identifiable at 100-1 10 mm when the testes became creamy white, but they were difficult to stage macroscopically. Females were 95-155 mm and 105- 165 mm in Early Developing and Late Developing stages, respectively. Most Gravid and Ripe females were 120-150mm, the smallest being 113mm. These sizes-at-maturity are supported by regressions of gonad weight on fork length (Fig. 3) in which x-intercepts for both sexes usually were 95-1 10 mm during September through February, a period that brackets much of the broad spawning period when fish should be maturing. Age compositions and sizes-at-age presented later in- dicate P. burti mature to first spawn at 9-16 months. Peprilus burti spawn primarily during the broad period of September through May. Fish 30-40 mm, which occurred mid-November to early July (Fig. 4), were 1.5 months of age April- July and mid-Novem- ber-December based on quadratic regressions of size- on-age for pooled Winter and Fall cohorts (Table 2). Little spawning of P. burti occurs June through August. Fish 30-40 mm, 1.5 months-old, were not cap- tured late July to early November (Fig. 4). No distinct, Murphy and Chittenden: Reproduction, age and growth, and movements of Pepnlus bum 105 80n 60 40 20 15- 10 5- 10 n 5 60! 45 30- 15 1 OCT 77 DAY N l',r-,l 5 NOV 77 DAY N = 481 -=l*v. 14 JUN 78 DAY N=36 15 JUL 78 DAY N=1076 15 SEP 78 DAY N=231 12 OCT 78 N=26 I DEC 78 NIGHT N=8 -A^N^V,^ . -W?8 - DEC 78 DAY N=107 .AW,.. 26 FEB 79 DAY N-161 . F -n 1„ rtTWTA. a -t-tv -J u\ 60 120 180 240 300 TOTAL LENGTH (mm) 40 80 120 160 200 FORK LENGTH (mm) °1 *-||W 7 B « 26 FEB 79 DAY N-161 5 A rVU-ift. a ^ W lA 60-, 45 30 15- -F, B _ 13 MAR 79 NIGHT N=567 80, 60 40 20 , /NwW\, 7 APR 79 NIGHT N-82 22 APR 79 DAY N=1464 16 MAY 79 NIGHT N=14 380 285 190 95 f~^*f*JS r./\7yv\ , JUN 79 NIGHT N=49 23 JUN 79 NIGHT N=4889 w„ 7 JUL 79 NIGHT N=4 15 10 5 40 30 20- 10- 30 20 10 21 JUL 79 DAY N=232 (Jj 24 AUG 79 DAY N=495 ^^ 24 SEP 79 DAY N = 939 4 OCT 79 NIGHT N = 4 60 120 180 240 300 TOTAL LENGTH (mm) ~40 80 120 ' 160 ' 200 FORK LENGTH (mm) 106 Fishery Bulletin 89(1), 1991 > 5 O z LJJ 8 » 160 120 80 40- 30 20 10 500 375 250 125 10 W79 7 •" F 78-« 4 OCT 79 NIGHT N=4 18 OCT 79 DAY N=295 1 , 1= r- 1 -=C — '- ,"->-"■. 1 , 1 5 NOV 79 NIGHT -W79- f 7e-. N=13 1 "T 11 ^" 1 1 1 1 -*„-< 17 NOV 79 DAY N=1174 /Aa - 3 DEC 79 NIGHT , W 79 -F T8 , 17 DEC 79 DAY N = 289 Aaa" A 5 JAN 80 NIGHT N = 37 18 JAN 80 DAY some F 7B N=4039 _ :'".w\ 8 FEB 80 NIGHT N = 490 18 FEB 80 DAY N-9B88 7 MAR 80 NIGHT N=129 60 120 180 240 300 TOTAL LENGTH (mm) 40 80 120 160 200 FORK LENGTH (mm) F ' 9 l-w\. 7 MAR 80 NIGHT N«129 60 45 30 15 21 MAR 80 DAY N=1097 IC l-F79 7 -« -^>-^ \ 3 APR 80 NIGHT N=93 80! 60 40 20 18 APR 80 DAY N-129 8 MAY 80 NIGHT N-157 21 MAY 80 DAY N-2641 --^A/U A JUN 80 NIGHT N=221 260 195 130 65 20 15 10 5 22 JUN 80 DAY N=5735 9 JUL 80 NIGHT N^520 A ~" 23 JUL 80 DAY N-282 t — *60 At- zA \n 60 120 180 240 300 TOTAL LENGTH (mm) ~40 80 120 ' 160 200 FORK LENGTH (mm) Figure 4 (continued) Murphy and Chittenden: Reproduction, age and growth, and movements of Pepnlus burti 107 abundant groups of fish 30-60 mm originated then, and the smallest fish caught then usually reflected the more slowly growing Winter- spawned individuals. The few fish 50-60 mm, about 2.5-3 months of age on average, caught September to mid-November, and the one Gravid fish captured in July, probably indicate some sum- mer spawning (Fig. 5). Peprilus burti spawn primarily during discrete Winter (late January-mid-May) and Fall (early September-late October) periods. Length com- positions were consistently bimodal off Texas, and modal groups originated from Winter and Fall spawning periods (Fig. 4). Winter-spawned fish first appeared mid-April to early July at lengths of 30-75 mm at an average age of 1.5-4.5 months, which indicates spawning from about late January to mid-May. Fall-spawned fish first ap- peared mid-November to early December at lengths of 30-75 mm and an average age of 1.5- 4.5 months, which indicates spawning from about early September to late October. Modes for Winter cohorts are readily followed in the periods: (1) mid-April 1978-late February 1979, (2) mid- May 1979-mid-June 1980, and (3) late May-late July 1980. Modes for Fall cohorts can be followed in the periods: (1) early December 1977-mid-July 1978, (2) December 1978-mid-October 1979, and (3) mid-November 1979-late July 1980. Calcu- lated mean hatching dates occurred during late January to March for Winter groups and during September for Fall groups (Table 2). In contrast to length frequencies, gonad weight and maturity data indicate P. burti spawns dur- ing much of the fall and winter. Gonad weight regressions had maximum slopes and elevations and usually were significant September through February (Fig. 3). Regressions had lower slopes and elevations and usually were not significant March through August (Murphy 1981, Table 2). Most Gravid and/or Ripe fish were captured November through February (Fig. 5). The end of the Fall spawning period is not clear, but length frequencies suggest low-level spawning, or spawning success, during late fall and early winter. The consistently bimodal length frequencies must reflect some temporal separation in spawning that originates then. Fall 7 y fish recruited in abundance by mid- Janu- ary to mid-February 1980 when they were 60-105 mm long and 3-7 months of age (Fig. 4). Fall 78 fish re- cruited in abundance by mid-March 1979 when they were 90-110mm and 5.5-7.5 months of age. These data suggest peak fall spawning ends by about late October. No abundant, distinct groups of fish 30-60 mm and 1.5-3 months of age originated during any late-fall or 30 20- OCT 78 n=21 MAY 79 n=11 ■'•., 150 100 50 OCT 79 n=204 u*. DEC 78 n=76 -i^nF ]. 200-| w, 9 1 JUN 79 n=245 100- f 7e - *m B -i m . — i — i o FEB 79 n=93 40 1 20 J . MAR 79 n = 346 AUG 79 n=171 >L F* APR 79 n=271 JU JAN 80 n = 252 W, 9 l linnnn L 120-1 80 40 SEP 79 n=203 2 3 4 5 6 7 ' 7B I ' 1 I 79 200- 100 FEB 80 n=304 1 2 3 4 5 6 7 8 MATURITY STAGE 1 2 3 4 5 6 7 Figure 5 Monthly maturity stages (Table 1) of immature and female Peprilus burti. Bars indicate Fall (F, dark) and Winter (W, light) cohorts and mixtures of two cohorts (diagonals) in which cohort identity is not clear for individual fish. early- winter period studied (Fig. 4), although 30-75 mm fish— which we labeled Fall-spawned fish— persisted January through May in 1979 and 1980 and probably reflect some winter spawning. Cohort spawning and recruitment periodicity vary within and between years. Winter 78 fish appeared as an abundant, distinct group in mid- April, but few recruited thereafter (Fig. 4). In contrast, Winter 79 and 80 fish did not form abundant, distinct groups until June, although a few fish appeared in May. Fall groups first appeared mid-November to early Decem- ber. However, the bimodal size frequencies of the Fall 79 cohort in the period mid-February to early June 1980 suggest intragroup variation in spawning or recruitment success and periodicity similar to that for winter cohorts. We have interpreted the lower mode 108 Fishery Bulletin 89(1), 1991 of the Fall 79 cohort as actually being part of that cohort; the lower mode first became distinct in Feb- ruary 1980 when those fish were primarily 55-75 mm long and about 4 months of age, which would suggest early October to early November hatching. Recruitment, movements, and nurseries Fall cohorts of P. burti seemingly recruit in abundance at an older age (4-5 months) than Winter fish (2-4 months) off Texas. Winter fish formed abundant, distinct groups soon after first appearing mid-April to early June at 30-75 mm in length when 1.5-4.5 months old (Fig. 4). Fall fish did not form abundant, distinct groups until winter to early spring, although they first appeared mid-November to early December when 30-75 mm and 1.5-4.5 months old. Young P. burti recruit to the bottom off Texas primarily in 5-27 m depths when 2-5 months old. Winter 79 fish 2-4 months old and 35-70 mm were cap- tured only at 22 m in May 1979 (Figs. 4, 6). They oc- curred from 16-55 m during the period June through August 1979 (primarily June) but were most abundant at 22-27 m; few were shallower than 22 m or deeper than 36 m. Recently hatched Fall 79 fish 25-70 mm long were captured only at 5-9 m in mid-November 1979 (Figs. 4, 6). Fall 79 fish occurred only at 5-27 m (pri- marily 22-27 m) December 1979 through February 1980 when 3-5 months old. Similarly, Fall 78 fish were abundant at 5-27 m March through May when 7-9 months old, but few occurred in deeper water. Juvenile P. burti disperse offshore as they mature and approach age I. Winter 79 fish were most abundant at 13-27m depths September through November 1979 when 6-8 months old (Fig. 6). However, none occurred shallower than 22 m December 1979 through February 1980, when they were 9-11 months old; most were at 36- 100 m. The largest Fall 79 individuals occurred in the deepest water December 1979 through February 1980, the size gradient suggesting gradual movement offshore. Fall 78 fish were almost exclusively at 5-27 m from March through May 1979 when 7-9 months old, but they were at 36-100 m from June through August when 10-12 months old. Age determination using otoliths Whole otoliths of P. burti apparently cannot be read- ily aged. Only 984 of 2461 whole otoliths examined had apparent internal features. Many were entirely opaque or lacked a distinct boundary between opaque and hyaline zones, possibly due to initial preservation or storage fluids, though fresh otoliths showed similar features. Only 11 of the 984 legible otoliths had an ap- parent annulus. These 11 fish were 120- 160 mm in length and could have been about age I by length- frequency analysis. Annuli frequently were not ap- parent for fish that were age I by length frequencies. Growth and age determination by length frequency Length frequencies could be used to determine age of P. burti through at least 13-16 months of age in the northwestern Gulf and apparently 20-27 months of age in the northcentral Gulf. No more than two cohorts oc- curred off Texas in any one month, except in March and December 1978, November 1979 through January 1980, and May through June 1980 when a few members of a third group were present (Fig. 4). Each cohort was followed easily until it disappeared when 13-16 months old. In contrast, in the northcentral Gulf in April 1980 there were three cohorts west of the Mississippi River and four to the east (Fig. 7). Fish were abundant at 20 months of age west of the Mississippi and were as old as 27 months to the east (Table 3). Early sizes for P. burti average 25mm in length at 1 month of age, 42 mm at 2 months, 57 mm at 3 months, 72mm at 4 months, and 84mm at 5 months. These values are predicted from quadratic regressions of length-on-age in days pooling all Winter and Fall cohorts (Table 2). Similar size-age combinations may be predicted from quadratic regressions for individual cohorts, and for pooled Winter groups and pooled Fall groups. Peprilus burti average about 65- 100 mm in length at 6 months, 120-145 at age I, and about 170mm at age II, but fish in the northcentral Gulf were smaller at age than off Texas. Quadratic and von Bertalanffy growth equations both fit observed data from off Texas well, and they predict similar sizes-at-age over most of the observed size range (Fig. 8). For Winter fish quadratic and von Bertalanffy equations predicted lengths of 99 and 100 mm at 6 months, respectively, and 146 and 141 mm at age I (Fig. 8). Observed lengths show many Winter fish were as large as 120mm at 6 months and 155 mm at age I (Fig. 4; Murphy 1981, Table 1). For Fall fish, quadratic and von Bertalanffy equations predict lengths of 93 and 97 mm at 6 months, respectively, and 131 and 130 mm at age I. Observed lengths show many Fall fish were as large as 105 mm at 6 months and 145 mm at age I. Winter-spawned fish from the northcentral Gulf averaged 120-124 mm at Figure 6 (facing page) Length frequencies of Peprilun burti by depth: March-May 1979, June-August 1979, September-November 1979, December 1979-February 1980. Murphy and Chittenden Reproduction, age and growth, and movements of Pepnlus bum 109 o z o 380 285 190 95 MARCH-MAY 1979 S^*M\ 5-9 m N=I24 12 TOWS 13-18 ni N=682 36 TOWS ' — \T 22-27 m N=I316 100 TOWS W,9— / v\ 36-47 m N=2 16 TOWS 55-100 m N=3 6 TOWS JUNE-AUGUST 1979 5-9 m N = 0. 19 TOWS -/>- 13-18 m N=IS 53 TOWS 15 n 10 60 45 30 15 22-27 m N=4?82 95 TOWS 36-47 m N»226, 11 TOWS 55-100 m N=646. 30 TOWS 60 120 180 240 300 TOTAL LENGTH (mm) ~40 80 120 160 200 FORK LENGTH (mm) 10 5 560 4 20 280 140 SEPTEMBER-NOVEMBER 1979 H W 79 / \ N=750 1 1 19 TOWS ,_F 79 — J \ S» "',t' 13-18 m N 174 53 TOWS N=565 108 TOWS ►— w, 9 f„- 36-47 m N=216, 14 TOWS % V, 55- I00 m N=520. 29 TOWS DECEMBER 1979- FEBRUARY 1980 5-9 m N«185 23 TOWS 13-18 m N=212. 72 TOWS _yi i N=14217. 163 TOWS \ »- W 79 f 78 -. 36-47 m W 79 'F 7e N = 17, 24 TOWS N-151, 32 TOWS 60 120 180 240 300 TOTAL LENGTH(mm) 40 80 120 160 200 FORK LENGTH (mm) 10 Fishery Bulletin 89(1), 1991 10-19 APRIL. 1980 N=574 CD < O cr 3 o 20 40 60 80 100 120 140 160 180 200 FORK LENGTH (mm) Figure 7 Length-frequency and cumulative percentages of Peprilus burti from the northcentral Gulf, (top) east and (bottom) west of the Mississippi River, April-May 1980. Cohort identities are indicated. 13 months and 171mm at 27 months (Table 3). Fall- spawned fish averaged 66-73 mm at 7 months and 142-149mm at 20 months. Male and female P. burti reach a similar size off Texas. The largest male sexed was 173 mm and the largest female was 163 mm. Both sexes were equally abundant among fish greater than 150 mm. Peprilus burti showed little apparent somatic growth off Texas as spawning approached, but fish from the 160 a. • 140 • ^-^- — *- * ^-iS"-- 120 J^6 ° 100 80 °y ° f . d WINTER 78 . /• • WINTER 79 40 *B* « WINTER 80 1 » I 1- 50 100 150 200 250 300 350 400 450 500 z AGE (days) -1 160 b £ 140 ° o ° O -- n -- ~ i__ „ U_ • ^»^~ B T?^- 120 fr*^ - 100 ^°- 80 s? Ky7 » FALL 77 60 Y * D FALL 78 ,/n • FALL 79 40 20 50 100 150 200 250 300 350 400 450 500 AGE (days) Figure 8 Mean observed and predicted sizes-at-age of Winter and Fall cohorts of Peprilus burti off Freeport, Texas. Predicted sizes were estimated for pooled Winter and Fall data from regres- sions of mean sizes-on-age using quadratic (solid line) and von Bertalanffy growth (dashed line) models. Regressions were significant at a = 0.05. northcentral Gulf had large annual growth increments. Off Texas, fish grew little after reaching modal sizes of 110-160mm at 7-15 months in Winter groups and 120-150mm at 9-16 months of age in Fall groups (Figs. 4, 8). Growth ceased or greatly slowed between Table 3 Size-at-age (mm FL) data for Peprilus burti from the northcentral Gulf, east and west of the Mississippi River. Ages assume hatching dates assigned to cohorts off Freeport, Texas. West of Mississippi River East of Mississippi River Spawned Approx. group age (mo.) n Size range Mean FL n Size range Mean FL Fall 79 7 150 50-95 73 235 37-93 66 Fall 78 20 743 132-176 149 196 130-155 142 Fall 77 33 — — — — Winter,, 13 1029 98-139 120 105 110-135 124 Winter 7g 27 — — 61 155-200 171 Winter 77 38 — — — — Murphy and Chittenden: Reproduction, age and growth, and movements of Pepnlus bum 1 I n = 32,332 < 60 LU 60 80 100 120 140 160 180 200 220 240 TOTAL LENGTH (mm) 60 80 100 120 140 160 180 FORK LENGTH (mm) Figure 9 Length-frequency and cumulative percentage of Peprilus burti off Freeport, Texas. the Late Developing and Ripe stages when sizes formed a plateau (Fig. 2). In contrast, growth of Winter and Fall cohorts in the northcentral Gulf continued for fish 7-13 months of age and 20-27 months of age (Table 3). Table 4 Weight-length, girth-length, and total, fork, and standard length regressions for Peprilus burti. All regressions were significant at a = 0.05. Measures are grams and millimeters. Maximum size and age The maximum size of P. burti is about 200 mm in length. The largest of 32,332 fish we captured off Texas was only 180 mm, though a 198 mm specimen was measured among only 574 fish from the northcentral Gulf east of the Mississippi River (Fig. 7). Off Texas, 90% of the fish were less than 122 mm, 99% were less than 142mm, and 99.5% were less than 145mm (Fig. 9). The maximum age of P. burti typically was only 1-1.5 years in the northwestern Gulf but appeared to be 2-2.5 years in the northcentral Gulf where the largest individuals occur. A t L of 1-1.5 years seems reasonable off Texas, because fish larger than 142- 145 mm, which made up less than 0.5-1.0% of the catch, approximate the average size there at age I. A larger t L is appropriate for the northcentral Gulf (Fig. 7), because 99-99.5% of the fish west of the Mississippi River were less than 163-167 mm long and 99-99.5% of those to the east were less than 182-184 mm. These sizes approximate the average at age II in the northcen- tral Gulf and the oldest fish collected there was about age II, so a reasonable estimate of t L is 2-2.5 years. Total weight-length, girth-length, and length-length relationships Regressions of total weight on fork length (Table 4) were significantly different (at a = 0.05) in elevation between sexes (F = 9.19, 1, 1615 df) but not in slope (F = 3.25, 1, 1614 df). Calculated slopes did not sig- nificantly differ from p = 3.0 (males: t = 0.97, 789 df; females: t= 1.60, 825 df) except when all fish were pooled (t = 20.08, 2734 df). Girth- fork length and length-length regressions are in Table 4. Equation FL range logm TW = -4.5517 + 2.9640 log,„ FL (males) logm TW = -4.7095 + 3.0477 log 10 FL (females) log,„ TW = -4.8621 + 3.1201 logn (males, females, immatures) FL G FL FL TL SL TL FL SL 10.38 -4.31 8.35 -9.35 3.22 -2.88 5.05 -4.46 + 0.99 FL + 0.96 G + 0.73 TL + 1.35 FL + 0.69 TL + 1.42 SL + 1.05 SL + 0.95 FL 791 827 2736 2736 2736 2662 2662 2668 2668 2734 2734 100-163 94-164 25-164 25-164 25-164 25-164 26-164 25-164 25-164 25-164 25-164 0.89 0.93 0.99 0.95 0.95 0.99 0.99 0.99 0.99 0.99 0.99 Discussion Spawning periodicity and its regulation The broad primary spawning period of September through May we sug- gest for Peprilus burti is realistic. Our data agrees with reports of fish 20-40 mm in length from December through June (Gunter 1945, Hoese 1965, Miller 1965) and, in part, with a suggested fall and winter spawn- ing (Miller 1965). Moreover, Finu- cane et al. (1979) collected larvae off Texas from September through May. 112 Fishery Bulletin 89(1). 1991 We interpret our data to mean P. burti spawns pri- marily—or most successfully— in temporally separate, discrete, Winter (late January-mid-May), and Fall (early September-late October) periods. The consis- tently well-separated, bimodal length frequencies— on which we place great emphasis— must reflect a tem- poral separation in spawning activity or success, or growth and mortality, that originates during late fall and early winter. However, the well-developed^onads and Gravid/Ripe fish we observed suggest spawning could occur throughout that period, possibly at a low level. Allen et al. (1986) and Vecchione (1987) also found consistently bimodal size distributions of P. burti in the northcentral Gulf. Allen et al. (1986) considered them separate spawning peaks in a continuous, not temporally separated, spawning. The actual degree of temporal separation in spawning of P. burti may be im- portant to resolve, because it may influence (1) ap- propriate management practices, (2) how many popula- tions and stocks exist in P. burti, concepts which are not necessarily the same, and (3) how speciation occurs in Peprilus. Given properly randomized geographical sampling, age determination by daily otolith incre- ments (Jones 1986) might resolve the question of how intense is late-fall to early-winter spawning and whether or not, and to what degree, spawning is tem- porally separate. Although P. burti appear to spawn primarily in two main periods, Winter and Fall, it also appears that in each period there is much variation in cohort spawn- ing periodicity or success, or in recruitment periodicity. We observed Winter cohorts to appear as distinct abun- dant groups in April in one year but not until June in two other years. Similarly, one Fall cohort was distinct- ly bimodal over a several-month time period. A more exact method of age determination than length fre- quencies, however, is needed to more clearly interpret these phenomena. Our interpretation of spawning periodicity in P. burti is similar to findings that other Gulf species spawn in discrete Winter-Spring and Late Summer-Fall periods related to current transport, including Cynoscion are- narius (Shlossman and Chittenden 1981), C. nothus (DeVries and Chittenden 1982), Larimus fasciatus (Standard and Chittenden 1984), Menticirrhus ameri- canus (Harding and Chittenden 1987), and Polydac- tylus octonemus (Dentzau and Chittenden 1990). Spawning of P. burti in the northwestern Gulf, and for many of these other fishes, probably is timed to coin- cide with currents (Shlossman and Chittenden 1981) that transport eggs and larvae from spawning areas to nurseries, assuming P. burti has pelagic eggs and larvae like P. alepidotus and P. triacanthus (Martin and Drewry 1978). Spawning, or its absence, coincides with wind-induced, up- or downcoast, alongshore coastal currents which drive circulation in the northwestern Gulf with seasonal reinforcement from the Missis- sippi— Atchafalaya discharge (Kelly et al. 1981). Aver- age winds are downcoast (toward Mexico) during August/September-April/May but upcoast (toward Florida) during May/ June-July/ August. Nearshore cur- rents parallel the coast. Upcoast wind stress causes up- coast alongshore currents which (1) are reflected in high inshore salinity off Galveston/Freeport and fall- ing sea levels during early to midsummer (Marmer 1954, Kelly et al. 1981), and (2) coincide with the sum- mer period of little spawning we observed in P. burti and which is reported in other species just cited. Downcoast wind stress causes downcoast (toward Mex- ico) alongshore currents, onshore surface Ekman transport, and downwelling which (1) are reflected in rising sea levels February-May and August-October, (2) transport low-salinity water downcoast causing a salinity minimum off Galveston/Freeport during Sep- tember and October, and May and June, and (3) coin- cide with the two major spawning periods we suggest for P. burti and which are reported as major or minor periods in other species just cited. Alongshore currents continue downcoast from late fall to early winter. Seemingly, however, no distinct, abundant groups of P. burti originated then, which may reflect low-level spawning or spawning success. Similarly, little or no spawning occurs then in the other species just cited. Temporal variation in the average meteorological and hydrographic patterns may be the primary reason for the variation we noted in cohort spawning and recruit- ment periodicity between and within years. Age determination and growth, maximum size and age, and mortality Our findings on age and growth in P. burti are new, because this species has not been aged previously. It would be desirable to corroborate them by analysis of daily otolith increments (Jones 1986). However, that may not prove feasible, because recent studies using scales, opercula, vertebrae, and thin-sectioned otoliths, fail to consistently show clear daily increments or annuli (Allen et al. 1986). Therefore, it appears length frequencies are the only way to age P. burti at present. As in our study, supporting length collections must be frequent in time and over a long duration, because cohort boundaries and age are not clear every month. However, they are quite clear in certain months (for examples, the Winter 79 and 80 groups in May or June 1979 and 1980, the Fall 78 and 79 groups in December 1978 and 1979). From the clear groups, one can work chronologically backward and forward in time and gradually assign age and cohort boundaries with reasonable certainty. This process, however, is not as Murphy and Chittenden: Reproduction, age and growth, and movements of Pepnlus bum 13 simple in species with a complex life history like P. burti as it is in species that spawn during one major period a year. The apparent cessation of somatic growth as P. bur- ti approach spawning in the northwestern Gulf con- trasts with large annual increments in the northcentral Gulf. This difference, combined with their disap- pearance from the northwestern Gulf at 13-16 months of age and their smaller maximum sizes and younger maximum ages there (see below), indicate fish from the northwestern Gulf (1) spawn and die at age I, or more probably (2) in some presently unclear combination emigrate offshore and to the northcentral Gulf prior to, or immediately after, spawning. Few P. burti apparently exist larger than 190-200 mm in length. The largest we captured (180 mm, Texas; 198mm, northcentral Gulf) are similar to maxima in other studies' sampling to at least 80-100m depths (173mm, Hildebrand 1954; 184mm, Franks etal. 1972; 169 mm, Chittenden and McEachran 1976; 193 mm, Allen et al. 1986). Maxima are even smaller from estuaries or the shallow Gulf (154mm, Gunter 1945; 131mm, Miller 1965; 122 mm, Perret et al. 1971; 133 mm, Christmas and Waller 1973), which agrees with our findings that P. burti disperse to deep water as they mature. The largest records were from the northcentral Gulf (Allen et al. 1986; our study). Our estimate that t L = 1-1.5 years in the north- western Gulf agrees with Chittenden and McEachran (1976) who suggested a 1-2 year maximum age. A higher t L (2-2.5 years) in the northcentral Gulf ap- pears realistic, because the largest records there are not much larger than the mean size at age II. Based on these maximum ages, theoretical estimates (Royce 1972, Hoenig 1983) of total annual mortality rate (1 - S) are nearly 100% in the northwestern Gulf and 82-90% in the northcentral Gulf. Murphy (1981) cal- culated similar values of 1 - S for the northwestern Gulf (99%) from observed time-specific abundance data for consecutive Winter or Fall cohorts. If mature fish from the northwestern Gulf emigrate offshore and to the northcentral Gulf, as we suggest, our values for maximum age and total mortality are under- and over- estimates, respectively, for P. burti in the north- western Gulf. The presence of the largest P. burti in the north- central Gulf follows a pattern in other marine and estuarine, demersal and pelagic species (Cynoscion nothus, DeVries and Chittenden 1982; Stenotomus caprinus, Geoghegan and Chittenden 1982; Micro- pogonias undulatus, Rivas and Rothmayr 1970, Gutherz et al. 1975, White and Chittenden 1977; Bre- voortia patronus, Nicholson 1978; Larimus fasciatus, Standard and Chittenden 1984; and probably C. are-' narius, Shlossman and Chittenden 1981). Small differ- ences between areas also exist in other population at- tributes of P. burti, as in C. nothus and L. fasciatus : younger age compositions and maximum ages, smaller maximum sizes, and higher total annual mortality rates occur in the northwestern Gulf. At least three explana- tions could account for this: (1) There may be no basic differences between areas, just much greater biomass (Moore et al. 1970) at age in the northcentral Gulf; (2) differences may be real, not related to biomass, imply- ing slight, but fundamental, population dynamics dif- ferences between areas; and (3) differences reflect some presently unclear combination of an offshore and spawning or postspawning movement of older, larger fish from the northwestern to the northcentral Gulf. The first implies Chittenden and McEachran (1976) and Chittenden (1977) are correct that shrimp communities on the Gulf continental shelf have a common popula- tion dynamics pattern. The other explanations imply that their arguments need modification for slightly longer life spans and lower mortality rates in the north- central Gulf, and that shrimp communities are a little more sensitive to fishing than Chittenden's (1977) simulations indicate. Movements, recruitment, and spawning areas Peprilus burti probably spawn offshore. We found fish congregate in 36-100 m depths as they mature and size gradients that indicate an offshore dispersal. In agree- ment, Allen et al. (1986) found a positive correlation between mean size and depth at 200-290 m. Finucane et al. (1979) collected larvae in water 22-182 m deep on the continental shelf off Texas. The young make their way inshore to 5-27 m depths off Freeport, Texas— the white shrimp community (Hildebrand 1954, Chittenden and McEachran 1976)— where they recruit to the bottom. At least two mechanisms could explain their arrival inshore. For one, Ekman surface trans- port, associated with prevailing downcoast alongshore currents in the spawning season, could, in theory, bring the young inshore. However, Dentzau and Chittenden (1990) rejected this idea. It implies fishes of the brown shrimp community would also recruit inshore, but, in actuality, there is a clear separation between the two communities (Chittenden and McEachran 1976). A more likely alternative, suggested for Polydactylus octonemus (Dentzau and Chittenden 1990), is based on the hydrography and cyclonic gyre of the northwestern Gulf (Kelly et al. 1981, 1983, 1984): The eastward counterflow of the gyre is diverted inshore at the Mississippi River delta and ultimately extends down- coast as an alongshore current. Members of the white shrimp community could spawn anywhere in this flow, in the offshore northeastward flowing arc or in the 114 Fishery Bulletin 89(1), 1991 alongshore southwestward flow: the young just need to be entrained in waters already in the white shrimp community, or enter them using the eastward counter- flow, and be transported in a "downstream manner" (see next paragraph). This interpretation is similar to the current transport model Shaw et al. (1985) suggest for B. patronus. Besides being offshore, spawning grounds for P. burti found off Texas may lie towards or in thenorth- central Gulf. We suggest some unknown combination of offshore and spawning or postspawning movement to the northcentral Gulf explains between-area popula- tion dynamics differences. Such upcoast movement may be required within the white shrimp community, given the coincidence of spawning with downcoast cur- rents. Mean alongshore surface current speed is 23 cm/second in water 22 m deep off Freeport, Texas from September through June (Kelly et al. 1981), so spawn- ing areas could lie 320 nmi upcoast assuming passive transport for 30 days before the young recruit to the bottom. This estimate depends on many poorly known factors, including (1) routes followed, (2) alongshore current speeds, (3) duration of the transport period, and (4) behavior of the young. However, alongshore cur- rents could transport young great distances— concep- tually "downstream"— and Texas recruits could be spawned off Louisiana where low-salinity waters prop- agate along Texas when alongshore wind components turn downcoast during August and September (Kelly and Randall 1980, Kelly et al. 1981). Mature fish must move toward spawning areas, conceptually in an up- stream, contranatant direction from the northwestern Gulf, to maintain a fixed, proven spawning ground following Harden Jones (1968). Zoogeographic considerations Peprilus burti fill a niche in the Gulf similar to the one P. triacanthus occupies on the Atlantic coast. However, their population dynamics differ, and this may reflect zoogeographic variation suggested for other taxa whose ranges traverse the Cape Hatteras area, in- cluding Micropogonias and Alosa (White and Chitten- den 1977), Cynoscion (Shlossman and Chittenden 1981), and Stenotomus (Geoghegan and Chittenden 1982). It appears for P. burti that (1) maturity occurs at 100- 160 mm in length as they approach age I and spawn, (2) maximum size is only about 200 mm, but most are much smaller, (3) maximum age is no more than 2.5 years, and (4) total annual mortality rate is not lower than about 82%. The life history of P. triacanthus north of Cape Hat- teras is complicated by north-south and onshore- offshore movements (Horn 1970), but they appear to (1) mature at lengths of 110-130 mm in their second year (Hildebrand and Schroeder 1927, Bigelow and Schroeder 1953, Horn 1970, DuPaul and McEachran 1973), (2) reach maximum sizes of 300 mm (Murawski and Waring 1979), (3) have maximum ages of 3-6 years (Draganik and Zukowski 1966, DuPaul and McEachran 1973, Waring 1975, Kawahara 1977), and (4) have total annual mortalities of 67-84% (Murawski and Waring 1979). Little has been published for P. triacanthus south of Cape Hatteras, but the largest fish collected in exten- sive trawling in this area was 150mm (Wenner et al. 1979). This size range is more similar to P. burti than P. triacanthus north of Cape Hatteras and may reflect an intrageneric, Carolinian Province similarity in sizes, maximum ages, and mortality. Acknowledgments We are much indebted to M. Burton, T. Crawford, P. Geoghegan, J. Pavela, M. Rockett, J. Ross, P. Shloss- man, B. Slingerland, G. Standard, H. Yette, and Cap- tains Hollis, Mike, and Robby Forrester, P. Smirch, and A. Smircic for assistance in field collections. T. Fehr- man and R. Grobe recorded the data. R. Case was of invaluable help for writing and assisting with computer programs. B. Rohr and E. Gutherz allowed the senior author to participate in NMFS groundfish survey 106. R. Darnell, G. Grant, E. Klima, D. Stilwell, K. Strawn, and K. Sulak reviewed drafts of the manuscript. Finan- cial support was provided, in part, by the Texas Agricultural Experiment Station; by the Strategic Petroleum Reserve Program, Department of Energy; and by the Texas A&M Sea Grant College Program, supported by the NOAA Office of Sea Grant, U.S. Department of Commerce. Final preparation and revi- sions of this manuscript were made while the authors were at Florida Marine Research Institute, Florida Department of Natural Resources (MDM) and College of William and Mary, Virginia Institute of Marine Science (MEC). This manuscript was based on a thesis submitted by the senior author as partial fulfillment of the M.S. degree, Texas A&M University. Citations Allen, R.L., J.H. Render, A.W. Liebzeit. and G.W. Bane 1986 Biology, ecology, and economics of butterfish and squid off the northern Gulf of Mexico. Final Rep. LSU-CFI-86-30, Coastal Fish. Inst., Center for Wetland Resources, Louisiana State Univ., Baton Rouge, 175 p. Alverson, D.L., and M.J. Carney 1975 A graphic review of the growth and decay of population cohorts. J. Cons. Cons. Int. Explor. Mer 36:133-143. Bagenal, T.B., and E. Braum 1971 Eggs and early life history. In Ricker, W.E. (ed.), Murphy and Chittenden: Reproduction, age and growth, and movements of Pepnlus bum 115 Methods for assessment offish production in fresh waters, p. 166-198. Blackwell Sci. Publ., Oxford. Bigelow, H.B., and W.C. Schroeder 1953 Fishes of the Gulf of Maine. Fish. Bull., U.S. 53(74), 577 p. Caldwell, D.K. 1961 Populations of the butterfish, Poronotus triacantkus (Peck), with systematic comments. Bull. South. Calif. Acad. Sci. 60:19-31. Chittenden, M.E. Jr. 1977 Simulations of the effect of fishing on the Atlantic croaker, Mitropogon undulatus. Proc. Gulf Caribb. Fish. Inst. 29:68-86. Chittenden, M.E. Jr., and J.D. McEachran 1976 Composition, ecology, and dynamics of demersal fish com- munities on the northwestern Gulf of Mexico continental shelf, with a similar synopsis for the entire Gulf. Sea Grant Publ. TAMU-SG-76-208, Texas A&M Univ., College Station, 104 p. Christmas, J.Y., and R.S. Waller 1973 Estuarine vertebrates, Mississippi. In Christmas, J.Y. (ed.), Cooperative Gulf of Mexico estuarine inventory and study, Mississippi, p. 320-434. Gulf Coast Res. Lab., Ocean Springs, MS, 434 p. Collette, B.B. 1963 The systematic status of the Gulf of Mexico butterfish, Poronotus burti (Fowler). Copeia 1963:582-583. Dentzau, M.W., and M.E. Chittenden Jr. 1990 Reproduction, movements, and apparent population dynamics of the Atlantic threadfin, Polydactylus octonemus, in the Gulf of Mexico. Fish. Bull., U.S. 88:439-462. DeVries, D.A., and M.E. Chittenden Jr. 1982 Spawning, age determination, longevity, and mortality of the silver seatrout, Cynoscion nothus, in the Gulf of Mex- ico. Fish. Bull., U.S. 80:487-500. Draganik, B., and Cz. Zukowski 1966 The rate of growth of butterfish, Peprilus triacantkus, and ocean pout, Macrozoarces americanus, from the region of Georges Bank. Int. Comm. Northwest Atl. Fish. Res. Doc. 66/42, 3 p. DuPaul, W.D., and J.D. McEachran 1973 Age and growth of the butterfish, Peprilus triacantkus, in the lower York River. Chesapeake Sci. 14:203-207. Finucane, J.H., L.A. Collins, L.E. Barger, and J.D. McEachran 1979 Environmental studies of the South Texas outer continen- tal shelf, 1977: Ichthyoplankton/mackerel eggs and larvae. Final Rep. to Bur. Land Manage., by Natl. Oceanic Atmos. Admin., Interagency Agreement AASSO-1A7-21. Galveston Lab., Natl. Mar. Fish. Serv., NOAA, Galveston, TX 77550, 504 p. (avail. NTIS, Springfield, VA). Franks, J.S., J.Y. Christmas, W.L. Siler, R. Combs, R. Waller, and C. Burns 1972 A study of the nektonic and benthic faunas of the shallow Gulf of Mexico off the State of Mississippi. Gulf Res. Rep. 4, 148 p. Gallucci, V.F., and T.J. Quinn HI 1979 Reparameterizating, fitting, and testing a simple growth model. Trans. Am. Fish. Soc. 108:14-25. Geoghegan, P., and M.E. Chittenden Jr. 1982 Reproduction, movements, and population dynamics of the longspine porgy, Stenotomus caprinus. Fish. Bull.. U.S. 80:523-540. Gledhill, C.T. Unpubl. A preliminary estimate of Gulf butterfish (Peprilus burti) MSY and economic yield. Mississippi Lab., Southeast Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Pascagoula, MS, 16 p. Gulland, J.A. 1969 Manual of methods for fish stock assessment. Part I. Fish population analysis. FAO Manuals in Fish. Sci. 4, 154 p. Gunter, G. 1945 Studies on marine fishes of Texas. Publ. Inst. Mar. Sci., Univ. Tex. 1, 190 p. Gutherz, E.J., G.M. Russell, A.F. Serra, and B.A. Rohr 1975 Synopsis of the northern Gulf of Mexico industrial and foodfish industries. Mar. Fish. Rev. 37(7):1-111. Harden Jones, F.R. 1968 Fish migration. Edw. Arnold Publ., Ltd., London, 325 p. Harding, S.M., and M.E. Chittenden Jr. 1987 Reproduction, movements, and population dynamics of the southern kingfish, Mentieirrhus americanus, in the north- western Gulf of Mexico. NOAA Tech. Rep. NMFS 49, 21 p. Hildebrand, H.H. 1954 A study of the fauna of the brown shrimp (Penaeus aztecus Ives) grounds in the western Gulf of Mexico. Publ. Inst. Mar. Sci., Univ. Tex. 3:103-109. Hildebrand, S.F., and W.C. Schroeder 1927 Fishes of Chesapeake Bay. Bull. U.S. Bur. Fish. 43 (pt. 1), 388 p. Hoenig, J.M. 1983 Empirical use of longevity data to estimate mortality rates. Fish. Bull., U.S. 82:898-903. Hoese, H.D. 1965 Spawning of marine fishes in the Port Aransas, Texas area as determined by the distribution of young and larvae. Ph.D. diss., Univ. Tex., Austin, 144 p. Horn, M.H. 1970 Systematics and biology of the stromateoid fishes of the genus, Peprilus. Bull. Mus. Comp. Zool. Harv. Univ. 140: 165-261. Jearld, A. Jr. 1983 Age determination. In Nielson, L.A., and D.L. Johnson (eds.), Fisheries techniques, p. 301-324. Am. Fish. Soc, Bethesda, MD. Jones, C. 1986 Determining age of larval fish with the otolith increments technique. Fish. Bull., U.S. 84:91-103. Kawahara, S. 1977 Age and growth of butterfish, Peprilus triacantkus (Peck), in the ICNAF subarea 5 and statistical area 6. Int. Comm. Northwest Atl. Fish. Res. Doc. 77/VI/27, 13 p. Kelly, F.J. Jr., and R.E. Randall 1980 Physical oceanography. In Hann, R.W. Jr., and R.E. Randall (eds.), Evaluation of brine disposal from the Bryan Mound site of the Strategic Petroleum Reserve Program. Final report of predisposal studies, p. 1-93. Contract DOE/P010114-2, Texas A&M Univ., College Station (avail. NTIS, Springfield, VA). Kelly, F.J., J.E. Schmitz, R.E. Randall, and J.E. Cochrane 1981 Physical oceanography. In Hann, R.W. Jr., and R.E. Randall (eds.), Evaluation of brine disposal from the Bryan Mound site of the Strategic Petroleum Reserve Program. Final report of twelve-month post-disposal studies, p. 1-105. Contract DOE/P010114-4, Texas A&M Univ., College Station (avail. NTIS, Springfield, VA). 1983 Physical oceanography. In Hann, R.W. Jr., and R.E. Randall (eds.), Evaluation of brine disposal from the Bryan Mound site of the Strategic Petroleum Reserve Program: An- nual report for September 1981 through August 1982, p. 1-134. Contract DOE/P010114-6, Texas A&M Univ., College Station (avail. NTIS, Springfield, VA). 116 Fishery Bulletin 89(1), 1991 1984 Physical oceanography. In Hann, R.W. Jr., C.P. Giam- mona, and R.E. Randall (eds.), Offshore oceanographic and en- vironmental monitoring services for the Strategic Petroleum Reserve: Annual report for the Bryan Mound Site from September 1982 through August 1983, p. 1-122. Contract DOE/P010850-2, Texas A&M Univ., College Station (avail. NTIS, Springfield, VA). Knight, W. 1968 Asymptotic growth: An example of nonsense disguised as mathematics. J. Fish. Res. Board Can. 25:1303 T 1307. Lagler, K.F. 1956 Freshwater fishery biology, 2d ed. Wm. C. Brown Co. Publ., Dubuque, Iowa, 421 p. Marmer, H.A. 1954 Tides and sea level in the Gulf of Mexico. In Galtsoff, P.S. (ed.), Gulf of Mexico, its origin, water and marine life, p. 101-118. Fish. Bull, U.S. 89, 604 p. Martin, F.D., and G.E. Drewry 1978 Development of fishes of the mid-Atlantic Bight, Vol. VI. U.S. Fish Wild!. Serv. Biol. Serv. Prog. FWS/OBS-78/12, 416 p. Miller, J.M. 1965 A trawl study of the shallow Gulf fishes near Port Aran- sas, Texas. Publ. Inst. Mar. Sci., Univ. Tex. 10:80-108. Moore, D., H.A. Brusher, and L. Trent 1970 Relative abundance, seasonal distribution, and species composition of demersal fishes off Louisiana and Texas, 1962-1964. Contrib. Mar. Sci., Univ. Tex. 15:45-70. Murawski, S.A., and G.T. Waring 1979 A population assessment of butterfish Peprilus triacan- thus, in the northwestern Atlantic Ocean. Trans. Am. Fish. Soc. 108:427-439. Murphy, M.D. 1981 Aspects of the life history of the Gulf butterfish, Peprilus burti. M.S. thesis, Texas A&M Univ., College Station, 76 p. Nicholson, W.R. 1978 Gulf menhaden, Brevoortia patronus, purse seine fishery, catch, fishing activity, and age and size composition, 1964-1973. U.S. Fish Wildl. Serv.. Spec. Sci. Rep. Fish. 722, 8 p. Perret, W.S., W.R. Latapie, J.F. Pollard, W.R. Mock, G.B. Adkins, W.F. Gaidry, and C.J. White 1971 Fishes and invertebrates. In Perret, W.S. (ed.), Coop- erative Gulf of Mexico inventory and study, Louisiana. Phase IV, Biology, p. 39-197. La. Fish. Wildl. Comm., Baton Rouge. Perschbacher. P.W., K.J. Sulak. and F.J. Schwartz 1979 Invasion of the Atlantic by Peprilus burti (Pisces: Stromateidae) and possible implications. Copeia 1979: 538-541. Rivas, L.R., and CM. Roithmayr 1970 An unusually large Atlantic croaker, Micropogon un- dulatus, from the northern Gulf of Mexico. Copeia 1970: 771-772. Rohr, B.A., A.J. Kemmerer, and W.H. Fox Jr. 1980 FRS Oregon II Cruise 106 4/10-5/1/80, Cruise Report. Mississippi Lab., Southeast Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Pascagoula, MS, 12 p. Roithmayr, CM. 1965 Industrial bottomfish fishery of the northern Gulf of Mex- ico, 1959-63. U.S. Fish. Wildl. Serv., Spec. Sci. Rep. Fish. 518, 23 p. Royce, W.F. 1972 Introduction to the fishery sciences. Academic Press, NY, 351 p. Saila, S.B., C.W. Reckseik, and M.H. Prager 1988 Basic fishery science programs, a compendium of micro- computer programs and manual of operation. Elsevier Sci. Publ., NY, 230 p. Shaw, R.F., W.J. Wiseman Jr., R.E. Turner, L.J. Rouse Jr., R.E. Condrey, and F.J. Kelly Jr. 1985 Transport of larval gulf menhaden Brevoortia patronus in continental shelf waters of western Louisiana: A hypothesis. Trans. Am. Fish. Soc. 114:452-460. Shlossman, P. A., and M.E. Chittenden Jr. 1981 Reproduction, movements, and population dynamics of the sand seatrout, Cynoscion arenarius. Fish. Bull., U.S. 79:649-669. Standard, G.W., and M.E. Chittenden Jr. 1984 Reproduction, movements, and population dynamics of the banded drum, Larimus fasciatus, in the Gulf of Mexico. Fish. Bull., U.S. 82:337-363. Tesch, F.W. 1971 Age and growth. In Ricker, W.E. (ed.), Methods for assessment of fish production in fresh waters, 2d ed., p. 98-130. Blackwell Sci. Publ., Ltd., London. Vecchione, M. 1987 Commercial fishing for gulf butterfish, Peprilus burti, in the Gulf of Mexico. Mar. Fish. Rev. 49(4):14-22. Waring, G. 1975 A preliminary analysis of the status of the butterfish in ICNAF subarea 5 and statistical area 6. Int. Comm. North- west Atl. Fis. Res. Doc. 75/74, 27 p. Wenner, C.A., C.A. Barans, B.W. Stender, and F.H. Berry 1979 Results of MARMAP otter trawl investigations in the South Atlantic Bight. III. Summer, 1974. Tech. Rep. 41, S.C. Mar. Resour. Cent., Charleston, 62 p. White, M.L. and M.E. Chittenden Jr. 1977 Age determination, reproduction and population dynam- ics of the Atlantic croaker, Micropogonias undulatus. Fish. Bull, U.S. 75:109-123. Abstract. - Annual assessments of the Northwest Atlantic mackerel stock have occurred every year since 1973, providing useful advice to fish- ery managers involved in the deci- sion making process for this impor- tant pelagic resource. Since 1985, assessment advice based on an F A management strategy has indicated that catches in the 300,000 mt range are feasible because stock biomass has increased greatly after the col- lapse of the fishery in the mid-1970s. However, indications from previous research are that compensatory pro- cesses are very important, so a sto- chastic simulation model with den- sity-dependent growth, maturity, and natural mortality was constructed to study how these mechanisms might affect our ability to provide short- and long-term advice for this impor- tant stock. Model results suggest that our present assessments may be too optimistic relative to yield pro- jections and that minimum spawn- ing-stock biomass levels may be dif- ficult to maintain even with an F 0] fishing strategy. Model results also reveal that natural mortality rates are probably much higher than pre- viously thought and are important in determining trends in abundance in this stock. Impact of Compensatory Responses on Assessment Advice for the Northwest Atlantic Mackerel Stock William J. Overholtz Steven A. Murawski William L. Michaels Woods Hole Laboratory, Northeast Fisheries Science Center National Marine Fisheries Service. NOAA. Woods Hole, Massachusetts 02543 Manuscript accepted 19 September 1990. Fishery Bulletin, U.S. 89:117-128 (1991). The Northwest Atlantic stock of At- lantic mackerel Scomber scombrus has historically been important to the U.S. domestic fishery; records from the early 1800s to the 1980s suggest that cumulative landings have been 7-8 million mt over that time-period (Sette and Needier 1934, Hoy and Clark 1967, Anderson 1985). A thriv- ing domestic industry utilized mack- erel well into the 1940s until landings dropped because of declines in abun- dance, availability, and increased pro- duction of fresh and frozen white fish products (e.g., haddock, cod) (Hoy and Clark 1967, Jenson 1967). A re- surgence of the fishery occurred in the 1970s when distant water fleets from eastern Europe and the Soviet Union landed an average of 310,000 mt annually from 1970 to 1976 (Fig. 1A). Since many of the important groundfish species in the region have declined recently, the U.S. industry has become more interested in mack- erel as a volume (high-catch, low- price) fishery. The fishery has been managed under the auspices of the Mid- Atlantic Fish- ery Management Council since 1977. Current management objectives for this stock include maintenance of a minimum spawning stock (600,000 mt), annual quotas based on an F .i catch strategy and a recognition of the necessity for keeping the total stock at some reasonably high level to insure that the recreational fishery remains viable. Recent assessments suggest that the stock has increased since collaps- ing in the late 1970s. A succession of moderate to good year-classes from 1981 to 1985 promoted rapid recov- ery of the stock to levels observed in the early 1970s (Fig. IB, C). Assess- ment advice during the last several years based on an F .i management strategy has indicated that annual catches in the 300,000 mt range are feasible in the short term. Allocations to joint ventures have increased over the last several years, amounting to about 75,000 mt in 1987, but recent landings have remained well below the 300,000 mt level and fishing mor- tality has averaged only about 0.07 since 1980 (Overholtz and Parry 1985). Pelagic fishes such as mackerel are important in the trophic dynamics of fishery ecosystems, supporting popu- lations of predatory fish, birds, and marine mammals. Additionally, these species may also increase to densities that inhibit their own population pro- cesses (e.g., growth, reproduction) and those of competitors. Evidence exists that Atlantic mackerel exhibit density-dependent growth (McKay 1979, Lett 1980, Overholtz 1989). Other factors such as maturation rates, fecundity-at-age, and preda- tion mortality rates may also vary 117 Fishery Bulletin 89(1), 199) uu- ~ BO- E / age 3 60- 40- yv 20- / age 2 _ i 1 1 1 1 1967 1970 1973 1976 1979 1982 1985 Year Figure 1 Descriptive data for the Northwest Atlantic mackerel stock, 1967-85. (A) Yield (ooo's mt); (B) Recruitment at age 1 (billions of fish); (C) Spawning-stock biomass (ooo's mt); (D) Mean weight at ages 1, 2, and 3 from commercial samples (g); (E) Percent maturity for ages 2 and 3. with stock size. An analysis of the food habits of mackerel predators suggested that natural mortality rates (M2) for this stock were higher during 1973-75 when relatively large numbers of juvenile mackerel were available, and declined during 1976-80 when there were few small fish (NEFC 1987). This study examines the impact that compensatory changes in growth, sexual maturity, and natural mor- tality rates may have on the Northwest Atlantic mack- erel stock. Implications of responses in these factors on catch and spawning-stock biomass are evaluated using a simulation model. The model was designed so that changes in these compensatory factors as well as the influence of fishing mortality patterns and strate- gies could be assessed. Model background Data on potential density-dependent population reg- ulatory mechanisms were obtained from research ves- sel survey cruises and commercial fishing operations conducted on the U.S. eastern coast. Information from spring groundfish surveys conducted by the Northeast Fisheries Center (NEFC) during 1973-85 and a com- mercial fishery conducted by Poland during 1981-86 were examined to quantify compensatory relationships. Analyses were performed to study changes in growth, maturation rates, and natural mortality (Overholtz et al. 1988, Overholtz 1989). Significant negative relationships between mean weight-at-age and stock size were confirmed for both research and commercial data sources (Overholtz et al. 1988, Overholtz 1989). Mean size-at-age for recent year-classes was also found to be significantly different; large year-classes grew more slowly (Overholtz 1989). In addition mean weight-at-age was also negatively related to year-class size, indicating that large cohorts may depress growth rates of individuals from that par- ticular year (Overholtz 1989) (Fig. ID). These analyses helped us to quantify the relationship between growth and density for this stock in our modeling exercises. Maturity data from 1981-86 were evaluated to ascer- tain if percent maturity at age 2 and 3 changed over the time period. No fish were mature at age 1, and all fish were mature at age 4 + (Overholtz et al. 1988). Per- cent maturity at age 2 ranged from 17% in 1981 to 53% in 1983 (Fig. IE). Percent mature at age 3 ranged from 67% in 1986 to 98% in 1983 and 1984 (Overholtz et al. 1988) (Fig. IE). The maturity data were collected in conjunction with age sampling and were not a priority item. During the critical time of gonadal development in mid- to late April, maturity samples were often sparse because age sampling requirements had been fulfilled; no additional maturity samples were collected because of this. A cur- sory examination of the data revealed that there was an apparent negative relationship with increased stock size at age 2 and no consistent pattern at age 3 (Over- holtz et al. 1988) (Fig. 1). The null hypothesis of no im- pact of density on maturity rates could not be accepted or rejected with the data at hand; we therefore included this potential mechanism in our modeling studies to ascertain its overall importance. During 1967-85 the northwest Atlantic mackerel stock underwent profound changes in recruitment with a subsequent decline in biomass (Fig. IB, C). We con- sidered this a good time period to study changes in predation on mackerel and associated possible changes in natural mortality rates (M). For the purposes of discussion in this paper we define natural mortality rate (M = Ml + M2), where Ml = sources of natural mortal- Overholtz et al.: Assessment advice for Northwest Atlantic mackerel stock 19 o <5 < >- C CD O 1— a. C/3 C o -Q E 3 10-i 10 15 20 25 30 35 40 Length (cm) Year Figure 2 (A) Percent by age group of Atlantic mackerel sampled as prey in Atlantic cod, silver hake, and spiny dogfish stomachs in 1982 (n = 8), 1983 (n = 12), 1984 (w = 32). (B) Length frequency of mackerel sampled in Atlantic cod, silver hake, and spiny dogfish stomachs, 1982-84. (C) Cumulative numbers of ages 1 and 2, and the total stock of mackerel from VPA stock size estimates for 1970-85. ity other than predation, and M2 = predation mortal- ity (ICES 1987). Since there was no multispecies vir- tual population analysis (MSVPA) available to examine annual trends in predation mortality rates (M2), we decided to use another method to examine possible changes in M2. We wanted to investigate possible predation mortality models for mackerel for this period of time. Summaries of NEFC food habits data indicate that spiny dogfish Squalus acanthias, Atlantic cod Gadus moruha, and silver hake Merluccius bilinearis are the most important fish predators on mackerel (Langton and Bowman 1980, Bowman and Michaels 1984, Bow- man et al. 1984). Food habits data were collected from 1973 to 1980, but did not include individual lengths of prey items from these predators. However, maximum, minimum, and average lengths of fish prey were re- Table 1 Percentage of mackerel by weight in stomach samples of silver hake, Atlantic cod, and spiny dogfish, and number of stomachs collected for 1973-76 and 1977-80. 1973-76 1977-80 Species % N % N Silver hake 4.21 2622 Cod 11.50 1009 Spiny dogfish 3.30 389 0.82 1657 0.10 457 0.10 2662 corded. Almost all the mackerel consumed in these years were less than 30 cm (Overholtz et al. 1988). To study the problem in more detail, food habits data from 1982-84 were examined to determine the size and age distribution of mackerel as prey items in the three fish predators. These data were chosen since detailed records of predator and prey length were available. Mackerel up to 35 cm were taken as prey by the three species, but fish 30 cm or less composed the bulk of the prey. These fish were predominantly ages 1 and 2 from the 1981-83 year-classes (Fig. 2A, B). Mackerel appeared to be consumed roughly in proportion to their abundance in the sea during 1982-84 (Fig. 2A, C). Our analysis thus centered on predation by these three predators on age-1 and -2 mackerel. Total food consumed (all species) by each predator was calculated and the average percentage of mackerel by weight com- prising the diet of each predator was estimated separ- ately for the periods 1973-76 and 1977-80 (Table 1). A period average was used because there was not enough information available for annual estimates. These two time-periods were chosen because the design of the food-habits sampling regime was different in each period, and because the abundance of small mackerel was much different in each of the periods (Fig. 2C). The method used to calculate consumption was based on residence times of the predator and prey, percent by weight of mackerel in the predator diet, daily ra- tion estimates for the predator by season, if available, abundance of 1- and 2-year-old mackerel in the sea, and biomass of predators of the correct size distribution (Bowman et al. 1984, Rexstad and Pikitch 1986). This method assumed that the estimate of predator biomass was an appropriate measure of the average standing stock present during the year, that mackerel consumed were only age-1 and -2 fish, and that predators con- sumed mackerel relative to their abundance in the sea. The estimates were made for 1973-80 for each pred- ator species and the total number of age 1- and -2 fish consumed annually was estimated (Table 2). 120 Fishery Bulletin 89(1), 1991 Total consumption of mackerel by each predator was calculated as B, • %BW • %Rmj • Tr (1) where Ci = Bi %BW %Rm Tr = i = consumption of mackerel by predator i, biomass of predator i, daily ration estimate, percent of total ration com- posed of mackerel for pred- ator i, residence time of predator and prey (days), and 1,3. Table 2 Annual consumption of age- 1 and -2 Atlantic mackerel b,V spiny dogfish , silver hake, and Atlantic cod, landings of mackerel for 1973 -80 , and total for both categories (millions of fish). Year Consumption Age 1 Age 2 Landings Tota Age 1 Age 2 Age 1 Age 2 1973 590.0 224.1 161.8 283.2 751.8 507.3 '1974 823.0 135.1 95.9 242.2 918.9 377.3 1975 616.0 170.0 373.7 431.4 989.7 601.4 1976 202.8 278.3 12.5 353.5 215.3 631.8 1977 8.4 15.1 2.0 27.0 10.4 42.1 1978 11.1 6.6 0.1 0.2 11.2 6.8 1979 14.1 1.2 0.4 0.6 14.5 1.8 1980 6.1 8.4 1.2 10.9 7.3 19.3 The estimated consumption in weight was then converted to numbers eaten on the basis of information on the abundance of age-1 and -2 mackerel in the sea and the mean weights of each age group. Con- sumption estimates were combined with landings-at-age for 1973-80 and a new VPA was completed for these years (Overholtz et al. 1988). A residual natural mortality rate (Ml) of 0.20 was used in this analysis to account for other sources of mortality at ages 1 and 2 and for all the other age groups (3-14) in the anal- ysis. A similar assumption has been used by ICES in the multispecies VPA model of the North Sea (ICES 1987). The mortality rates from the VPA were a proxy for F and M2 and were appor- tioned by using the ratios of consumption and landings to total deaths (numbers). This gave an estimate of F and M2 mortalities for 1973-80 (Table 3). Consump- tion (numbers) of age-1 mackerel exceeded landings of that age group in all years from 1973 to 1980 and was generally smaller than landings at age 2 (Table 2). Sizes of incoming year-classes increased up to a factor of two in the revised VPA over the 1973-80 period (Overholtz et al. 1988). When mortality rates from the VPA were apportioned by consumption estimates and landings, natural mortality rates (M = Ml + M2) were generally higher in the 1973-76 period, when mackerel were abundant, than in 1977-80 (Table 3). These values and the new VPA stock size-at-age estimates were used in regressions to study the rela- tionship between M2 and year-class strength. The results of this analysis suggest a positive relationship between M2 for ages 1 and 2 and year-class size (R = 0.37, 0.78, P = 0.363, 0.0234), respectively (Fig. 3A, B). An examination of the scatter plots from these two regressions revealed that the M2 value from 1976 was Table 3 Mortality rates of age-1 and -2 Atlantic mackerel for 1973-80. Year Age 1 Age 2 Z F M2 Ml Z F M2 Ml 1973 0.78 0.12 0.46 0.20 0.75 0.31 0.24 0.20 1974 0.66 0.05 0.41 0.20 0.86 0.42 0.24 0.20 1975 0.71 0.19 0.32 0.20 0.83 0.45 0.18 0.20 1976 0.81 0.04 0.57 0.20 0.93 0.41 0.32 0.20 1977 0.40 0.04 0.16 0.20 0.42 0.14 0.08 0.20 1978 0.61 0.00 0.41 0.20 0.39 0.01 0.18 0.20 1979 0.30 0.00 0.10 0.20 0.31 0.04 0.07 0.20 1980 0.44 0.04 0.20 0.20 0.38 0.10 0.08 0.20 high relative to the number of age-1 and -2 fish in the stock. There were fewer age-1 fish in the stock in 1976 (Fig. 2C); furthermore, food habits data indicated that mackerel was not present in the diet of the three pred- ators in 1976. The 1976 data point was dropped and a new regression was fitted for age 1 (R = 0.60, P = 0.157, Fig. 3A). We were cognizant of the fact that results of this analysis may have been influenced by the assumption of proportional feeding. However, the suggestion that natural mortality rates may change with year-class size is an interesting research question. A positive relation- ship between year-class size and predation mortality rate has obvious implications for assessment and man- agement, and thus the potential impacts of density- dependent predation mortality were a major focus of our modeling studies. Model structure A simulation model addressing changes in growth, per- cent maturity-at-age, and predation mortality rates Overholtz et al.: Assessment advice for Northwest Atlantic mackerel stock 121 04 • age 1 ~~ I I I l~~ age 2 1000 2000 3000 YEAR CLASS SIZE (millions) Figure 3 Relationship between predation mortality (M2) and year-class size for age-1 and -2 Atlantic mackerel from calculations of mackerel consumption by Atlantic cod, silver hake, and spiny dogfish for 1973-80. (M2) was constructed to evaluate the potential impact of these population regulatory mechanisms in the con- text of single-species assessment advice. The model was a basic fishery simulator much in the same style as many other common fishery models (Walters 1969, Sissenwine 1977). An age-structured Baranov catch equation was used to compute annual fishery yields, and the negative exponential relationship was used to update stock size for 14 age groups. A stochastic recruitment function was used to model recruitment for the mackerel stock. A three-parameter model (Shephard 1982; Table 4) with a lognormal white- noise multiplier was used to generate an annual esti- mate of recruitment as: R = (a*SSB)/[l + (SSB/K) b ] * lnm where R = recruitment at age 1, SSB = spawning-stock biomass, (2) Table 4 Parameters used in stochastic simulations of the Atlantic mackerel stock; e-4 = io- 4 , e-5 = 10" 5 , e-6=10" 6 . age a b K Weight (Wt, ) 1 0.122 -1.24e-5 — Growth (Gj) 2 0.187 -1.42e-5 — increment 3 0.154 -1.08e-5 — 4 0.132 -1.03e-5 — 5 0.102 -8.62e-6 - Recruitment (R) 1 5.800 1.7 600.0 Maturity (PM) 2 0.543 -2.14e-4 — 3 1.043 -2.14e-4 - Predation 1 0.600 3.00 2.5 mortality (M2) 2 0.500 2.00 2.5 a,b,K = parameters, lnm = lognormal multiplier with u and s from the SR data. Recruitment estimates were scaled upward by a fac- tor of 1.5 to account for the fact that the original VPA age-1 stock size estimates do not reflect higher natural mortality rates due to predation (Overholtz et al 1988). Growth-at-age was based on a two-stage model that related life-history characteristics and year-class size to growth increment for a year-class (Overholtz 1989). Since age-1 fish maintain a separate distribution from the adult stock (Sette 1950) a relationship between age-1 growth and corresponding age-1 year-class size was used to predict weight at age 1. The relationship was parameterized (Table 4) with available empirical data, such that size at age 1 varied from 48 g for slow- growing fish to 122 g for fast-growing fish (Overholtz 1989). Weight of age-1 mackerel was calculated as Wtj = a - bN, (3) where Wt = average weight (g) at age, Nj = year-class size at age 1 (numbers), a,b = parameters. Age 2-5 growth was determined by relationships between adult stock size (numbers) and growth-at-age (Table 4) calculated as Wt, = Wtj_! + Gi (4) where Wt; = average weight at age i, i = 2,5, G; = annual age-specific growth increment (g), and 122 Fishery Bulletin 89(1), 1991 G; = a, - bjSS (5) where SS = total adult stock size (numbers), a,b = age specific parameters. The increment from ages 2 to 5 was smaller the larger the adult stock. Fish weight at ages 1-5 was the result of growth in the first year and subsequent in- crements at ages 2-5; thus, age-1 growth partially determined the average weight of a fish throughout its lifetime. If the stock was reduced in any given year, the cohort could recover and grow faster. Growth at ages 6 + was assumed to follow trends in the recent data, since by this time cumulative mortality has usual- ly been sufficient to reduce a cohort to lower levels. Percent maturity at ages 2 and 3 was assumed to vary based on a relationship between the fraction mature and spawning stock size calculated as PM, = a - b(SSB) (6) where PMj = percent mature at age i, i = 2,3, SSB = spawning-stock biomass, a,b = parameters. This submodel was parameterized (Table 4) so that the maturity of age-2 fish can vary from 20 to 50%, while maturities for age-3 fish range from 70 to 100%. Natural mortality due to predation (M2) for ages 1 and 2 fish was estimated from a relationship between M2 and year-class size-at-age (Fig. 4; Table 4) calcu- lated as r 12 3 4 5 6 7 8 Year Class Size (billions) Figure 4 Type 3 functional relationship between predation mortality rate (M2) and year-class size used in the density-dependent simulation model (DDM). sponses in regulating this stock. Monte Carlo simula- tions were produced for a variety of different scenarios, and average results from 1000 annual data points were summarized. Results from the model were compared with forecasts from the current standard assessment model (STD). M2j = (a*YCi)/[l+(YCi/K) h ] (V) where M2 natural mortality due to predation on age i, i = 1,2, YCj = year-class size at age i, i = 1,2, a,b,K = parameters. M2 mortalities on age-1 and -2 fish could reach approx- imately 1.0 and 0.6, respectively, with this model. This relationship was used since it approximates the findings of our mortality study over an initial range of stock size (Fig. 3A, B), and since it appears to be an appropriate predator prey response model (Holling 1965, Murdoch 1973). Although this model does not produce a typical type-3 response (Holling 1965) exact- ly, since there is no inflection point over the initial stock sizes (Fig. 4), it serves as a sufficient functional model to study the natural mortality mechanism. The density-dependent simulation model (DDM) was used to study the impact of different levels of fishing mortality, management strategies, and to investigate hypotheses concerning the role of compensatory re- Model sensitivity and validation The sensitivity of model results to the different density- dependent mechanisms was investigated by compar- ing catch in 1987 and spawning stock in 1988 and 1991 for all the different combinations of growth, maturity, and natural mortality at a reference fishing mortality of 0.05. In runs where only a single mechanism was examined, 1987 catch was most affected by changes in the growth pattern (mean weights) of the stock (Fig. 5-B). Spawning stock in 1988, on the other hand, was almost equally sensitive to maturity and natural mor- tality. Density-dependent natural mortality influenced 1991 spawning stock to the greatest degree (Fig. 5-C). When the results of pairing the mechanisms were ex- amined, weight and natural mortality had the greatest effect on catch, percent maturity, and natural mortality on SSB in 1988 and weight and natural mortality on SSB in 1991 (Fig. 5-D, E, F). When the three mechan- isms were all operating there was no change in the im- pact on 1987 catch, but spawning stock in 1988 and 1991 was several percentage points lower (Fig. 5-G). Overholtz et al.: Assessment advice for Northwest Atlantic mackerel stock 123 10 F=0.05 u -io LU DC LU -20- Q -30 -40 -50 WTO m as ^ ^ ^ i i i r A B C D E F G F=0.29 A B C D E F G Figure 5 Impact on 1987 catch, 1988 spawning-stock biomass (SSB), and 1991 SSB of running the density-dependent simulation (DDM) model with all possible combinations of growth, matur- ity, and predation mortality. (A) maturity; (B) growth; (C) predation mortality; (D) growth, maturity; (E) maturity, preda- tion mortality; (F) growth, predation mortality; (G) maturity, growth, predation mortality. A validation run of the density-dependent model was produced for comparison with the observed time-series of catch and VPA biomass for 1967-85 (Fig. 6). The simulated series was produced by using the same fish- ing mortality series as in the VPA, recruitment scaled upward by a factor of 1.5, and density-dependent growth, maturity, and natural mortality. The pattern of simulated versus observed catch and biomass is quite comparable in terms of trend and magnitude, except for a few years in the early 1970s. This occurs even though the density-dependent model has greatly dif- ferent natural mortality rates at ages 1 and 2 and much higher recruitment. Another run with the same F pat- •_. 500 400^ 1500 -I o O / / X \ — 1000 ■•/' 1 m , | 500 -' O "V_/ m n-L 1967 1 ] 1 1 1 1 1 1 1 1 1 I 1 I 1 1 1970 1973 1976 1979 1982 1985 Year Figure 6 Comparison of observed (- -) catch and biomass versus simulated (— — ) catch and biomass obtained from running the density-dependent simulation (DDM) model with historic estimates of fishing mortality, historic recruitment estimates scaled by a factor of 1.5, and density-dependent growth, maturity, and predation mortality. tern, recruitment scaled by 1.5, and no density depen- dence produced the same trend in catch and biomass for 1967-85, but the values were approximately a fac- tor of two larger than the observed series. Thus, not including the density-dependent component resulted in values of catch and SSB that were greatly different than the observed series. The model components were validated by comparing the different outputs produced by the various mechan- isms with available empirical data. In some cases, rela- tionships were re-parameterized or tuned to produce results in the same ranges as observed in the empirical database. The model was used to investigate a variety of different problems. Runs from the STD model were compared with the results of the different density- dependent model outputs to gauge the changes that oc- curred in catch, total stock, spawning stock, mean weights, and other factors. STD runs were parameter- ized with the same data as that used in the 1986 assess- ment (Overholtz et al 1988). 124 Fishery Bulletin 89|1), 1991 100- Frequency W ^ O) CO o o o o ill ji \ \ ft V f I \ n I i ^~~ — ~ - ! i 1 100 200 300 400 Yield (ooo's mt) 100 80 >■ o g 60 °~ 40- it 20 / ;/ V/L x - 1000 2000 3000 4000 Biomass (ooo's mt) Figure 7 Comparison of standard assessment (STD) model ( ) and density-dependent simulation (DDM) model ( ) catch and spawning-stock biomass (SSB) for 5000 iterations of each model, with fishing mortality set at 0.05. 100 Frequency M -P> O) CO 3 O O O O I I I I 100 200 300 400 Yield (ooo's mt) 100 80 > S 60 °" 40 - 20- / ' \ 1 \ N / ' \ S / ' \ ' 1000 2000 3000 4000 Biomass (ooo's mt) Figure 8 Comparison of standard assessment (STD) model ( ) and density-dependent simulation (DDM) model ( ) catch and spawning-stock biomass (SSB) for 5000 iterations of each model, with fishing mortality set at 0.29 (F , ). Model results Short-term perspective Since fishing mortality has been so low in recent years and the stock has increased, an attempt was made to investigate the effect of further stockpiling of fish on expected catches and spawning stock size. To address this problem, a series of 5-year simulations were used to compare the density-dependent model results with projections from the standard model. Fishing mortal- ity was set at 0.05, a value close to the average rate over the last several years. These runs suggest that advice based on the standard model would have over- predicted catch by about 8% in 1987 and spawning stock size in 1988 by about 13% (Fig. 5-G). Further- more, if the standard model were used to project more than a few years into the future, the estimate of spawn- ing stock could possibly be too large; the 1991 estimate would be lower by 37% for the DDM model (Fig. 5-G). If the frequency of yields over this 5-year series is calculated, the STD model is more optimistic with a mean catch of 74,892 mt versus 63,686 mt for the DDM model (Fig. 7). Similarly for spawning-stock biomass (SSB), the STD model suggests a higher mean SSB of 1.5 million mt versus 1.1 million mt for the DDM model (Fig. 7). Since F .i is an important benchmark fishing-mor- tality rate in the present management plan, another 5-year summary with F = 0.29 (F, u ) was also pro- duced. As with the previous example, mean yield for the STD model was considerably higher, 242,939 mt versus 197,116 mt for the DDM model (Fig. 8). Spawn- ing-stock biomass would be considerably different under the two perspectives with an estimated mean SSB of 1.0 million mt under the STD model and 725,887 mt under the DDM model (Fig. 8). Long-term perspective To develop a longer-term perspective, simulations at the same two levels of fishing mortality (F = 0.05, 0.29) were produced for 15-year series to allow sufficient Overholtz et al.: Assessment advice for Northwest Atlantic mackerel stock 125 150 E 125 ■o ™ 2 75 sz 50 o ro 25 U /*^= . , , , | , , i i | i i i i | 2000- £ co 1500 "o O _ _ -. — 1000 Biomass en o , , i 5 10 15 Year Figure 9 Comparison of standard assessment (STD) model ( ) and density-dependent simulation (DDM) model ( ) catch and spawning-stock biomass (SSB) for 15-year runs of each model, with fishing mortality set at 0.05. E *o o o o CD u 300 250 200 150 100 50 2000 E co 1500 "o O O — 1000 CO E Q m 500 5 10 15 Year Figure 10 Comparison of standard assessment (STD) model ( ) and density-dependent simulation (DDM) model ( ) catch and spawning-stock biomass (SSB) for 15-year runs of each model, with fishing mortality set at 0.29 (F , ). time for the stock to reach an equilibrium point. At F = 0.05, and after roughly 7-8 years, catch would be higher under the STD model than for the DDM model (Fig. 9). Spawning-stock biomass would also be con- siderably higher, about 1.7 million mt for the STD model versus about 1.1 million mt for the DDM model (Fig. 9). Fishing the stock at F .i for a 15-year period would result in an equilibrium catch of roughly 250,000 mt (STD) versus 150,000 mt (DDM) after about 7-8 years (Fig. 10). The spawning stock would appear to be much higher, 1.1 million mt versus 615,000 mt, for the STD and DDM models, respectively (Fig. 10). To evaluate the overall differences between the two models, two measures of performance and one measure of risk were produced for fishing mortality values rang- ing from 0.1 to 0.6, in 0.1 increments for a total of 1000 runs of each model. Mean catch, coefficient of varia- tion (CV) of catch, and proportion of time the spawn- ing stock fell below 600,000 mt, the management plan benchmark, were calculated for year 10 of the simula- tion. Based on the previous model runs, the tenth year of the simulation appeared to be a reasonable choice for a summary year (Figs. 9, 10). Mean catch would continue to increase dramatically under the STD model until F reached about 0.40 and would remain relatively constant thereafter. Yields would also steadily increase for the DDM model at F's ranging from 0.1 to 0.3, but would remain constant after that point (Fig. 11 A). A comparison of expected yields under the two models suggests that the STD model is always more optimistic, especially at the higher levels of fishing mortality (Fig. 11). Mean catch at F = 0.5 would be about 125,000 mt higher under the STD model. Variability as measured by CV's of the catch in- creased with increases in fishing mortality for the STD model, but remained relatively constant at about 50% for the DDM model. Thus the STD model is more op- timistic at lower levels of fishing mortality and becomes 126 Fishery Bulletin 89(1), 1991 300 -i CD U c CO CD o CO (J > 2 3 4 5 Fishing Mortality Figure 1 1 (A) Mean catch, (B) coefficient of variation of catch, and (C) proportion of the time spawning-stock biomass (SSB) falls below 600,000 mt for year 10 of a set of 10-year simulations of the standard assessment (STD) (darkened bar) and density- dependent simulation (DDM) (cross-hatched bar) models, with fishing mortalities of 0.1-0.6 in 0.1 increments. progressively more variable as fishing mortality in- creases (Fig. 11B). Both models have relatively high CV's at all levels of fishing mortality, suggesting a large range in possible catches. The minimum spawning-stock criteria of 600,000 mt is a threshold biomass defined in the present fishery management plan. This value was chosen by managers because a relatively clear demarcation point between low and high recruitment is evident in the 1962-85 stock-recruit data series (Anderson 1985). When the simulation results are expressed relative to the pro- portion of times the SSB drops below the 600,000 mt level for each model, the STD model appears to be much less prone to risk and, therefore, is much more optimistic than the DDM model results. The propor- tion ranges from 0.0 at F = 0.1 to about 0.52 at F = 0.6 for the STD model, while P ranges from 0.01 to 0.94 for the DDM model (Fig. 11C). The STD model results suggest that there is little risk of the SSB dropping below the threshold, even at fishing mortality rates of 0.4-0.5 (Fig. 11C). The DDM model results are much less optimistic, suggesting the need for some concern at fishing mortality levels of 0.3 and greater (Fig. 11C). Discussion Biological interactions may play an important role in regulating marine ecosystems (Sherman et al. 1981, Walters et al. 1986, Overholtz and Tyler 1986, Over- holtz et al. In press). Species interactions are becom- ing an important fishery management issue, and as- sessment advice is increasingly contingent on these mechanisms (Anderson and Ursin 1977; Pope 1976, 1979; Shephard 1984; ICES 1987,1988). This study in- dicates that the stock dynamics of Atlantic mackerel are not only influenced by fishing, but that predation and intraspecific compensatory mechanisms including density-dependent growth are strong influences that probably effect yield forecasts and management advice in the short and long term. Large differences in mackerel growth suggest that year-class size partially influences the initial pattern of growth during a cohort's first several years. Adult stock size probably plays an important role in regu- lating growth after a year-class recruits to the adult portion of the population (Overholtz 1989). Declines in growth are probably significant as stock biomass in- creases (Overholtz 1989). Model results, although not presented in this paper, suggest that mackerel would reach a larger size if the mackerel stock were fished more heavily. Larger catches of Atlantic mackerel would probably cause growth to stabilize at higher rates, and more of the annual production would be available for harvest. Predation mortality rate has usually not been ac- counted for in the past in most assessment work, but recent studies have shown the importance of including this mechanism to enhance stock assessments (Ander- son and Ursin 1977; Shephard 1984; ICES 1987,1988; Overholtz et al. 1990). Our analysis suggests that predation probably has a major influence on the dynamics of Northwest Atlantic mackerel. Predation mortality is probably the largest component of natural mortality on this stock, since other sources such as general diseases, parasitism, and epizootics are not thought to be important sources of mortality on most fish stocks on an annual basis (Anderson 1979). Strong year-classes of mackerel may attract elevated levels of predation, in contrast with the usual assumption of con- stant natural mortality. Other studies have suggested that predation mortality rates should continually decline or remain constant as abundance increases (Sparre 1984). Our model results indicate that preda- Overholtz et al.: Assessment advice for Northwest Atlantic mackerel stock 127 tion mortality rates on Atlantic mackerel are probably much higher than previously thought. The results of the model projections show that un- less the impacts of compensatory mechanisms are ac- counted for, evaluations of current stock status using the current standard assessment methodology may, in fact, be optimistic and risky if catches are increased to high levels in the future (Fig. 11). The differences in results between the two models are, of course, con- tingent on the parameterizations of the growth and predation mortality submodels and how recruitment is scaled in the density-dependent model. Two advances in research on mackerel would help to improve our ability to assess the stock: An MSVPA to provide cor- rectly scaled estimates of recruitment, and a general predation mortality model that would provide useful estimates of M2's for forecasting purposes. Although recent assessment advice indicates that catches can be increased on the mackerel stock (Overholtz and Parry 1985), it perhaps needs to be modified to accommodate the results of this study. The current management regime relies on catch and stock size projections based on an F .i strategy. The use of a reference point such as F .i is probably not very useful for mackerel since growth, sexual matur- ity, and natural mortality rates appear to fluctuate con- siderably. This concept is best applied in situations where these important variables are stable in the long term. A more appropriate approach might be to remove a moderately large sustainable catch annually or apply an appropriate constant effort level over several years, preserve a reasonable amount of spawning-stock bio- mass, and monitor the results. This method would be keyed to some of the uncertainties in stock dynamics that we have investigated in this study and would pro- vide information on stock responses with a low prob- ability of stock collapse (i.e., F = 0.2-0.3; Fig. 11C). Additional analyses are necessary to confirm the population processes that were modeled in this study. Weights of individual fish should be monitored closely to assess future changes. Sexual maturities of ages 2-3 fish should also be followed annually. Collection of these data would also allow better parameterization of the growth and maturity models. Sufficient samples must be collected at the correct times to assess whether these two variables, particularly percent maturity, are continuing to change with stock density. Additional food habits sampling at critical times and places would help confirm and quantify the relationships found in this analysis. Obtaining some information on predation mortality on age-0 mackerel would be valuable. Preliminary data suggest that predator preference may play an important role in determining the levels of predation on available prey species. Recent declines in sand lance Ammodytes dubius populations may in- crease predation mortality on mackerel and Atlantic herring Clupea harengus. This points to the need for a multispecies VPA where simultaneous impacts of predation may be investigated. Improved predation models that account for predator preference and prey abundance would allow for more accurate predictions of the impacts of these important factors, and better management advice could be provided (Livingston 1986). Larger mackerel are preyed upon by marine mammals, large pelagic fishes, and sea birds (Stillwell and Kohler 1982, 1985; Payne and Selzer 1983; Payne et al. 1984; Overholtz et al. 1990). The impact of these predators is no doubt important, but was not evaluated in this study. Acknowledgments We thank the personnel from the Northeast Fisheries Center and other institutions who have collected data on research surveys over the last 25 years. We are grateful to the foreign fishery observers and scientists who collected data from the Polish commercial fishery. Special thanks to Louise Dery who provided the age data for the study. We would also like to thank Brian Rothschild for his guidance and insight in the initial phases of this research. Citations Anderson, E.D. 1985 Status of the Northwest Atlantic mackeral stock— 1984. Ref. Doc. 85-03, Woods Hole Lab., Northeast Fish. Sci. Cent., Natl. Mar. Fish. Serv.. NOAA, Woods Hole, MA, 46 p. Anderson, K.P., and E. Ursin 1977 A multispecies extension to the Beverton and Holt theory of fishing, with accounts of phosphorous circulation and primary production. Medd. Dan. Fisk. Havunders. 7:319-435. Anderson, R.M. 1979 The influence of parasitic infection on the dynamics of host population growth. In Anderson, R.M., B.D. Turner, and L.R. Taylor (eds.), Population dynamics, p. 245-281. Blackwell Sci. Publ., Oxford. Bowman, R.E., and W.L. Michaels 1984 Food of seventeen species of northwest Atlantic fish. NOAA Tech. Memo. NMFS-F/NEC-28, Northeast Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA, 183 p. Bowman, R.E., R. Eppi, and M.D. Grosslein 1984 Diet and consumption of spiny dogfish in the northwest Atlantic. ICES Demersal Fish Comm., ICES CM 1984/G:27, 16 p. Holling, C.S. 1965 The functional response of predators to prey density and its role in mimicry and population regulation. Mem. Entomol. Soc. Can. 91:293-320. Hoy, D.L., and G.M. Clark 1967 Atlantic mackerel fishery, 1804-1965. Fish. Leaf]. 603, U.S. Dep. Inter., Wash. DC, 9 p. 128 Fishery Bulletin 89(1), 1991 ICES (International Council for the Exploration of the Sea) 1987 Report of the ad hoc multispecies assessment working group. ICES CM 1987/Assess:9, 130 p. 1988 Report of the multispecies assessment working group. ICES CM 1988/Assess:23, 154 p. Jenson, A.C. 1967 A brief history of the New England offshore fisheries. Fish. Leafl. 594, U.S. Dep. Inter., Wash. DC, 14 p. Langton, R.W., and R.E. Bowman 1980 Food of fifteen northwest Atlantic gadiform fishes. NOAA Tech. Rep. NMFS SSRF 740, 23 p. Lett, P.F. 1980 A comparative study of the recruitment mechanisms of cod and mackerel, their interaction, and its implication for dual stock assessment. Can. Tech. Rep. Fish. Aquat. Sci. 988, 45 p. Livingston, P. A. 1986 Incorporating fish food habits data into fish population assessment models. In Simenstad, C.A., and G.M. Cailliet (eds.), Contemporary studies on fish feeding, p. 225-234. Dr. Junk Publ., Dordrecht, Netherlands. McKay, K.T. 1979 Synopsis of biological data of the northern population of Atlantic mackerel (Scomber scombrus). Can. Fish. Mar. Serv. Tech. Rep. 885, 26 p. Murdoch, W.W. 1973 The functional response of predators. J. Appl. Ecol. 10: 335-342. NEFC (Northeast Fisheries Center) 1987 Report of the fourth NEFC stock assessment workshop. Ref. Doc. 87-07, Woods Hole Lab., Northeast Fish. Cent., Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA, 102 p. Overholtz, W.J. 1989 Density dependent growth in the northwest Atlantic stock of Atlantic mackerel. J. Northwest Atl. Fish. Sci. 9:127-135. Overholtz, W.J., and B.L. Parry 1985 Update of the status of the northwest Atlantic mackerel stock for 1985. Ref. Doc. 85-113, Woods Hole Lab., Northeast Fish. Cent., Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA, 5 p. Overholtz, W.J., and A.V. Tyler 1986 An exploratory simulation model of competition and predation in a demersal fish assemblage on Georges Bank. Trans. Am. Fish. Soc. 115:805-817. Overholtz, W.J., S.A. Murawski, W.L. Michaels, and L.M. Dery 1988 The effects of density dependent population mechanisms on assessment advice for the northwest Atlantic mackerel stock. NOAA Tech. Memo. NMFS F/NEC-62, Northeast Fish. Cent., Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA, 49 p. Overholtz, W.J., S.A. Murawski, and K.L. Foster In press Impact of predatory fish, marine mammals, and sea- birds on the pelagic fish ecosystem of the Northeastern USA. In Daan, N., and M. Sissenwine (eds.), Symposium on multispecies models relevant to management of living resources, The Hague, Netherlands, October 1989. Rapp. P.- V. Reun. Cons. Int. Explor. Mer. Payne, P.M., and L.A. Selzer 1983 Population distribution, abundance, and prey require- ments of the harbor seal in southern New England. NMFS Contract Rep. NA-82-FA-00007, by Manomet Bird Obser- vatory, Manomet, MA. Northeast Fish. Cent., Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA, 51 p. Payne, P.M., L.A. Selzer, and A.R. Knowlton 1984 Distribution and density of cetaceans, marine turtles, and seabirds in the shelf waters of the northeastern United States, June 1980-December 1983, based on shipboard observa- tions. NMFS Contract Rep. NA-81-FA-C-00023, by Manomet Bird Observatory, Manomet, MA. Northeast Fish. Cent., Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA, 246 p. Pope, J.G. 1976 The effect of biological interactions on the theory of mixed fisheries. Int. Comm. Northwest Atl. Fish. Sel. Pap. 1:163-170. 1979 A modified cohort analysis in which constant natural mor- tality is replaced by estimates of predation levels. ICES Pelagic Fish Comm., ICES CM 1979/H:16, 15 p. Rexstad, E.A., and E.K. Pikitch 1986 Stomach contents and food consumption estimates of Pacific hake, Merluccius productus. Fish. Bull., U.S. 84: 947-956. Sette, O.E. 1950 Biology of the Atlantic mackerel (Scomber scombrus) of North America: Part 2: Migrations and habits. Fish. Bull.. U.S. 51:251-358. Sette, O.E., and W.H. Needier 1934 Statistics of the mackerel fishery off the coast of North America. Invest. Rep. 19, U.S. Dep. Commer., Wash. DC. 48 p. Shepherd, J.G. 1982 A versatile new stock-recruitment relationship for fish- eries and the construction of sustainable yield curves. J. Cons. Cons. Int. Explor. Mer 40:67-75. 1984 A promising method for the assessment of multispecies fisheries. ICES Demersal Fish Comm., ICES CM 1984/G:4, 12 p. Sherman, K., C. Jones, L. Sullivan, W. Smith, P. Berrien, and L. Ejsymont 1981 Congruent shifts in sand eel abundance in western and eastern North Atlantic ecosystems. Nature (Lond.) 291: 486-489. Sissenwine, M.P. 1977 A compartmentalized simulation model of the Southern New England yellowtail flounder, Limandafemujinea, fishery. Fish. Bull., U.S. 75:465-482. Sparre, P. 1984 A computer program for estimation of food suitability coefficients from stomach content data and multispecies VPA. ICES Demersal Fish Comm.. ICES CM 1984/G:25, 60 p. Stillwell, C.E., and N.E. Kohler 1982 Food, feeding habits, and estimates of daily ration of the shortfin mako (Isurus oxyriwhus) in the northwest Atlan- tic. Can. J. Fish. Aquat. Sci. 39:407-414. 1985 Food and feeding ecology of the swordfish Xiphias gladius in the western north Atlantic ocean with estimates of daily ration. Mar. Ecol. Prog. Ser. 22:239-247. Walters, C.J. 1969 A generalized computer simulation model for fish popula- tion studies. Trans. Am. Fish. Soc. 98:505-512. Walters, C.J., M. Stocker, A.V. Tyler, and S.J. Westrheim 1986 Interaction between Pacific cod [Gadus macrocephalus) and herring (Clupea harengus pallasi) in the Hecate Strait, British Columbia. Can. J. Fish. Aquat. Sci. 43:830-837. Abstract.- The equilibrium con- tribution of hatchery-released juve- niles to a rockfish fishery is evalu- ated by using a yield-per-recruit model. Hatchery-released juveniles may be worth up to an estimated US$0.16 per juvenile to the fishery. The use of hatchery releases to re- store a depleted population of Pacific ocean perch Sebastes alutus is exam- ined with the Deriso-Schnute model. This model indicates that hatchery releases have the potential to sub- stantially increase a stock's yield and rate of recovery during the recovery period. Evaluation of Hatchery Releases of Juveniles to Enhance Rockfish Stocks, with Application to Pacific Ocean Perch Sebastes alutus * Jeffrey J. Polovina Honolulu Laboratory, Southwest Fisheries Science Center National Marine Fisheries Service. NOAA 2570 Dole Street. Honolulu. Hawaii 96822-2396 There is a long history of attempts worldwide to enhance marine fisher- ies by releasing hatchery -reared juve- niles. However, few attempts have had long-term success (Yatsuyanagi 1982, Botsford and Hobbs 1984, Isi- basi 1984, Ulltang 1984). In theory, hatchery releases can enhance fish- eries in two ways. First, juvenile re- leases can be added to the natural stock on a long-term basis to support a higher level of fishery harvest than that achieved from the natural stock alone. Secondly, juvenile releases can be used on a short-term basis to in- crease a depleted natural stock more rapidly, then discontinued once the natural stock has recovered. Pilot releases of the rockfish Sebas- tes schlegeli indicate that large-scale rearing and releases of rockfish may be biologically and technically possi- ble (Sakai et al. 1985, Kusakari In press). The merit of hatchery releases of juveniles for fishery enhancement is examined in the present paper with mathematical models. Specifically, the Beverton and Holt (1966) yield- per-recruit model will be used to eval- uate the sustainable increase in fish- ery catches from a long-term release of hatchery-reared juveniles. The De- riso-Schnute delay-difference age- structure model (Zheng and Walters 1988) will be used to evaluate the fishery benefits from short-term ju- venile releases to increase the recov- ery of depleted rockfish stocks. Models and methods To evaluate the equilibrium contribu- tion of hatchery-released juveniles to a rockfish fishery, the Beverton and Holt (1966) equation was used to ex- press the equilibrium yield-per-re- cruit as a function of the following ratios: instantaneous natural mortal- ity to von Bertalanffy growth (M/K), length at recruitment to asymptotic length, and fishing mortality to natural mortality (F/M) (Beverton and Holt 1966). If a recruit in the yield-per-recruit model is taken to represent a hatchery-released juve- nile rather than a recruit from the natural population, then the equilib- rium yield per hatchery-released ju- venile can be computed from yield- per-recruit tables (Beverton and Holt 1966). Specifically, let Y/R denote the yield per recruit where the re- cruits are of some reference age, say the age at which they have just be- come demersal. Then the yield per released fish at age t is just »Mt Rn Manuscript accepted 7 August 1990. Fishery Bulletin, U.S. 89:129-136 (1991). 'Presented at U.S. -Japan Symposium on Re- production and Early Life History in the Genus Sebastes, Honolulu, Hawaii, June 1989. where M is the natural mortality rate from the reference age until release age t. The values of Y/R are tabled as a function of M/K and F/M (Bever- ton and Holt 1966). The equilibrium yield per hatchery-released juvenile 129 130 Fishery Bulletin 89(1), 1991 Table 1 Von Bertalanffy growth (K), natural mortality (M) M/K, and maximum age for seven commercially important species of rockfish (parameter estimates taken from Pacific Fishery Management Council 1989). Max. age Common name M/yr K/yr M/K (yr) Pacific ocean perch 0.05 0.09 0.56 70+ Yellowtail 0.07 0.16 0.44 64 Shortbelly 0.25 0.21 1.19 12 Widow 0.15-0.20 0.15 1.17* 58 Canary 0.01-0.09 0.16 0.31* 75 Chilipepper 0.20 0.18 1.11 16 Bocaccio 0.25 point of the 0.11 range. 2.27 36 *M taken as the mic will be estimated for values of release age t, M, M/K, and F/M, which are representative estimates for rock- fish populations. To evaluate the short-term releases of hatchery- reared juveniles to restore a depleted rockfish popula- tion, a Deriso-Schnute delay-difference age-structure model is used (Zheng and Walters 1988). Deriso (1980) derived a population model that combined simple sur- plus production models with more detailed age-struc- ture models. Schnute (1985) modified this model with a three-parameter Brody growth model. This Deriso- Schnute model depends on a Brody growth parameter (p), the age of recruitment to a fishery (k), body weights at ages k and k - 1 (W k and W k _ ] , respectively), and total annual survival in year t (s t ). Thus, the biomass in year t + 1 (B t + 1) is described as B t+ i = (l + p)s t B t - pstSt.jBt.j + R t + 1 PSt— — Rt, Wk where R t is the recruitment to the fishery in year t (Schnute 1985). A Ricker stock-recruitment relation- ship is used to model the recruitment (in weight) from the natural rockfish population and from hatchery- released 1 -year-old juveniles to the fishery to obtain the total recruitment (in weight) function (R t ): R t = AS t _ k exp(-BS t _ k )exp(z) + H t - k + 1 W k exp(-M(k-l)); where S t is the spawning biomass, H t is the number of hatchery-released 1-year-old juveniles, M is natural 0.05 0.1 10.18 0.25 0.33 0-43 0.54 0.67 0.62 1.00 1.22 1. 50 1 86 2.33 3 00 4.00 5.67 9 0019-00 F • M" 1 Figure 1 Beverton and Holt yield-per-recruit as a fraction of asymp- totic weight at optimum size at entry to the fishery as a func- tion of F/M for three levels of M/K. mortality, and A and B are constants. This model assumes hatchery juveniles are added to the natural population without any density-dependence. The term exp(z) is used to add a stochastic element to the natural recruitment function when the variable z represents a random variable, which has a normal distribution, a mean of 0, and a variance of a 2 . When o 2 is set to 0, the stochastic term is eliminated, and deterministic recruitment is assumed. The Ricker stock-recruitment relationship appears to represent an appropriate model of recruitment in a natural rockfish population (Archi- bald et al. 1983). The Deriso-Schnute model is fit to catch and fishing mortality data for Pacific ocean perch Sebastes alutus from Queen Charlotte Sound, Canada, during 1963-77 when the population was fished from an estimated biomass of 82,000 metric tons (t) to 13,000 t (Archibald et al. 1983). The model with its estimated parameters is then used to estimate the equilibrium yield curve to determine the biomass and fishing mortality that achieve maximum sustainable yield (MSY). Then the model with stochastic recruitment is used to simulate the recovery of this depleted stock to the biomass level that supports MSY under several management strate- gies, including the release of hatchery-reared juveniles. Results Values of M for rockfishes typically range from 0.01 to 0.25 per year, K from 0.09 to 0.21 per year, and M/K ratios from 0.44 to 2.27 (Table 1). Values of yield- Polovina: Hatchery releases of juvenile Sebastes alutus 131 a Actual -t- Estimated O 63 64 '65 '66 67 '68 '69 70 71 72 73 74 75 76 77 Year Figure 2 Historic catches (10 3 metric tons) of Pacific ocean perch from Queen Charlotte Sound and estimated catches from the fit of the Deriso-Schnute model, 1963-77. per-recruit (expressed as a fraction of the asymptotic weight of the fish, as a function of F/M, and for M/K equaling 0.5, 1.0, and 2.0) are shown in Figure 1. These yield curves assume that the ratio of the length-at- recruitment to the asymptotic length is optimum to achieve maximum yield-per-recruit. For M/K ranging from 0.5 to 2.0, the optimum length at harvest will be 45-86% of the asymptotic length (Beverton and Holt 1966). Thus, in the case of Pacific ocean perch, which has an asymptotic weight of about 1.4 kg, if M/K = 0.5, F/M = 1.0, and M = 0.05 per year, the contribution to the fishery of an individual juvenile released from the hatchery at age 0.25 is calculated as the product of 1.4 kg x 0.17 (from Fig. 1), multiplied by exp(0.05 • 0.25) or 0.24 kg. The average 1988 ex-vessel price for all rockfishes was US$0.67 per kg (National Marine Fisheries Service 1989); therefore, each 3-month-old, hatchery-released juvenile on the average is worth $0.16 to the fishery. The contribution of a juvenile to the fishery is strongly inversely related to M/K (Fig. 1). For example, if M/K is 1.0 rather than 0.5, then the contribution is only about one-half as much. Also, as long as natural mortality is assumed to be low and con- stant, changes in the release age have little influence on the contribution of the juvenile to the fishery. How- ever, it is quite possible that natural mortality of young juveniles varies considerably by age and an optimum release age exists, although we have no data to docu- ment either case. Pacific ocean perch in Queen Charlotte Sound, Can- ada, underwent heavy exploitation from 1963 to 1977 (Fig. 2) (Archibald et al. 1983). A reconstruction of the Table 2 Parameters for the Deriso-Schnute model fit to Pacific ocean perch in Queen Charlotte Sound Parameter Value Source Recruitment age 9 years Archibald et al. (1983) (k) Natural mortality 0.05 per Archibald et al. (1983) (m) year Weight at entry 0.614 kg Archibald et al. (1983) (W k ) Weight at age 0.572 kg Archibald et al. (1983) k-l(W k _,) Body growth (p) 0.52 per Estimated as mean of year range: Recruitment Ricker Archibald et al. (1983) model Recruitment A = 0.29 From fit of model to parameters 1963-77 data A. B B = 1.6x10- history of this exploitation by using a catch-at-age model estimates annual exploitable biomass and average fishing mortality during this period (Archibald et al. 1983). The Deriso-Schnute model is fit to the catch history of this fishery during 1963-77 by using esti- mates of fishing mortality from the catch-at-age model (Fig. 2). All but two of the parameters required for the model are from published values (Table 2). Two param- eters (A and B in Table 2) in the stock-recruitment rela- tionship part of the model are estimated by fitting the model to the 1963-77 catch series. Based on the Deriso-Schnute delay-difference model with the parameters used to fit the catch and effort series (1963-77) and to estimate the equilibrium yield curve, the maximum sustainable yield (MSY) of about 1800 t is achieved at F MSY = 0.06 per year (Fig. 3). At that level of F MSY , the corresponding equilibrium bio- mass (Bmsy) is estimated at about 35,000 t. At B MSY , the recruitment of 1 -year-old juveniles is estimated at 3.5 million fish, whereas an estimated 1.3 million 1-year-olds recruit if biomass is at the 1977 depleted levels. Assuming that the goal of restoring the Pacific ocean perch stock is to increase the biomass to 35,000 t so it can be harvested at F = 0.06 to achieve MSY, three approaches to stock recovery are examined. One ap- proach has as its goal to increase the biomass to B MSY as quickly as possible by setting F at until the bio- mass reaches 35,000 t, then the stock will be fished at 132 Fishery Bulletin 89(1), 1991 (0 O Figure 3 Equilibrium yield curve (10 3 metric tons) for Pacific ocean perch from Queen Charlotte Sound estimated from the Deriso- Schnute model. F = 0.06. The second approach to stock recovery is to eventually achieve B MSY while also maximizing the yield to the fishery. Under this approach, the stock is fished at F = 0.06 from the beginning. A third strategy, which represents a compromise between these two philosophies, will set F at 0.03 until the biomass reaches 35,000 t, then the stock will be fished at F = 0.06. Each of these three sequences of F can be evaluated with and without hatchery releases. Catch and biomass series with and without hatchery releases for each sequence of F can be simulated with the stochastic Deriso-Schnute model with the parameters used to fit the population decline. When hatchery releases are used, 5 million 1 -year-old juveniles will be stocked an- nually. This number represents almost four times the natural recruitment for the stock at the 1977 depleted level and about 40% more than the natural recruitment at the B MSY level. The variance of the random variable z in the stochastic recruitment component is assumed to have a variance of 0.3 (Archibald et al. 1983). Based on 100 simulations of each of the six management ap- proaches over 100 years, the mean annual catches, mean cumulative catches and the biomass distributions after 20 years can be computed (Figs. 4-6, Table 3). When the management strategy closes the stock to fishing to restore it as quickly as possible, B MSY (35,000 1) is achieved in 21 years without stocking and 14 years if 5 million juveniles are stocked annually for 6 years (Table 3). After 20 years, the mean biomass levels are about the same for the stocked and non- stocked cases, so annual catches are about the same and the difference in cumulative yield remains con- 1 ' [ i ' ' ' ' i ' ' ! ' i ' ' ' ' i ' ' ' ' i ' ' ' ' i ' ' " i ' ' ' ' i ' ' 5 10 15 20 25 30 35 40 45 YEARS A F = 0, y-ars 1-21; F = 0.06, years " 22-50 ('0— a) F = 0, years 1-14; F = 0.06, years 15-50, wnlh slocking of 5 million per year for 6 years (H >■) □ F = 0.03. years 1-35; F = 0.06, D years 36-5'0 (fj— Q) F = 0.03. years 1-16; F = 0.06, years 17-50, »ilh slockini; of 5 mil- lion per year for 10 years ( + y-) C F = 0.06, years 1-50 (Q — D) F = 0.06. years 1-50, with stocking of 5 million per year for 12 years (+— +) Figure 4 Simulated annual catches of Pacific ocean perch, with and without hatchery releases, for three management strategies. Polovina: Hatchery releases of juvenile Sebastes alutus 133 100- A 80- 0) 60 " jr \ O 40- SyS O 20- h- Ul 0- 9— 1 ; Q CO O X o is o LU > _! O 100- 80 60 40 20 100 80- 60- 40 20 B 10 20 30 40 YEARS SINCE 1977 50 A F = 0, years 1-21; F = 0.06, years n 22-50 145mm CW) female crabs was assessed by sampling the catch of commercial fishermen onboard their vessels. The number and carapace width of crabs caught were recorded as were the date and location of the sample. We report two in- dices of female abundance: w/trap • soak day and vir- tual catch rate (Smith and Jamieson 1989a). Virtual catch rate (n/trap day) is the rate at which crabs would enter a trap if initial entry rates were not reduced by changes in bait effectiveness over time and agonistic interactions among crabs. Fishing effort survey The number of traps fished by all local fishermen, in- cluding the three fishermen who agreed to maintain records of all tagged crabs recovered in the study site, was determined by interviewing fishermen at least monthly from April 1985 until November 1986. The number of traps hauled each month in each of the seven zones demarcated in Figure 1 was estimated by incor- porating trap-count data with haul-frequency data. We also maintained a record of the number of our own research trap hauls. We validated our fishing effort survey by comparing the estimated number of traps in Lemmens Inlet each month, as obtained in fishermen interviews, with the number of trap buoys we counted in Lemmens Inlet each month. With few exceptions each trap buoy indicated one trap. Trap buoys were easily counted from a moving boat on calm water in this enclosed body of water. Figure 2a compares the estimated number of traps fished in Lemmens Inlet by each method and indicates that the fishermen's inter- views provided an acceptable census of trap abundance. Mark-recovery We tagged and released 4038 sublegal-sized male (x 142 mm CW, with 95% between 106 and 162 mm CW) and 1246 female (x 150mm CW, with 95% between 135 and 160 mm CW) Dungeness crabs from April 1985 to May 1986. Blue, individually numbered, 4.1-cm T-bar anchor tags (Floy Tag and Manufacturing Company, Seattle, WA) were inserted through the right posterior epimeral suture line, taking care not to puncture in- ternal organs. Crabs to be tagged were obtained from research traps, fishermen, or beam trawls. Before being released, the date, location and tag number were recorded, the crabs were sexed, and their carapace width measured. Release and recovery locations were < IOOO- 800- 600- 400- 200- • TRAP COUNT INTERVIEWS WITH FISHERMEN 1 — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I — I K 2000 1 -ST o LEMMENS INLET 600 A & INDIAN ISLAND 1200- 800- 400- rUDY SITE (excluding o,a □ ) BROWNING PASSAGE • ELSEWHERE A M J J 1986 DATE Figure 2 (A) Time-series comparison of the number of trap buoys counted in Lemmens Inlet and the number of traps estimated to be in Lemmens Inlet (zones 1-3) from interviews with fishermen. (B) Time-series of the number of traps estimated to be fishing in different regions within the study site from interviews with fishermen. 'Elsewhere' refers to traps fished outside the study site (zone 7) by fishermen who also fish within the study site. determined with a grid identification system (0.9 x 1.2km) and landmarks. Most recoveries were obtained from fishermen and research traps. Three fishermen recorded the date, location, and tag number of all tagged crabs they recovered on special recovery forms. Females and sublegal-sized males were returned to the water, but legal-sized males (i.e., > 154 mm CW) were generally retained. We asked other fishermen to ignore the recoveries of sublegal-sized crabs, i.e., return them to the water, but to retain tags from legal-sized males. Tags were either given to us or our associates, or we had permission to board a fisherman's vessel at the dock and retrieve tags set aside. From April until August of 1985 and 1986 we saw most fishermen at least biweekly, at other times monthly. 140 Fishery Bulletin 89|l). 1991 Table 1 Zone area (km 2 ), fishing effort (trap hauls/month), fishing in- tensity (trap hauls/month km 2 ), and number of tagged male and female Dungeness crabs released for geographical zones 1-7 (Fig. 1). Zone Area Fishing effort Fishing intensity No. of tagged crabs released Male Ferrfale 1 1.23 2234 1734 1438 160 2 0.98 8655 8823 1685 286 3 0.40 16165 40718 537 76 4 0.68 9606 14106 101 95 5 1.31 1053 801 84 11 6 7.56 25294 3345 193 618 7 10 6 15736 0.0157 Analysis of movement and mortality We simultaneously analyzed crab movement and mor- tality using a model structure that facilitated estima- tion of monthly proportional transfer rates of crabs among the seven geographical zones demarcated in Figure 1. Zone 7 essentially represents the universe outside the study site and provides the option for crabs to vacate the study site. Maximum-likelihood param- eter estimates for the transfer rates and mortality were obtained by minimizing the discrepancy between ob- served and expected number of recoveries in each zone during consecutive one-month intervals of time-at- large. Estimation of expected numbers of recoveries required our fishing-effort survey data. This model structure allowed us to search for persistent directional movement patterns and obtain male and female mor- tality estimates that were not confounded by move- ment of tagged crabs from the study site or into a zone where fishing effort was low. We organized 920 male and 103 female tag recov- eries by the three fishermen and ourselves who re- corded all tag-recovery information, and whose fishing effort over time was measured, into frequency cells (Oijk) by zone released (i=l,...,7), zone recovered (j = l,...,7), and months-at-large (k = 1,..., 18). Therefore, if the observed number of recoveries in cell 04,1.2 = 5, then 5 crabs released in zone 4 were recovered in zone 1 after 2 months (30-60 days)-at- large. We ignored absolute time, so our analysis was not sensitive to seasonal movement patterns. We also assumed an equal number of trap hauls each month. Although the total number of trap hauls peaks in sum- mer due to an increase in the number of part-time fishermen (Fig. 2b), our assumption is reasonable for the three full-time fishermen, and ourselves, whose tag- recovery and trap-haul data are the only trap-haul data incorporated into this analysis. Monthly fishing inten- sity (Ricker 1975) for each zone was calculated by dividing the number of trap hauls each month by the area (km 2 ) of the zone (Table 1). Zone 7 was deliber- ately defined to represent an extremely large area (1 million km 2 ) to ensure that we did not underestimate the rate of movement of tagged crabs from the study site. The number of tagged crabs initially released in zone i that are present in zone j after k months-at-large (Njj k ) is described by the following series of difference Equations (la-g). N n ,k + i = (Nii k (l-<2 12 ) + G 2 iN i2k ) S N i2lk+ i = (N i2k (l - 2*i - 2 23 ) + S2i 2 N !lk + S 32 N i3k )s N i3 .k + i = (Ni3 k (l-Q3 2 -Q3 5 -Q 36 ) + Q 23 N i2k + ^N^k + G 63 N i6k )s N i4 .k + i = (Ni4k(l-2«) + S54N.sk )s N i5 .k + i = (N 1Bk (l - Q 53 - Q54) + SisNuk + «35N i3k )s N i6 . k+ i = (N i6k (l-Q 63 -Q 67 ) + Q 36 Ni3 k )s N i7 ,k + i = (N,7k + SeyNiek ) s (la) (lb) (lc) (Id) (le) (If) dg) In these equations, Q l} represents the proportion of tagged crabs in zone i that move to zone j during any one-month interval. Note that when i=j, N ij0 is number of tagged crabs released in zone i (Table 1), and that Qjj's exist only for adjacent zones (Fig. 1). The parameter Q 76 could not be estimated because no tagged crabs were released in zone 7. The parameter s is monthly survival rate. The annual instantaneous rate of disappearance of tagged crabs (S) is related to s by Smith and Jamieson: Movement and mortality of Cancer magister 141 121og e [s]. (2) Consequent to Equations (1 a-g), the expected num- ber of recoveries of tagged crabs released in zone i and recovered in zone j after k months-at-large (E ijk ) is E ijk = N ijk qlj (3) where Ij is monthly fishing intensity (Table 1) in zone j, and q is the catchability coefficient (Ricker 1975). We employed a multinomial negative log-likelihood function, the separation statistic f\ = 1 O ijk log e [O ijk /E ijk ], for all O ljk >0 (4) ijk of Schnute and Fournier (1980), without their factor 2, to evaluate the parameter estimates. In total, the 11 Qij's in Equations (la-g), q and S required estima- tion. The model was structured to constrain the Qjj's to values between and 1, and to assure the multi- nomial likelihood condition of lo ijk ijk = 1 ijk E ijk- (5) Equation (4) measures the discrepancy between ob- served (O ijk ) and expected (Ejj k ) frequencies over all frequency cells, and yields maximum-likelihood param- eter estimates when f 1 is minimized. We used the SIMPLEX algorithm of Nelder and Mead (1965) as im- plemented by Mittertreiner and Schnute (1985) to minimize f l , while approximate standard errors for the estimates were calculated using the numerical method supplied with Mittertreiner and Schnute (1985). This method uses the matrix of second partial deriva- tives of fi with respect to the parameters (calculated numerically) to generate the asymptotic covariance matrix (Kendall and Stuart 1979). For males we applied the analysis only to recoveries obtained after at least one full month-at-large (k>l), because tagged males were released a short time before fisherman were prepared for the mark-recovery program. Consequent- ly our analysis of male movement and mortality is based on a sample size of 864 recoveries rather than the complete sample of 920 recoveries. Acoustic tagging Freshly activated acoustic tags (Smith-Root Inc., Van- couver, WA) were attached longitudinally to the cara- pace of recently trap-caught hard-shelled male and female Dungeness crabs using fast-drying epoxy. Care was taken to maintain each crab cool and moist while the carapace was allowed to air-dry for a short period prior to attachment of the tag. Once the epoxy set and we were confident the bond was secure (about one-half hour), each crab was placed in a bucket with fresh sea- water to assess its vitality. No crabs appeared to suf- fer an obvious detriment from the tagging procedure, so all crabs released were anticipated to survive and transmit location information. Once activated, each 61 x 14 mm capsule-shaped acoustic tag emits an acoustic signal characterized by a unique transmitting frequency and pulse rate. The bulk of the tag structure is a battery with a transmit- ting life of =60-90 days. With our direction-finding equipment we were able to identify a strong signal up to =lkm away and subsequently home-in on the loca- tion of a tag by audibly or electronically evaluating changes in signal strength. In our study site we were generally able to define the location of a tag to within =10-25 m of a chart reference. During 7-13 August 1986 two male and three female crabs were tagged with acoustic tags, then released near the mouth of Lemmens Inlet (zone 3, Fig. 1). Similarly on 15 November 1986 two male and three female crabs were again tagged and released in the same area. Following the first series of releases we attempted to monitor the location of each crab at least daily, but other commitments, and occasionally being unable to locate the transmitted signal, resulted in the time between observations often exceeding one day. The same limitations applied to the second series of releases, and additionally we only attempted to monitor each crab every three days. All crabs released pro- vided location information for 21-86 days, with the exception of a 140-mm CW male released on 15 November 1986 which was never located after its release. Mean daily displacement rates for males and females were estimated under the assumption that movement was random from the point of release. We could not entertain a more sophisticated hypothesis with our limited data. Also because of our limited data, we report our results for males and females, and by series (August-October, November-February), with all data from individual crabs combined. Mean daily displace- ment is the mean expected distance between the loca- tion of a crab at the same time on two consecutive days. The maximum-likelihood estimates for mean daily displacement rates (A) were obtained by minimizing the negative log-likelihood function ft = Z {(Di 2 /t,A) + logeftA*] - log e [2Di]}, i = l for all Dj>0 (6) 142 Fishery Bulletin 89(1), 1991 where n is the number of observations, and D ; is the linear displacement after t; days for crab i. When our best measurement for any Dj was zero, we assumed a minimum Dj of lm so that Equation (6) was defined for all data points i = l,...,n. Our Equation (6) is the maximum-likelihood solution to Equation (3) of Skellam (1951), or Equation (11.6) of Pielou (1977), both of whom develop a mathematical framework for describ- ing random dispersal of individuals from a release point. As for our analysis of movement and mortality, f 2 was minimized and approximate standard errors for A were determined using Mittertreiner and Schnute's (1985) SIMPLEX package. Results Beam-trawl and trap sampling Beam-trawl surveys in 1985 and 1986 captured few Dungeness crabs except in the selected locations con- sidered in the following paragraphs. In 46 beam trawl samples, 2- to 3-year-old (=75- 145 mm CW) male and female crabs (Butler 1961, Stevens and Armstrong 1984, Smith and Jamieson 1989c) were found at den- sities generally less than 10/hectare (ha). Where much higher densities were found, there were significant seasonal differences, and differences in the relative pro- portions of males and females. Sampling since 1986 in the same locations (G. Jamieson, unpubl. data) has in- dicated that the overall abundance of crabs in the 75-145 mm CW size range has dropped to a low level, thus we believe the pattern of spatial and temporal distribution we describe herein is based on observations of essentially one strong year-class. In upper Lemmens Inlet the densities of male and female 2-year-olds in summer 1985 were initially low and continued to decline throughout 1985 and 1986 (Fig. 3a). This suggests this area is a poor crab habitat since this cohort was abundant elsewhere. The highest density of males was observed in middle Lemmens In- let (Fig. 3b). During autumn 1985 and the subsequent winter, male densities steadily increased to greater than 1200/ ha. Toward the mouth of Lemmens Inlet male densities generally declined (Fig. 3c, d). Because densities of 2-year-old males elsewhere were con- sistently low, the increase in the number of males in middle Lemmens Inlet is suspected to be due to move- ment away from exposed shallow water during winter. We suspect that these males concentrated in middle Lemmens Inlet because of poor habitat further up the inlet. This high density of males eventually decreased rapidly during late spring 1986. Most males were in the normally distributed instar with a mean carapace width of 129 mm and a standard deviation of 12 mm (Smith and Jamieson 1989c), and many molted to legal UPPER LEMMENS INLET 300' in £- ui Q 900 MIDDLE LEMMENS INLET MALE • FEMALE ( 75-l45mm CW ) 1 1 1 1 1 — I 1 1 1 1 1 1 1 — r I I LOWER LEMMENS INLET CO 600 300 600 300- ~1 I I I I T I I ENTRANCE TO LEMMENS INLET t i i — i — i — i — i — i — i — i — i — i — rn — I -\ MAIN CHANNEL 1 1 "T I I 1 I I I I 1 I I 1 1 1 1 1 M J J A SON D|JFMAMJ J A S 1986 DATE 1985 Figure 3 Time-series of densities for 2-3 year-old (=75-145 mm CW) male and female Dungeness crabs in upper Lemmens Inlet (zone 1), middle Lemmens Inlet (zone 2), lower Lemmens In- let (zone 2), the entrance to Lemmens Inlet (zone 3), and in the main channel out to sea (zone 6). size in 1986 (i.e., to the normally distributed instar with a mean carapace width of 156 mm and a standard devia- tion of 13 mm). Following their molt to legal size they were quickly caught in an intense fishery (Smith and Jamieson 1989b). About 25-35% of the fishing effort in the study site was concentrated in Lemmens Inlet during spring 1986 (Fig. 2b). A high density of 2-year-old females was observed in the narrow channel at the lower end of Lemmens Inlet in June 1985 (Fig. 3c). Density declined after this date, but increased during autumn in the entrance to Lemmens Inlet (Fig. 3d) 0.5-1. Okm seaward of the nar- row channel. Female density further up Lemmens Smith and Jamieson: Movement and mortality of Cancer magister 143 100 -i VIRTUAL ENTRY RATE CATCH RATE / (n.10 trap"'-d"' ) ~~\ — I — i — I — I — r MAMJJASO 1985 DATE Figure 4 Two indices of abundance for female Dungeness crabs > 145 mm CW in zone 6, where females were most abundant (Table 2). The virtual catch rate is the rate at which females would enter a trap if entry rates were not modified by changes in bait effectiveness over time and agonistic interactions among crabs (see Smith and Jamieson 1989a). Inlet continued to decline, suggesting seaward move- ment. The increase in female abundance in September 1986 where the main channel out of the study site meets the open coast (Fig. 3e) also suggests seaward movement of females. These females were inferred to be 3-year-olds because they were mainly in the normally distributed instar with a mean carapace width of 137 mm and a standard deviation of 9 mm (Smith and Jamieson 1989c). The relative abundance of larger females in the study site was assessed by trap sampling. Two indices of abundance obtained from commercial trap samples (Fig. 4), show that females were most readily caught in spring, perhaps because they forage more actively after a winter of incubating eggs. The highest observed abundance of the larger females (>145mm CW) in spring 1985 and 1986 (Table 2) was in the main chan- nel out of the study site. This is also where (presumably) 3-year-old females in the 137 mm instar were collected in abundance by beam trawl in September 1986. This is consistent with the suggestion from beam-trawl sampling of seaward movement from local inlets as females mature. Analysis of movement and mortality Our estimate of the catchability coefficient q for fe- males of 0.198 x 10~ 5 recoveries per trap haul was about one-seventh that for males (0.138 x 10~ 4 recov- Table 2 Abundance (n / 1 00 traps • soak day) of female Dungeness crabs > 145 mm CW in commercial traps during May and June of 1985 and 1986. Area samples Female abundance 1985 1986 Upper Lemmens Inlet (zone 1) Lemmens Inlet (zones 2 and 3) Browning Passage (zone 5) Near Indian Island (zone 4) Main channel (zone 6) 7 13 15 8 22 4 10 36 429 138 Table 3 Maximum-likelihood estimates and standard errors (SE) of the proportion (Q IS ) of tagged male and female Dungeness crabs in zone l transferring to an adjacent zone j during aone- month interval. Asterisks indicate transfer proportions sig- nificantly (P<0.05) greater than zero Parameter Male Female Proportion SE Proportion SE a u 0.15* 0.02 0.84* 0.25 Q 21 0.50* 0.09 0.68* 0.13 «23 0.22* 0.02 0.10* 0.03 »32 0.16* 0.03 0.07 0.07 «35 0.44* 0.06 0.22 0.12 « 5 3 0.11* 0.02 0.36 0.27 °36 0.40* 0.04 0.63* 0.12 2 63 <0.01 <0.01 <0.01 <0.01 «4 5 0.12* 0.03 0.01 0.01 2 6 4 <0.01* <0.01 0.09 0.08 Q fi7 0.04* 0.01 0.02 0.01 eries per trap haul), indicating males were much more readily caught than females by commercial traps. This difference is mainly due to commercial traps being more efficient at retaining larger crabs which are predominantly males. Consequently it is apparent in Table 3 that our transfer proportion estimates for females are much less confidently made than those for males because of the much smaller number of tagged females recovered. The proportional transfer rates for males do not sug- gest any persistent directional movement. We note that there is little transfer of crabs between Indian Island (zone 4) and Browning Passage (zone 5) when compared with the apparent mixing among zones 1, 2, 3, 5, and 6, so the movement of tagged crabs through the deeper water between Browning Passage and Indian Island appears limited. Male crabs appear to vacate middle and lower Lemmens Inlet (zones 2 and 3) at propor- tional rates of about 0.22 and 0.40 per month with 144 Fishery Bulletin 89(1). 1991 only a small proportion per month (0.01) entering Lemmens Inlet from zone 6. The monthly proportional transfer rate of crabs from zone 2 to zone 1 (£21) of 0.50 seems high compared with transfer of crabs from zone 1 to zone 2 (Q 2 i) of 0.15, but this might be ex- plained by zone 1 (upper Lemmens Inlet) being poor crab habitat and the possibility that only a small pro- portion of male crabs are successful in escaping this habitat. The proportional transfer rates for females have large standard errors and must be interpreted cautious- ly. However, a notable result is the 0.63 proportional transfer rate for £ 36 compared with <0.01 for Q es , thus indicating that a greater proportion of tagged females vacated lower Lemmens Inlet than entered Lemmens Inlet. The monthly proportional transfer rate of 0.84 for females escaping the poor habitat in zone 1 (Q12) is high compared with males, but might indi- cate females are more capable of recognizing the en- vironmental clues leading to more suitable substrate. The transfer rates of females in both directions be- tween zone 1 and zone 2 are notably higher than the male transfer rates, and since the largest release of tagged male and female crabs was in zones 1 and 2, this result might indicate females are generally more mobile than males. Proportional transfer rates alone are an incomplete interpretation of movement trends, since the zones differ in area and therefore also in crab abundance. Consequently, the high transfer rate for Q 36 when compared with Q 63 is in large part a result of zone 6 having an area of 7.56km 2 , whereas zone 3 has an area of 0.40km 2 . Relative movement vectors were therefore calculated from the transfer-rate estimates under the assumption that crab densities were equal for all zones. For example, if crab density is A crabs per km 2 , then the estimated number of crabs transfer- ring from zone 6 to zone 3 in a one-month period is 7.56AQ 63 . Figure 1 diagrams net movement tendencies, under the assumption of equal crab densities in all zones, across those zone boundaries where transfer rates in both directions were estimated. Two notable features for males are the strong indication of net movement of males into upper Lemmens Inlet, and a tendency for males to vacate lower Lemmens Inlet and the waters near Indian Island. As previously mentioned, the net movement into upper Lemmens Inlet might be a result of a low proportion of males escaping this poor habitat. It might also indicate dispersion from zone 2 where many males were tagged and released in spring 1986. A large number of males 125-140 mm CW were tagged in zone 2 in spring 1986 because they occurred in high density (> 1200/ ha) and were readily captured by beam trawl. Beam trawling indicated these crabs apparent- Table 4 Maximum-likelihood estimates and standard errors (SE) of mean daily disp acements (A, m/d) of male and female Dungeness crabs Male Female n Estimate SE n Estimate SE Aug 86-Oct 86 31 321 29 30 500 45 Nov 86-Feb 87 8 74 13 15 161 21 Combined 39 288 23 45 419 31 ly dispersed, or molted to legal size and were caught, by summer 1986 (see previous section). For females there is a general, although non-significant, tendency to vacate Lemmens Inlet and Browning Passage. This result is consistent with our beam-trawl and trapping results (previous section) which also demonstrate a net seaward movement of females from Lemmens Inlet. We estimated the annual rate of disappearance of tagged crabs with reasonable confidence at S = 2.54 (SE 0.13) and S = 1.28 (SE 0.27) for males and females, respectively. Corroborating these high estimates for S, our analysis estimated that only about 2% and 7% of tagged males and females, respectively, left zones 1-6, i.e., moved to zone 7, over the time-period of our study. In other words, the distribution over time of tag recoveries in zones 1-6 could not be well explained by a high transfer rate into zone 7 where crabs might not be recovered due to a low fishing intensity. Thus these estimates for S were obtained despite our model struc- ture providing an exaggerated opportunity for crabs to disperse to zone 7 as an alternative explanation for the disappearance of tagged crabs, i.e., as an alter- native to mortality. This conclusion appears to validate our original assumption that few tagged crabs left our study site during the study period (Smith and Jamieson 1989b). Based on a double-tagging study, Smith and Jamie- son (1989b) concluded that tag loss by sublegal-sized crabs is low and therefore unlikely to be an important source of tag disappearance. Observations of tank-held crabs over several months did not reveal differential mortality of tag and untagged crabs. Thus we conclude that the values for S estimated in this study represent mainly natural mortality and, in the case of males, tag disappearance due to molting to legal size with subse- quent exploitation. However, because only about 5% of tagged sublegal-sized males were reported caught as legal-sized crabs (210 recovered when legal-sized of the 4038 released when sublegal-sized), despite high fishing mortality (F = 5.1-6.9) and apparently good compliance in reporting recovered tags (=87%), we con- Smith and Jamieson: Movement and mortality of Cancer magister 145 elude that our male estimate for S measures mainly natural mortality of sublegal-sized males. We recog- nize, however, that there is likely some degree of mor- tality due to trapping and handling by fishermen. Similarly, and in consideration that females are not commercially fished, we are confident that our female estimate for S also mainly measures natural mortality. Acoustic tagging From the means and standard errors of the movement rate estimates (Table 4) we conclude that for both the August and November 1986 releases of acoustically tagged crabs, females tended to move about significant- ly (P<0.01) more than males. Similarly, during Aug- ust-October 1986 both males and females seemed to move at rates about 3 to 4 times faster (P<0.01) than those for November 1986-February 1987. We realize, however, that our male results for the latter time period are based on observations of just one crab. Slower movement during winter for both male and female crab is consistent with the poikilothermic habit of marine invertebrates in that respiratory activity is closely related to temperature. Our analysis of move- ment based on the mark-recovery data also hints that females are more mobile than males. We cannot conclude that our estimate of a faster movement rate for females, relative to males, repre- sents a general phenomenon for all Dungeness crab populations. Our acoustic tagging results are possibly peculiar to our study site, since the greater movement rate for females is consistent with our interpretation of the trapping and beam trawl results which suggest that females tend to vacate the coastal inlets in search of the exposed coast. Further, we caution that our dispersal rate estimates for males and females from the acoustic tagging data should be considered minimum estimates because the archipelago in which the acous- tically tagged crabs were released will tend to restrict the potential for movement. Also, our assumption of random dispersal from the point of release of an acoustically tagged crab is probably too strict. Our beam trawl and trapping data indicate that both males and females might move in response to environmental clues such as tides and currents; and our beam-trawl, trapping, and mark-recovery data all suggest seaward movement of females from Lemmens Inlet. These limitations to our interpretation of the acoustic tagging data preclude us from obtaining precise rates of population dispersal over time. However, we think that there is some value in presenting approximate dispersal rates since they can be compared with the results of our analysis of movement and mortality, and to some degree with Dungeness crab movement rates that might be obtained from other regions. Conse- quently, the mean daily displacement rates (A) of 288 m/day and 419 m/day obtained for males and females, respectively, suggest that after one year of random dispersal, in the absence of geographical boundaries, 95% of males and females would be within radii of 9.5 and 13.9 km, respectively, of the point where they were one year previous. These radii were determined from r 2 = -log e [p rt ]A 2 t (7) where r is the radius within which the proportion 1 - p rt of a randomly dispersing population is expected after t days (Pielou 1977). These estimates of r agree favorably with the results of our simultaneous analysis of movement and mortality from our mark-recovery data in that about 2% and 7% of male and female crabs, respectively, were estimated to leave the study site dur- ing the 18-month study period. The study site encom- passes an area of about 5-10km radius around Tofino (Fig. 1). Discussion The four methodologies we employed to assess male and female Dungeness crab movement provided in- sights at different levels of resolution and from differ- ent perspectives. The methodologies were complemen- tary and together allowed us to document a coherent description of Dungeness crab movement near Tofino, British Columbia. For example, acoustic tagging gave us a general indication of the dispersal rates for in- dividual male and female crabs (summer and winter) which was consistent with the results of our simul- taneous analysis of movement and mortality using mark-recovery data. The former analysis suggested that the dispersal rates would maintain 95% of the male and female populations within about 10 km and 14 km, respectively, of points of release, while the latter analysis suggested that only about 2% of males and 7% of females would escape the study site (about 10-15km radius) during our 18-month study period. In addition, both the beam-trawl and trap samples documented seasonal changes in the relative distributions and abun- dances of males and females which could not be gleaned from the acoustic tagging and mark-recovery data alone. Overall, our four sources of movement information suggest that male Dungeness crab undergo only limited movements within the local archipelago. There was no evidence of migratory movement, but males were in- ferred to move to shallower water (=10m) during sum- mer, then to retreat to more sheltered habitat in autumn. Others have observed, or inferred, similar be- havior for this species. Stevens and Armstrong (1984) 146 Fishery Bulletin 89(1), 1991 reported that juvenile males and females of the 1980 year-class in Grays Harbor, Washington, disappeared during the winter of their first year, then reappeared the following spring. Gotshall (1978) observed move- ment of sublegal- and legal-sized male crabs in northern California to deeper water in winter, and a return to shallower water in spring. It is reasonable to surmise that the autumn movement to deeper water is to avoid rough shallow water during the winter. Returning to shallower, warmer, and more productive water during summer could enhance growth and survivorship. Our estimated average daily displacement rate for male Dungeness crabs of =300m/day, is consistent with our inferences on movement from our mark-recovery, beam-trawl, and trap sampling results. Our general conclusion of limited movement is also consistent with the results of other tagging studies. Both Butler (1957) for Dixon Entrance, British Columbia, and Gotshall (1978) for northern California, suggested that Dunge- ness crab populations remain local. With the exception of the apparent seasonal shift in habitat, no studies suggest migratory movements for males; however, Gotshall (1978) noted that males seem to move in the direction of prevailing currents off northern Califor- nia. Bennett and Brown (1983) report that most tagged males of the closely related crab Cancer pagurus re- mained near where they were released in the English Channel. Our acoustic tagging and mark-recovery data sug- gest that female Dungeness crab undergo only limited movement. Diamond and Hankin (1985) similarly ar- gued that mature female Dungeness crab off the coast of northern California undergo limited movements and suggested that females constitute localized stocks. Dia- mond and Hankin (1985) do suspect that females move short distances to shallower water in spring to mate and molt. Our analyses provided no evidence of this, but it is quite conceivable that both males and females could improve mating opportunities by concentrating in shallow water (Butler 1960). We inferred, mainly from our beam trawl and trap samples, that females tended to move from coastal in- lets to an area more exposed to the open coast (zone 6, Fig. 1). In the inlets the substrate ranged from mud to a mud/sand mix, whereas in the more exposed area the bottom was mainly sand or a sand/gravel mix. Wild (1980) states that females must be at least partially buried in sandy substrate to extrude and incubate eggs so our inference is consistent with the current under- standing of the life history of Dungeness crab females. Stevens and Armstrong (1984) noted that egg-bearing females were rare in Grays Harbor, and speculated that most mature females left the harbor to incubate and release their eggs in a preferred environment. Our study yielded an average daily displacement rate for females (=400m/day) which was significantly (P<0.01) more than the rate for males (=300 m/day), and which might be explained by females undergoing deliberate migratory movements to locate suitable substrate for incubating eggs. Similar movement behavior has been reported for females of other crab species. Hyland et al. (1984) ob- served the movement of female portunid crab Scylla serrata from an estuarine environment, where they lived as juveniles, to the open ocean where they re- leased their eggs. Some females returned to inshore waters after the hatching season. Bennett and Brown (1983) demonstrated that female C. pagurus undergo extensive movements, apparently to locate habitat more suitable for egg incubation and release. While SCUBA diving, Howard (1982) observed egg-bearing female C. pagurus congregated in relatively deep (24 m), quiet water. Since they were rare elsewhere, he concluded this was a preferred habitat. Dinnel et al. (1987) observed a similar behavior for Dungeness crab in Puget Sound, Washington, from the Govern- ment of Canada submersible Pisces IV. Our simultaneous analysis of Dungeness crab move- ment and mortality using mark-recovery and fishing- effort data diminished the confounding of these two processes and yielded a revised estimate of the natural mortality rate originally proposed by Smith and Jamie- son (1989b). Our (mainly) natural mortality rate esti- mate for males of 2.5 (95% CI of 2.3-2.8) is moderate- ly lower than our previous estimate of 2.9-4.5, probably because of dispersal of tagged crabs into zones (espe- cially zone 6) with low fishing intensities. Our estimate of female natural mortality of 1.3 (95% CI of 0.8-1.8) is significantly lower than that of males but in general agreement with the mortality estimates of Hankin et al. (1985, 1989) for females in northern California. They estimated annual instantaneous nat- ural mortality for females greater than 140 mm CW at 2.0 and 2.5 for two different periods of release in a mark-recovery experiment. For females 125-140mm CW their rough estimate was =0.7, a more precise estimate being unobtainable due to females this size having a high probability of molting and changing vulnerability to traps. Our estimate of 1.3 is based on a group of tagged females whose carapace widths at release were 135-171 mm (x 150mm), thus our esti- mate seems consistent with those of Hankin et al. (1985, 1989). The estimates of male and female natural mortality from this study, and of female natural mortality from Hankin et al. (1985, 1989), for crabs near the Canadian minimum legal size limit of 154 mm CW (165 mm spine- to-spine CW) increases our confidence that mortality of mature Dungeness crab is indeed high. For exam- ple, a mortality rate of 2.0 means only 13.5% annual Smith and Jamieson: Movement and mortality of Cancer magister 147 survivorship. The recognition of such a high mortality rate for adults of this commercially important species immediately prompts questions regarding the appro- priateness of both the Canadian and American (=159 mm CW) minimum legal carapace width limits for optimizing two of the most basic stock-management guidelines: yield-per-recruit and eggs-per-recruit. Acknowledgments We are grateful to Messrs. A. Phillips, D. Heritage, and W. Harling who provided most of the field support, and to Dr. L.J. Richards and Mr. J. Fulton who re- viewed a previous version of this paper. We extend special thanks to the fishermen of Tofino who par- ticipated in the mark-recovery program and expressed a keen interest in our activities. Citations Armstrong, D.A., and D.R. Gunderson 1985 The role of estuaries in Dungeness crab early life history: A case study in Grays Harbor, Washington. In Melteff, B.R. (ed.), Proceedings of the symposium on Dungeness crab biology and management, p. 145-169. Alaska Sea Grant Rep. 85-3, Univ. Alaska, Fairbanks. Bennett, D.B., and C.G. Brown 1983 Crab (Cancer pagurus) migrations in the English Chan- nel. J. Mar. Biol. Assoc. U.K. 63:371-398. Botsford, L.W. 1986 Population dynamics of the Dungeness crab (Cancer magister). In Jamieson, G.S., and N. Bourne (eds.), North Pacific workshop on stock assessment and management of in- vertebrates, p. 140-153. Can. Spec. Publ. Fish. Aquat. Sci. 92. Botsford, L.W., and D.E. Wickham 1978 Behavior of age-specific, density-dependent models and the northern California Dungeness crab (Cancer magister) fishery. J. Fish. Res. Board Can. 35:833-843. Botsford, L.W., R.D. Methot Jr.. and W.J. Johnston 1983 Effort dynamics of the northern California Dungeness crab (Cancer magister) fishery. Can. J. Fish. Aquat. Sci. 40: 337-346. Butler, T.H. 1957 The tagging of the commercial crab in the Queen Char- lotte Islands region. Fish. Res. Board Can. Pac. Prog. Rep. 109:16-19. 1960 Maturity and breeding of the Pacific edible crab, Cancer magister Dana. J. Fish. Res. Board Can. 17:641-646. 1961 Growth and age determination of the Pacific edible crab Cancer magister Dana. J. Fish. Res. Board Can. 18:873-889. Diamond, N., and D.G. Hankin 1985 Movements of adult female Dungeness crabs (Cancer magister) in northern California based on tag recoveries. Can. J. Fish. Aquat. Sci. 42:919-926. Dinnel, P.A., D.A. Armstrong, R.R. Lauth, and G.S. Jamieson 1987 Use of the PISCES IV submersible for determining the distributions of Dungeness crab, shrimp, and bottomfish in Port Gardner, Washington. Fish. Res. Inst. Final Rep. FRI- UW-8709, Univ. Wash., Seattle, 16 p. Gotshall, D.W. 1978 Northern California Dungeness crab, Cancer magister, movements as shown by tagging. Calif. Fish Game 64: 234-254. Gunderson, D.R., and I.E. Ellis 1986 Development of a plumb staff beam trawl for sampling demersal fauna. Fish. Res. (Amst.) 4:35-41. Gunderson, D.R., D.A. Armstrong, and C. Rogers 1985 Sampling design and methodology for juvenile Dungeness crab surveys. In Melteff, B.R. (ed.). Proceedings of the sym- posium on Dungeness crab biology and management, p. 135-144. Alaska Sea Grant Rep. 85-3, Univ. Alask, Fairbanks. Hankin, D.G., N. Diamond, M. Mohr, and J. Ianelli 1985 Molt increments, annual molting probabilities, fecundity and survival rates of adult female Dungeness crabs in northern California. In Melteff, B.R. (ed.), Proceedings of the sympo- sium on Dungeness crab biology and management, p. 189-206. Alaska Sea Grant Rep. 85-3, Univ. Alaska, Fairbanks. Hankin, D.G., N. Diamond, M.S. Mohr, and J. Ianelli 1989 Growth and reproductive dynamics of adult female Dungeness crabs (Cancer magister) in northern California. J. Cons. Cons. Int. Explor. Mer 46:94-108. Howard, A.E. 1982 The distribution and behavior of ovigerous edible crabs (Cancer pagurus), and consequent sampling bias. J. Cons. Cons. Int. Explor. Mer 40:259-261. Hyland. S.J., B.J. Hill, and C.P. Lee 1984 Movement within and between different habitats by the portunid crab Scylla serrata. Mar. Biol. (Berl.) 80:57-61. Jamieson, G.S., and A.C. Phillips 1988 Occurrence of Cancer crab (C magister and C. oregonen- sis) megalopae off the west coast of Vancouver Island, British Columbia. Fish. Bull., U.S. 86:525-542. Kendall, M., and A. Stuart 1979 The advanced theory of statistics, Vol. 2. Inference and relationship, 4th ed. MacMillan, NY, 748 p. McKelvey, R., D. Hankin, R. Yanosko, and C. Snygg 1980 Stable cycles in multistage recruitment models: An ap- plication to the northern California Dungeness crab (Cancer magister) fishery. Can. J. Fish. Aquat. Sci. 37:2323-2345. Methot, R.D. Jr. 1986 Management of Dungeness crab fisheries. In Jamieson, G.S., and N. Bourne (eds.), North Pacific workshop on stock assessment and management of invertebrates, p. 326-334. Can. Spec. Publ. Fish. Aquat. Sci. 92. Methot, R.D. Jr., and L.W. Botsford 1982 Estimated preseason abundance in the California Dunge- ness crab (Cancer magister) fisheries. Can. J. Fish. Aquat. Sci. 39:1077-1083. Mittertreiner, A., and J. Schnute 1985 Simplex: A manual and software package for easy non- linear parameter estimation and interpretation in fishery research. Can. Tech. Rep. Fish. Aquat. Sci. 1384, 90 p. Nelder, J. A., and R. Mead 1965 A simplex method for function minimization. Computer J. 7:308-313. Pielou, E.C. 1977 Mathematical ecology. John Wiley, NY, 384 p. Ricker, W.E. 1975 Computation and interpretation of biological statistics offish populations. Bull. Fish. Res. Board Can. 191, 382 p. Schnute, J., and D. Fournier 1980 A new approach to length frequency analysis: Growth structure. Can. J. Fish. Aquat. Sci. 37:1337-1351. 148 Fishery Bulletin 89(1), 1991 Skellam, J.G. 1951 Random dispersal in theoretical populations. Biometrika 38:196-218. Smith, B.D., and G.S. Jamieson 1989a A model for standardizing Dungeness crab (Cancer magister) catch rates among traps which experienced different soak times. Can. J. Fish. Aquat. Sci. 46:1600-1608. 1989b Exploitation and mortality of male Dungeness crabs (Cancer magister) near Tofino, British Columbia. Can. J. Fish. Aquat. Sci. 46:1609-1614. / 1989c Growth of male and female Dungeness crabs near Tofino. British Columbia. Trans. Am. Fish. Soc. 118:556-563. Stevens, B.G., and D.A. Armstrong 1984 Distribution, abundance, and growth of juvenile Dunge- ness crabs, Cancer magister, in Grays Harbor estuary, Wash- ington. Fish. Bull., U.S. 82:469-483. Wild, P.W. 1980 Effects of seawater temperature on spawning, egg devel- opment, hatching success, and population fluctuations of the Dungeness crab, Cancer magister. Calif. Coop. Oceanic Fish. Invest. Rep. 21:115-120. Abstract.- A logbook program was initiated to determine the rela- tive abundance of selected fish spe- cies around oil and gas platforms off the Louisiana coast. Logbooks were maintained by 55 anglers and 10 charterboat operators from March 1987 to March 1988. A total of 36,839 fish were caught representing over 46 different species. Principal component analysis (PCA) grouped the seventeen most abun- dant species into reef fish, pelagic fish, bluefish-red drum, Atlantic croaker- silver/sand seatrout, and cobia-shark- blue runner associations. Multiple regression analyses were used to compare PCA groupings to physical platform, temporal, geological, and angler characteristic variables and their interactions. Reef fish, Atlan- tic croaker, and silver/sand seatrout abundances were highest near large, structurally complex platforms in relatively deep water. High spotted seatrout abundances were correlated with small, unmanned oil and gas platforms in shallow water. Pelagic fish, bluefish, red drum, cobia, and shark abundances were not related to the physical parameters of the platforms. Factors Affecting the Abundance of Selected Fishes near Oil and Gas Platforms in the Northern Gulf of Mexico* David R. Stanley Charles A. Wilson Coastal Fisheries Institute. Center for Wetland Resources Louisiana State University, Baton Rouge, Louisiana 70803 Louisiana has long been recognized as having abundant fisheries resources as evidenced by the large number of recreational and commercial fishing opportunities. This is particularly true of its offshore waters which Moore et al. (1970) characterized as having high densities of demersal fishes and Gunter (1963) postulated were the most productive waters on earth based on fishery harvests. Saltwater sportfishing off Loui- siana is concentrated around oil and gas platforms with an estimated 37% of all saltwater angling trips (Witzig 1986) and over 70% of all recrea- tional angling trips in the Exclusive Economic Zone (more than 3 miles from shore) occurring around plat- forms (Reggio 1987). All of the 3700 oil and gas platforms off the coast of Louisiana are thought to act as ar- tificial reefs and have contributed to Louisiana's designation as a fishing "paradise." Gallaway (1984) estimated that oil and gas platforms constitute 28% of the known hard substrate off Loui- siana and Texas. This is of particular importance off the Louisiana coast since the nearest natural hard bottom habitat is approximately 92 km from shore (Sonnier et al. 1976); therefore, oil and gas platforms provide the only source of hard-bottom habitat close to shore. The continued growth of the Manuscript accepted 28 September 1990. Fishery Bulletin, U.S. 89:149-159 (1991). * Contribution No. LSU-CFI-89-04, Louisiana State University, Coastal Fisheries Institute. offshore oil and gas industry in Loui- siana has provided habitat expansion for organisms dependent on hard substrate (Sonnier et al. 1976, Galla- way et al. 1981a, Continental Shelf Associates 1982). Oil and gas platforms are unique as artificial reefs because they extend throughout the entire water column. Their effects are not confined to ben- thic and demersal fishes; pelagic fishes also benefit (Gallaway et al. 1981a, Continental Shelf Associates 1982). For example, pelagic baitfish (i.e., round scad Decaptarus puncta- tus, Spanish sardine Sardinella an- chovia, and scaled sardine Harengula pensacolae) often maintain a position from nearsurface to mid-depth with- in or upcurrent from oil and gas structures feeding on plankton and zooplankton, while large predatory pelagic fishes (i.e., king mackerel and blue runner) are reported to swim from the surface to mid-depth around structures, rarely venturing within the structure (Hastings et al. 1976, Gallaway et al. 1981a). Although oil and gas structures, like most artificial reefs, are con- sidered to increase production and attract fish, there are few accepted techniques to assess their effective- ness. A method of testing the success or importance of an artificial reef is to track the number of fish caught over time. Due to the complex con- struction of oil and gas platforms, sampling with traditional fisheries 149 150 Fishery Bulletin 89(1). 1991 gear (e.g., trawls, gillnets) is difficult at best; there- fore, the number of fish caught per angler/hour (CPUE) was used to estimate the relative abundance of fish near oil and gas platforms. Artificial reefs have been reported to concentrate scattered fishes and/or elevate secondary production by increasing the growth and survival of new individ- uals. However, few studies have examined the tropho- dynamics of these systems (Bohnsack and Sutherland 1985). The attraction/production paradigm should not be viewed as a black and white issue, but as a gradient depending upon species, life-history stage, type of ar- tificial reef, etc. (Bohnsack 1989). Many species offish around oil and gas platforms are trophically indepen- dent of the structure (e.g., pelagic fishes), but may use the platform for other purposes (e.g., optical stimulus, shelter, protection from predation, seasonal move- ments, spawning and orientation) (Gooding and Mag- nuson 1967, Hunter and Mitchell 1967, Klima and Wickham 1971, Wickham et al. 1973, Gallaway et al. 1981a, Continental Shelf Associates 1982). Factors that may explain the congregation of fish around artificial reefs are poorly known (Grove and Sonu 1983). Some theories on factors influencing the abundance and attraction of fish to artificial reefs in- clude shape and complexity (Hunter and Mitchell 1968, Luckhurst and Luckhurst 1978, Grove and Sonu 1983, Chandler et al. 1985), size of the artificial reef (Hunter and Mitchell 1968, Huntsman 1981, Grove and Sonu 1983, Turner et al. 1969, Rousenfell 1972, Ogawa 1982, Vik 1982), age of the artificial reef and seasonality (Turner et al. 1969, Molles 1978, Stone et al. 1979, Smith 1979, Lukens 1981, Stephens and Zerba 1981). Colonization of natural and artificial reefs did not follow the MacArthur and Wilson (1967) model of species equilibrium for insular biotas according to Smith (1979) and Lukens (1981). They found the strong seasonal ef- fects in the northern Gulf of Mexico produced seasonal- ly stable communities with regular fluctuations in diversity and abundance. The objective of this study was to determine if a rela- tionship exists between the relative abundance of selected fish species near oil and gas platforms off the Louisiana coast and (1) physical platform variables (e.g., water depth, submerged surface area, volume of water enclosed by the platform, mode of platform operation, platform age), (2) temporal variables (e.g., linear, quadratic, and cubic functions of date), (3) meteorological and geological variables (e.g., air tem- perature, wind speed and direction, mean sediment size), and (4) angler characteristic variables (e.g., fish- ing method, boat length, total engine horsepower, presence of electronic fishing aids). Materials and methods Between September 1986 and March 1987 we solicited 120 recreational anglers from fishing clubs across Loui- siana to maintain logbooks. In addition, 23 of the charterboat operators listed in National Marine Fish- eries Service records and Coleman (1984) volunteered to maintain logbooks. Logbook data were collected from March 1987 to March 1988. The design of the logbook and data collected were based on the Lake Erie Angler Diary Program (Sztramko 1986) and logbook criteria outlined by Demory and Golden (1983). Infor- mation obtained from the logbooks included: date of trip, number of anglers, oil and gas platform fished, fishing time (not including travel time), fishing method, bait used, and the species and number of fish caught. Due to the difficulty in identification of some fish species, snapper other than red snapper, groupers, sharks, and silver and sand seatrout were classified as other snapper, groupers, sharks, and silver/sand sea- trout respectively. Other data acquired from logbook participants included boat length (m), total engine horsepower, and the presence of electronic fishing aids (e.g., LORAN, graph recorders, and echosounders) which assisted in the capture of fish. We also measured characteristics of the platform, surrounding sediments, and weather which we con- sidered important. Submerged surface area (m 2 ), vol- ume of water enclosed by the structure (m 3 ), and the number of legs, wells, and structural crossmembers for each platform utilized by the logbook participants were calculated from drawings and information provided by offshore oil operators. Water depth (m) and age of the structures were supplied by the Minerals Management Service. Surface sediment sizes (^m) adjacent to oil and gas structures were taken from Coleman et al. (1986). Meteorological data, including average daily wind speed (km/hour), direction, and temperature (°C), for the New Orleans International Airport were obtained from the Louisiana State Climatology Office. To account for seasonal differences in abundance, linear, quadratic, and cubic orthogonal polynomials of the 12 months of the study were used. Orthogonal polynomial contrasts are by definition uncorrelated, thus enabling the unique contribution of the linear, quadratic, or cubic effects of time to be identified. CPUE was calculated as the number of fish caught per angler per hour of fishing. Prior to any analysis that assumes data normality, the distribution of CPUE data was tested and found not to be normal. Therefore, in order to approximate the normal distribution, the CPUE data were transformed by ln(CPUE + 1) due to the large number of zero values in the original data (Pennington 1983, 1985). Stanley and Wilson: Fish abundance near oil and gas platforms Two multivariate analysis techniques were utilized to determine the relationships between species abun- dance and geological, physical, temporal, and meteor- ological variables. Principal Component Analysis (PCA) was used on the individual fish species as a data- reduction technique. The PCA transforms the original set of variables into a smaller set of orthogonal linear combinations of species that account for a major por- tion of the variance in the original set (Chatfield and Collins 1980, Dillon and Goldstein 1984). The CPUE data from 17 species or species groups were reduced to 5 principal components (PC's) using the FACTOR procedure (SAS 1985). Only PC loadings greater than 0.35 were considered; although the value of 0.35 is arbitrary, it implies at least 12% of the variance of the species variable was accounted for by the PC. The com- ponent scores of the five PC's were used in subsequent multiple-regression analyses. Stepwise multiple-regression analyses (MRA) were performed with spotted seatrout ln(CPUE + 1) and the component scores of each PC on the angler character- istics, meteorological, temporal, geological, and phys- ical platform data and their interactions (predictor variables) (Table 1). An MRA of the predictor variables and ln(CPUE + 1) of spotted seatrout was treated as a separate analysis because spotted seatrout repre- sented 24.8% and 28.3% of the total number of fish caught by anglers and charterboat operators, respec- tively (Table 2), and because they did not load positively with the other species in the PCA. The MRA was ex- ecuted using the STEPWISE procedure with the MAXR option in SAS (1985). Unless otherwise stated, all differences discussed are significant at the a = 0.01 level of significance. Results A total of 55 anglers and 10 charterboat operators returned logbooks with usable information, a 45.8% and 43.5% return rate, respectively. The participants fished at 467 different oil and gas platforms a total of 1196 separate times. Anglers fished at platforms on 666 occasions and caught a total of 20,559 fish repre- senting over 46 different species (Table 2). Charterboat operators fished at platforms 530 times and caught a total of 16,280 fish representing over 42 different species (Table 2). A five-factor PCA explained 45.7% of the variance of the original data set and allowed us to reduce the data from the 17 separate species or species groups into a smaller data set of presumably related species (Table 3). The first factor was defined as a reef fish factor which included high positive loadings for greater amberjack, grey triggerfish, grouper, other snapper Table 1 Temporal, meteorological, angler characteristic, physical plat- form, and geological variables and their interactions used in the multiple regression analysis. Angler Fishing method Boat length Engine horsepower Presence of echosounder Presence of LORAN Presence of graph recorder Geological Mean sediment size Interactions Boat length x Hp Quadratic structure age Structure age x number of legs Structure age x number of crossmembers Structure age x submerged surface area Number of legs x number of cross members Number of legs x number of wells Number of legs x enclosed volume Number of legs x submerged surface area Structure manned x structure in production Water depth x volume of water enclosed Water depth x submerged surface area Submerged surface area x volume of water Physical Structure age Number of crossmembers Number of legs Number of wells Water depth Submerged surface area Volume of water enclosed Structure manned Structure in production Temporal/meteorological Linear date Quadratic date Cubic date Wind speed Wind direction Air temperature and red snapper, and a negative loading for spotted seatrout (Table 3). The pelagic fish factor consisted of positive loadings for dolphin, king mackerel, little tunny and Spanish mackerel, and a negative loading for silver/sand seatrout (Table 3). The third factor was composed of high positive loadings of Atlantic croaker and silver/sand seatrout (Table 3). The fourth factor was composed of high positive loadings of bluefish and red drum (Table 3). The fifth consisted of positive loadings for cobia and sharks, and a high negative loading for blue runner (Table 3). The strongest eco- logical relationships within a PC existed for reef fish and pelagic fish PC's. These groupings included species with similar life histories, habits, and abundances. The biological relationships between the species in the other PC's were more tenuous; however they did provide in- formation on factors relating to the species relative abundance. Results of the MRA of ln(CPUE + 1) of spotted sea- trout with the predictor variables indicated spotted seatrout abundances were highest near small, non- producing structures in shallow water. Fourteen 152 Fishery Bulletin 89|l), 1991 Table 2 Composition of catch around oil and gas platforms of angler and charterboat operator logbook participants, March 1987-March 1988. Species/Group Angler Charterboat operator No. caught Percent No. caught Percent Atlantic croaker (Micropogonias undulatus) 385 1.9 327 2.0 Atlantic spadefish (Chaetodipterus faber) 16 0.1 3 0.0 Bearded brotula (Brotula barbata) 14 0.1 - - Black drum (Pogonias cromis) 118 0.6 8 0.0 Blackfin tuna (Thunnus atlanticus) 20 0.1 5 0.0 Bluefish (Pomatomus saltatrix) 699 3.4 460 2.8 Blue marlin (Makaira nigricans) 2 0.0 2 0.0 Blue runner (Caranx fusus) 209 1.0 70 0.4 Cobia (Rachycentron canadum) 216 1.1 203 1.2 Crevalle jack (Caranx hippos) 41 0.2 19 0.1 Cubbyu (Equetus umbrosus) 5 0.0 - - Dolphin (Coryphaena hippurus) 209 1.0 172 1.1 Florida pompano (Trachinotus carolinus) 5 0.0 324 2.0 Flounder (Paralichthys sp.) 1 0.0 7 0.0 Gafftopsail catfish (Bagre marinus) 56 0.3 17 0.1 Great barracuda (Sphyraena barracuda) 19 0.1 27 0.2 Greater amberjack (Seriola dumerili) 625 3.0 1086 6.7 Grey triggerfish (Batistes capriscus) 635 3.1 211 1.3 Grouper (Family: Serranidae) 422 2.1 583 3.6 Grunts (Haemulon sp.) 44 0.2 5 0.0 Hake (Urophycis sp.) 1 0.0 2 0.0 Hardhead catfish (Arius felis) 301 1.5 133 0.8 King mackerel (Scomberomorus cavalla) 198 1.0 292 1.8 Ladyfish (Elops saurus) 20 0.1 3 0.0 Little tunny (Euthynnus alletteratus) 183 0.9 147 0.9 Lookdown (Selene vomer) 9 0.0 — — Other jacks (Caranx sp.) 49 0.2 14 0.1 Other snapper (Family: Lutjanidae) 443 2.2 809 5.0 Pinfish (Lagodon rhomboides) 66 0.3 70 0.4 Puffer (Family: Tetraodontidae) 1 0.0 - - Rainbow runner (Elagatis bipinnulata) 1 0.0 — — Rays (Family: Dasyatidae) 1 0.0 1 0.0 Red drum (Sciaenops ocellatus) 622 3.0 637 3.9 Red snapper (Lutjanus campechanus) 7315 35.6 3977 24.4 Sharks (Order: Selachii) 165 0.8 236 1.4 Sheepshead (Archosargus probatocephalus) 31 0.2 4 0.0 Shrimp eel (Ophichthus sp.) — — 10 0.1 Silver/sand seatrout (Cynoscion sp.) 1716 8.3 1407 8.6 Skipjack tuna (Euthynnus pelamis) 3 0.0 157 1.0 Spanish mackerel (Scomberomorus maculatus) 484 2.4 211 1.3 Spotted seatrout (Cynoscion nebulosus) 5108 24.8 4605 28.3 Squirrelfish (Holocentrus sp.) — - 12 0.1 Tarpon (Megalops atlanticus) 3 0.0 1 0.0 Tripletail (Lobotes surinamensis) 65 0.3 7 0.0 Wahoo (Acanthocybium solanderi) 19 0.1 10 0.1 White spotted soapfish (Rypticus maculatus) 9 0.0 - — Yellowfin tuna (Thunnus albacares) Total 5 0.0 — 20,559 16,280 significant predictor variables explained 42.2% of the variance contained in spotted seatrout ln(CPUE + 1) (Table 4). Water depth had the highest partial correla- tion coefficient (r 2 ) and a negative regression coeffi- cient (Table 4), indicating that spotted seatrout were more prevalent in shallow water. Regression coeffi- cients were negative for submerged surface area x volume of water enclosed, submerged surface area, Stanley and Wilson: Fish abundance near oil and gas platforms 153 Table 3 Common factor analysis using principal component analysis for the first five factors of ln(CPUE + 1) for the 17 most frequently caught species by logbook participants. March 1987- •March 1988. Loadings below 0.35 are marked with a dash. Principal component Cobia Atlantic croaker Bluefish Bluerunner Species/Group Reef fish Pelagic fish Silver/sand seatrout Red drum Shark Atlantic croaker _ — 0.755 — — Bluefish — — — 0.766 — Bluerunner — — — — -0.681 Cobia — — — - 0.379 Dolphin - 0.406 - - - Greater amberjack 0.664 — — — — Grey triggerfish 0.515 - - - — Grouper 0.691 — — — — King mackerel — 0.636 — — — Little tunny — 0.600 — - — Other snapper 0.582 - - - - Red drum — — — 0.589 — Red snapper 0.613 - — - - Sharks — — — — 0.350 Silver/sand seatrout — -0.401 0.671 — — Spanish mackerel - 0.600 - - - Spotted seatrout -0.474 — — — — Eigenvalue 2.248 1.691 1.448 1.235 1.141 Proportion of variance explained 0.134 0.103 0.082 0.073 0.065 Cumulative variance explained 0.134 0.236 0.319 0.392 0.457 Table 4 Significant variables of a multipk regression of spotted sea- trout ln(CPUE + 1) from logbook participants. March 1987- March 1988. based on physical, temporal, geological, meteor- ological, and angler characteristic variables and their interactions. Variable b value Partial r 2 Water depth -0.03 0.230 Surface area x volume -1.1 x 10- 8 0.168 Water depth x volume 1.6x10-' 0.162 Water depth x surface area 2.2 xlO" 6 0.132 Number of legs 0.01 0.060 Surface area -5.7 x 10~ 5 0.038 Echosounder presence -0.21 0.037 Fishing method -0.26 0.033 Quadratic data -0.01 0.025 Boat length 0.08 0.017 LORAN presence -0.17 0.012 Engine horsepower -7.2 x 10- 4 0.008 Structure in production -0.14 0.006 Number of legs x wells -2.7 x 10 4 0.006 Number of wells — N.S. Volume - N.S. Intercept 0.93 Model r 2 0.42 t the 1% level. N.S. = not significant a number of legs x number of wells, and whether the structure was in production; and coefficients were positive for the number of legs, water depth x sub- merged surface area, and water depth x volume of water enclosed. Since the regression coefficients were negative for water depth, fishing method, quadratic date, boat horsepower, and the presence of LORAN and echosounder, catch rates of spotted seatrout were highest while bottom-fishing by relatively small vessels without sophisticated electronics. Based on the plot of the sum of the linear and quadratic components of month, spotted seatrout were most abundant in the late spring and early summer (Fig. 1). Multiple-regression analysis suggested reef fish were most abundant near large complex structures at inter- mediate water depths (Fig. 2). Ten significant predic- tor variables accounted for 35.2% of the variance in the reef fish factor (Table 5). The regression coeffi- cients of water depth, volume of water enclosed, and the interactions of submerged surface area x volume of water enclosed, and the number of legs x number of crossmembers were positive, while negative regres- sion coefficients were observed of the interactions of water depth x volume of water enclosed, and water 154 Fishery Bulletin 89(1), 1991 - Atlantic croaker- silver/sand seatrout ' Spotted \ seatrout — * Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec. Jan. Feb. 1987 ,o 88 Figure 1 Calculated values of Atlantic croaker-silver/sand seatrout, bluefish-red drum, and spotted seatrout abundance on ortho- gonal of month from multiple regressions. depth x submerged surface area (Table 5). Platform age did not affect the abundance of reef fish as evi- denced by the negative regression coefficient for quad- ratic age of the platform (Table 5). Angler character- istics were not good predictors of reef fish catches, with only the presence of graph recorders and fishing method being significant (Table 5). Since the regres- sion coefficient of fishing method was negative, reef fish catches were highest while bottom fishing. Catches of pelagic fish were higher while trolling in relatively small, well-equipped vessels near large un- manned structures in intermediate water depths (Fig. 3). Retention of 10 significant predictor variables accounted for 31.5% of variance in the pelagic fish factor (Table 5). Regression coefficients for fishing method, LORAN presence, linear date, submerged sur- face area, water depth, and the interaction of sub- merged surface area and volume of water enclosed were positive from the MRA, while negative regres- sion coefficients were found for the interactions of structure manned x structure in production, water depth x submerged surface area, and water depth x volume of water enclosed and boat length (Table 5). Highest abundances of Atlantic croaker and silver/ sand seatrout were found near small, manned plat- forms in deep water. Angler characteristic variables Water Depth (m) Volume Enclosed (•100,000 m 3 ) Figure 2 Response surface plot of water depth, volume of water en- closed by oil and gas platforms, and reef fish abundance from the multiple regression of principal component scores for catch data from logbook participants. Note: stippled area represents realistic values of volume of water enclosed at various depths for oil and gas platforms. had little influence on their catch rates. Eight signifi- cant predictor variables explained 18.4% of the Atlan- tic croaker-silver/sand seatrout factor (Table 5). Tem- poral variables (quadratic date and linear date) had the highest partial r 2 values (Table 5) and indicated Atlan- tic croaker-silver/sand seatrout were most abundant in the early spring (Fig. 1). Positive regression coeffi- cients were found for water depth and the interaction of structure manned x structure in production, and negative regression coefficients were found for volume of water enclosed and structure age x submerged sur- face area. LORAN presence and fishing method were the only significant angler characteristic variables re- tained in MRA (Table 5). Based on MRA results, highest catches of bluefish and red drum occurred while trolling from late winter to early spring near platforms with complex construc- tion. Retention of five significant predictor variables explained 14.5% of the variance in the bluefish-red drum factor (Table 5). The positive regression coeffi- cient and the high partial r 2 (Table 5) for the interac- Stanley and Wilson: Fish abundance near oil and gas platforms 155 Table 5 Significant variables of multipl e regressions of reef fish, pelagic fish, Atlantic croaker -silver/sand seatrout, bluefish-red drum, and blue runner-shark-cobia principal-component scores from logbook participants, March 1987-March 1988, based on physical, temporal, geological, meteorological, and angler characteristic variables and their interactions. Atlantic croaker- Bluefish- Blue runner- Reef fish Pelagic fish silver/sand seatrout red drum shark-cobia Partial Partial Partial Partial Partial Variables b value r~ b value r 2 b value r 2 b value r 2 b value r 2 Water depth 0.04 0.17 0.02 0.03 0.01 0.01 -0.01 0.04 Surface area x volume 1.6 x 10" 9 0.12 1.1 x 10- B 0.05 — — — — — — Water depth x volume -3.2 x lO" 7 0.08 -1.1 xlO" 7 0.04 — — — — — — Water depth x surface area - -2.3xl0" 6 0.06 -1.8x 10- 6 0.04 — — — — — — Number of legs -0.01 0.01 -0.01 0.04 - — — — — — Surface area — — 5.5 x 10" 5 0.02 — — — — 2.4 xlO 5 0.02 Echosounder — — — — — — -0.19 0.01 -0.21 0.01 Fishing method -0.51 0.06 1.13 0.20 0.35 0.02 0.45 0.04 -0.39 0.03 Quadratic date — — — — 0.02 0.08 0.01 0.01 — — Boat length — — -0.07 0.03 — — — — — — LORAN presence — — 0.43 0.04 -0.33 0.02 — — — — Number of legs x crossmembers 2.5 xlO- 5 0.02 — — — — 1.9 xlO" 5 0.08 — — Graph recorder -0.23 0.02 — — — — 0.23 0.01 0.17 0.01 Quadratic age of platform -2.8 x 10" 6 0.01 - — — - — - — — Volume 1.8 x lO" 5 0.01 — — -3.5 x 10" 6 0.01 — — — — Linear date — — 0.04 0.03 0.04 0.03 — — -0.02 0.01 Structure manned x production — — -0.14 0.01 0.11 0.01 — — — — Age x surface area — — — — -2.4 x 10" 7 0.01 — — — — Intercept -0.51 -0.11 -0.11 0.45 0.31 Model r 2 0.35 0.32 0.18 0.15 0.12 Volume Enclosed (•100,000 m 3 ) Figure 3 Response surface plot of water depth, volume of water enclosed by oil and gas platforms, and pelagic fish abun- dance from the multiple regression of principal component scores for catch data from logbook participants. Note: stippled area represents realistic values of volume of water enclosed at various depths for oil and gas platforms. tion of the number of legs x number of crossmembers indicated that bluefish and red drum were most abun- dant near platforms with complex construction. Based on the plot of the orthogonal components of month, bluefish and red drum catch rates were highest in the early spring and late winter (Fig. 1). A positive regres- sion coefficient for fishing method and negative regres- sion coefficients for the presence of echosounders and graph recorders (Table 5) indicated that highest catches of bluefish and red drum occurred while trolling and that sophisticated electronic equipment did not increase catch rates. 156 Fishery Bulletin 89(1), 199! Shark and cobia catches were highest while bottom fishing in the spring near large platforms in shallow water. The MRA with six significant predictor vari- ables accounted for 11.5% of the variance in the shark- cobia factor (Table 5). The negative regression coeffi- cient for water depth and positive regression coefficient for submerged surface area (Table 5) provide evidence that shark and cobia abundances were highest in shallow waters near large structures. Highest cajtches of shark and cobia occurred while bottom fishing in the spring, based on the negative regression coefficient for fishing method and linear date (Table 5). Conflicting results on the presence of electronic gear were found with a positive regression coefficient for presence of graph recorders and a negative regression coefficient for the presence of echo sounders (Table 5). Discussion Anglers and charterboat operators utilized the entire range of sizes and operational types of platforms available off the coast of Louisiana (single-well caissons, steel template platforms, and mobile semisubmersible drilling platforms), although certain trends in platform size utilization and fishing method were evident. Near- shore fishermen most often fished at the small struc- tures (i.e., caissons) in shallow water (i.e., <10m), while offshore bottom-fishing and trolling fishermen fished near much larger steel template platforms in deeper water (i.e., >30m). Charterboat operators had larger vessels and were able to fish in deeper waters and far- ther offshore than anglers. Anglers and charterboat operators caught a total of 36,839 fish representing at least 46 different species, providing evidence for the high diversity of fish around the oil and gas platforms. Fishes caught ranged from relatively common and highly desirable species such as spotted seatrout and red snapper to relatively rare fishes such as hake, bearded brotula, and squirrel fish. Highly sought-after gamefish such as tarpon, blue marlin, king mackerel, and yellowfin tuna were also caught. Catches by angling are selective and biased towards larger species because of the hook-and-line gear utilized, and usually only carnivorous species are susceptible to the gear. Consequently species not susceptible to angling were not represented (Grimes et al. 1982). The associations of fish identified by the reef fish and pelagic fish PC's were in agreement with fish classifica- tion systems using direct observation around natural and artificial reefs in the Gulf of Mexico by Starck (1968) and Lukens (1981). This confirms that these groupings have an ecological basis, and were not an artifact of the sampling or analysis techniques. Factors affecting the abundances of fish Physical platform variables Generally, the highest abundances of spotted seatrout were found in shallow water (i.e., <10m) around small, non-producing plat- forms such as caissons. These results were expected, as this estuarine-dependent species (Johnson and Seaman 1982) would likely have its highest abundances in shallow water near estuaries. Our results suggest that reef fish, Atlantic croaker, and silver/sand seatrout abundances increased with size and complexity of the artificial reef, agreeing with past studies (Turner et al. 1969, Grove and Sonu 1983). However, an optimal artificial reef size occurred for reef fish based on the response surface plot of water depth, volume enclosed, and fish abundance as highest reef-fish abundances occurred at intermediate depths (i.e., 70-100m) near relatively large platforms (i.e., mean volume enclosed 150,000-250,000 m 3 ). The op- timal size range of oil and gas platforms acting as artificial reefs was significantly larger than the optimal artificial reef sizes reported in past studies. This dif- ference could be due to the open construction and lack of interstitial spaces on oil and gas platforms which may not be as efficient at attracting or increasing secondary production of fish as the large rubble or prefabricated artificial reef units on which past esti- mates were calculated. Also, oil and gas platforms ex- tend throughout the entire water column, and because many reef fish are demersal, a large portion of the plat- form may not be suitable reef-fish habitat. Reef-fish abundances were lowest at the largest platforms in extremely deep water, probably because the water depths exceeded the preferred ranges for these species. Physical platform variables were not important predictor variables of pelagic fish abundance. Our results are consistent with Wickham et al. (1973) and Grove and Sonu (1983) who concluded that pelagic fishes respond to the visual attraction of artificial reefs and not to the overall size or surface area. Bluefish, red drum, cobia, and shark abundances were highest around large, structurally complex plat- forms. These species were probably not trophically dependent on the structures, but were attracted to plat- forms by an optical stimulus as reported by Wickham et al. (1973) for cobia. Age of the structure was not a significant factor in explaining fish composition or abundance around oil and gas platforms. Therefore it appears that the species-equilibrium model, which suggests that the number of species present and their abundances in- crease rapidly over a colonization period eventually stabilizing (MacArthur and Wilson 1967), was not ap- plicable, or that the platforms may have been fully Stanley and Wilson: Fish abundance near oil and gas platforms 157 colonized at the start of the study and we were sam- pling after species stabilization occurred. The latter ex- planation seems most likely, because the age of plat- forms in our study ranged from 8 months to 30 years, and full colonization of artificial reefs in the northern Gulf of Mexico can occur in as little as 15 months (Lukens 1981). Seasonal variation Season was an important factor affecting the abundances of fish around oil and gas plat- forms. Smith (1979) and Lukens (1981) reported large fluctuations in species composition and abundance around natural and artificial reefs in the northern Gulf of Mexico, and based on our results this would include fish populations around oil and gas platforms off the Louisiana coast. The apparent higher abundances of spotted seatrout in the spring and summer may be a result of a tem- perature-induced increase in feeding rate and/or the aggregation of the fish into spawning schools (Johnson and Seaman 1982). The increase in pelagic fish catches from winter through fall indicated that abundance may have been related to water temperature. This is consistent with the findings by Fable et al. (1981), as they found that the charterboat catches of king mackerel and other pelagic species in the northeast Gulf of Mexico in- creased with water temperature. Lassuy (1983) reported that Atlantic croaker abun- dance was highest in the spring and summer, while Sut- ter and Mcllwain (1982) reported silver/ sand seatrout abundance to be highest in the winter and spring. This may explain the apparent conflicts in the results from the MRA with respect to seasonal abundances of these two species. Silver/sand seatrout may have been abun- dant in the winter and spring, and Atlantic croaker in the spring and summer. However, when these species were grouped the seasonal abundance results explained only silver/sand seatrout and not the entire group. Our results, along with those of others (Gallaway et al. 1981b, Reagan 1982) provided evidence that bluefish and red drum were seasonal transients with highest abundances from fall through spring. Shark and cobia abundances appeared to be highest in the spring and decreased thereafter. Angler characteristic variables Overall, fishing power was not equal because fishermen with larger vessels and engines had access to deeper water, and fishermen with sophisticated electronics could more easily locate fish. Because reef fish, Atlantic croaker, and silver/ sand seatrout are found in deep water, anglers with large vessels and engines had the highest catches for these species, while cobia, sharks, and pelagic species were more frequently caught by vessels having sophis- ticated electronics. Since spotted seatrout were found in shallow water, large vessels or sophisticated elec- tronics did not increase spotted seatrout catches. Geological and meteorological variables Geological and meteorological variables were not significant pre- dictors of fish abundance around oil and gas platforms. Meteorological variables were probably not important because fishermen generally fish only in good weather, consequently catch rates during poor weather condi- tions were not reported. Conclusions The physical construction of oil and gas platforms precludes sampling of the associated sportfish popula- tions using traditional methods (e.g., gillnets, trawls). Based on the results of this logbook program, the col- lection of CPUE data over long periods of time may be an effective technique of monitoring the fish popula- tions susceptible to angling associated with artificial reefs. Although the data supplied by the logbooks is an index of relative abundance of fish susceptible to angling and is biased towards larger individuals, it pro- vides a valuable source of data which is otherwise dif- ficult to obtain. With the advent of the "Rigs to Reefs" initiative in Louisiana, biological criteria to determine where to locate retired oil and gas platforms as artificial reefs was needed. Information derived from this study sug- gests that optimal artificial reef configurations exist, but vary depending on the species. To effectively site artificial reefs for reef fish, Atlantic croaker, and silver/sand seatrout there should be large complex plat- forms at intermediate depths; for pelagic species ar- tificial reef size is not a factor, although they should again be placed at intermediate depths. Optimal siting of artificial reefs for spotted seatrout should include small structures in shallow water. Gallaway and Lewbel (1982) suggested that abundances of some species were directly proportional to the submerged surface area of oil and gas platforms. We believe that the relationship between fish abundances and artificial reefs is much more complex, with other factors, such as natural and temporal variability of species distribution and abun- dance interacting with physical platform variables and water depth to determine overall species abundances. Acknowledgments The authors thank the 55 anglers and 10 charterboat operators for maintaining logbooks; the oil and gas operators for providing information and drawings on 158 Fishery Bulletin 89(1), 1991 the platforms; Mr. Villere Reggio, Minerals Manage- ment Service, for additional platform information; Louisiana State Climatology Office for meteorological data; Drs. James Geaghan and Gary Shaffer for statis- tical assistance; Dr. Linda Jones and two anonymous reviewers for critical review of the manuscript and im- proving its contents. This study was part of a Master's thesis by David R. Stanley and was supported bv the Department of Marine Sciences, LSU; the Louisiana Artificial Reef Initiative; and Dupont Inc. Educational Aid Program. Citations Bohnsack, J. A. 1989 Are high densities of fishes at artificial reefs the result of habitat limitation or behavioral preference? Bull. Mar. Sci. 44:631-645. Bohnsack, J. A., and D.L. Sutherland 1985 Artificial reef research: A review with recommendations for future priorities. Bull. Mar. Sci. 37:11-39. Chandler, C.R., R.M. Sanders Jr., and A.M. Landry Jr. 1985 Effects of three substrate variables on two artificial reef communities. Bull. Mar. Sci. 37:129-142. Chatfield, C, and A.J. Collins 1980 Introduction to multivariate analysis. Chapman and Hall, NY, 246 p. Coleman, E. 1984 Coastal Louisiana: climate and recreation. Sea Grant Coll. Prog., La. State Univ., Baton Rouge, 24 p. Coleman, J.M., H.H. Roberts, and T.F. Moslow 1986 Sedentary sequences and seismic responses related to sea level change: Louisiana continental shelf. Report to Gulf of Mexico Research Consortium. School of Geoscience, La. State Univ., Baton Rouge, 96 p. Continental Shelf Associates, Inc. 1982 Study of the effect of oil and gas activities on reef fish populations in the Gulf of Mexico OCS area. OCS Rep./ MMS82-010, U.S. Dep. Inter., Minerals Manage. Serv.. Gulf of Mexico OCS Reg., New Orleans, 210 p. Demory, R.L., and J.T. Golden 1983 Sampling the commercial catch. In Nielson, L.A., and D.L. Johnson (eds.), Fisheries techniques, p. 421-430. Am. Fish. Soc, Bethesda, MD. Dillon, W.R., and M. Goldstein 1984 Multivariate analysis. John Wiley, NY, 587 p. Fable, W.A. Jr., H.A. Brusher, L. Trent, and J. Finnegan Jr. 1981 Possible temperature effects on charter boat catches of king mackerel and other coastal pelagic species in northwest Florida. Mar. Fish. Rev. 43(8):21-26. Gallaway, B.J. 1984. Assessment of platform effects on snapper populations and fisheries. In Proc, Fifth annual Gulf of Mexico informa- tion transfer meeting, New Orleans, Nov. 27-29, 1984, p. 130-137. OCS Study/MMS85-0008, U.S. Dep. Inter., Minerals Manage. Serv., Metarie, LA. Gallaway, B.J., and G.S. Lewbel 1982 The ecology of petroleum platforms in the northwestern Gulf of Mexico: A community profile. Open-file Rep. 82-03, U.S. Fish Wildl. Serv., Office Biol. Serv.. Wash. DC. 91 p. Gallaway, B.J., L.R. Martin, R.L. Howard, G.S. Boland, and G.D. Dennis 1981a Effects on artificial reef and demersal fish and macro- crustacean communities. In Middleditch, B.S. (ed.), Environ- mental effects of offshore oil production: The Buccaneer gas and oil field study, p. 237-299. Mar. Sci. Vol. 14, Plenum Press, NY. Gallaway, B.J., M.F. Johnson, L.R. Martin, F.J. Margraf, G.S. Lewbel, R.L. Howard, and G.S. Boland 1981b The artificial reef studies, Vol. 2. /nBedinger Jr., C.A., and L.Z. Kirby (eds.), Ecological investigations of petroleum production platforms in the central Gulf of Mexico. SWRI Proj. 01-5245, U.S. Dep. Inter., Bur. Land Manage., Gulf of Mexico OCS Reg., New Orleans, 199 p. Gooding, R.M., and J.J. Magnuson 1967 Ecological significance of a drifting object to pelagic fishes. Pac. Sci. 2:486-497. Grimes, C.B., C.S. Manooch, and G.R. Huntsman 1982 Rock and reef outcropping fishes of the outer continen- tal shelf of North Carolina and South Carolina and ecological notes on red porgy and vermillion snapper. Bull. Mar. Sci. 32:277-289. Grove, R.S., and C.J. Sonu 1983 Review of Japanese fishing reef technology. Tech. Rep. 83-RD-137, South. Calif. Edison Co., Rosemead, CA, 112 p. Gunter, G. 1963 The fertile fisheries crescent. J. Miss. Acad. Sci. 9: 286-290. Hastings, R.W., L.H. Ogren, and M.T. Mabry 1976 Observations on the fish fauna associated with offshore platforms in the Northeastern Gulf of Mexico. Fish. Bull., U.S. 74:387-401. Hunter, J.R., and C.T. Mitchell 1967 Association of fishes with flotsam in the offshore waters of Central America. Fish. Bull., U.S. 66:13-29. 1968 Field experiments on the attraction of pelagic fish to floating objects. J. Cons. Perm. Explor. Mer 31:427-434. Huntsman, G.R. 1981 Ecological considerations influencing the management of reef fishes. In Aska, D.Y. (ed.), Artificial reefs: Conference proceedings, p. 167-175. Rep. 41, Fla. Sea Grant Coll. Prog., Univ. Fla., Gainesville. Johnson, D.R., and W. Seaman Jr. 1982 Species profiles: Life history requirements and en- vironmental requirements of coastal fishes and invertebrates- spotted seatrout. U.S. Fish Wildl. Serv. Biol. Rep. 82(11.43), 18 p. Klima, E.F., and D.A. Wickham 1971 Attraction of coastal pelagic fishes with artificial struc- tures. Trans. Am. Fish. Soc. 100:86-99. Lassuy, D.R. 1983 Species profiles: Life history requirements and environ- mental requirements (Gulf of Mexico)- Atlantic croaker. U.S. Fish Wildl. Serv. FWS/OBS-82/11.3, 12 p. Luckhurst, B.E., and K. Luckhurst 1978 Analysis of the influence of substrate variables on coral reef fish communities. Mar. Biol. (Berl.) 49:317-323. Lukens. R.R. 1981 Ichthyofaunal colonization of a new artificial reef in the northern Gulf of Mexico. Gulf Res. Rep. 7:41-49. MacArthur, R.H., and E.O. Wilson 1967 The theory of island biogeography. Princeton Univ. Press, Princeton, NJ, 203 p. Stanley and Wilson: Fish abundance near oil and gas platforms 159 Molles, M.C. Jr. 1978 Fish species diversity on model and natural reef patches: Experimental insular biogeography. Ecol. Monogr. 48: 289-305. Moore, D., H.A. Brusher, and L. Trent 1970 Relative abundance, seasonal distributions and species composition of demersal fishes off Louisiana and Texas, 1962-1964. Contrib. Mar. Sci. 15:45-70. Ogawa, Y. 1982 The present status and future prospects of artificial reefs: Developmental trends of artificial reef units. In Vik, S.F. (ed.), Japanese artificial reef technology, p. 23-41. Tech. Rep. 604, Aquabio, Inc., Bellair Bluffs, FL. Pennington, M. 1983 Efficient estimators of abundance for fish and plankton. Biometrics 39:281-286. 1985 Estimating the relative abundance of fish from a series of trawl surveys. Biometrics 41:197-202. Reagan R.E. 1982 Species profiles: Life history requirements and environ- mental requirements of coastal fishes and invertebrates (Gulf of Mexico)— red drum. U.S. Fish Wildl. Serv. Biol. Rep. 82(11.36), 18 p. Reggio, V.C. Jr. 1987 Rigs-to-reefs: The use of obsolete petroleum structures as artificial reefs. OCS Rep./MMS87-0015, U.S. Dep. Inter., Minerals Manage. Serv., Gulf of Mexico OCS Reg., New Orleans, 17 p. Rousenfell, G.A. 1972 Ecological effects of offshore construction. J. Mar. Sci. 2:1-208. SAS Institute Incorporated 1985 SAS user's guide: Statistics, version 5 edition. SAS Inst., Cary, NC, 956 p. Smith, G.B. 1979 Relationship of eastern Gulf of Mexico reef fish com- munities to species equilibrium theory of insular biogeography. J. Biogeogr. 6:49-61. Sonnier, F., J. Teerling, and H.D. Hoese 1976 Observations on the offshore reef and platform fish fauna of Louisiana. Copeia 1976:105-111. Starck, W.A. 1968 A list of fishes of Alligator reef, Florida with comments on the nature of Florida reef fish fauna. Undersea Biol. 1: 5-40. Stephens, J.S. Jr., and K.E. Zerba 1981 Factors affecting fish diversity on a temperate reef. En- viron. Biol. Fishes 6:111-121. Stone, R.B.. H.L. Pratt. R.O. Parker Jr., and G.E. Davis 1979 A comparison of fish populations on an artificial and natural reef in the Florida Keys. Mar. Fish. Rev. 41:1-11. Sutter, F.C.. and T.D. Mcllwain 1982 Species profiles: Life history requirements and environ- mental requirements of coastal fishes and invertebrates (Gulf of Mexico)— sand seatrout and silver seatrout. U.S. Fish Wildl. Serv. Biol. Rep. 82(11.36), 16 p. Sztramko, L. 1986 Lake Erie angler diary program. 1985. Lake Erie Fish. Assess. Unit Rep. 1986-4. Ontario Minist. Nat. Resourc, Wheatley, Ontario, 45 p. Turner, C.F., E.E. Ebert, and R.R. Given 1969 Man-made reef ecology. Calif. Dep. Fish Game, Fish. Bull. 146, 221 p. Vik, S.F. (editor) 1982 Japanese artificial reef technology. Tech. Rep. 604, Aquabio, Inc, Bellair Bluffs, FL, 380 p. Wickham, D.A., J.W. Watson, and L.H. Ogren 1973 The efficacy of midwater artificial structures for attrac- ting pelagic sport fish. Trans. Am. Fish. Soc. 102:563-572. Witzig, J. 1986 Fishing in the Gulf of Mexico 1984 marine recreational fishing results. In Proceedings, Sixth annual Gulf of Mexico information transfer meeting. New Orleans, p. 103-105. OCS Study MMS86-0073, U.S. Dep. Inter., Minerals Manage. Serv., Gulf of Mexico OCS Reg., New Orleans. Abstract. - Serum progesterone and testosterone levels, as measured by radioimmunoassay, were used to estimate the mean length at attain- ment of sexual maturity (LSM) in a sample of 124 female and 31 male incidentally-killed Dall's porpoises Phocoenoides dalli. Females with serum progesterone levels greater than 1.34ng/mL were considered mature. The LSM for females was estimated at 169.0cm using Kasu- ya's "summation" technique. A tech- nique to fit a two-phase regression model to the male data produced an estimated LSM for males of 183.0 cm. Overall, the estimates for the LSM in this study agreed well with previously published reports using histological and morphological mea- sures of sexual maturity. Hormonal estimation of maturity is proposed as a rapid, inexpensive, and potential- ly non-lethal alternative technique in odontocete populations. Use of Serum Progesterone and Testosterone to Estimate Sexual Maturity in Dall's Porpoise Phocoenoides dalli Jonathan L. Temte Hatfield Marine Science Center, Oregon State University, Newport Oregon 97365 Present address. Department of Zoology, University of Wisconsin Madison, Wisconsin 53706 Manuscript accepted 17 October 1990. Fishery Bulletin, U.S. 89:161-165 (1991). Estimation of the mean age and length at the attainment of sexual maturity in odontocetes has traditionally relied upon morphological or histological parameters, thus necessitating whole animal preparations. Examination of reproductive tissue allows for the de- termination of maturity. When this information is coupled with age/length data, the age or length at attainment of sexual maturity for females and males in a population can be deter- mined using a variety of methods (DeMaster 1984). These methods, however, implicitly require the collec- tion of dead marine mammals in the field. Several authors have used radioim- munoassay (RIA) to assess reproduc- tive condition in living odontocetes. For example, RIA of progesterone has been used as an indicator of luteal function in Tursiops truncatus, Del- phinus delphis (Kirby and Ridgway 1984, Sawyer-Steffan et al. 1983), and Stenella longirostris (Wells 1984). Likewise, measurements of serum testosterone by RIA have been used to assess male sexual condition in T. truncatus (Harrison and Ridgway 1971) and S. longirostris (Wells 1984). Temte and Spielvogel (1985) dem- onstrated that serum progesterone concentration, as measured by RIA, was a good predictor of corpus lu- teum mean diameter in a sample of 32 incidentally-killed pregnant or lac- tating Dall's porpoises Phocoenoides dalli from the northwestern North Pacific Ocean. Furthermore, they noted that 17 sexually immature fe- males (no corpora present in ovaries) had very low concentrations of serum progesterone (Table 1). In this study the results of proges- terone and testosterone RIA in 124 female and 31 male Dall's porpoises, respectively, were used to estimate the mean length at attainment of sex- ual maturity (LSM) in this species. The results are compared to those ob- tained using traditional histological techniques to demonstrate the effec- tiveness of this inexpensive and non- lethal technique. Methods Blood samples During June and July of 1982, per- sonnel of the National Marine Mam- mal Laboratory (NMFS) obtained blood samples from a total of 105 Dall's porpoises that were inciden- tally-killed in Japanese salmon gill- nets in the North Pacific Ocean. The following groups were represented in the sample: 21 non-pregnant, non- lactating females in which the matur- ity status was not known (149-199 cm, standard body length); 30 preg- nant females (167-200 cm); 23 lac- tating females (167-200cm), and 31 males in which the maturity status was not known (101-210 cm). Sam- ples of whole blood were drawn and centrifuged using methods described 161 162 Fishery Bulletin 89(1), 1991 Table 1 Serum progesterone concentrations [P] in immature and mature female Dall's porpoises. Data from 1980 sample of Temte and Spielvogel (1985). Status No. Mean [P] (ng/mL) SD Range of [P] Mature Pregnant (P) Lactating (L) non-P, non-L Immature* 33 24 17 14.76 19.40 2.63 0.49 0.27 12.43 11.31 3.75 0.46 0.0-45.3 0.9-45,3 0.0-11.4 0.0-1.3 * Excluding one immature female with a large follicle and 31.1 ng/mL progesterone. by Temte and Spielvogel (1985). In these samples, how- ever, serum was decanted into 1.5-mL plastic ultra- centrifuge tubes and frozen at - 20° C until assays were performed. RIA for progesterone Progesterone RIA was identical to that reported by Temte and Spielvogel (1985). This assay had previ- ously been validated for porpoise serum, and chroma- tography had shown an absence of interference from other serum constituents. Triplicate volumes of por- poise serum (10/iL for pregnant, 50^L for lactating, and 100 mL for non-pregnant, non-lactating females) were doubly extracted utilizing a 1:2 mixture of benzene and hexane (Sawyer-Steffan and Kirby 1980). The antiserum used was anti-progesterone-11-BSA, No. 1337 (Gordon D. Niswender). The intraassay coef- ficient of variation (COV) was 5.5%, while the inter- assay COV was 5.9%. The sensitivity of this assay was 0.1 ng/mL. RIA for testosterone The procedure for testosterone RIA was nearly iden- tical to that for progesterone. Triplicate volumes of 25-^L porpoise serum were doubly extracted with 1:2 benzene:hexane. The antiserum used was anti-testos- terone, No. s-250 (Gordon D. Niswender). The com- petitor was [ 3 H]testosterone (NET-553, New England Nuclear). The intraassay coefficient of variation (COV) was 6.3%, while interassay COV was 14.6%. The assay sensitivity was 0.3ng/mL. Maturity criteria for females The mean serum progesterone for immature females reported in Temte and Spielvogel (1985) was 0.27 ng/mL. Assuming a normal distribution, 99% of imma- ture females would be expected to have serum pro- gesterone levels less than 1.34 ng/mL. Based upon this result, females with serum progesterone concentra- tions greater than 1.34 ng/mL were considered sexually mature. Whereas, this progesterone level is lower than the 3.0ng/mL used by Kirby and Ridgway (1984) as an indicator of ovulation in D. delphis and T. truncatus, 94% of the immature females and 96% of the pregnant females reported in Temte and Spielvogel (1985) were correctly classified using this criteria. Maturity criteria for males Since samples were collected near the peak breeding period for Dall's porpoises (Newby 1982; see also Jef- ferson 1989), and considering the seasonal flux of tes- tosterone in other odontocetes (Wells 1984), separation of immature and mature Dall's porpoises by testos- terone level alone was theoretically possible in this study. Although Wells (1984) regarded serum testos- terone concentrations of less than 8.0 ng/mL as baseline levels in S. longirostris, a natural break in the Dall's porpoise data occurred between 1.7 and 5.0ng/mL. However, as tissue samples were not available for maturity assessment, no hormonal maturity criteria could be confirmed. Estimation of LSM The proportions of mature females in each 10-cm in- crement (135-225cm) were calculated. Graphical anal- ysis determined the length at which the cumulative probability of maturity, I(mj), equaled the cumulative probability of not being mature at that length or longer, 1(1 -mj). This method was first used by Kasuya (1972) and has been termed the summation technique by DeMaster (1984). It is recommended when small samples of size- or age-classes exist. A small total sample precluded the use of the sum- mation technique in males. As a consequence, a modi- fication of the regression technique (see DeMaster 1984) was used to identify the LSM. Because testos- terone concentrations demonstrated a discontinuity (see above), a continuous two-phase regression model (Yeager and Ultsch 1989; review by Nickerson et al. 1989) was used to objectively identify the transition point in the data. The transition point in the length represents the point at which the discontinuity oc- curred, or in this case, the LSM. Temte: Hormonal assessment of porpoise maturity 163 Females — 30 E 0)25 ID ' O (30) (21) (23) lOn-pregnanl pregnanl lactaling Reproductive Status Figure 1 Serum progesterone concentrations for non-pregnant/non- lactating, pregnant, and lactating female Dall's porpoises. Ver- tical bars indicate ranges of values. Horizontal bars indicate means and standard errors. Sample sizes are in parentheses. Animals were collected June-July 1982 in the western North Pacific Ocean. Results Comparison of 1982 and 1980 progesterone data The results of the progesterone assays (1982 data: Fig. 1) were compared with results of Temte and Spiel vogel (1985) (1980 data). No significant differences were found between the mean progesterone concentrations for pregnant females (1980: n = 24, x = 19.40; 1982: n = 30, x = 18.09; P>0.5; Student's t-test), or for lac- tating females (1980: n = 8, x = 2.63; 1982: n = 23, 5=2.17; P»0.05; Wilcoxson rank sum). Mean pro- gesterone concentrations were significantly different for the two groups of non-pregnant, non-lactating por- poises (1980: n = 19, x= 1.91; 1982: n = 21, x = 2.92; P<0.01; Wilcoxson rank sum). However, the mean length of the 1982 sample was significantly greater than that of the 1980 sample (1982: x = 171.2cm; 1980: x= 163.1cm; P<0.05; Student's i-test), and the difference in mean progesterone could be due to a dif- ference in the proportion of mature females. Therefore, the results of progesterone analysis in pregnant, lac- tating, and non-pregnant/non-lactating females from the 1982 sample were pooled with the results from the 1980 sample of Temte and Spielvogel (1985). Females Serum progesterone is plotted against length for 124 female Dall's porpoises (Fig. 2). The results demon- Females 50 1 o np-1980 , 45- E 40 • • np-1982 en 35 • preg c lac 30 g- . • o • . . * i- 20- 15- * . o> „ o 10 • .• •. * '.•' • O- 5 • *. '*** : o V d/JBloaf VftifV %- 1 > 1 Length (cm) Figure 2 Serum progesterone concentration as a function of length in non-pregnant (1980 O; 1982 •), pregnant (*), and lactating ( + ) female Dall's porpoises. Females 130 140 150 160 170 180 190 200 210 Length (cm) Figure 3 Summation technique for female Dall's porpoises based on pro- gesterone values. Intersection of lines occurs when X(m,) = X(l-m,). strate the incidence of high progesterone levels at lengths greater than 165 cm. These elevated levels were seen not only in pregnant females, but also in non- pregnant, non-lactating females, indicating possible ovulation. The summation technique provided an estimated LSM of 169.0cm (Fig. 3). Males Serum testosterone displayed a marked increase with an increase in length. A natural break in the testos- terone data occurred between 175 and 180cm, when 164 Fishery Bulletin 89(1), 1991 Males 100 110 120 130 140 150 160 170 ISO 190 200 210 Length (cm) Figure 4 Serum testosterone concentration as a function of length for 31 male Dall's porpoises collected in 1982. Two-phase regres- sion lines are shown. testosterone increased from undetectable levels to relatively high levels. The best fitting two-phase regres- sion model (Fig. 4) indicated that the transition point occurred at 183.0cm. Hence, the LSM for males was estimated as 183.0cm. The mean (± SE) testosterone concentration was 1.00 ± 0.52ng/mL for males shorter than the LSM, and 10.46 ± 1.20 ng/mL for males longer than the LSM. Discussion Newby (1982), using the presence of ovarian scars and the 50th percentile method (see DeMaster 1984). esti- mated the LSM to be 170.5cm for females in this western North Pacific population. The use of hormonal data alone provided a similar, but slightly lower, LSM estimate of 169.0 cm. However, based on the growth curves for females of this population (fig. 31 in Newby 1982), this 1.5-cm difference translates into small dif- ferences in ages. The estimate for LSM in males of 183.0cm is in very close agreement with the previously estimated LSM of 182.6cm based on the 50th percentile method using testis-epididymal weight (Newby 1982). Direct com- parison with previous studies are not possible due to the lack of variance estimates for LSM by the methods used. Nevertheless, the hormonal estimates of LSM provided in this study agree quite well with previous, non-hormonal methods. Radioimmunoassay is a quick, economical, and accu- rate measure of reproductive condition. Small samples of blood (< 1 mL) provide adequate serum for replicate assays. Previous studies have shown the high predic- tive value of serum progesterone level to corpus luteum size and reproductive state (Temte and Spiel vogel 1985, Kirby and Ridgway 1984, Sawyer-Steffan etal. 1983). Moreover, such methods could allow the rapid collec- tion of samples for maturity status assessment from large numbers of animals in the field. Hormonal assessment of reproductive status may prove to be a non-lethal technique to estimate popula- tion parameters such as LSM. It is, at present, more applicable to captive animal studies. However, situa- tions in which animals are killed incidentally may pro- vide opportunities for obtaining concordant blood and reproductive tissue samples. Such sampling could allow the direct comparison of hormonal and histological methods of estimating maturity and establish appro- priate hormonal criteria for future studies. Seasonality in breeding may well limit the usefulness of this method in males (see Perrin and Donovan 1984) because testosterone undergoes seasonal fluctuation. In addition, the presence of environmental toxins may interfere with steroid hormone production. For exam- ple, Subramanian et al. (1987) have shown a significant negative relationship between testosterone concentra- tion and DDE residue level. There is potential for using progesterone, testos- terone, and other hormonal parameters, such as chori- onic gonadotropin, FSH, and LH, in the estimation of sexual maturity and reproductive status in marine mammal populations. As we expand our reproductive database in the cetacea, we also need to correlate histological and morphological states with hormonal parameters. Research protocols which include 10-mL samples of fresh blood and assessment of hormonal state should be encouraged. Acknowledgments I thank L.L. Jones, L.M. Tsunoda, and personnel of the National Marine Mammal Laboratory, NMFS, Seattle, Washington for collection of blood samples and exchange of information. F.L. Moore of Oregon State University Department of Zoology provided laboratory space for RIA. L.L. Jones and two anonymous re- viewers made very helpful comments that greatly im- proved this manuscript. This study was made possible, in part, by support to the author from the Lutheran Brotherhood Medical Research Fund. Temte: Hormonal assessment of porpoise maturity 165 Citations DeMaster, D.P. 1984 Review of techniques used to estimate the average age at attainment of sexual maturity in marine mammals. In Perrin, W.F., R.L. Brownell Jr., and D.P. DeMaster (eds.), Reproduction in whales, dolphins, and porpoises, p. 175-180. Rep. Int. Whaling Comm., Spec. Iss. 6, Cambridge. Harrison, R.J.. and S.H. Ridgway 1984 Gonadal activity in some bottlenose dolphins (Tursiops truncatus). J. Zool. (Lond.) 165:355-366. Jefferson, T.A. 1989 Calving seasonality of Dall's porpoise in the eastern North Pacific. Mar. Mammal Sci. 5:193-195. Kasuya, T. 1972 Growth and reproduction of Stenella coeruleoalba based on the age determination by means of dentinal growth layers. Sci. Rep. Whales Res. Inst., Tokyo 24:57-79. Kirby, V.L., and S.H. Ridgway 1984 Hormonal evidence of spontaneous ovulation in captive dolphins, Tursiops truncatus and Delphinus del-phis. In Perrin, W.F., R.L. Brownell Jr., and D.P. DeMaster (eds.), Reproduc- tion in whales, dolphins, and porpoises, p. 459-464. Rep. Int. Whaling Comm., Spec. Iss. 6, Cambridge. Newby, T.C. 1982 Life history of Dall's porpoise (Phocoenoides dalli, True 1885) incidentally taken by the Japanese high seas salmon mothership fishery in the northwestern North Pacific and western Bering Sea, 1978 to 1980. Ph.D. thesis, Univ. Wash., Seattle, 155 p. Nickerson, D.M., D.E. Facey, and G.D. Grossman 1989 Estimating physiological thresholds with continuous two- phase regression. Physiol. Zool. 62:866-887. Perrin, W.F., and G.P. Donovan 1984 Report of the workshop. In Perrin, W.F., R.L. Brownell Jr., and D.P. DeMaster (eds.), Reproduction in whales, dolphins, and porpoises, p. 1-24. Rep. Int. Whaling Comm., Spec. Iss. 6, Cambridge. Sawyer-Steffan, J.E., and V. Kirby 1980 A study of serum steroid hormone levels in captive female bottlenose dolphins, their correlation with reproduction status, and their application to ovulation induction in captivity. U.S. Mar. Mammal Comm. Rep. MMC-77/22, 21 p. Sawyer-Steffan, J.E., V. Kirby, and W.G. Gilmartin 1983 Progesterone and estrogen in the pregnant and nonpreg- nant dolphin, Tursiops truncatus, and the effects of induced ovulation. Biol. Reprod. 28:897-901. Subramanian A., S. Tanabe, R. Tatsukawa, S. Saito, and N. Miyazaki 1987 Reduction of testosterone levels by PCBs and DDE in Dall's porpoises of northwestern North Pacific. Mar. Pollut. Bull. 18:643-646. Temte J.L., and S. Spielvogel 1985 Serum progesterone and reproductive status of inciden- tally killed female Dall porpoises. J. Wildl. Manage. 49(1): 51-54. Wells, R.S. 1984 Reproductive behavior and hormonal correlates in Hawaiian spinner dolphins, Stenella longirostris. In Perrin, W.F., R.L. Brownell Jr., and D.P. DeMaster (eds.), Reproduc- tion in whales, dolphins, and porpoises, p. 465-472. Rep. Int. Whaling Comm., Spec. Iss. 6, Cambridge. Yeager, D.P., and G.R. Ultsch 1989 Physiological regulation and conformation: A BASIC pro- gram for the determination of critical points. Physiol. Zool.62:888-907. Detecting Differences in Fish Diets David A. Somerton Honolulu Laboratory, Southwest Fisheries Science Center National Marine Fisheries Service, NOAA, 2570 Dole Street Honolulu, Hawaii 96822-2396 Statistical comparison of the diet of a predator between areas or time- periods allows one to distinguish true dietary differences from sam- pling variability and may lead to a better understanding of a species' feeding habits. Despite the utility of statistical testing, few procedures appropriate for dietary comparisons have been developed. Perhaps one impediment to the development of a general approach to dietary com- parisons is the wide variety of ways in which diets have been expressed and the lack of consensus about which is best. For example, diets ex- pressed as the numeric or gravi- metric proportions of the total food consumed will require different ap- proaches to statistical comparison than those expressed either as the proportion of the samples contain- ing each of the various prey types, or as the index of relative impor- tance of each prey type (Pinkas et al. 1971). For cases in which diets are ex- pressed in terms of gravimetric pro- portions, Crow (1979) and Ellison (1979) have recommended statis- tical tests of between-sample differ- ences based on multivariate analysis of variance (MANOVA). Validity of such tests, however, requires that the prey proportions have a multi- variate normal distribution and that the variance-covariance structure of the prey proportions is identical among samples (Morrison 1976). Recognizing that dietary data are unlikely to have these properties, Crow (1979) further recommended using MANOVA that incorporates non-parametric procedures. Herein, this recommendation is followed, and a new approach for testing dif- ferences in diets using non-para- metric MANOVA is examined. This approach combines the usual mea- sure of between-sample differences employed in parametric MANOVA (i.e., Hotelling's T 2 statistic; Mor- rison 1976) and a non-parametric procedure (i.e., a randomization test; Edgington 1987) to determine the significance of T 2 . The method is then applied to determine wheth- er the diet of pelagic armorhead Pseudopentaceros wheeleri from the Southeast Hancock Seamount changed between two sampling periods. Materials and methods Testing for between-sample differ- ences is accomplished in three steps: (1) calculating for each sample the gravimetric dietary proportions and their variances and covariances, (2) calculating a measure of the statis- tical difference between samples, and (3) determining the statistical significance of the measure. The gravimetric proportion of the diet contributed by prey category j(pj) can be estimated as the total weight of prey category j in all stomach samples divided by the total weight of all prey categories combined (Hyslop 1980). Algebraically this is expressed as W; Pj = 2jSkW ]k w. (1) where Wjk is the weight of prey category j for individual k, wj is the total weight of prey category j summed across all individuals, and w is the total weight of all prey. Each Pj is transformed to Xj, where xj = arcsin ^Pj , so that it conforms more closely to a normal random variable (Sokol and Rohlf 1969). Be- cause Xj is estimated as a pooled proportion rather than the average of the proportions for individual fish, the variance of Xj and the co- variance between Xj and x, cannot be calculated directly and instead are approximated by using the delta method (Seber 1973). In the follow- ing, X; indicates the vector of Xj for sample i, and S ; indicates the ma- trix of variance and covariance esti- mates for X;. The measure of statistical differ- ence used is the Hotelling's T 2 sta- tistic, a multivariate extension of the t- statistic (Morrison 1976). In matrix notation, this statistic is ex- pressed as T 2 = (x 1 -x 2 )'S.- 1 (xi-x 2 ), (2) where S. _1 is the inverse of the pooled estimate of the variance- covariance matrix (Morrison 1976). S. is approximated, assuming rea- sonably large sample sizes (>50 stomachs with prey per sample), as 2(N 1 S 1 + N 2 S 2 ) o. = - , (o) Nj+No where Nj and N 2 are the sizes of the two samples. Once a value of T 2 is computed, its significance is determined from an empirical probability distribution of T 2 computed by using a tech- nique known as randomization (Ed- gington 1987). Computation of the empirical probability distribution using this technique proceeds as follows: (1) stomach content data from both time or area samples are Manuscript accepted 12 October 1990. Fishery Bulletin, U.S. 89:167-169 (1991). 167 168 Fishery Bulletin 89(1). 1991 Table 1 Probability levels associated with the randomization tests of the individual hypotheses. Because these are a posteriori tests, significance levels were adjusted according to the Bonferroni inequality to 0.2 5cr (*P< 0.0125, "P<0.0025). Mean proportion Probability Sample 1 Sample 2 t- value level Tunicates Crustaceans Fish "Others" 0.32 0.69 0.27 0.13 0.26 0.10 0.15 0.08 -9.i 200m depth; chi-square analysis). For the comparison, we used specimens ob- tained between the hours of 0600-1000 and 1400-1900 (we chose 0600 and 1900 because they were the start and the termination of sampling, and we arbitrarily selected 1000 and 1400). Results Four of the nine females (nos. 195, 566, 651, 695) in- strumented with TDRs and radio transmitters were deep divers, diving at all hours of the day; the other five (nos. 214, 308, 713, 767, 854) were shallow divers and dived primarily at night (Fig. 2). Seven of these females, three deep divers and four shallow divers, were located at sea at least once. Female 195 was located on the third day of a 12-day trip at the continental shelf break. At this location the bathymetry changes rapidly, and because aerial loca- tions are only approximate this female could have been Goebel et al.: Diving patterns and foraging locations of female Callorhinus ursmus 173 T Figure 2 Three-dimensional plots of diving records for nine female northern fur seals at St. Paul Island, Alaska, in July and August 1985. Each plot represents the diving record (all dives >20m in depth) of one trip to sea for each female. in water 140 m to 1400 m deep. On the day she was located, as well as on most days of her trip, she ex- hibited the deep-diving pattern. For a 24-hour period from 1200 on 4 August until 1200 on 5 August she made 70 dives (50% during the day) with a mean depth of 94.2m (SD 18.5). Two females (651 and 767) were located only once, on their return to St. Paul from a foraging trip (Fig. 1, Table 1). Female 651 was located 2.5 hours before she returned ashore and after she had completed 99% of her foraging trip. Her last dive was 6.9 hours before being located. Female 767 was located 8.2 hours before she returned ashore and 2.5 hours after her last dive. Female 695 was located on three separate days, each time she was 110-160km northwest of St. Paul Island, diving to depths greater than 100 m over the continen- tal shelf in water 1 10-140 m deep. This female ex- hibited the deep-diving pattern every day of her 10-day trip, but also made seven dive bouts which had mean dive depths of less than 75 m, five of which occurred during the night. Female 214 was located on the fifth day of a 12-day feeding trip in approximately 3500 m of water. On that day, as well as most other days, she exhibited the shallow diving pattern and dived primarily at night (Fig. 3). Female 713 was also located on three consecutive days. She exhibited a foraging pattern of deep diving, day and night, on the first and last days of a foraging trip and shallow diving at night on other days. She was located on 25 July, the fifth day of a 7-day trip, 80km south of St. Paul Island at the continental shelf break in approximately 200 m of water. On that day she dived exclusively at night with a mean depth of dive of 46.7 m (SD 17.1, n 171). This was the night following her loca- tion at the shelf break. On the sixth day and on the last day of her trip she was located over the shelf and dived both during the day and night to much greater depths (mean depth of dive 101.3m, SD 26.9, n 67 for the sixth day, and 103.5m, SD 11.0, n 11 for the last day). Female 854 was located on the first day of a 4-day trip over the continental shelf but within 18 km of the 200-m depth contour. From the time she left shore (2208) until she was located 20.8 hours later she had made 22 dives— all but three were greater than 75 m (x 75.6, SD 7.6). Of these dives, 77% were made be- fore dawn. Her second night at sea was spent actively diving to depths of less than 35 m (n 70, x 22.8, SD 19.6). She began diving at 2353 and dived continuous- ly until 0340. This female was in transit across the con- tinental shelf during the night, making occasional dives 174 Fishery Bulletin 89(2). 1991 Table 1 Mean depth (m) of dive for a 24-hour period (1200-1200) on the date of locatior radio transmitters and time-depth recorders, St. Paul Island, Alaska, 1985. for seven female northern fur seals equipped with Female ID Date Time % time into trip Depth of dive 1 Nearest dive 2 Approximate depth at location X SD N Depth Time Deep divers 195 8/04 1411 21.7 *■ 94.2 18.5 70 70.5 + 1.63 140-1400 651 7/28 1613 99.0 - - - 71.5 -7.10 50 695 7/28 1340 47.8 107.6 28.4 85 131.5 -0.45 120 7/29 1231 56.9 103.0 36.9 51 104.0 + 0.07 110 7/30 1217 66.4 118.1 10.3 52 137.5 -0.05 120 Shallow divers 214 8/04 1313 39.0 35.5 20.9 159 85.6 -1.57 3500 713 7/25 1758 69.7 46.7 17.1 171 60.6 + 6.35 200 7/26 1414 80.9 101.3 26.9 67 89.5 + 7.40 110 7/27 1717 96.1 103.5 11.0 11 99.5 -1.47 80 767 7/26 1113 72.8 — — — 45.8 -2.67 20 854 8/03 1855 23.1 39.3 18.2 d 651 had no dives during 1200-1200 on the date of location, e dive closest to the time of location and the number of hours 78 since 71.5 the time of locat -1.23 on. 120 'Females 767 ar - Depth (m) of th Time of day 12 12 12 12 12 12 12 12 12 12 12 12 12 12 9 566 August 6 15 Zr I 'i i "'111 i ! wilHIir r*~ n» up w "'i' ii i i r g> 214 July 30-August 12 i r 120 ~~f fr |»| v^n fi i V i II r INI I M 9 195 August 1 14 »j "'"WTimn fi mil n'i i"m n n mr 9 695 July 23 August 3 120 n nr"T iFiiiir | Mil I "HP! I pr i ' 9 308 July 30 August 3 ' !"» 9 651 July 20 28 9 713 July 20 27 1111111' '^ T nilflP Jf'JT PHI"H 120 9 854 August 2 6 ti n 'iir Dm 120 m m-l 9 767 July 22 26 mr ■-' i l T Figure 3 Two-dimensional plots of diving records for nine female northern fur seals at St. Paul Island, Alaska, in July and August 1985. Each plot represents the diving record (all dives >20m in depth) of one trip to sea for each female. Goebel et al.: Diving patterns and foraging locations of female Callorhmus ursinus 175 Table 2 Record lengths, times to first and from last diving bouts, and mass of nine female northern fur seals instrumented with time depth recorders in July and August 1985, St. Paul Island, Alaska. Start Time to Time from Mass (ke) 3 fir^t last dive bout 2 Female ID Date Time Total hours dive bout 1 1 2 Gain Deep divers 195 8/01 2308 290.22 4.72 12.25 34.0 37.6 3.6 566 8/06 1425 225.25 34.08 22.67 41.3 45.4 4.1 651 7/20 1442 196.33 14.72 12.75 31.3 33.6 2.3 695 7/23 1204 253.17 20.52 18.18 48.5 49.9 1.4 Shallow divers 214 7/30 1820 295.42 11.87 14.63 31.3 33.1 1.8 308 7/24 2042 242.50 2.98 15.13 — 39.5 ? 713 7/20 1342 178.28 23.03 8.12 — 34.0 i 767 7/22 0041 114.75 18.98 18.73 39.5 — 1 854 8/02 2208 3 bout is the time from 91.01 4.65 departure from shore until the first dive 3.00 bout measured in 40.1 40.4 0.3 hours. A dive bout is defined as 'Time to first div< any series of dives greater than five dives with surface intervals of less than 40 minutes (Gentry et al. 1986c). -Time from the last dive bout (in hours) until the female returns to shore. 'Mass 1 is the mass taken upon deployment (the first capture , and mass 2 is the mass taken upon recapture (the second capture). to depths greater than 75 m and then once near the shelf break, she began diving to shallow depths. Female mass and depth of dive Seven of the nine females were weighed at instrument deployment, and eight were weighed when the instru- ment was recovered (Table 2). All females weighed twice showed a mass gain (range 0.3-4.1 kg). Time at sea ranged from 91 to 295 hours. Length of time at sea was not correlated with diving patterns or mass of the female (r 2 0.0). No correlation was found be- tween mass of individuals and their diving patterns (r 2 0.0). The mean maximum depth of dive for each deep diver was greater than 75 m whether considering day dives or night dives (Table 3). The mean maximum depth of dive for the shallow-diving females was less than 75 m (Table 3). When considering only night dives, shallow- diving females showed mean maximum depths of less than 50 m. Two of the deep divers (195 and 651) showed no dif- ference in depth of dives between day and night. The other seven females showed a significant difference in depth of dives between day and night. Six females dived deeper during the day than at night. The seventh (566) had significantly deeper dives at night than dur- ing the day (Table 3). Time from shore to the first dive bout and return The time between departure from shore and the first dive bout was highly variable, as was the time between the last dive bout and the return to shore (Table 2). These times ranged from about 3 to 34 hours. Mean time to first dive bout was 15.06 hours (SD 10.26) and the mean time from last dive bout to shore was 13.94 hours (SD 5.88). There was no significant difference between the time to first dive bout and the time from the last dive bout (Mann-Whitney P = 0.86). There were no differences in either the time to the first dive bout or the time from the last dive bout when comparing shallow divers with deep divers. The mean time to the first dive bout was 12.30 hours (SD 8.74) for shallow divers and 18.51 hours (SD 12.26) for deep divers (Mann- Whitney P = 0.46). The mean time from the last dive bout to shore for shallow divers was 11.92 hours (SD 6.29) and for deep divers was 16.46 hours (SD 4.93; Mann-Whitney P = 0.54). Foraging patterns Deep divers showed a much lower percentage of dives at night than the shallow divers (Table 4). Females 566 and 695 made less than 50% of their dives at night. In general, shallow divers made greater than 70% of their dives at night (females 214 and 713 >90%) and deep divers less than 70%. 176 Fishery Bulletin 89|2). 1991 Table 3 Mean depth of dive for each dive record and for each record by day and night of nine female northern fur seals from St. Paul Island, Alaska, during July and August 1985. Female ID Record total Day hours Night hours P- value 1 Mean depth SD N Mean depth SD N Mean depth SD N Deep divers 195 101.5 22.0 440 H)2.1 18.3 205 101.0 24.8 235 0.590 566 75.3 31.2 439 72.4 28.0 285 80.8 35.9 154 0.007 651 76.0 27.3 496 75.3 27.0 193 76.5 27.4 303 0.632 695 85.1 34.1 504 93.0 30.8 259 76.7 35.3 245 0.001 Shallow divers 214 38.7 23.1 1369 87.1 15.9 134 33.5 16.9 1235 0.001 308 60.6 42.1 832 94.8 41.6 230 47.5 34.3 602 0.001 713 45.4 21.7 900 85.1 25.4 55 42.8 18.7 845 0.001 767 33.3 27.7 431 72.9 22.6 90 22.8 17.6 341 0.001 854 40.6 28.2 one-way 312 63.6 ANOVA for differences in 15.2 80 32.6 27.3 mean depth of dive between day and night 232 hours. 0.001 'P- values are derived from a Table 4 Total dives and dives/hour for day and night di /es of nine female northern fur seals from St. Paul Island, Alaska, during July and August 1985. Female ID \ dives % total dives Dives/hour Day Night Total Day Night Deep divers 195 440 46.6 53.4 1.8 1.1 3.5 566 439 65.1 34.9 1.9 1.7 2.8 651 496 38.9 61.1 2.5 1.3 6.3 695 504 51.4 48.6 2.0 1.3 4.0 Shallow divers 214 1369 9.8 90.2 4.7 0.6 16.8 308 832 27.6 72.4 3.4 1.3 10.0 713 900 6.1 93.9 5.0 0.4 20.1 767 431 20.9 79.1 3.8 1.0 14.6 854 312 25.6 74.4 3.4 1.2 9.7 The number of dives/hour for the entire trip was greater for shallow divers than for deep divers (Mann- Whitney, P = 0.02). The maximum number of dives/ hour for deep divers was 2.5 (x 2.0, SD 0.31), where- as the minimum number of dives/hour for shallow divers was 3.4 (x 4.1, SD 0.75). Although deep divers tended to execute more dives per hour at night than during the day, the difference was not significant (Mann- Whitney, P = 0.11). Shallow divers did, how- ever, make more dives per hour at night than during the day (Mann- Whitney, P = 0.02). Regardless of which diving pattern females exhibited, they always made more dives per hour at night than during the day. Diving bouts No differences in the total number of dive bouts per trip were detected between shal- low and deep divers (Table 5). Although the number of dives that did not meet the dive- bout criteria tended to be greater for the deep divers, this difference was not signifi- cant (Mann- Whitney, P = 0.07). Differences in bout durations, number of dives per bout, and number of dives per hour within bouts were all greater for shallow divers (ANOVA, P = 0.01 for duration, P = 0.01 for dive num- ber, P = 0.02 for dives/hour). When all bouts were divided into those less than 75 m and those greater than or equal to 75 m, the percent of time spent below the surface was significantly differ- ent for shallow- and deep-diving bouts (ANOVA, P = 0.01). A greater percent of time was spent below the surface for shallow bouts (<75m, x 0.35 hour, SD 0.11, n 59) than for deep bouts (>75m, x 0.29 hour, SD 0.12, n 102). Pelagic fur seal studies During the hours 0600-1000, the stomachs of 740 post- partum females were sampled, 504 over the shelf and 236 off the shelf (Table 6). From 1400 to 1900 hours, 750 postpartum females were sampled, 513 over the shelf and 237 off the shelf. Females collected in the morning hours off the con- tinental shelf were more likely to have food in their Goebel et al.: Diving patterns and foraging locations of female Callorhmus ursinus 177 Table 5 Dive bout 1 statistics and effort within dive bouts for nine female northern fur seals from St. Paul Island, Alaska, in July and August 1985. Dives excluded (%) Bout duration (hours) Dives per bout Dives/hour in bout Female ID Dive bouts X SD Max. X SD Max. X SD Max. Deep divers 195 26 4.8 2.5 2.3 9.2 16 13 51 8.9 5.7 28.0 566 24 10.5 2.8 2.0 7.9 16 13 54 6.6 4.2 16.7 651 15 4.4 3.8 3.0 11.7 32 26 93 8.4 4.0 16.2 695 26 9.3 2.7 1.8 7.7 18 14 63 7.0 4.2 17.8 Shallow divers 214 20 3.0 3.8 2.2 8.3 66 65 163 13.5 10.3 • 32.3 308 25 4.9 3.0 1.8 7.7 32 37 135 10.5 10.0 35.7 713 9 1.8 4.6 2.4 8.2 98 75 184 20.1 12.8 33.3 767 7 3.2 4.2 2.9 7.8 60 55 145 12.8 6.7 24.4 854 9 defined as any 2.6 3.0 1.9 8.6 34 24 series of dives greater than five dives with surface intervals 74 13.2 9.5 30.3 of less than 40 minutes (Gentry et al. 'A dive bout is 1986c). Table 6 The percent of stomachs with food of lactating northern fur seals collected over the continental shelf (in water <200m deep) and off the continental shelf (> 200 m deep). Collections were made from July through September. 1958-74. Time (hours) Over the shelf Off the shelf 0600-1000 1400-1900 80.0 (403/504) 54.0 (277/513) 95.3 (225/236) 39.7 (94/237) stomachs than females collected over the shelf (chi- square, P<0.001). Afternoon samples showed a reverse trend with a higher incidence of food in stomachs of seals collected on the shelf than off the shelf (chi- square, P< 0.001). Discussion Dive patterns and foraging locations The results of this study show that northern fur seal females diving in deep water beyond the continental shelf primarily exhibit the shallow-diving pattern and dive predominantly at night. Females feeding at or near the shelf break may exhibit both diving patterns. Females located on the continental shelf (which had not already completed their feeding trip) were more likely to exhibit the deep-diving pattern and dived to depths of greater than 75 m throughout the day and night. However, females found over the continental shelf had, at times, shallow dive bouts (<30m) at night. The pelagic fur seal database provided insight into the relationship between diving patterns and feeding locations of fur seals in the Bering Sea. If the deep- diving pattern is associated with feeding on the con- tinental shelf, and deep divers feed during the day, one can test the hypothesis that females found over the con- tinental shelf during the day have a greater probabil- ity of having food in their stomachs than those females found off the shelf, which would be diving predominant- ly at night. The analyses of the northern fur seal pelagic database support our finding that the deep-diving pat- tern is associated with the continental shelf by show- ing that females collected over the continental shelf were more likely to be feeding during the day. Dive patterns and probable prey The feeding locations and dive patterns observed in this study are consistent with the known distribution of fur seal prey items. Kajimura (1984) summarized the varia- tion in principal forage species for fur seals, depending on location. Fur seals feeding in the Bering Sea beyond the continental shelf over deep water fed on oceanic squid of the family Gonatidae (primarily Goncitus spp., Berryteuthis magister, and Gonatopsis borealis) or deep-sea smelts of the family Bathylagidae. These prey species, and fish with swim bladders, exhibit diel ver- tical migration and are at relatively shallow depths at night as they move vertically in synchrony with the deep scattering layer (Roper and Young 1975, Pearcy et al. 1977). It is during the night that they are fed upon by fur seals which rarely dive beyond 200 m (Gentry Fishery Bulletin 89(2). 1991 et al 1986b). Fur seals foraging over the shelf were likely to feed on walleye pollock Theragra chalco- gramma, Pacific herring Clupea harengus pallasi, and capelin Mallotus villosus (Kajimura 1984). Each of these prey items is distributed throughout the water column over the shelf, depending on the sex and age of the individual and time of day; however, they are principally found near the bottom (Bakkala and Waka- bayashi 1985). Even when prey are near the bottom over most of the shelf floor, they are shallower than the maximum diving depths observed for most fur seals and are accessible during all hours of the day. The results of foraging effort for shallow and deep divers in this study are consistent with the results of Gentry et al. (1986c) in their study of diving of females from St. George Island. Costa and Gentry (1986) used isotopic turnover methods to measure food intake and metabolic rate in female fur seals instrumented with time-depth recorders. In that study of two deep-diving and two shallow-diving fur seals, the deep divers ate less food and expended less energy but gained similar body mass on a single trip to sea. Our results confirm their conclusion. In this study, using four measures- bout duration, dives per bout, dives per hour, and time spent below the surface-deep divers expended less effort on foraging than shallow divers. Deep divers apparently obtain greater energy per dive. The difference in foraging effort of diving types may be explained by differences in energy content of prey, success rate for prey capture, and/or average size of prey captured. If success rate per dive and average size of prey were similar for both diving types, then energy content for the prey of deep divers would have to be greater. If prey were of similar size and energy content, then the prey of deep divers would have to be easier to capture. If success rate and energy content were similar, then the prey captured by deep divers would have to average a greater size per dive. Unfortunately, no data exist on success of individual dives. Data on the energy content of prey items of northern fur seals suggest that Gonatus sp. and walleye pollock have similar energy content (M.A. Perez and T.R. Loughlin, NMFS Natl. Mar. Mammal Lab., Seattle, unpubl. data). No data exist for energy content of bathylagids. Perez and Bigg (1986) report on the size range of prey found in the stomachs of northern fur seals collected during 1958-74: walleye pollock, 4-40cm (1721 prey from 71 stomachs); gonatid squid, 5-24 cm (>59 prey from 10 stomachs); deep-sea smelts, 8-12cm (986 prey from six stomachs). The size range for Myctophiform fish prey and the mean size of prey were not reported. Transit times The most frequently used measure of transit time, derived from time-depth records, has been the time from shore to the first diving bout and from the last dive bout to shore (Gentry and Kooyman 1986). Data from the current study show that female fur seals fre- quently feed while traveling to very distant feeding locations and that the time from shore to the first diving bout is only a subset of the total time in transit. There was no difference in the time from shore to the first dive bout for the two types of diving patterns observed in this study. Similarly, the times from the last dive bout to arrival ashore were not different. Fur- thermore, no correlation was found between these times and' feeding trip duration. It has been suggested, however, that a correlation exists between feeding trip duration and distance to feeding area (Gentry et al. 1986a). Gentry et al. (1986a) found that regressing transit time of individuals upon their trip duration resulted in a poor fit (r 2 0.357). When they compared species averages using two tropical and two sub-polar otariid species, the correlation was much greater (r 2 0.761). They concluded that transit time may largely determine trip duration for the species, but its effect is partly obscured by the large variation in transit times and trip duration of some individual animals. The results of Loughlin et al. (1987) and of this study show that females feed while in transit to primary foraging areas. A correlation may exist between trip duration and total time spent in transit; however, measuring the time from shore to the first diving bout and from the last diving bout to shore is only a subset of actual time spent in transit and therefore an inade- quate measure of total transit time. Without knowledge of either the swim velocity or the location of females (either through radio-tracking or with some instrument carried by the animal to record location) it is not possi- ble to discern from a record of a time-depth recorder the actual time spent in transit. Classification of diving patterns It is important to point out that any classification of diving patterns gives the impression that they are more discrete than they really are. It is more accurate to view any particular diving record as fitting into a continuum from strictly shallow diving at night to exclusively deep diving at all hours. The terminology used to categorize these diving patterns may overemphasize the impor- tance of depth. It should be remembered that though deep- and shallow-divers are classified as such, the depth of dives may not be as important for identifying the pattern as the time of day in which diving occurs. Goebel et al Diving patterns and foraging locations of female Callorhmus ursinus 179 Before the advent of dive-recording instruments, one of the most important measurements used to assess changes in prey availability for pinnipeds and sea birds was the length of foraging trips. Foraging trips, how- ever, may be useful only in documenting large-scale changes in prey availability (e.g., El Nino events). Dive- recorder instruments provide the opportunity to measure changes in the foraging environment on a finer scale. This work underscores the importance of quantifying dive patterns in elucidating differences in foraging strategies and in changes in the foraging environment. Acknowledgments We wish to thank the following people for their gen- erous assistance in the field: P. Dawson, H. Kajimura, R. Merrick, Y. Nomura, and K. Yoshida. Special thanks are extended to R. Merrick, who was instrumental dur- ing the aerial tracking of animals at sea, and C. Fish, the pilot during aerial surveys. We thank the follow- ing for their valuable comments on the manuscript: G. Antonelis, H. Huber, L. Jones, W. Roberts, E. Sin- clair, and A. York. Citations Bakkala, R.G., and K. Wakabayashi (editors) 1985 Results of cooperative U.S. -Japan groundfish investiga- tions in the Bering Sea during May-August 1979. Int. North Pac. Fish. Comm. Bull. 44. Costa, D.P., and R.L. Gentry 1986 Free-ranging energetics of northern fur seals. In Gen- try, R.L.. and G.L. Kooyman (eds.), Fur seals: Maternal strategies on land and at sea, p. 41-60. Princeton Univ. Press, Princeton, NJ. Gentry, R.L., and J.R. Holt 1982 Equipment and techniques for handling northern fur seals. NOAA Tech. Rep. NMFS SSRF-758, 15 p. 1986 Attendance behavior of northern fur seals. In Gentry, R.L., and G.L. Kooyman (eds.), Fur seals: Maternal strategies on land and at sea, p. 41-60. Princeton Univ. Press. Princeton, NJ. Gentry, R.L., and G.L. Kooyman 1986 Methods of dive analysis. In Gentry, R.L., and G.L. Kooyman (eds.), Fur seals: Maternal strategies on land and at sea, p. 28-40. Princeton Univ. Press, Princeton, NJ. Gentry, R.L., D.P. Costa, J.P. Croxall, J.H.M. David, R.W. Davis, G.L. Kooyman, P. Majluf, T.S. McCann, and F. Trillmich 1986a Synthesis and conclusions. In Gentry, R.L., and G.L. Kooyman (eds.), Fur seals: Maternal strategies on land and at sea, p. 209-219. Princeton Univ. Press, Princeton, NJ. Gentry, R.L., M.E. Goebel, and W.R. Roberts 1986b Behavior and biology, Pribilof Islands, Alaska. In Kozloff, P. (ed.), Fur seal investigations, 1984, p. 29-40. NOAA Tech. Memo. NMFS F/NWC-97, Natl. Mar. Mammal Lab., Seattle 98115-0070. Gentry, R.L., G.L. Kooyman, and M.E. Goebel 1986c Feeding and diving behavior of northern fur seals. In Gentry, R.L., and G.L. Kooyman (eds.), Fur seals: Maternal strategies on land and at sea, p. 61-78. Princeton Univ. Press, Princeton, NJ. Kajimura, H. 1984 Opportunistic feeding of the northern fur seal, Callo- rhinus ursinus, in the eastern North Pacific Ocean and eastern Bering Sea. NOAA Tech. Rep. NMFS SSRF-779. 49 p. Kooyman, G.L., R.L. Gentry, and D.L. Urquhart 1976 Northern fur seal diving behavior: A new approach to its study. Science (Wash. DC), 193:411-412. Loughlin, T.R., J.L. Bengtson, and R.L. Merrick 1987 Characteristics of feeding trips of female northern fur seals. Can. J. Zool. 65:2079-2084. Pearcy, W.G., E.E. Krygier, R. Mesecar, and F. Ramsey 1977 Vertical distribution and migration of oceanic micronekton off Oregon. Deep-Sea Res. 24:223-245. Perez, M.A., and M.A. Bigg 1986 Diet of northern fur seals, Callorkinus ursinus, off western North America. Fish. Bull., U.S. 84:959-973. Roper, C.F.E., and R.E. Young 1975 Vertical distribution of pelagic cephalopods. Smithson. Contrib. Zool. 209:1-51. Abstract.- In Chesapeake Bay in June, the predatory lobate cteno- phore Mnemiopsis leidyi and the eggs of the bay anchovy Anchoa mit- chilli typically reach seasonal and localized abundance together. When examined at small vertical (1-3 m), horizontal (10-50m), and temporal (6-hour) scales, the co-occurrence of M. leidyi and fish eggs (32.3-74.2% of which were A. mitchilli) was great- est in the northern reaches of the mouth of Chesapeake Bay, where the water column was well mixed, than in the southern reaches where the water column was stratified. Stratifi- cation to the south was effected by the Chesapeake Bay plume. With es- timates of ctenophore clearance rate reported elsewhere and observed densities of ctenophores and fish eggs, potential predation was judged to be greatest in the northern reaches of the Bay mouth. The observation that co-occurrence and potential pre- dation are greatest in areas where Chesapeake Bay water mixes with coastal shelf water implies that those fishes that spawn in low-salinity sur- face waters of well-stratified water columns may afford protection of their eggs from ctenophore predation. Potential Predation on Fish Eggs by the Lobate Ctenophore Mnemiopsis leidyi Within and Outside the Chesapeake Bay Plume' John J. Govoni Beaufort Laboratory, Southeast Fisheries Science Center National Marine Fisheries Service. NOAA, Beaufort. North Carolina 28516 John E. Olney Virginia Institute of Marine Science, School of Marine Science College of William and Mary. Gloucester Point, Virginia 23062 Predation is probably the leading cause of mortality for fertilized fish eggs and yolksac larvae because star- vation is not relevant for these early- life-history stages and because the short duration of egg incubation and yolk absorption for most teleosts limits transport to areas inimical to development (Bailey and Houde 1989). Assessments of the impact of predation on cohorts of fish eggs and larvae in the ocean, however, have been hindered by three problems: two practical, the third inferential. Eggs and larvae leave little identifi- able residue in the guts of predators, and, as a result, direct estimates of the extent of predation are difficult. Predators and prey, moreover, are concentrated together in collecting devices, a situation that can result in artifically high feeding rates and in- flated estimates of predation. Last- ly, predation is often spuriously in- ferred from the inverse abundance of predators and prey, when presence and absence may actually reflect spatial and temporal segregation rather than removal of prey by pred- ators. Such misinterpretations result from failure to consider the small- scale temporal and spatial distribu- Manuscript accepted 12 December 1990. Fishery Bulletin, U.S. 89:181-186 (1991). * Contribution no. 1635 of the Virginia Insti- tute of Marine Science and School of Marine Science, College of William and Mary. tion of predator and prey in differing water masses (Frank and Leggett 1982, 1985). Among the known invertebrate predators of fish eggs and larvae, coelenterates and ctenophores are likely candidates for significant pre- dation because of their high rates of ingestion and population growth (Al- ldredge 1984, Purcell 1985, Monte- leone and Duguay 1988). Lobate ctenophores, in particular, are major predators of small zooplankton of limited mobility (Kremer 1979, Pur- cell 1985, Monteleone and Duguay 1988). They capture prey by pump- ing water past lobes lined with mucus and secondary tentacles (Larson 1988), a feeding mechanism that is seemingly well suited for the capture of fish eggs. In Chesapeake Bay, a lobate cteno- phore Mnemiopsis leidyi and the eggs of the bay anchovy Anchoa mit- chilli reach seasonal and localized abundance together, thereby provid- ing a predator and prey pair that is ideal for an evaluation of potential predation. Mnemiopsis leidyi is pres- ent from late fall through midsum- mer, and episodically explodes in abundance between May and July (Bishop 1967, Miller 1974, Kremer and Nixon 1976, Mountford 1980). Mnemiopsis leidyi can exhibit ap- preciable predation on fish eggs (A. 181 182 Fishery Bulletin 89(2). 1991 MNEMIOPSIS LEIDYI DISTRIBUTION FISH EGG (ALL SPECIES) DISTRIBUTION 1 1 \ ;i i y w Chesapeake Bay ^ (27)»E4 50' 75° 40' 75° 30' (1)»E1 '37° 00'" (0)»FT4 ^f^\-J Atlantic Ocean 36 _ 00' . . A . . ''i i 1 \ rn v^ 1 1 1 i 1 — 76°00'«J. . ^5° 50' 75° 40' 75° 30' Chesapeake Bay Ca (32)»E4 y- 37 c 00' (28)»FT4 Atlantic Ocean 36 _ 00' 1 I Figure 1 Positions of stations and mean station densities (numbers/m :i ) of Mnemiopsis leidyi and fish eggs at the mouth of Chesapeake Bay. Dotted line indicates position of the Chesapeake Bay plume. mitchilli) in the laboratory (Johnson 1987, Monteleone and Duguay 1988), but while it consumes some fish lar- vae in Chesapeake Bay (Burrell and Van Engel 1976), its predation on fish eggs in the field is not documented. Anchoa mitchilli spawns in the Bay in spring and sum- mer and its eggs typically account for over 90% of all fish eggs present between May and August (Olney 1983). The mouth of the Bay is characterized by water masses that differ spatially in both the vertical and horizontal dimensions (Boicourt et al. 1987) and pro- vides hydrographic structure capable of shaping the spatial distribution of planktonic animals. Its complex hydrography is dominated by a buoyant plume char- acterized by a horizontal scale of 10-100km, a vertical scale of 5-20m, and a temporal scale of 1-10 days (Boicourt et al. 1987). As a result, the small-scale ver- tical and horizontal distributions of predator and prey can be observed synoptically in water columns of dif- ferent structure within a confined study area. Here we describe the small-scale spatial and temporal co-occurrence of M. leidyi and fish eggs at the mouth of Chesapeake Bay and assess potential predation. Methods Sampling protocol Three stations were allocated across the mouth of Chesapeake Bay with two additional stations on the continental shelf (Fig. 1) such that some stations were within and others outside of the typical boundaries of the Chesapeake Bay plume (Boicourt et al. 1987). Each station was occupied for 30 hours between 11 and 21 June 1985 (the sampling period at station El was in- terrupted for 24 hours by vessel failure). At each sta- tion, hydrographic profiles (temperature, salinity, and specific gravity anomaly o t ) and plankton collections at three nominal depths (surface, within the pycnocline, and below the pycnocline) were obtained once at four diel intervals (dawn, noon, dusk, and midnight). Fish eggs and ctenophores were collected with a 1-m Tucker trawl equipped with three 202-/^m mesh nets, General Oceanic flow meters, and an Applied Microsystems Limited temperature, salinity, and depth recorder and towed at approximately lOOcm/second. Nets were opened at depth and fished along a horizontal trajec- tory for 30-60 seconds each; for subsurface strata, the trawl was lowered while the vessel was stopped and its nets were fished along a horizontal trajectory at depth. The trawl was positioned at nominal depth strata by the trigonometry of the warp angle and length. Triplicate samples were obtained at the surface; duplicate, discrete-depth samples were obtained within and below the pycnocline. With these sampling pro- cedures, the trawl sampled on small vertical (1-3 m) and horizontal (10-50m) scales. All plankton collections were passed through a 6.4-mm mesh screen to separate ctenophores from ichthyoplankton. Ctenophores retained on this screen were fixed to prevent dissolution following the methods of Gosner (1971), then rinsed and preserved in 5% Govoni and Olney Predation on fish eggs by Mnemiopsis leidyi in and around Chesapeake Bay 183 formalin solution. Ichthyoplankton was preserved in either 5% formalin or 95% ethanol. All M. leidyi and fish eggs were counted except in those samples of ex- ceptionally high ctenophore volume, where ctenophore number was estimated by volumetric subsampling and multiplication. Counts of ctenophores and fish eggs were averaged for replicate collections taken at a depth stratum and diel interval. Estimation of co-occurrence Our intention was to assess the small-scale co-occur- rence of ctenophores and eggs relative to the water masses overlying these stations and to then evaluate potential predation. Because the depth of each sample occasionally varied from the nominal and the trawl con- sequently fished through hydrographic discontinuities, some collections were omitted from consideration. Col- lections omitted were those in which salinity values, recorded during each 30-60 second fishing interval, varied outside a range of 1.5%o. This procedure elim- inated seven of 35 collections at station El and none at E4, the two stations where ctenophores and fish eggs were consistently present and where we focused our assessment of potential predation. Estimates of potential predation We estimated potential predation, for each depth and diel interval, as the product of clearance rate (the volume of water cleared of all prey per unit time per ctenophore), times the end points of the range of den- sity of ctenophores (the number of ctenophores per unit volume, averaged for replicates), times the end points of the range in density of fish eggs (again averaged for replicates). A clearance rate of 168L/day was used from Monteleone and Duguay (1988), who found that the clearance rate of fish eggs was independent of egg density (as well as the presence of alternate prey) and was positively and linearly related to experimental vessel size. This clearance rate was the highest rate observed for ctenophores 4.5-5.0 cm in length feeding in the largest vessels employed and falls roughly within the range of values reported elsewhere (Larson 1987). A sample of 10 preserved ctenophores from our col- lections averaged 8.5mL in volume which converts to an average length of 4 cm (Kremer and Nixon 1976). We did not account for shrinkage. Results Distribution and co-occurrence Mnemiopsi leidyi and fish eggs were consistently pres- ent only at stations El and E4 (Fig. 1). Pulses in den- sities of M. leidyi were evident, but did not conform to specific diel intervals or tidal phases (Figs. 2, 3). Egg density showed a diel pattern, with peak densities from dusk to dawn. Eggs of Anchoa mitchilli accounted for an average of 74.2% (range 23.0-98.5%) of the fish eggs at station El and 32.3% (range 0-62.9%) at E4. Mnemiopsis leidyi and fish eggs were, for the most part, vertically segregated at station El, but co- occurred, particularly in surface water, at E4. Vertical segregation at El (Fig. 2) reflected the physical stratification of the water column with a warm, low- salinity, surface-layer characteristic of the Chesapeake Bay plume overlying a cool, higher-salinity, bottom- layer characteristic of coastal shelf water (Boicourt et al. 1987). At El, in the southern reaches of the mouth of the Bay, surface collections within the plume yield- ed higher egg densities, while subsurface collections yielded higher M. leidyi densities. Station E4, in the northern reaches and outside the plume, was unstratified with no thermo-, halo-, or pycnocline (Fig. 3). Water at this station apparently was a mixture of Chesapeake Bay water and coastal shelf water, likely the result of tidal, rather than wind, mixing. Winds, often responsible for mixing at the mouth of the Bay (Ruzecki 1981), were light to moderate during this sampling period (l-8m/second). Potential predation Overall, potential predation was greater in the un- stratified northern reaches of the mouth of the Bay out- side the plume (E4) than in the southern reaches stratified by the plume (El), because of greater tem- poral and spatial co-occurrence of M. leidyi and fish eggs there. Range estimates of potential population predation were 0.1-14.7 eggs per m 3 /day at El, and 0-174.3 at E4 (Table 1). Discussion The assessment of ichthyoplankton predation in the field has been based historically on the examination of predator gut contents or on the strength of a negative correlation of predator and prey densities, even though biases may result from the lability of fish eggs and lar- vae in the guts of predators, from the feeding of pred- ators within the collecting device used to sample predator and prey (Purcell 1985), and from the spurious inference of cause and effect drawn from correlation analysis (Frank and Legget 1982, 1985). Few have resolved successfully the first two problems (Bailey and Houde 1989, Purcell 1989, Purcell and Grover 1990). In regard to the latter, the importance of small-scale spatial and temporal distribution of predator and prey 184 Fishery Bulletin 89[2), 1991 i Q- 10 15 in evaluating predation is apparent across the mouth of Chesapeake Bay. Potential predation in the southern reaches where the Chesapeake Bay plume overlays coastal shelf water was low because of the relative lack of ver- tical co-occurrence there. In the north- ern reaches where the water column was well mixed, M. leidyi and fish eggs co-occurred in a more or less well-mixed water column, and as a result our estimates of potential predation were high. The application of parameter esti- mates derived from laboratory preda- tion experiments to the evaluation of the impact of gelatinous planktivores on their prey in nature, an approach that avoids field sampling errors, has other pitfalls (Purcell 1985). These problems relate to the unrealistic confines of ex- perimental vessels, which constrain movement and small-scale hydrodynam- ics, and to unnaturally high experimen- tal densities of predator and prey (Sulli- van and Reeve 1982, de Lafontaine and Leggett 1988). The result is often arti- ficially low estimates of clearance rate, values that are then used as functions in mathematical operations that range from simple multiplication of clearance rate and predator density (e.g., Reeve et al. 1978) to complex models that in- volve the swimming and foraging velo- cities and ambit geometries of motile predators and prey, and the turbulence of the environment in which they are embedded (e.g., Bailey and Batty 1983, Rothschild and Osborn 1988, Evans 1989). The simple approximation used herein was justified, in part, by the be- havior of M. leidyi feeding on immobile fish eggs. Lobate ctenophores feed as a moving pump, pumping water con- tinuously through mucus- and tentacle- lined lobes, while either swimming ver- tically or hovering (Larson 1988), and changing position in response to low prey density (Reeve et al. 1978). While the geometry of the pred- atory field of M. leidyi is unknown, we assume, given forage velocities of from 1-3 mm/second for its con- gener M. mccradyi (Larson 1987), that it encounters new water continuously. Although the gut capacity of lobate ctenophores is small, M. leidyi egests super- fluous food in a mucus bolus when its gut is full and EBB FLOOD MIDNIGHT DAWN EBB NOON FLOOD DUSK EBB MIDNIGHT 10 10 10 15 Cteno- phores 1 j>.h Fish Eggs C! a 73 10 100/10 100/10 — f -f- 100 10 1 I ' 100/10 u 100 NUMBERS • m" Figure 2 Temporal hydrographic sections (temperature, salinity, sigma-t) and densities (numbers/m 3 ) of Mn^miopsis leidyi and fish eggs at station El at the mouth of Chesapeake Bay. Vessel failure caused a 24-hour interruption in sampling between noon and dusk. continues feeding; egested fish eggs, embedded in this bolus, are either dead or moribund (Johnson 1987). The observation that co-occurrence of M. leidyi and fish eggs, and consequently potential predation, is greatest in areas where Chesapeake Bay water mixes with coastal shelf water, coupled with the observation that M. leidyi are more abundant in regions of higher salinity within other estuaries, implies that those fishes Govoni and Olney: Predation on fish eggs by Mnemiopsis leidyi in and around Chesapeake Bay 185 FLOOD EBB SLACK EBB FLOOD DUSK MIDNIGHT DAWN NOON DUSK Or 5- 10 I I I I I 10 100/10 100/10 100/10 100/10 100 NUMBERS • m" 3 Figure 3 Temporal hydrographic sections (temperature, salinity, sigma-t) and den- sities (numbers/m 3 ) of Miiemiopsis leidyi and fish eggs at station E4 at the mouth of Chesapeake Bay. Table 1 Potential predation of the lobate ctenophore Mnemiopsis leidyi on fish eggs at the mouth of Chesapeake Bay. Values are the end points of the range over five diel intervals. Potential predation rate (eggs/m 3 x day) Strata Station El Station E4 Surface Pycnocline Below pycnocline 0.1-14.7 1.6-4.4 0.5-3.8 21.0-174.3 0-68.8 0-25.1 that spawn in low-salinity surface waters of well-strati- fied water columns may afford protection of their eggs from ctenophore predation. Estuaries typically fluc- tuate between stratified and unstratified conditions as a result of lunar periodicity and meteorological forcing. A survival advantage may be afforded by spawning in association with the predator free surface waters of the Chesapeake Bay plume. Acknowledgments Laboratory, field, and computer assistance was provided by J. Brubaker, B. Comyns, P. Crew, J. Fields, H. Johnson, E. Maddox, and J. Posenau, all of the Virginia Institute of Marine Science. A.J. Chester, J.H. Cowan, J.C. McGovern, and J. A. Musick reviewed the manuscript. This study was supported by a grant from the National Science Foundation (OCE-850144). Citations Alldredge, A.L. 1984 The quantitative significance of gelatinous zooplankton as pelagic consumers. In Fasham, M.J.R. (ed.), Flows of energy and materials in marine ecosystems, p. 407-433. Plenum Press, NY. Bailey, K.M., and R.S. Batty 1983 A laboratory study of predation by Aurelia aurita of larval herring (Clupea harengus): Ex- perimental observations compared with model predictions. Mar. Biol. (Berl.) 72:295-301. 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. Bishop, J.W. 1967 Feeding rates of the ctenophore, Mnemiopsis leidyi. Chesapeake Sci. 8:259-264. Boicourt, W.C., S.-Y. Chao, H.W. Ducklow, P.M. Gilbert, T.C. Malone, M.R. Roman, L.P. Sanford, J.A. Fuhrman. C. Garside, and R.W. Garvine 1987 Physics and microbial ecology of a bouyant estuarine plume on the continental shelf. EOS-Riv. Immunol. Immuno- farmacol. 68(31):666-668. 186 Fishery Bulletin 89(2), 1991 Burrell, V.G., and W.A. Van Engel 1976 Predation by and distribution of a ctenophore, Mnemiop- sis leidyi A. Agassiz, in the York River estuary. Estuarine Coastal Mar. Sci. 4:235-242. de Lafontaine, Y., and W.C. Leggett 1988 Predation by jellyfish on larval fish: An experimental evaluation employing in situ enclosures. Can. J. Fish. Aquat. Sci. 45:1173-1190. Evans, G.T. 1989 The encounter speed of moving predator and prey. J. Plankton Res. 11:415-417. Frank, K.T., and W.C. Leggett 1982 Coastal water mass replacement: Its effects on zoo- plankton dynamics and the predator-prey complex associated with larval capelin (Mallotus villosus). Can. J. Fish. Aquat. Sci. 39:991-1003. 1985 Reciprocal oscillations in densities of larval fish and poten- tial predators: A reflection of present or past predation? Can. J. Fish. Aquat. Sci. 42:1841-1849. Gosner, K.L. 1971 Guide to identification of marine and estuarine in- vertebrates, Cape Hatteras to the Bay of Fundy. John Wiley, NY, 693 p. Johnson, H.D. 1987 Potential fish egg predation by Mnemiopsis leidyi deter- mined by hydrography at the Chesapeake Bay mouth. M.S. thesis, College of William and Mary, Williamsburg, VA, 52 p. Kremer, P. 1979 Predation by the ctenophore Mnemiopsis leidyi in Nar- ragansett Bay, Rhode Island. Estuaries 2:97-105. Kremer, P., and S. Nixon 1976 Distribution and abundance of the ctenophore, Mnemiop- sis leidyi in Narragansett Bay. Estuarine Coastal Mar. Sci. 4:627-639. Larson, R.J. 1987 In situ feeding rates of the ctenophore Mnemiopsis mccradyi. Estuaries 10:87-91. 1988 Feeding and functional morphology of the lobate cteno- phore Mnemiopsis mccradyi. Estuarine Coastal Shelf Sci. 27:495-502. Miller, R.J. 1974 Distribution and biomass of an estuarine ctenophore population. Mnemiopsis leidyi (A. Agassiz). Chesapeake Sci. 15:1-8. Monteleone, D.M,, and L.E. Duguay 1988 Laboratory studies of predation by the ctenophore Mnemiopsis leidyi on the early stages in the life history of the bay anchovy, Anchoa mitchilli. J. Plankton Res. 10:359-372. Mountford, K. 1980 Occurrence and predation by Mtiemiopsis leidyi in Barnegat Bay, New Jersey. Estuarine Coastal Mar. Sci. 10:393-402. Olney, J.E. 1983 Eggs and early larvae of the bay anchovy, Anchoa mit- chilli, and the weakfish, Cynoscion regalis, in lower Chesapeake Bay with notes on associated ichthyoplankton. Estuaries 6:20-35. Purcell. J.E. 1985 Predation on fish eggs and larvae by pelagic cnidarians and ctenophores. Bull. Mar. Sci. 37:739-755. 1989 Predation on fish larvae and eggs by the hydromedusa (Aequorea victoria) at a herring spawning ground in British Columbia. Can. J. Fish. Aquat. Sci. 46:1415-1427. Purcell, J.E., and J.J. Grover 1990 Predation and food limitation as causes of mortality in larval herring at a spawning ground in British Columbia. Mar. Ecol. Prog. Ser. 59:55-61. Reeve, M.R., M.A. Walter, and T. Ikeda 1978 Laboratory studies of ingestion and food utilization in lobate and tentaculate ctenophores. Limnol. Oceanogr. 23: 740-751. Rothschild, B.J., and T.R. Osborn 1988 Small-scale turbulence and plankton contact rates. J. Plankton Res. 10:465-474. Ruzecki, E.P. 1981 Temporal and spatial variation of the Chesapeake Bay plume In Campbell, J.W., and J.P. Thomas (eds.), Chesapeake Bay Plume Study: Superflux 1980, p. 111-130. NASA Conf. Publ. 2188, Natl. Aeronautics & Space Adm., Wash., DC. NOAA NEMP III (Northeast Monit. Prog.. Natl. Mar. Fish. Serv., NOAA, Woods Hole, MA). Sullivan, B.K.. and M.R. Reeve 1982 Comparison of estimates of the predatory impact of ctenophores by two independent techniques. Mar. Biol. (Berl.) 68:61-65. Abstract.- Spring ichthyoplank- ton surveys in the tidal freshwater reaches of Virginia rivers were used to document the temporal and spatial occurrence of spawning by striped bass Morone saxatilis (Walbaum). Single river systems were intensive- ly surveyed in 1980 (York River sys- tem), 1981 (James River system), and 1982 (Rappahannock River). In spring 1983, all three river systems were sampled at approximately weekly intervals. Some spawning oc- curred in all years, including those yielding poor year-classes (1980 and 1981). Spawning occurred largely within the first 40 km of tidal fresh- water in major rivers, except when drought conditions displaced spawn- ing upstream in advance of encroach- ing saltwater. Eggs appeared in sharp, brief peaks of abundance, usually between the third week in April and the first week in May. Peak densities coincided with rapid- ly rising water temperatures in the range 13.7-19.5°C. Distribution of Striped Bass Morone saxatilis (Walbaum) Eggs and Larvae in Major Virginia Rivers* George C. Grant John E. Olney Virginia Institute of Marine Science, School of Marine Science College of William and Mary, Gloucester Point, Virginia 23062 The importance of Chesapeake Bay spawning grounds to Atlantic coast stocks of striped bass Morone saxa- tilis (Walbaum) has long been recog- nized (Merriman 1941; Raney 1952; Briggs 1962; Alperin 1966; Schaefer 1967, 1968, 1972; Berggren and Lieberman 1978; Kohlenstein 1981; Fabrizio 1987a, 1987b). Striped bass spawning in Chesapeake Bay and its tributaries has been documented from collections of eggs and larvae (Tresselt 1952, Rinaldo 1971, John- son and Koo 1975, Polgar et al. 1976, Conte et al. 1979, Kernehan et al. 1981, Setzler-Hamilton et al. 1981), and survey data on the distribution of juveniles or presence of ripe adults (Vladykov and Wallace 1952, Grant and Joseph 1969, Markle and Grant 1970, Grant 1974). Direct documentation of striped bass spawning in Virginia rivers based on plankton collections of eggs and larvae was provided by consecu- tive single surveys of five rivers in April and May 1950 (Tresselt 1952); a single river survey during the en- tire 1966 spawning season (Rinaldo 1971); and a single river survey dur- ing the entire 1985 spawning season (McGovern and Olney 1988). This paper documents temporal and spa- tial occurrence of striped bass eggs and larvae in Virginia's major river Manuscript accepted 19 October 1990. Fishery Bulletin, U.S. 89:187-193 (1991). * Contribution no. 1626 of the Virginia Insti- tute of Marine Science and School of Marine Science, College of William and Mary. systems. The York, James, and Rap- pahannock river systems, respec- tively, were surveyed in the years 1980-82; all three rivers were sam- pled in 1983. Methods and materials Lower tidal freshwater portions of the James, Chickahominy, Pamun- key, Mattaponi, and Rappahannock Rivers were divided into 3-mile (5-km) strata, from which single sta- tions were randomly selected prior to each sampling trip. Number of strata (in parentheses) and cruise dates were: the Pamunkey River (10), 16 April- 13 June 1980 and 5 April- 11 May 1983; Mattaponi River (6), 18 April-14 June 1980; James River (10), 22 April-19 June 1981 and 8 April-8 May 1983; Chickahominy River (7), 21 April-18 June 1981; and Rappahannock River (9), 5 April-6 June 1982 and 9 April-13 May 1983. Stations were sampled semi-week- ly to weekly within strata extending upstream from the limits of brackish water (~0.5"A») to beyond the ob- served occurrence of striped bass eggs. Collections at each station were stepped-oblique, daylight tows of a 60-cm bongo sampler, equipped with 333-^m mesh nets and flowmeters. Length of tows were 2-6 minutes; tows in deep water were of longer duration and tows encountering ex- cessive detritus loads were short- ened. Catches from the paired nets 187 188 Fishery Bulletin 89(2). 1991 \ V^^ \5Z 39 1983 {80% of all eggs collected. Morone larvae lacking yolk material and larger than 8.5 mm standard length were cleared and stained for positive identification (Fritzsche and Johnson 1980, Olney et al. 1983). Results York River System, 1980 and 1983 Spawning extended approximately 40 km upstream from the limit of brackish water in the Pamunkey River in 1980 and 1983 and 27 km in the Mattaponi River in 1980. Peak spawning was recorded over 13km on the Mattaponi River, while on the Pamunkey River it was observed in two disjunct regions (miles 27-41 and 45-47) in both years (Fig. 1). In 1980, eggs were found on initial surveys (April 16 and 18), indicating that spawning activity had already begun (Figs. 2, 3). Peak egg densities (> 1/m 3 ) occurred on 22 and 25 April in the Pamunkey and Mattaponi rivers, respectively. In 1983 (Fig. 4), sampling on the Pamunkey River was initiated on 5 April, but few eggs were collected until 27 April, when peak densities for the season were encountered. Abundance of larval striped bass was cen- tered somewhat further upstream than eggs on the initial 1980 surveys, and all were yolk- sac larvae. Peaks in egg and larval abundance coincided more closely on subsequent sampling dates (Figs. 2, 3). Few larvae were collected after the 8 and 9 May surveys of the Pamunkey and Mattaponi rivers, none after 31 May. Modal length of larvae in both rivers was 5 mm NL (notochord length) and most were yolksac larvae. All larvae captured after 16 May were 12 mm SL (standard length) or larger. In 1983, all larvae from the surveys of 4 and 11 May (the only collections examined for larvae) were yolksac larvae less than 6 mm in length. Grant and Olney: Spawning of Morone saxatilis in Virginia rivers 189 20 30 10 20 APRIL MAY Figure 3 Spatial and temporal abundance (numbers/100 m :i ) of striped bass eggs and larvae in the Mattaponi River, 1980. River miles are nau- tical miles above the York River mouth. James River System, 1981 and 1983 Location of peak spawning and maximum den- sities of eggs differed considerably between 1981 and 1983 (Fig. 5). Occurrence of eggs much far- ther upstream in 1981 was probably related to drought and intrusion of saltwater in winter and spring, 1980-81. Spawning extended over 35 km in 1981, and over 40km in 1983. In both years 45 40- CC LU > 35 RAPPAHANNOCK RIVER PAMUNKEY RIVER JAMES RIVER Figure 4 Spatial and temporal abundance (numbers/ 100 m 3 ) of striped bass eggs in three Virginia rivers, 1983. River miles are nautical miles. 1981 1983 Figure 5 Spatial extent of observed striped bass egg occurrence, James River, 1981 and 1983. Over 80% of eggs occurred in solid black areas. Inset shows location in Chesapeake Bay region. 190 Fishery Bulletin 89(2). 1991 20 30 APRIL 20 20 30 10 20 30 10 20 MAY APRIL MAY JUNE Figure 6 Spatial and temporal abundance (numbers/lOOm 1 ) of striped bass eggs and larvae in the James River, 1981. River miles are nautical miles above the river mouth. peak spawning appeared to be bimodally distributed along the river channel. Eggs were present in both rivers on initial surveys of 1981 (21 April in the Chicka- hominy River and 22 April in the James River) in fresh to slightly brackish water and temperatures of 15.6- 18.7°C. Maximum egg densities in the James River were only 46/100 m 3 in 1981 (13/100 m 3 in the Chicka- hominy), compared with 195/100 m 3 in 1983. These maxima occurred in both years during the first week in May, in tidal freshwater and similar temperatures (18.5-19.5°C), but at sites separated by 16km (Fig. 6). In the Chickahominy River, eggs were restricted to the river's junction with the James and found on only two surveys, suggesting that eggs could have been spawned in the James River and tidally advected into the Chickahominy. Larval striped bass were not present in collections from the Chickahominy River in 1981, and the river was not sampled in 1983. James River collections in 1981 yielded primarily yolksac larvae, found in max- imum densities similar to those of eggs (Fig. 6). The modal length of larvae during the first 3 weeks of surveys in 1981 was 5.0-5.9 mm, but a few smaller lar- vae (4.0-4.9 mm) were hatched as late as 21-22 May. The maximum size of larvae increased from 7.0-7.9mm in late April to over 13 mm on 4-5 June, the last survey in which striped bass larvae were captured. Only a dozen larvae larger than 9 mm were collected. In 1983, larvae from the early-May surveys attained densities of 31/100m 3 . Figure 7 Spatial extent of observed striped bass egg occurrence. Rap- pahannock River, 1982 and 1983. Over 80% of eggs occurred in solid black areas. Inset shows location in Chesapeake Bay region. Rappahannock River, 1982 and 1983 Eggs were found within the first 40 km of tidal fresh- water in both 1982 and 1983 (Fig. 7). Peak spawning was observed between river miles 51 and 62 in 1982 and miles 48 and 59 in 1983. Eggs occurred in densities ranging between 1.4-330/100m 3 during April 1982 and 0.7-50/100 m 3 in May of that year. None were found after the 1 1 May survey, and the maximum den- sity occurred on 21 April in temperatures 15.7-16.6°C. In 1983, eggs were collected in all weekly surveys from 9 April to 13 May, but increasing densities were re- versed by a cold snap in the third week of April, when only scattered eggs were found (water temperatures had declined from 13-14°C to 11-12°C). Egg densities peaked once water temperature rose in the last week of April, when a maximum of 477 eggs/100 m 3 was recorded. Eggs were rare by the final sampling date (May 13). Grant and Olney: Spawning of Morone saxatilis in Virginia rivers 191 10 20 30 10 APRIL MAY Figure 8 Spatial and temporal abundance (numbers/100 m :i ) of striped bass eggs and larvae in the Rappahan- nock River, 1982. River miles are nautical miles above the river mouth. Peak larvai densities were recorded on the Rappa- hannock River in 1982 during late April and early May (Fig. 8). These densities were the highest observed in our 3-year study of Virginia rivers. Modal lengths of larvae were 5-6 mm NL until mid-May; all larvae from 21 May to the conclusion of sampling in June were at least 8mm in length. In 1983, larvae from the early May surveys were all less than 8 mm in length and at- tained a maximum density of 103 larvae/100 m 3 . Discussion Tresselt (1952), Rinaldo (1971), McGovern and Olney (1988), and this study constitute the only direct obser- vations on striped bass eggs and larvae in major Vir- ginia rivers. Striped bass spawning in Virginia rivers occurred from early April through the first week or two in May 1980-83, and within the first 40 km of tidal freshwater, generally confirming earlier observations of Tresselt (1952) and Rinaldo (1971). Tresselt (1952) observed spawning on the two major rivers of the York River system, the Pamunkey River (4-13 April 1950) and the Mattaponi River (13-30 April 1950). He found largest numbers of eggs 27 km above the mouth of the Pamunkey and 14 km above the mouth of the Mattaponi. Historic centers of successful spring fishing for striped bass were in these tidal freshwater rivers. Spawning in the Pamunkey River in 1966 oc- curred 8-48km above West Point during 24 April-13 May 1966 (Rinaldo 1971). Surveys of Virginia striped bass spawning grounds in 1950 (Tresselt 1952) were inadvertently somewhat late in the Chickahominy and James rivers, but provide the only direct documenta- tion of striped bass spawning prior to the present study. He found only three eggs in 30 collections in the Chickahominy River, 5-8 May 1950, and 57 eggs in 38 collections from the James River during 9-10 May. No direct observations on striped bass larvae were made prior to the present study. Use of the Rappahannock River as a spawning site by striped bass was also documented by Tresselt (1952), although his limited survey (four sampling dates in May 1950) yielded only five eggs. Temporal occurrence of striped bass spawning is similar throughout Chesapeake Bay. In Virginia rivers, peaks in spawning were sharp and of limited duration. They occurred in the fourth week of April in York River tributaries in both surveyed years (1980 and 1983) and in the Rappahannock River in 1983; peak spawning was a week earlier in the Rappahannock River in 1982. Spawning in the James River was somewhat later, peaking in the first week of May in both 1981 and 1983. Peak spawning in this time-period is also typical for the Potomac River (Setzler-Hamilton et al. 1981) and in the upper Chesapeake Bay and Chesapeake and Delaware Canal (Johnson and Koo 1975, Kernehan et al. 1981). Although striped bass eggs were found in a wide range of temperatures (8.0-21.2 c C), peak den- sities were limited to rapidly rising temperatures in the range 13.7-19.5°C and nearly always to freshwater. Annual differences in time of spawning within a given river system are most likely a result of differences in temperature and the rate of vernal warming. Areas of peak spawning and the spatial extent of the spawn are remarkably similar among years and river systems, but differences can occur in years of drought. The present data from the James River contrasted a year of severe drought (1981) with one of near-average rainfall (1983). The estimated streamflow from the James River system into Chesapeake Bay during March, April, and May averaged only 69,000cfs in 1981 compared with 180,000 cfs in 1983 (monthly summary reports of estimated streamflows entering Chesapeake Bay, 12/30/81 and 12/30/83, U.S. Geol. Surv., Towson, MD). Peak egg production was displaced 15 km upriver in 1981, in advance of encroaching saltwater, whereas interannual differences in location of peak spawning in the other river systems where drought was not a fac- tor were insignificant (<2km). The tidal freshwaters of the Rappahannock River were surveyed for striped bass eggs and larvae in the spring of 1982, during production of what later was determined to be the strongest year-class measured in Virginia during the period 1971-82 (Colvocoresses 192 Fishery Bulletin 89(2). 1991 1989). The observed high density of eggs and larvae in 1982 and subsequent success of the 1982 year-class support the concept that attributes strong year-classes to exceptional survival of eggs and larvae on the spawn- ing grounds, rather than absolute size of spawning stock (management actions designed to prevent over- fishing and to increase size of spawning stock were not yet in effect). After 10 years of poor year-classes, i.e., 1972-81, the stock producing this 1982 year-class must necessarily have been small. Chapman's (1987, 1989, 1990) discovery of differences between genotype fre- quencies of 1982 year-class males and those of older Chesapeake Bay females is more direct evidence that strong year-classes can stem from successful spawn- ing by relatively few females. Unfortunately, we have yet to determine which factors govern survival and growth of striped bass eggs and larvae on Chesapeake Bay spawning grounds, although recent research ef- forts are addressing this important question (e.g., McGovern and Olney 1988, Chesney 1989, Uphoff 1989). Acknowledgments This research was supported in part by the National Marine Fisheries Service (NOAA) grants NA80FAD- VA1B, NA81FAD-VA3B, NA81FAD-VA5B, and con- tract AFC-14-1, and by NOAA Office of Sea Grant, grant no. 5-29346. We particularly owe thanks to Bruce H. Comyns, James E. Price, Cathy J. Womack, and Patricia A. Crewe for their assistance in the study, and to Daniel W. Sved, James A. Colvocoresses, and John C. McGovern for early manuscript reviews. Citations Alperin, I.M. 1966 Dispersal, migration and origins of striped bass from Great South Bay, Long Island, N.Y. N.Y. Fish Game J. 13: 79-112. Berggren. T.J., and J.T. Lieberman 1978 Relative contribution of Hudson, Chesapeake and Roa- noke striped bass, Morone saxatilis, stocks to the Atlantic coast fishery. Fish. Bull., U.S. 76:335-345. Briggs, P.T. 1962 The sport fisheries of Great South Bay and vicinity. N.Y. Fish Game J. 9:1-36. Chapman, R.W. 1987 Changes in the population structure of male striped bass, Morone saxatilis, spawning in the three areas of the Chesa- peake Bay from 1984 to 1986. Fish. Bull., U.S. 85:167-170. 1989 Spatial and temporal variation of mitochondrial DNA haplotype frequencies in the striped bass (Morone saxatilis) 1982 year class. Copeia 1989:344-348. 1990 Mitochondrial DNA analysis of striped bass poulations in Chesapeake Bay. Copeia 1990:355-366. Chesney, E.J. Jr. 1989 Estimating food requirements of striped bass larvae, Morone saxatilis: Effects of light, turbidity and turbulence. Mar. Ecol. Prog. Ser. 53:191-200. Colvocoresses, J. A. 1989 Striped bass research, Virginia. Part I: Juvenile striped bass seining program. Annu. Rep., Proj. AFC 18, Segment 2, July 1987-December 1988. Va. Inst. Mar. Sci., Gloucester Point, 43 p. Conte, M.H., R.G. Otto, and P.E. Miller 1979 Short term variability in surface catches of ichthyoplank- ton in the upper Chesapeake Bay. Estuarine Coastal Mar. Sci. 8:511-522. Fabrizio, M.C. 1987a Contribution of Chesapeake Bay and Hudson River stocks of striped bass to Rhode Island coastal waters as esti- mated by isoelectric focusing of eye lens proteins. Trans. Am. Fish. Soc. 116:588-593. 1987b Growth-invariant discrimination and classification of striped bass stocks by morphometric and electrophoretic methods. Trans. Am. Fish. Soc. 116:728-736. Fritzsche, R.A., and G.D. Johnson 1980 Early osteological development of white perch and striped bass with emphasis on identification of their larvae. Trans. Am. Fish. Soc. 109:387-406. Grant, G.C. 1974 The age composition of striped bass catches in Virginia rivers, 1967-1971, and a description of the fishery. Fish. Bull.. U.S. 72:193-199. Grant. G.C, and E.B. Joseph 1969 Comparative strength of the 1966 year class of striped bass, Roceus saxatilis (Walbaum), in three Virginia rivers. In Proc. 22d Annu. Conf., Southeast. Assoc. Game Fish Comm., p. 501-509. Johnson, R.K., and T.S.Y. Koo 1975 Production and distribution of striped bass (Morone saxa- tilis) eggs in the Chesapeake and Delaware Canal. Chesa- peake Sci. 16:39-55. Kernehan, R.J., M.R. Headrick, and R.E. Smith 1981 Early life history of striped bass in the Chesapeake and Delaware Canal and vicinity. Trans. Am. Fish. Soc. 110: 137-150. Kohlenstein. L.C. 1981 On the proportion of the Chesapeake Bay stock of striped bass that migrates into the coastal fishery. Trans. Am. Fish. Soc. 110:168-179. Markle, D.F., and G.C. Grant 1970 The summer food habits of young-of-the-year striped bass in three Virginia rivers. Chesapeake Sci. 11:50-54. McGovern, J.C., and J.E. Olney 1988 Potential predation by fish and invertebrates on early life history stages of striped bass in the Pamunkey River, Virginia. Trans. Am. Fish. Soc. 117:152-161. Merriman, D. 1941 Studies on the striped bass Roceus saxatilis of the Atlantic Coast. Fish. Bull, U.S. 50:1-77. Olney, J.E., G.C. Grant, F.E. Schultz, C.L. Cooper, and J. Hageman 1983 Pterygiophore-interdigitation patterns in larvae of four Morone species. Trans. Am. Fish. Soc. 112:525-531. Polgar, T.T., J.A. Mihursky, R.E. Ulanowicz, R.P. Morgan II, and J.S. Wilson 1976 An analysis of 1974 striped bass spawning success in the Potomac estuary. In Wiley, M. (ed.), Estuarine processes, vol. 1 Uses, stresses, and adaptation to the estuary, p. 151-165. Academic Press, NY. Grant and Olney: Spawning of Morone saxatilis in Virginia rivers 1 93 Raney. E.C. 1952 The life history of the striped bass, Roccus saxatilis (Walbaum). Bull. Bingham Oceanogr. Coll. 14:5-97. Rinaldo, R.G. 1971 Analysis of Morone saxatilis and Morone americana spawning and nursery area in the York and Pamunkey rivers, Virginia. M. A. thesis, College of William and Mary. Williams- burg, VA, 56 p. Schaefer, R.H. 1967 Species composition, size and seasonal abundance of fish in the surf waters of Long Island. N.Y. Fish Game J. 4:1-46. 1968 Size, age composition and migration of striped bass from the surf waters of Long Island (Roccus saxatilis). N.Y. Fish Game J. 15:1-51. 1972 A short-range forecast function for predicting the relative abundance of striped bass in Long Island waters. N.Y. Fish Game J. 19:178-181. Setzler-Hamilton, E.M., W.R. Boynton, J. A. Mihursky, T.T. Polgar, and K.V. Wood 1981 Spatial and temporal distribution of striped bass eggs, larvae and juveniles in the Potomac estuary. Trans. Am. Fish. Soc. 110:121-136. Tresselt, E.F. 1952 Spawning grounds of the striped bass, Roccus saxatilis (Walbaum), in Virginia. Bull. Bingham Oceanogr. Coll. 14: 98-110. Uphoff, J.H. Jr. 1989 Environmental effects on survival of eggs, larvae, and juveniles of striped bass in the Choptank River, Maryland. Trans. Am. Fish. Soc. 118:251-263. Vladykov, V.D., and D.H. Wallace 1952 Studies of striped bass, Roccus saxatilis (Walbaum), with special reference to the Chesapeake Bay region during 1936- 1938. Bull. Bingham Oceanogr. Coll. 14:132-177. Abstract. — Growth and mortal- ity rates were compared for juvenile California halibut ParaMckthys cali- fornicus from bay and open coast habitats. Growth was estimated by determination of size-at-age using daily increments in otoliths. No sig- nificant difference was observed in size-at-age for juvenile halibut be- tween 6 and 41mm from the bays and open coast. However, age-specific mortality rates estimated for halibut <70 days were highest for newly- settled halibut on the open coast. California halibut settled either in bays or on the open coast, but ulti- mately nearly all of the halibut that settled on the coast entered and used the bays as nursery areas during their first year of life or else they died. The advantages of bays as nur- sery areas may be a decrease in risk of mortality of newly-settled halibut and an increase in growth of larger juveniles that feed upon the abun- dant small fishes in the bays. Growth, Mortality, and Movements of Juvenile California Halibut Paralichthys californicus in Shallow Coastal and Bay Habitats of San Diego County, California Sharon Hendrix Kramer Southwest Fisheries Science Center. National Marine Fisheries Service. NOAA PO Box 271, La Jolla. California 92038 Present address: MBC Applied Environmental Sciences 947 Newhall Street, Costa Mesa, California 92627 Manuscript accepted 12 December 1990. Fishery Bulletin, U.S. 89:195-207 (1991). The utilization of specialized nursery habitats by juvenile fish is a common phenomenon (Boehlert and Mundy 1988, Miller et al. 1986). Many of the fish species that utilize bays as nur- sery areas spawn in offshore waters, and move into bays as late larvae and early juveniles (Boehlert and Mundy 1988, Miller et al. 1986). The migra- tion, location, and entry of larvae and juveniles into the bays involve com- plex behaviors that are particularly important on the Pacific coast of North America, where only 10-20% of the coastal habitat consists of estuaries and lagoons, compared with 80-90% on the Atlantic and Gulf coasts (Emery 1967). Possible conse- quences of the use of bays as nursery areas include faster growth because of high food production, warm tem- peratures, and decreased predation (Miller et al. 1986, Kneib 1987, Kry- gier and Pearcy 1986). The California halibut Paralichthys californicus is a commercially impor- tant flatfish found in southern Cali- fornia coastal waters and bays (Frey 1971, Haaker 1975, Allen 1988, Love et al. 1986, Plummer et al. 1983). Eggs and larvae occur over the shelf and seaward, with greatest densities in waters less than 75 m deep and within 6km of shore (Frey 1971, Gruber et al. 1982, Barnett et al. 1984, Lavenberg et al. 1986, Walker et al. 1987, Moser and Watson 1990). In past studies juvenile halibut rare- ly were taken on the open coast, sug- gesting that bays and lagoons might be the only significant nursery habi- tat (Plummer et al. 1983, Allen 1982, Kramer 1990). The objective of this study was to determine the relative importance of bays as nursery areas and to evaluate the movements between bay and open coast habitats. To meet these objectives, I estimated habitat-spe- cific distribution, abundance, and growth and mortality rates of juve- nile halibut from both bay and open coast habitats. Materials and methods Distribution and abundance California halibut were collected dur- ing a 2-year survey (September 1986- September 1988) of the open coast between Mission Bay and San Ono- fre, and two bays, Mission Bay and Agua Hedionda Lagoon (Fig. 1). A stratified random sampling design was used, consisting of four open coast blocks each with three depth strata (5-8m, 9-llm, and 12-14m), and five blocks in Mission Bay and three blocks in Agua Hedionda La- goon, each with three depth strata (0-lm, l-2m, and 2-4m) (Fig. 1). 195 196 Fishery Bulletin 89(2), 1991 Figure 1 Location of sampling blocks. Open coast blocks are (1) San Onofre, (2) adjacent to Agua Hedionda Lagoon, (3) Torrey Pines, and (4) adjacent to Mission Bay. The two bays sampled are Agua Hedionda Lagoon and Mission Bay, with sampling blocks denoted. For further description of the sampling design and the habitats see Kramer (1990). Three gear types were used, all lined or made of 3-mm mesh: a 1.0-m wide beam trawl, a 1.6-m wide beam trawl, and a 1 x 6-m beach seine. The 1.6-m beam trawl, set from a 15-m research vessel, was used to sample the open coast and Mission Bay (Fig. 1). I sampled Agua Hedionda Lagoon and the areas of Mis- sion Bay that were inaccessible to the larger vessel with the 1.0-m beam trawl, set from a 6-m skiff. The 1.0-m beam trawl and the beach seine were pulled along the bottom by two people to sample the shallow shoreline (< 1 m) in the bays. The trawls were fitted with a wheel and revolution counter to determine the distance traveled by the trawl along the bottom, allowing a quantitative assessment of fish density since the trawls had a fixed mouth opening (Krygier and Horton 1975). All trawls and seines were fished during the day. Table 1 Gear weighting coefficients and their variances by length-class for conversion of shoreline collections by beach seine and 1-m beam trawl. Coefficients determined by 3-way ANOVA be- tween gear types, blocks, and months of sample on density for each length-class. Correction terms are given for length- classes with significant gear effects (P< 0.05). There were no significant gear effects in the 1.0-1. 6m beam trawls for open water tows. Length class (SL, mm) Correction term Variance 26-30 31-35 36-40 41-45 3.291 4.398 2.752 4.699 0.124 0.319 0.099 0.359 All flatfishes taken in trawls and seines were mea- sured to standard length (SL) in mm. Density of halibut in 5-mm standard length-classes was determined for juveniles <70mm SL. The thirteen length-classes used were: SL<10mm, ll-15mm, 16-20mm, continuing to 66-70 mm SL. Abundance was determined by multiply- ing the mean density for each habitat by the area of each habitat. Gear comparison Densities based on the 1.0-m beam trawl collections did not differ significantly from those of the 1.6-m beam trawl for any length class (ANOVA, P>0.05, n 826). However, the beach seine captured significantly fewer small halibut (26-45 mm SL) than the 1.0-m beam trawl (Table 1). Since significant biases existed, density and abundance estimates of halibut were corrected for the differences in gear efficiency by weighting the mean density and variance for each length class where signifi- cant differences in catchability were found (Table 1). The weighted mean density for each gear type was calculated as d w = (d 1 + gd,)/(l + g) where dj = unweighted density, d 2 = weighted density, and g = weighting coefficient. Estimated variance of the weighted mean d w was calculated as V(d w ) = V(d 1 ) + g 2 V(d 2 ) + d 2 2 V(g) + V(g)V(d 2 ) where V(dj) = variance of unweighted density, V(d 2 ) = variance of weighted density, and V(g) = variance of Kramer: Growth and mortality rates of juvenile Paralichthys cahfornicus 197 weighting coefficient. Variance of the weighted mean was underestimated because the covariance terms were not included. Resampling techniques to estimate vari- ance (e.g., bootstrap) were impractical because of the large size of the database. Age validation and determination Laboratory-reared halibut larvae of known age were measured to standard length in mm and their sagittae excised and mounted in resin (Eukitt, 0. Kindler, West Germany) on a microscope slide. Age was estimated using the methods of Methot (1981) and Butler (1987). A microcomputer interfaced to an electronic digitizer was used to measure and count increments on a pro- jected image of the otolith from a high-resolution video camera mounted on a compound microscope. Incre- ment counts of 45 larvae (3.1-9.1 mm SL) that were reared at 16-20°C in the laboratory were regressed against the known age of the larvae to establish a rela- tionship between estimated and known age. Incre- ments were formed daily: the slope of the relationship (0.969) did not differ significantly from unity (P>0.05). The regression of the number of increments on age of halibut larvae (5-29 days) was Age (days) = 3.496 + 0.969 x (no. increments) where r 2 = 0.981, SE constant = 1.055, SE slope = 0.018, and range of increment counts = 1-26) (Fig. 2). Daily formation of rings has also been found in juveniles 30-70 mm SL (Kicklighter 1990). The first increment is deposited about 3.5 days after hatching, coinciding with the day of first feeding (Gadomski and Peterson 1988). I added 3.5 to the number of increments counted on the otolith so that age was equivalent to the number of days from hatching. Ageing of field-caught halibut Juvenile halibut from field collections were measured alive and either frozen or preserved in 80% ethanol. Sagittae were dissected and increments counted using the techniques described above. Sagittae from juveniles >20mm SL were polished with 400- and 600-grit wet sandpaper before counting. A total of 120 field-caught halibut were aged: 50 from Mission Bay, 19 from Agua Hedionda Lagoon, and 51 from the open coast. Larval sagittae are sym- metrical and nearly circular (Fig. 3A), but after meta- morphosis additional foci develop and the sagittae became asymmetrical, with maxiumum deposition along the rostral axis (Karakiri et al. 1989) (Fig. 3B). This shift in the axis of sagittal growth produces areas KNOWN AGE (days) Figure 2 Age validation of California halibut sagittae. Number of in- crements counted on the sagittae are compared with the known age of laboratory-reared larval halibut (n 45). Straight line represents a one-to-one relationship of increment number and known age. that are difficult to interpret (Fig. 3). These areas correspond to a period of about 7 days after metamor- phosis. I estimated the number of increments in regions of transition between foci by counting the number of increments that occurred in an adjacent area on a dif- ferent axis (Fig. 3). The relationship between standard length (mm) and otolith radius (^m) was linear for halibut MOmrn SL (Fig. 4). Mortality estimates I did not use data from the 1987 survey for estimating mortality because nearly all of the 1987 year-class oc- curred in bays and comparisons of mortality between bay and coast habitats were an essential step in the analysis. The relationship between abundance and age (estimated from the length-at-age relationship) of the 1988 year-class was used to estimate age-specific mor- tality rates. I used seven different models to estimate age-specific instantaneous mortality rates. Three of the models were estimates based on the following assumptions regarding the relationship between survival rates and age (Barlow 1982): (1) Age-specific survival rates in- crease linearly with age; (2) age-specific survival rates increase exponentially with age; and (3) age-specific survival rates approach an asymptote with age. The two daily production models estimated age-specific in- stantaneous mortality rates based on the relationship between daily production (abundance of length class/ duration of length class) and age (Lo 1985). The last 198 Fishery Bulletin 89(2). 1991 Figure 3 Photomicrographs of California halibut otoliths. (A) Sagitta from halibut 6.94 mm SL, estimated age 22 days. Distance of drawn radius, 70(jm. (B) Sagitta from halibut 48 mm SL, estimated age 109 days. Distance of drawn radius, 890 ^m. . B 22 J 18 - . . E 3. 1,6 - . . ' 5 ? 1 4 - • :■-' - < a tr g I £ 1 - . • J _J »■ 08 - • .* 0£ - 04 - #' 02 - y 20 40 60 80 100 '20 H STANDARD LENGTH (mm) two models were simple linear esti- mates, using a linear regression of In-transformed abundance-at-age on age (constant mortality rate with age) and on ln(age) (age-specific mortality rates). The sum of squared deviations of observed abundance- at-age from the calculated or trans- formed abundance-at-age predicted by each model was used to deter- mine the model that best fit the data. Mortality estimates for halibut from the open coast include loss of juveniles from the open coast popu- lation due to emigration into the bays. These estimates are used to calculate age-specific emigration rates by comparing the apparent mortality (= mortality + emigration) on the open coast to the total mor- tality calculated for the population on the open coast and in the bays. Results Effects of season and location on size-at-age Relationships between halibut length and age did not vary significantly between seasons or between habi- tats. Analysis of covariance indicated no significant dif- ference in length-at-age between fish that had birth- dates in the spring and those with birthdates in the late summer and fall, but the sample size was small for fall fish (n 9) (Table 2). The common slope was 0.6206 (SE 0.0905). Figure 4 Relationship between otolith radius (/im) and standard length (mm) of California halibut (n 120). Kramer: Growth and mortality rates of juvenile Paralichthys californicus 199 Analysis of covariance also indicated no significant difference between juveniles from the bays and the open coast in the relationship between length and age (Table 2). The comparison was made be- tween fish from the bays and the open coast. The common slope was 0.471 (SE 0.0238). Therefore, I used the pooled data for all estimates of growth (n 120). Length-at-age The relationship between standard length (mm) and age (days) was best described with the Gompertz growth function, Table 2 Regression analys s and analysis of covariance (ANCOVA) of size-at-age by season and by habitat. Slopes and intercepts were compared using ANCOVA; all tests were not significant at P>0.05. Size range Equality of F-statistic Covariates N (mm) Intercept Slope Intercept Slope Seasons Spring 26 27-80 -7.6 0.637 1.68 0.051 Fall 9 27-83 -8.6 0.594 Habitats Coast 51 6.8-41 -7.2 0.468 3.87 0.035 Bays 26 8.3-41 -4.7 0.478 Length = P ; x exp(P 2 (l - exp(-P 3 x age))) with Pj = 2.13, P 2 = 4.77, and P 3 = 0.011, and an esti- mated mean square error of 0.99 (2SE Pl = 0.34, 2SE P , = 0.137, 2SEp 3 = 0.0013) (Fig. 5A). The param- eter P^ closely estimates the length-at-hatching, which is 2.0 for halibut (Ahlstrom et al. 1984). The relationship of age-at-length was determined with the function, Age = -88.347 x In (In (standard length x 251.07)/ -4.769) derived from the Gompertz relationship for size-at-age (Methot 1981) (Fig. 5B). The variance in the estimate of age-at-length increases with increasing length; the 95% confidence interval (CI) for a halibut age 25 days is ± 6 days, but for a 90-day-old halibut the 95% CI is ± 19 days (Fig. 5). This relationship was used to con- vert length-classes into age-classes using the mean of each length-class (Lo 1985). I used the method outlined by Methot (1981, equa- tions 1-5) to compute the age-specific daily growth rates. Length-specific daily rate of growth and the variability in growth rate increased with increasing length: the slowest growth occurred just after transfor- mation (SL 6-10 mm), with daily growth < 0.3 mm/day, and maximum growth rates of about 1 mm/day oc- curred in juveniles 70-120mm SL (between 110 and 160 days) (Fig. 6). These growth rates are similar to those measured by Allen (1988) who estimated that juveniles 21-29 mm SL grew at 0.36 mm/day, and juve- niles 19-47 mm SL grew at 0.99 mm/day. Distribution and abundance Juvenile California halibut 16-70 mm SL were present in the bays during January-July 1987 and March- < 200 - 190 - 180 - 170 - A 160 - 150 - 140 - 130 - 120 - 110 - 100 - j V, 90 - 80 - 70 - ' 60 - 50 - 40 - MP 30 - 20 - 10 - - 80 100 120 AGE (days) 40 60 80 100 STANDARD LENGTH (mm) Figure 5 Gompertz relationship with 95% confidence intervals fitted to length-at-age of California halibut. (A) Standard length- at-age; (B) estimated age-at-length (mm). September 1988, and on the open coast between May and September 1988. The distribution of transforming larvae and juveniles on the open coast differed for the 1987 and 1988 year-classes, with very few larvae and no small juveniles taken on the open coast in 1987, 200 Fishery Bulletin 89(2), 1991 40 60 STANDARD LENGTH (mm) 80 100 120 AGE (days) Figure 6 Growth rates of juvenile California halibut estimated from the Gompertz parameters for length-at-age, shown with 95% con- fidence intervals. (A) Relationship between growth rate (mm/day) and length (mm); (B) relationship between growth rate (mm/day) and age (days). whereas transforming larvae and newly-settled juve- niles were common in 1988. Only the 1988 year-class was used to compare growth and mortality rates for juvenile halibut in bays and on the open coast. Further information on the distribution patterns of juvenile California halibut can be found in Kramer (1990). The length distribution of transforming larval and juvenile halibut varied with depth. The smallest length- class of halibut (<7mm SL) was taken at an average depth of 9.6m (SD 3.08, N 54). The mean depth of oc- currence decreased with increasing length up to a mean length of 67.8 mm SL (Fig. 7). At this size, the trend reversed, with mean depth of occurrence increasing with increasing length (Fig. 7). This pattern of length- at-depth indicates that transforming and newly-settled halibut move into shallower water along the open coast and into the bays, and halibut >70mm SL move into MEAN BOTTOM DEPTW — 2SE 40 80 120 160 200 240 MEAN OF LENGTH CLASS (mm) Figure 7 Mean depth of capture of California halibut by standard length- class (mm). Length-classes (mm) were < 7 (n 54), 8 (52), 9 (47), 10 (38), 11-15 (116), 16-20 (198), 21-25 (161), 26-30 (161), 31-35 (116), 36-40 (116), 41-45 (138), 46-50 (127), 51-55 (124), 56-60 (92), 61-65 (81), 66-70 (92), 71-75 (74), 76-80 (45), 81-85 (53), 86-90 (50), 91-95 (42), 96-100 (53), 101-110 (74), 111-120 (103), 121-130(101), 131-140(154), 141-150(154), 151-160 (109), 161-170(129), 171-180(143), 181-190(121), 191-200 (135), 201-210 (128), 211-220 (88), 221-230 (73), 231-240 (60), and 241-250 (47). OPEN COAST BAYS 2SE 70 90 AGE (DAYS) Figure 8 Abundance of juvenile California halibut by age-class for open coast and bay habitats in 1988. deeper water habitats within the bays (maximum bay depths were <5m) and eventually move out of the bays to the open coast (Fig. 7) (Plummer et al. 1983). The age of peak abundance of juvenile halibut in the bays in 1988 was equivalent to the average time re- quired for newly-hatched larvae to move from the continental shelf to the bays. Peak abundance was at about 70 days in both Mission Bay and Agua Hedionda Kramer: Growth and mortality rates of juvenile Paralichthys californicus 201 190 - 180 - i 1988 HALIBUT ABUNDANCE AT AGE MODEL 170 - 160 - \ ln(ABUNDANCE)-18 78-1 958-|n(AGE) 150 - 140 - 130 - NUMBER (Thousands) o o o o 70 - 60 - X. 50 - 40 - 30 - 20 - ■^~-*^^ 30 50 70 90 110 AGE (Days) Figure 9 Total abundance of juvenile California halibut by age-class summed over both open coast and bay habitats, with the fitted model used to estimate mortality rates. LagOOn (N mssion Bay 42,067, SE 8543; -/Va^ H edionda Lagoon 14,432, SE 3312), but there was a second large peak in Mission Bay for individuals at about 90 days (N 43,697, SE 7850) (Fig. 8). The class composed of transforming larvae (age 30 days) was the most abundant age class on the open coast in 1988 (N 191,553, SE 17,339) (Fig. 8). Abun- dance rapidly decreased with age on the open coast, with essentially no halibut 70-180 days of age present on the open coast (Fig. 8). The decline in abundance of halibut on the open coast corresponded to an increase in the bays (Fig. 8). Mortality Total age-specific abundance was determined by com- bining data from the bay and open coast habitats (Fig. 9). In the survey area, the total loss of juvenile halibut ages 30-115 days was estimated at 183,250 (95% CL of 148,800 and 210,350) (Fig. 9). Instantaneous mortality rates (z (t) ) were calculated by age-class using abundance-at-age, with age obtained from the linear regression of In-transformed abundance on In (age) (Table 3, Fig. 9), and the duration of each age-class calculated from the age-at-size relationship (Table 4) (Lo 1985). Instantaneous mortality rates (z (t) ) were highest (0.044) for the youngest juveniles, and decreased with increasing age but became constant (x 0.0124, SD 0.001) for juveniles 70 days of age and older (Table 4). I also calculated habitat-specific instantaneous mor- tality rates (z (t) ) for juveniles < 70 days of age that were taken only on the open coast, and for those Table 3 Sum of squared deviations (SS) between observed and calcu- lated or transformed abundance-at-age predicted from seven regression models applied to abundance-at-age for juvenile California halibut from 1988. Residual SS Model (xlO 9 ) 1 Survival rate (SR) increases linearly 6.48 SR = 0.264 + 0.0072 • age 2 Survival rate increases exponentially 7.99 SR = 0.174 * exp((0.0070 * age) + 1.0035) 3 Survival rate asymptotic 7.49 SR = 0.025 » age »exp(- 0.897) 4 Daily production 12.09 lnfdaily production) = 17.1 - 1.54 * In (age) 5 Daily production 12.60 In (daily production) = 9.55 - 0.0123* age 6 Log-transformed abundance on age 10.87 In (abundance) = 12.3 - 0.0229 * age 7 Log-transformed abundance on In (age) 4.38 ln(abundance) = 18.78 - 1.958 *ln(age) 88-115 days of age taken only in the bays (i.e., im- migration completed (Table 5). The apparent mortality in bays was much higher than that predicted from the combined bay and coast data, ranging from 0.043 to 0.037 for the bay model and from 0.011 to 0.014 for the total mortality model (Tables 4, 5). The age-specific mortality of halibut from the bays declined with increasing age, and was not constant as predicted by the total mortality model (Tables 4, 5). To test for differences between the age-specific in- stantaneous mortality rates (z (t) ) of the total popula- tion, and of the open coast and bays, I used ANCOVA on the age-specific mortality coefficient, Beta. Beta is related to the instantaneous mortality rate (z (t) ) by the equation: z (t) = Beta/t (Lo 1985). The Beta coefficients for the total population (1.94, SE 0.22) and the open coast juveniles (ages <70 days, Beta 3.58, SE 1.10) were significantly different (P<0.01) (Table 6). The dif- ference in the Beta coefficient between juvenile halibut on the open coast and the total halibut abundance-at- age is probably due to movement of halibut from the coast to the bays. Nearly half of the decline in abun- dance of juveniles along the coast could be caused by their movement into the bays (1.94/3.58 = 0.54). The Beta coefficients for the total population and the bay juveniles (ages 94-115 days) also differed significant- ly (P< 0.05), with a Beta of 0.69 (SE 0.77) for the total population, and 2.96 (SE 0.65) for the juveniles from the bays (Table 6). Mortality rates in the bays appear to be underestimated by the abundance-at-age model for the total halibut population. 202 Fishery Bulletin 89(2), 1991 Table 4 Instantaneous mortality rates for the 1988 year-class of juvenile halibut < 1 15 days (z(t, )) by age in days (t, ) computed from daily juvenile production estimates (P tl ) and age (t,) for 1988. Daily production estimates were obtained from log-linear model estimates of abundance- in-age classes adjusted for the number of days juveniles remained in the age-class. Percent of the total population in the bays by age- class is also given. Length-class (SL, mm) Age-class (t|) (days) Estimated abundance (») Percent in bays' Estimated daily production P« Pti-i-Pti ti-t,., t, = (t, + t,_ 1 )/2 z(t,) <10 30.3 181385.1 3.08 18138.5 11-15 43.3 90161.8 40.37 7706.1 10432.4 13.0 36.8 0.044 16-20 53.3 60025.3 69.92 6454.3 1251.8 10.0 48.3 0.016 21-25 61.8 44927.3 86.16 5615.9 838.4 8.5 57.5 0.015 26-30 69.3 35901.3 95.24 5056.5 559.4 7.5 65.5 0.013 31-35 76.2 29812.5 100.00 4586.5 469.9 6.9 72.7 0.013 36-40 82.5 25518.1 95.93 4183.3 403.2 6.3 79.3 0.013 41-45 88.4 22290.1 98.13 3910.5 272.7 5.9 85.4 0.011 46-50 94.1 19723.2 93.30 3586.0 324.5 5.7 91.2 0.014 51-55 99.5 17681.8 95.68 3336.2 249.8 5.4 96.8 0.012 56-60 104.8 15973.4 96.17 3132.0 204.2 5.3 102.1 0.011 61-65 109.9 14554.3 100.00 2910.8 221.2 5.1 107.3 0.013 66-70 114.9 13340.1 100.00 2722.4 188.4 5.0 112.4 0.012 Table 5 Habitat-specific instantaneous mortality rates of juvenile halibut < 115 days(z(t,))by age in days (t, ) computed from daily juvenile production estimates (P n ) and age (t,) for 1988. Halibut 30.3-69.3 days of age were from the open coast habitat, and halibut >88.4 days were from the bays. Daily production estimates were obtained from log-linear model estimates of abundance by habitat adjusted for the number of days juveni es remained in each age -class. Estimated Estimated daily production Age-class ft) (days) Habitat (») Pt, Pt.-i-Pti ti- t,^ t, = (t, + t,_ )/2 z(t,) Open coast 30.3 185384.8 18538.5 43.3 32540.9 2781.3 15757.2 13.0 36.8 0.065 53.3 11820.4 1271.0 1510.3 10.0 48.3 0.054 61.8 5746.9 718.4 552.6 8.5 57.5 0.051 69.3 3288.5 463.2 255.2 7.5 65.5 0.048 76.2 82.5 Bays 88.4 45824.8 8039.4 94.1 33279.7 6050.9 1988.6 5.7 91.2 0.043 99.5 25010.6 4718.9 1331.9 5.4 96.8 0.041 104.8 19175.6 3759.9 959.1 5.3 102.1 0.038 109.9 15035.1 3007.0 752.9 5.1 107.3 0.039 114.9 11972.7 2443.4 563.6 5.0 112.4 0.037 Bay abundance-at age estimate: In (abundance) = 33.68 - 5.12 *1 n(age) (residual SS = 4.567 x 10 7 ). Open coast abundance-at-age < estimate: In (abundance) - = 28.76 - 4.87*1 n(age) (residual SS = 6.219 xlO 7 ). Analysis of covariance where Beta is defined and age (z t = Beta/t). Table 6 (ANCOVA) on the age-specific mortality rate Beta by habitat, by the relationship between the instantaneous mortality rate Covariates N Age range (days) Beta SE F P Total population Coast population Total population Bay population . 4 4 5 5 43-69 43-69 94-115 94-115 1.94 3.58 0.69 2.96 0.22 1.10 0.76 0.65 19.62 <0.01 8.31 <0.05 Kramer: Growth and mortality rates of juvenile Paralichthys californicus 203 Rate of movements into bays I estimated the proportion of the population by age- class emigrating each day from the open coast to the bays by calculating the difference between the percent- age of juvenile halibut lost daily from the total popula- tion and from the open coast using age-specific instan- taneous mortality rates (Tables 4, 5) in the following equation: % emigrating/day = U\ — g-z(total population) _ Q — e~ z (°P en coast; )Y) x 100. The decline in abundance of juvenile halibut on the open coast between days 30 and 70 was 182,100, and for the total population was 145,500 (Tables 4, 5). During this time, the daily emigration rate increased from 1.99% for juveniles from age 30-43 days, to 3.67% from age 43-53 days, then declined slightly to 3.35% by 70 days. Discussion Extent of bay utilization Juvenile halibut appear to be dependent upon bays as nursery areas, since nearly all halibut between 76 and 115 days of age occurred in the bays rather than the open coast (Fig. 8). Transforming larvae and newly- settled juvenile halibut < 70 days old occurred on the open coast (97% of the transforming larvae were on the open coast), but over 95% of the total population of halibut >70 days were in the bays (Table 4). An alternative explanation for the decline in abun- dance of juvenile halibut on the open coast is that they move somewhere other than the bays, or suffer heavy mortality. If halibut moved offshore, one would expect a positive relationship between size of juvenile halibut (31-70 mm SL, or 76-115 days) and bottom depth. This is contrary to the observed size-structured distribution pattern (Fig. 7). The decrease in abundance of juvenile halibut on the open coast may have included higher in situ mortality rates, but the corresponding increase in abundance in the bays suggests that movement from the coast to the bays probably accounts for about half of the coastal decline. Advantage of bays as nursery areas Growth The potential advantages of using bays as nursery areas are increased growth and decreased mor- tality. Increased growth was not observed for juvenile English sole Parophrys vetulus in Oregon estuaries: they grow at about the same rate as juveniles on the Oregon coast, but were more variable in size-at-age than those on the coast (Rosenberg 1982). Similarly, growth rates of juvenile California halibut < 40 mm SL on the coast and in the bays were not significantly different. California halibut 70- 120 mm SL grew faster than all other length-classes with rates approaching 1 mm/ day (Fig. 6). These fast and variable growth rates occurred during the period when juvenile halibut oc- curred only in the bays (>115 days of age). Unfor- tunately, comparisons could not be made between open coast and bay habitats during this period of fast growth, which coincides approximately with a change in the food habits of halibut >55mm SL, from a diet composed primarily of small crustaceans (copepods, amphipods, mysids, and cumaceans) to one composed of an increasing proportion of fish by weight (mostly gobies) (Haaker 1975, Allen 1988, Drawbridge 1990). Juvenile halibut feeding on gobies in the laboratory re- main partially buried in the substrate, only striking at gobies passing within a distance of three headlengths (Haaker 1975). Gobies are abundant in bays (mean den- sity of Ilypnus gilberti in Mission Bay, 8.1/m 2 ), but not in shallow coastal waters <30m (Brothers 1975, Allen 1985, Plummer et al. 1983). The diet of larger juvenile halibut becomes increasingly piscivorous: juvenile halibut > 150 mm SL on the open coast eat primarily northern anchovies by weight (Plummer et al. 1983, Allen 1982). Predation risk Predation risk may be higher for small halibut on the open coast than in the bays. At least six fish species on the open coast are known to eat flat- fishes: these include California halibut, thornback ray Platyrhinoidis triseriata, fantail sole Xystreurys lio- lepis, bigmouth sole Hippoglossina stomata, speckled sanddab Citharichthys stigmaeus, and California lizard- fish Synodus lucioceps (Ford 1965, Allen 1982). Ford (1965) found many small halibut (TL <10mm) in the stomach contents of thornback rays, with a maximum of 15 newly-settled halibut in the stomach of one ray alone. The combined density of rays Platyrhinoidis triseriata, Urolophus halleri, and Gymnura mar- morata) on the shallow open coast (<10m) is about 100/hectare (Ford 1965). Speckled sanddab is the most abundant flatfish in shallow open coast waters, with a mean density of 950/hectare at Torrey Pines (Ford 1965, Allen 1982, Love et al. 1986, DeMartini and Allen 1984, Kramer 1990). Although the diet of speckled sanddab is composed primarily of mysids, they are probably capable of eating newly-settled halibut, since small unidentified flatfish juveniles have been found in their stomachs (Ford 1965). In the bays, two potential predators include the round stingray Urolophus halleri, and the staghorn sculpin Leptocottus armatus (Allen 1985, Tasto 1975, Babel 204 Fishery Bulletin 89(2), 1991 1967). Both occur along the shallow open coast as well, but are most abundant in bays (Allen 1985). Staghorn sculpin feed primarily on crustaceans (>50% by weight), but small diamond turbot Hypsopsetta guttu- lata have also been found in their stomachs (frequen- cy of occurrence 0.5%) (Tasto 1975). Over 94% of the diet by volume of round stingray is composed of mol- luscs, polychaetes, and crustaceans, but gobies--also have been found in their stomachs (Babel 1967). Other predators found both in the bays and on the open coast include barred sand bass Paralabrax nebulifer, spotted sand bass P. maculatofasciatus, and kelp bass P. clathratus. Spotted sand bass occur predominantly in bay habitats, barred sand bass occur ubiquitously in the bays and on the open coast, and kelp bass are associated with rock reef and kelp bed habitats on the open coast, but also have been taken as juveniles in bays (Allen 1985, Lane 1975). Kelp bass on the open coast feed mostly on northern anchovies and crabs, and have been found occasionally with flatfishes in their stomachs (Quast 1968). The diet of barred sand bass taken from bottom depths of 8-30 m on the open coast indicates that they forage close to the substrate, feeding on brachyuran crabs, mysids, pelecypods, and epibenthic fishes (mostly Porichthys notatus) (Roberts et al. 1984, Feder et al. 1974). Spotted sand bass oc- cur predominantly in bay habitats, feeding on crabs and other crustaceans, and on small kelpfish (Feder et al. 1974, Allen 1985). The juveniles of all three species are found commonly in Mission Bay, and are considered highly probable goby predators (Brothers 1975). The sand basses probably eat juvenile halibut also, as gobies and halibut share the same habitats. Comparison of predation risk must also include a measure of abundance or biomass of predators by habitat. The estimated density of the potential bay predators (round stingray, Paralabrax spp., and stag- horn sculpin) based on otter trawl surveys is 61/hectare in Agua Hedionda Lagoon, and only 3/hectare on the open coast (San Diego Gas and Electric 1980). The estimated density of two open-coast predators, the speckled sanddab and the thornback ray, is >1000/ hectare (Ford 1965). Based on this scanty information, it appears that predators are more abundant on the open coast than in the bays. Thus the possible advantages of using bays as nur- sery areas by juvenile halibut appear to be at least two- fold: (1) Decreased risk of predation on newly-settled juveniles, since fewer predators are known to occur there; and (2) increased potential for faster growth of juveniles >55mm SL because small fishes (gobies) are more abundant in bays than on the open coast (Haaker 1975, Allen 1985). Migration to bays The migration of larvae from spawning areas over the continental shelf to their juvenile nursery areas in em- bayments is thought to consist of two phases (Boehlert and Mundy 1988): Accumulation of larvae in the near- shore zone (Boehlert and Mundy 1988, Miller et al. 1986), and location and entering of the bays by trans- forming larvae and juveniles (Boehlert and Mundy 1988). The nearshore accumulation of larvae prior to movement to the bays is probably facilitated by the timing of spawning, the short duration of pelagic stages, and the vertical distributions of the postflex- ion and transforming larval stages. California halibut spawn throughout the year, with peak spawning dur- ing the winter and spring (Lavenberg et al. 1986, Walker et al. 1987). The spawning peak coincides with the period of minimum offshore transport of surface water in the Southern California Bight (Parrish et al. 1981, Jackson 1986). Offshore transport increases in late spring and summer due to increasing upwelling ac- tivity (Parrish et al. 1981, Jackson 1986). The seasonal shift in upwelling activity has been correlated with a seasonal cross-shelf shift in the zooplankton assemblage off San Onofre: from February to early April the com- munity was shifted onshore, and from mid-April to July the shift was offshore, corresponding to the period of increased upwelling (Barnett and Jahn 1987). The size distribution of California halibut larvae taken in plankton tows indicates that they move in- shore as they approach metamorphosis. Preflexion and flexion larvae (~2-6mm SL) occur in mid water >2km offshore, whereas transforming larvae occur at night in the neuston within 1km of shore (Moser and Wat- son 1990). My collections indicated that transform- ing larvae occur on the bottom during the day; thus transforming larvae appear to undergo a daily vertical migration, occurring at the surface at night and at the bottom during the day. Larvae of other Paralichthys species, yellowtail flounder Limandaferruginea, stone flounder Kareius bicoloratus, and the larval stages of several crustacean taxa have similar diurnal activity patterns (Weinstein et al 1980, Tsuruta 1978, Shanks 1988, Penn 1975, Smith et al. 1978). Postflexion and transforming halibut larvae may be transported shoreward by internal waves at night when they are in the neuston, with very little movement dur- ing the day while they are on the bottom, resulting in accumulation of larvae nearshore (Moser and Watson 1990). Surface slicks associated with internal waves may transport neustonic larval fishes and crustaceans onshore (Shanks 1988, Kingsford and Choat 1986). Recovery of drift bottles released <20 miles offshore in the Southern California Bight region is greatest Kramer: Growth and mortality rates of juvenile Paralichthys cahfornicus 205 between March and October, also suggesting increased onshore transport of surface water (Schwartzlose 1963). Once nearshore, transforming larvae or settled juveniles may search for bays by using longshore transport (Boehlert and Mundy 1988). Net longshore transport of shallow shelf waters in the Southern California Bight is to the south (Winant and Bratkovich 1981). Longshore current speed measured in shallow water (15 m) averages less than 5 cm/second; at this speed, after 12 hours longshore movement could be as great as 2km (Winant and Bratovich 1981). My data on the abundance of transforming larvae and newly-settled juveniles provide an estimate of the time required for halibut to locate and enter the bays from the open coast. The time required can be considered to be equivalent to the difference in the age of peak abundance between the coast and the bays. Halibut reached peak abundance in the bays at an age of about 70 days in 1988, whereas they were most abundant at age 30 days (transformation) on the open coast (Fig. 8). Thus the time required to locate and enter the bays was about 40 days in 1988 (70-30 = 40 days) (Fig. 8). Over a 40-day period, halibut potentially could be transported about 80 km alongshore (40 days x (2km at 12 hours in the neuston)), which is greater than the total distance between the northern sampling block at San Onofre and Mission Bay (64 km) (Fig. 1). I mea- sured the maximum distance between adjacent bays in southern California at less than 60 km, thus larvae using this transport mechanism would probably en- counter a bay within 30 days of reaching the shallow- water coastal environment. The potential cues used to find the entrances to bays include temperature, currents, odor, turbidity, and bot- tom substrate (Boehlert and Mundy 1988). A probable cue in southern California is temperature: during spring and summer, when larvae and juveniles are mov- ing into the bays, the temperature is as much as 5°C warmer in the bays than on the open coast (Kramer 1990). Once a bay entrance is located, the mechanism used to migrate into the bay probably is tidal transport, using incoming tidal currents to aid movement into the bay, and remaining at the bottom to avoid transport out of the bay (Weinstein et al. 1980, Boehlert and Mundy 1988, Fujii et. al. 1989, Tsuruta 1978, Weihs 1978, Runsdorp et al. 1985). To use tidal stream trans- port to move into bays, individuals must be able to orient to currents, control vertical movements, and re- main on the bottom during unfavorable currents. These abilities probably develop by the time larvae reach transformation (Boehlert and Mundy 1988, Weinstein et al. 1980). Only a few tidal cycles may be required for halibut to move from the entrance into the bay. Larval flounder {Paralichthys sp.) on the North Caro- lina coast use tides to augment movement into marshes, migrating to the surface during night flood tides and remaining on the bottom during ebb tides and during the day (Weinstein et al. 1980). In conclusion, California halibut settle either in bays or on the open coast, but ultimately nearly all halibut settling on the coast enter and use the bays as nursery areas during their first year of life. The advantages of bays as nursery areas may be a decrease in risk of mor- tality of newly-settled halibut, and an increase in growth of larger juveniles that feed upon the abundant small fishes in the bays. Acknowledgments This paper represents part of a dissertation submitted to University of California San Diego, Scripps Institu- tion of Oceanography. The research was supported by funds provided by the Habitat Program of the National Marine Fisheries Service and by the Southwest Fish- eries Science Center of the National Marine Fisheries Service. I thank John Hunter, Mike Mullin, and Richard Rosenblatt for their support and comments. Sandor Kaupp reared halibut larvae in the laboratory and generously provided samples. John Butler provided assistance with the apparatus and programs used for interpreting otoliths. Darlene Ramon assisted in pre- paring and photographing otoliths. Mike Davis, Steve Swailes, and many others were invaluable in the field. I also thank Douglas Chapman, Richard Charter, Nancy Lo, Geoffrey Moser, and William Watson for their advice and assistance. Citations Ahlstrom, E.H., K. Amaoka, D.A. Hensley, H.G. Moser, and B.J. Sumida 1984 Pleuronectiformes: Development. In Moser, H.G. , etal. (eds.), Ontogeny and systematics of fishes, p. 640-670. Spec. Publ. 1, Am. Soc. Ichthvol. Herpetol. Allen Press, Lawrence, KS. Allen, L.G. 1985 A habitat analysis of the nearshore marine fishes from southern California. Bull. South. Calif. Acad. Sci. 84:133-155. 1988 Recruitment, distribution, and feeding habits of young- of-the-year California halibut (Paralichthys cahfornicus) in the vicinity of Alamitos Bay-Long Beach Harbor, California, 1983-1985. Bull. South. Calif. Acad. Sci. 87:19-30. Allen, M.J. 1982 Functional structure of soft-bottom fish communities of the southern California shelf. Ph.D. thesis, Univ. Calif. San Diego, 577 p. Babel, J.S. 1967 Reproduction, life history, and ecology of the round stingray, Urolophus halleri Cooper. Calif. Dep. Fish Game, Fish. Bull. 137, 104 p. 206 Fishery Bulletin 89(2), 1991 Barlow, J.P. 1982 Methods and applications in estimating mortality and other vital rates. Ph.D. thesis, Univ. Calif. San Diego, 177 p. Barnett, A.M., and A.E. Jahn 1987 Pattern and persistence of a nearshore planktonic eco- system off southern California. Continental Shelf Res. 7:1-25. Barnett, A.M., A.E. Jahn, P.D. Sertic, and W. Watson 1984 Distribution of ichthyoplankton off San Onofre, Califor- nia, and methods for sampling very shallow coastal waters. Fish. Bull., U.S. 82:92-111. 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. Brothers, E.D. 1975 The comparative ecology and behavior of three sympatric California gobies. Ph.D. thesis, Univ. Calif. San Diego, 370 p. Butler, J.L. 1987 Comparison of the early life history parameters of Pacific sardine and northern anchovy and implications for species interactions. Ph.D. thesis, Univ. Calif. San Diego, 242 p. DeMartini, E.E., and L.G. Allen 1984 Diel variation in catch parameters for fishes sampled by a 7.6-m otter trawl in southern California coastal waters. Calif. Coop. Oceanic Fish. Invest. Rep. 15:119-134. Drawbridge, M.A. 1990 Feeding relationships, feeding activity and substrate preferences of juvenile California halibut, Paralichthys califor- nicus, in coastal and bay habitats. M.S. thesis, San Diego State Univ., 214 p. Emery, K.O. 1967 Estuaries and lagoons in relation to continental shelves. In Laugg, G.H. (ed.), Estuaries, p. 9-11. Am. Assoc. Adv. Sci. Publ. 83. 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, Cithari.chthys stigmaeus. Ph.D. thesis, Univ. Calif. San Diego, 243 p. Frey, H. 1971 California's living marine resources and their utiliza- tion. Calif. Dep. Fish and Game, Sacramento, 148 p. Fujii, T., H. Sudo, M. Azeta, and M. Tanaka 1989 Settling process of larvae and juveniles of Japanese flounder, in Shijiiki Bay, Hirado Island. Nippon Suisan Gak- kaishi 55:17-23 [in Jpn., Engl, abstr.]. Gadomski D.M., and J.H. Peterson 1988 Effects of food deprivation on the larvae of two flat- fishes. Mar. Ecol. Prog. Ser. 44:103-111. Gruber.D., E.H. Ahlstrom, and M.M. Mullin 1982 Distribution of ichthyoplankton in the Southern Califor- nia Bight. Calif. Coop. Oceanic Fish. Invest. Rep. 23:172-179. Haaker, P.L. 1975 The biology of the California halibut, Paralichthys califor- nicus (Ayres) in Anaheim Bay. In Lane, E.D., and C.W. Hill (eds.), The marine resources of Anaheim Bay, p. 137-151. Calif. Dep. Fish Game, Fish. Bull. 165. Jackson, G.A. 1986 Physical oceanography of the Southern California Bight. In Eppley, R.W. (ed.), Plankton dynamics of the Southern California Bight. Lecture notes on coastal and estuarine studies, p. 13-52. Springer-Verlag, NY. 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. Kicklighter, W.T. 1990 The relationship between somatic growth and tempera- ture in the formation of otoliths by juvenile California halibut, Paralichthys californicus. M.S. thesis, San Diego State Univ., 76 p. Kingsford, M.J., and J.H. Choat 1986 Influence of surface slicks on the distribution and onshore movements of small fish. Mar. Biol. (Berl.) 91:161—71. Kneib, R.T. 1987 Predation risk and use of intertidal habitats by young fishes and shrimp. Ecology 68:379-386. Kramer, S.H. 1990 Habitat specificity and ontogenetic movements of juvenile California halibut, Paralichthys californicus, and other flat- fishes in shallow waters of southern California. Ph.D. thesis, Univ. Calif. San Diego, 266 p. Krygier, E.E., and H.F. Horton 1975 Distribution, reproduction, and growth of Crangon nigri- cauda and Crangon franciscorum in Yaquina Bay, Oregon. Northwest Sci. 49:216-240. Krygier, E.E., and W.G. Pearcy 1986 The role of estuarine and offshore nursery areas for young English sole, Parophrys vetulus Girard, of Oregon. Fish. Bull., U.S. 84:119-132. Lane, E.D. 1975 Early collections of fishes from Anaheim Bay made be- tween 1919 and 1928. In Lane, E.D., and C.W. Hill (eds.), The marine resources of Anaheim Bay, p. 13-15. Calif. Dep. Fish Game, Fish. Bull. 165. Lavenberg, R.J., G.E. McGowan, A.E. Jahn, J.H. Peterson, and T.C. Sciarrota 1986 Abundance of southern California nearshore ichthyo- plankton: 1979-1984. Calif. Coop. Oceanic Fish. Invest. Rep. 27:53-64. 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 California Bight: An overview. Calif. Coop. Oceanic Fish. Invest. Rep. 27:84-106. Lo, N.C.H. 1985 Egg production of the central stock of northern anchovy, Engraulis mordax, 1951-1982. Fish. Bull., U.S. 83:137-150. Methot, R.D. 1981 Growth rates and age distributions of larval and juvenile northern anchovy, Engraulis mordax, with inferences on lar- val survival. Ph.D. thesis, Univ. Calif. San Diego, 203 p. Miller J.M., L.B. Crowder, and M.L. Moser 1986 Migration and utilization of estuarine nurseries by juve- nile fishes: An evolutionary perspective. In Rankin, M.A. (ed.), Migration: Mechanisms and adaptive significance. Contrib. Mar. Sci. Univ. Tex. suppl. vol. 27:338-352. Moser, H.G., and W. Watson 1990 Distribution and abundance of early life history stages of the California halibut, Paralichthys californicus and com- parisons with the fantail sole, Xystreurys liolepis. Calif. Dep. Fish Game, Fish. Bull. 174 (in press). Parrish, R.H., C.S. Nelson, and A. Bakun 1981 Transport mechanisms and reproductive success of fishes in the California Current. Biol. Oceanogr. 1:175-203. Kramer: Growth and mortality rates of juvenile Paralichthys cahfomicus 207 Penn, J.W. 1975 The influence of tidal cycles on the distributional pathway of Penaeus latisulcatus Kishinouye in Shark Bay, Western Australia. Aust. J. Mar. Freshwater Res. 26:93-102. Plummer, K.M., E.E. DeMartini, and D.A. Roberts 1983 The feeding habits and distribution of juvenile-small adult California halibut (Paralichthys californicus) in coastal waters off northern San Diego County. Calif. Coop. Oceanic Fish. Invest. Rep. 24:194-201. Quast, J.C. 1968 Observations on the food and biology of the kelp bass, Paralabrax clathratus with notes on its sportfishery at San Diego, California. In North, W.J., and C.L. Hubbs (eds.), Utilization of kelp-bed resources in southern California, p. 81-108. Calif. Dep. Fish Game, Fish. Bull. 139. Roberts, D.A., E.E. DeMartini, and K.M. Plummer 1984 The feeding habits of juvenile-small adult barred sand bass (Paralabrax nebulifer) in nearshore waters off northern San Diego county. Calif. Coop. Oceanic Fish. Invest. Rep. 25: 105-111. Rosenberg, A. A. 1982 Growth of juvenile English sole, Parophrys vetulus, in estuarine and open coastal nursery grounds. Fish. Bull., U.S. 80:245-252. Runsdorp, A.D., M. van Stralen, and H.W. van der Veer 1985 Selective tidal transport of North Sea plaice larvae Pleu- ronectes platessa in coastal nursery areas. Trans. Am. Fish. Soc. 114:461-470. San Diego Gas and Electric 1980 Encina power plant cooling water intake system demon- stration. San Diego Gas and Electric, San Diego, CA. Schwartzlose, R.A. 1963 Nearshore currents of the western United States and Baja California as measured by drift bottles. Calif. Coop. Oceanic Fish. Invest. Rep. 9:15-22. Shanks, A.L. 1988 Further support for the hypothesis that internal waves can cause shoreward transport of larval invertebrates and fish. Fish. Bull., U.S. 86:703-714. Smith, W.G., J.D. Sibunka, and A. Wells 1978 Diel movements of larval yellowtail founder, Limanda ferruginea, determined from discrete depth sampling. Fish. Bull., U.S. 76:167-178. Tasto, R.N. 1975 Aspects of the biology of Pacific staghorn sculpin, Lepto- cottus armatus Girard, in Anaheim Bay. In Lane, E.D., and C.W. Hill (eds.). The marine resources of Anaheim Bay, p. 123-135. Calif. Dep. Fish Game, Fish. Bull. 165. Tsuruta, Y. 1978 Field observations on the immigration of larval stone flounder into the nursery ground. Tohoku J. Agric. Res. 29:136-145. Walker, H.J. Jr., W. Watson, and A.M. Barnett 1987 Seasonal occurrence of larval fishes in the nearshore Southern California Bight off San Onofre, California. Estu- arine Coastal Shelf Sci. 25:91-109. Weihs, D. 1978 Tidal stream transport as an efficient method for migra- tion. J. Cons. Cons. Int. Explor. Mer 38:92-99. Weinstein, M.P., S.L. Weiss, R.G. Hodson, and L.R. Gerry 1980 Retention of three taxa of postlarval fishes in an inten- sively flushed tidal estuary, Cape Fear River, North Carolina. Fish. Bull., U.S. 78:419-436. Winant, CD., and A.W. Bratkovich 1981 Temperature and currents on the southern California shelf: A description of the variability. J. Phys. Oceanogr. 11:73-86. Abstract. -The larvae and juve- niles of two platytroctid species, Holt- byrnia latifrons and Sagamichthys abei, from the northeastern Pacific are described. Young individuals were collected at a broad range of mesopelagic depths, indicating that they occupy similar depths as adults. The relatively large larvae (~15mm) have well-developed teeth and fins while carrying the yolksac. Presence of <20mm sizes from most months of the year may indicate year-round spawning. Description of Young of the Mesopelagic Platytroctids Holtbyrnia latifrons and Sagamichthys abei (Pisces, Alepocephaloidea) from the Northeastern Pacific Ocean Tetsuo Matsui Marine Life Research Group Scnpps Institution of Oceanography A-027 La Jolla, California 92093 Manuscript accepted 10 January 1991. Fishery Bulletin, U.S. 89:209-219 (1991). The only published description of larval platytroctids is that of Beebe (1933). However, Parr (1960) in his review of the family (then known as Searsidae) believed larval platytroc- tids of 11- 17 mm could not be iden- tified even to genus, and showed that Beebe's detailed description of the larvae identified as Bathytroctes ros- tratus was based on several species. In a recent revision of the family, Matsui and Rosenblatt (1987) recog- nized five species off California. None are congeners. The earliest larvae of the two commonest species, Holtbyr- nia latifrons * and Sagamichthys abei, are identifiable by their photophores. This report describes the young of H. latifrons and S. abei and presents catch data from the Scripps Institu- tion of Oceanography (SIO) Fish Col- lection and depth of capture data from a series of opening-closing net tows. Material and methods Material used in this study is from the SIO Fish Collection and from samples collected on seven cruises sponsored by the Marine Life Re- * Holtbyrnia latifrons may be a junior synonym of//, baucoti Mayer and Nalbant 1972. How- ever//, baucoti was inadequately described, and I have failed to gain additional informa- tion on the holotype. search Group (MLRG) of SIO. Most SIO Fish Collection samples and some from the MLRG cruises were captured in 3-m Isaacs-Kidd mid- water trawls (IKMT). Other MLRG collections were made with 2-m Stra- min nets and 1-m plankton nets. The 1-m nets were attached to modified Leavitt devices (Leavitt 1938) that allowed the nets to be opened and closed by messengers to sample dis- crete depth intervals. Their sampling depths were either recorded by TSK depth-distance recorders on the nets and activated when the nets were sampling, or estimated from records taken on a Benthos time-depth re- corder attached to the bottom net in the cast. Eight depth intervals from surface to nearbottom were sampled at stations 850-1350 m deep, and 12 depth intervals at stations of 1370 m to ~1800m. Two samples in the Fish Collection were taken in opening- closing Bongo nets (McGowan and Brown 1966). A total of 167 (young and adult) H. latifrons and 220 5. abei were ex- amined in the study. The description of H. latifrons is based on measure- ments and counts of 157 individuals (1 larva, 12 transitional specimens, and 144 juveniles). For S. abei, 1 lar- va, 4 transitional specimens, and 37 juveniles were measured and counted. Samples of H. latifrons were gener- 209 210 Fishery Bulletin 89(2), 1991 ally collected near the coast at 22-38°N, off the west coast of California and Baja California (but mainly 28-33°N). Samples of S. abei were also collected near the coast at 28-38°N, with one individual from 4°N, 142°W. Most specimens were initially preserved in formal- dehyde and later transferred to 70% isopropanol. Length measurements of small specimens were taken with an ocular micrometer of a dissecting microscope and by dividers for larger measurements. All S. abei and H. latifrons examined had a flexed notochord. Since the notochord extended more than 1 mm beyond the hypurals in the earlier stages, the standard length (SL) measurements of the larvae were taken from the snout to posterior tip of notochord, or to the most posterior extension of the hypurals, whichever was greater. Head length (HL) measurements were taken from the tip of the snout to the posterior margin of opercle. In individuals with torn or curled gill covers, the anterior base of pectoral fin was substituted as the posterior reference point for HL. Photophore nomenclature follows Parr (1960) as modified in Matsui and Rosenblatt (1971). Key to young platytroctids off California la Photophores (or melanophores in the shape of photophores) in gular region (G0 2 ; Figs. 1, 2) from yolksac through later juvenile stages 2 lb Neither photophores nor patch of melanophores in gular region 3 2a Photophores and silvery reflector present on anterior dorsal margin of eye (00) and on subopercle (SBO) from yolksac stage; intraventral photophore (IVO) present from yolksac stage; opercular open- ing extending dorsally to about mideye; body coloration generally whitish blue Sagamichthys abei 2b 00 and SBO photophores absent; IVO appearing near end of yolksac stage; opercular opening ex- tending dorsally to top of eye; epidermal layer lightly pigmented from yolksac stage, becoming dark- brown in juvenile Holtbynia latifrons 3a Broad edentulous space between innermost tooth of each premaxilla; small photophore between bases of pelvic fins (IVO) present after yolk is resorbed; gill opening extending dorsally to a level with top of eye Maulisia argipalla 3b Only narrow edentulous space between innermost tooth of each premaxilla; photophores absent; gill opening on a level with mideye 4 4a Nasal sac nearly bordering maxilla; premaxilla not meeting medially with part extending laterally Pellisolus eubranchus 4b Nasal sac midlength of snout; premaxilla meeting medially, with none along lateral margin of mouth Mirorictus taningi Description Holtbyrnia latifrons (Fig. 1 J Pigmentation The least developed individual exam- ined has pigmented eyes, pigmentation on the shoulder organ and at the site of the posterior gular organ (G0 2 ); the posterior margin of opercle and the dorsal portion of yolksac are lightly pigmented (Fig. 1A). This is the only specimen examined that is considered a lar- va; the remaining yolksac stages are already beginning to form the juvenile pigment pattern, and are termed transitional individuals. The most prominent pigmen- tation in these is the black tissue lining the digestive tract from the mouth and branchial chamber to the anus. Melanophores are concentrated at the dorsal and ventral margins, and on the fins and fin bases. Pigmen- tation increases with size, and in late juveniles the en- tire body as well as the head (except for the translu- cent top of skull) is nearly black in color. Morphometries The larva (Fig. 1A) has a small head and mouth and an oblong eye that is nearly twice as long horizontally as vertically. The transitional speci- men (Fig. IB) is much more adult-like with the head and eyes nearly doubling in size, and the eyes almost round. Much of the growth in transition stages is in the head region, and although the body length of 12-17 mm Matsui: Description of young platytroctids from northeastern Pacific Ocean 21 I A YOLK c Figure 1 Camera lucida drawings of the young of Holtbyrnia latifrons: (A)Ml-2-l, 15-mm larva; (B) SIO 65-439, 14-mm transitional specimen; (C) 65-443, 14-mm juvenile. GO, = posterior gular organ; IVO = intraventral organ. juveniles is shorter than that of the larva, their head length is two times greater (Table 1). Head length in- creased from 15% of SL in the larva to the adult pro- portion of 30% of SL at the end of the transitional period. Head length of juveniles is proportionately larger than that of adults even from the earliest stages, measuring 29-37% of SL in the 12-18 mm SL range, with maximum of >40% SL at 36-45 mm SL (Table 2). Head depth increased from 9% SL in the larva, to 10-17% SL in transitional specimens, to 16-23% SL in juveniles as large as 50 mm. 212 Fishery Bulletin 89(2), 1991 Table 1 Measurements (in mm) and counts of Holtbyrnia lat ifrons. SL = standard length, HL = head length, Hd = head depth at angular, asterisks = SIO Sample nc . Ml-2-1 Mll-1-3 M2-A7 M10-1R-2 M13stl * 67-62 *63-447 *65-439 *63-447 •66-398 •65-443 M7st3 •67-101 Lengths — SL 15 15 15 14 14 13 14 14 14 14 14 16 16 HL 2.3 3.8 3.3 2.4 2.3 3.0 4.2 4.8 3,8 4.3 4.5 5.1 6.0 SL minus HL 13 11 12 12 12 10 10 8.9 11 9.3 9.2 10 9.7 Hd 1.4 1.4 1.7 1.4 1.2 1.4 2.0 2.3 1.8 2.2 2.4 2.7 2.9 Yolksac 3.7 3.6 3.3 3.2 3.0 3.0 2.6 2.5 2.3 Eye (long.) 0.72 0.92 0.91 1.1 0.76 1.0 1.3 1.6 1.3 1.5 1.4 1.6 2.0 Eye (vert.) 0.48 0.60 0.58 1.0 0.44 0.52 0.83 0.83 0.52 0.83 0.92 0.75 1.2 Maxilla 1.1 1.4 1.2 1.2 1.0 1.3 1.5 2.1 1.7 1.7 2.1 2.1 2.5 Counts Dorsal rays 11 15 14 14 12 16 13 16 15 16 18 19 Anal rays 10 13 10 — 13 13 13 14 15 14 16 12 Pectoral rays + 2 — 4 + ? 4 + ? 4 + ? Branchiostegals 2 4 5 8 8 8 8 8 8 8 8 8 8 Gill rakers + 2 + 3 1 + 10 1 + 11 + 8 2 + 12 2+12 2 + 13 3 + 13 Pseudobranchiae 2 2 9 2 2 2 2 2 2 2 2 2 2 Dentition Vomerine 1 2 2 2 1 2 2 2 Palatine 1 + 1 1 + 1 1 + 1 1 + 1 1 + 2 1 + 1 1 + 1 Basihyal 1 1 1 Dentary 3 + 3 1 + 1 7 + 9 12 + 12 7 + 7 10 + 10 10+12 13+17 10 + 12 Mid-dentary 1 + 1 1 + 1 + 1 2 + 2+1 Maxillary 2+1 1 + 1 2 + 1 3 + 3 Premaxillary 1 + 1 2 + 2 3 + 4 4 + 3 3 + 5 4 + 5 5 + 5 4 + 3 Table 2 Summmarized counts and morph smetrics (% SL) of Holtbyrnia latifrons. SL = standard length, HL = head length, He = head depth at angul ar. SL Gill Pseudo- Dentition % SL Vomer- Pala- Pre- Max- Mid- Basi- Basi- Max- (mm) rakers branchiae ine tine maxillary illary Dentary dentary hyal branchia HL Hd Eye illary Yolksac 0-1 + 0-11 2 0-2 0-1 0-4 0-2 0-12 0-1 0-1 15-35 8-17 5-11 7-15 12-15 1-3+11-14 2 1-2 1 4-5 0-5 7-12 0-2 0-1 29-33 15-19 9-12 12-15 16-20 2-5 + 13-15 2-4 2-3 1-2 4-6 2-8 10-24 0-6 0-4 29-42 14-19 7-13 11-16 21-25 2-5 + 13-17 2-4 2-5 1-2 4-8 5-16 15-27 0-2 0-4 33-39 16-20 10-12 14-17 26-30 5-7+16-19 4-5 1-2 1-2 4-9 5-19 16-30 0-5 0-3 35-45 17-23 11-13 14-20 31-35 6-7+16-17 5 2-4 1-2 7-10 5-23 15-26 0-5 0-2 36-39 16-19 11-12 17-20 36-40 6-8+18-19 6 1-2 1-4 6-9 11-33 19-23 4-5 0-2 0-1 38-42 19-20 10-12 18-21 41-45 7-9+18-21 5-6 2-3 1-4 7-11 11-33 20-27 3-6 0-1 0-5 38-42 19-22 10-13 17-20 46-50 7-8+18-19 8 2-3 1-4 10-12 15-32 33-40 2-4 0-1 0-4 36-42 19-21 10-11 18-21 51-55 7 + 19 7 2-3 2 9 27-29 30-34 1-4 0-1 0-5 37 20 11 21 Eye length increased from about 5% SL in the larva to as much as 11% SL in some of the transitional specimens and to 11-13% SL in 20-45 mm juveniles. Maxilla length increased from 7% of SL in the larva, to 9-15% SL during transition, to 17-21% SL in 30-50 mm juveniles. Fins All specimens had the notochord flexed with 19 principal caudal rays present. In larvae, narrow finfolds extend anteriorly along the dorsal (to head) and ven- tral (to anus) body margins. Only the basal parts of the dorsal and anal rays are discernible on the larva (Fig. 1A), but there are 10-13 anal and 11-16 dorsal rays in Matsui: Description of young platytroctids from northeastern Pacific Ocean 213 the transitional specimens (Table 1). Two pelvic rays are discernible in the most advanced transitional speci- men (SI065-439). Adult counts of dorsal (17-20), anal (14-16), and pelvic (9) fin rays are usually present in 30-mm juveniles. Pectoral fin rays form last, with the first rays appearing at approximately 23 mm SL and adult counts (16-20) by 45 mm SL. Branchial region Branchiostegal rays appear early and only the larval specimen and the least developed transitional specimen had fewer than the adult count of 8. Gill rakers are absent in the larva, but as many as 12 are on one side of the 1st gill arch in the transitional specimen. Nearly all of these gill rakers are on the lower arch. Only one transitional specimen (SI065-439) had gill rakers on the 1st epibranchial. The lowest count for juveniles was 1 on the epibranchial and 12 total on the 1st arch, and ranged from 1 to 5 on the upper arch and 12 to 19 total in juveniles <20mm SL. The smallest individual with the adult gill raker count of 25 was 27 mm SL and all individuals 42 mm and larger had counts in the adult range of 25-30. Medial gill rakers of the 3d and 4th arches are in an uninterrupted row in juveniles as small as 19 mm, but no medial gill rakers form on the 1st and 2d arches until after 20 mm. By 45mm SL, there are about 6 medial rakers on the epibranchial and on the ceratobranchial of the 2d arch, and 5 on these elements of the 1st arch. There are 2 pseudobranchiae in the larva and tran- sitional specimens. The smallest individual with a 3d pseudobranchium measured 17 mm and the smallest with a 4th was 20 mm. No specimen <50mm SL had attained the highest adult count of 8. Counts varied as much as 4 between individuals of similar lengths. Dentition In the study material, only the larva is toothless. Teeth on dentary, premaxilla, maxilla, vomer, palatines, basihyal, and on the lateral face of the dentary (mid-dentary teeth) appear during tran- sition. Teeth are easily dislodged, contributing sub- stantially to individual variation in counts. The most advanced transitional specimen (SI065-439; Table 1) had a single medial tooth on the basihyal, a tooth on each palatine, and a total of 7 premaxillary, 2 maxil- lary, 24 dentary, and 2 vomerine teeth. Only one tran- sitional specimen had mid-dentary teeth. There are fewer maxillary than dentary teeth in the early stages, but this gradually changes and the numbers are about even in individuals 25 mm and larger (Table 2). Highest count of premaxillary teeth among transitional indi- viduals was 4 on a side. The numbers increased to 4-6 in 13-20 mm juveniles, with counts as high as 10 at 50mm SL. The inner pair of premaxillary teeth point horizontally beeinninff from about 25mm SL. Fre- quently, a second smaller tooth forms adjacent to the 1st. Mid-dentary teeth are probably more susceptible to damage than other dentition and most individuals smaller than 30 mm were without them, although counts of 4 on one side occurred as early as 17 mm. In- dividuals 30-50 mm long usually had 4-5 mid-dentary teeth. In the youngest stages, a single tooth was usually present on each palatine. The number variably in- creased to as many as 4 in individuals over 30 mm. Most prejuveniles and early juveniles had a medial tooth on the basihyal, with occasional individuals with 2-4 in a medial row. Basibranchial and mesopterygoid teeth appeared after 40 mm, and ectopterygoid teeth were only found in the adults. Photophores Photophores of young platytroctids are oriented horizontally (Matsui and Rosenblatt 1971). The posterior gular organ (G0 2 ; Fig. 1) is the only photo- phore present during most of the yolksac stage. It is an opaque spot outlined by dark pigment in the larva, and forms at the posterior, narrow end of a black, conically-shaped pouch. Near the end of the yolksac stage, the intraventral organ (IVO) develops inside a silver-lined, anteriorly facing pouch. The photophore at the subopercle (SBO) appears later in juveniles. However, it is considered rudimentary as it is sur- rounded by opaque tissue and is without an anterior opening. These photophores are covered over and lost in larger individuals. Additional photophores begin appearing in some juveniles of 26 mm but are uncommon until after 28mm. Unlike earlier photophores, they face ventral- ly and persist in adults. Earliest to appear are (1) a transversely barred thoracic organ (THO) located on the ventral body margin midway between the pectoral and pelvic fins, (2) two elliptical supraventral organs (SVO) located anterolateral to the bases of the ventral fins, and (3) two elliptical supraanal organs (SAO), lateral to the anal opening. Other adult photophores appear soon after and include a transversely barred midventral organ (MVO) located anteroventral to the SVO, the elliptical branchiostegal organs (BRO), the infracaudal organ (ICO) located on the ventral margin of the caudal peduncle, the pectoral organ (PO) located on the ventralmost ray of pectoral fins, and a longi- tudinally barred jugular organ (JO) located between the bases of the pectoral fins. All adult photophores are usually present by 50 mm. Sagamichthys abei (Fig. 2) Pigmentation The single larval specimen is nearly unpigmented. Most heavily pigmented areas are the eye, shoulder organ, subopercular photophore (SBO), 214 Fishery Bulletin 89(2). 1991 YOLK c Figure 2 Camera lucida drawings of the young of Sagamichthys abei: (A) M13-1A, 16-mm larva; (B) SIO66-390, 14-mm transitional specimen; (C) M7 Sta 3, 16-mm juvenile. 00 = orbital organ; GO., = posterior gular organ; SBO = subopercular organ; IVO = intraventral organ. and posterior gular photophore (G0 2 ), with light pig- mentation in the mouth, gill chamber, and on the dor- sal region of the yolksac. In the transitional specimen, the mouth, stomach cavity, and intestine are lined with black tissue. These blackened areas show through the translucent muscu- lature, darkening the ventral half of the head and body anterior to the vent. Muscles between the anal and dor- sal fins take on a bluish-black tinge, which spreads anteriorly and posteriorly from that area in larger in- dividuals, with the area around the caudal peduncle darkening last. Among the more advanced transitional Matsui: Description of young platytroctids from northeastern Pacific Ocean 215 Table 3 Measurements (in mm) and counts of Sagamichthys ibei. SL = standard length HL = head length, Hd = head depth at angular; asterisks = SIO. 1 Sample no. 113stlA M4-stl * 66-390 *66-390 ♦66-422 •63-165 M7-st3 *66-371 ♦70-8 •65-439 •57-41 ♦75-472 •54-122 Lengths SL 16 16 14 14 14 13 16 16 16 17 17 18 20 HL 2.8 4.2 4.5 4.7 4.5 3.7 5.3 5.1 5.0 5.9 5.5 5.8 7.2 SL minus HL 13 12 9.5 9.3 9.5 9.3 11 11 11 11 12 12 13 Hd 1.5 2.0 2.3 2.3 2.1 2.1 2.7 2.3 2.6 2.4 2.4 2.3 3.2 Yolksac 4.0 4.2 2.5 2.5 3.2 2.6 Eye (long.) 0.92 1.1 1.1 0.92 1.0 0.76 1.7 1.8 1.3 1.7 1.3 1.6 2.2 Eye (vert.) 0.54 0.58 0.64 — 0.72 0.60 1.2 0.83 0.80 1.2 1.2 1.2 1.6 Maxilla 1.1 1.5 1.8 — 1.7 0.92 — 2.2 2.1 2.4 2.4 2.3 3.0 Counts Dorsal rays 16 16 14 13 13 14 16 14 16 15 16 15 Anal rays 14 13 14 13 12 14 15 13 15 14 14 14 Pectoral rays 3 3 5 - 5 7 4 7 5 Branchiostegals 8 8 8 8 8 8 8 8 8 8 8 8 Gill rakers + 9 0+10 0+11 + 8 + 7 1 + 12 1 + 11 3 + 13 2 + 11 2 + 10 3 + 12 3 + 13 Pseudobranchiae 2 2 2 2 2 2 2 3 3 3 3 3 3 Dentition Vomerine 2 2 2 2 2 1 2 2 2 2 Palatine 1 + 1 + 1 1 + 1 + 1 1 + 1 1 + 1 1 + 1 1 + 1 1 + 1 1+0 Basihyal 1 3 3 1 1 4 1 4 7 4 6 5 Basibranchial 1 2 4 Dentary 3 + ? 3 + 5 7 + ? 8 + ? 5 + 5 ? + 14 10 + 8 15 + 11 17 + ? 14 + 15 17 + ? 12 + ? Mid-dentary 3 + 2 4 + ? 2 + 3 2 + ? ? + 3 Maxillary 4 + 6 — 9 + 6 10 + ? 9 + ? 11 + 14 12 + ? Premaxillary 2 + ? 3 + 3 — 1 + 1 2 + ? 6 + 5 7 + 7 5 + 5 ? + 5 5 + 5 6 + 8 8 + ? specimens, sparse epidermal pigment is found around the lower jaw, with a broken line at the midline be- tween the anal and dorsal fins. This pigmentation spreads and intensifies in the juveniles, resulting in the blue-gray to black coloration. Morphometries A large yolk mass extends from the cleithrum to about halfway to the anus in the single larval specimen, which has a small head (headlength = 17% of SL) and mouth and undeveloped fins. Head length in transitional specimens nearly doubles to 26-33% of SL. Maxillary length increased from 7% SL to 9-13% SL during transition. Body length shortened or remained about the same (Table 3), as body length behind the head shortened. Head length and depth in transitional specimens are similar proportionately to adults, but the mouth is smaller than in adults and ex- tends only to mideye, instead of behind the eye. Max- illary length is 9-13% SL during transition and 14-16% SL in adults. Eye length in the larva is proportionate- ly similar to eye diameter in adults; however, eye depth is only half of the length in the former. Head, mouth, and eye are proportionately largest in the 20-60 mm juveniles. Head length was mainly 35-38% of SL at this range (Table 4) but <30% of SL in individuals larger than 150mm SL. Fins All specimens have a flexed notochord and 19 principal caudal rays. Dorsal and anal fin rays are ab- sent in the larval specimen, but pterygiophores of 11 anal and 13 dorsal rays are present. Counts of 13-16 dorsal, and 12-15 anal rays of the transitional speci- mens (Table 3) were nearly in the adult range of 16-18 dorsal and 14-16 anal finrays. Most juveniles over 20 mm SL had the adult counts. Pectoral fin rays are absent in specimens <18mm. There were 0-7 pelvic rays at that length. Adult counts on all fins are found in juveniles >30mm SL. There are 9-10 (usually 9) pelvic and 14-18 pectoral fin rays in the adult. Branchial region Branchiostegal rays and gill rakers are absent in the larval specimen. Except for one indi- vidual with 7 rays, transitional specimens have the adult count of 8 branchiostegal rays. There are 8-11 216 Fishery Bulletin 89(2). 1991 Table 4 Summarized counts and morphometries '% SL) of Sagamichthys abei. SL = standarc length HL = head length, Hd = head depth at angu. ar. SL Gill Pseudo- Dentition % SL Vomer - Pala- Pre- Max- Mid- Basi- Basi- Max- (mm) rakers branchiae ine tine maxillary illary Dentary dentary hyal branchial HL Hd Eye illary Yolksac + 0-11 2 0-2 0-1 0-3 - 3-8 0-3 18-33 9.3-16 6-8 7-13 16 1-3 + 11-13 2-3 2 1 5-7 6-9 10-15 0-3 0-4 31-33 14-16 8-15 13-14 17-20 2-4 + 11-13 2-3 2 1 5-9 7-14 12-18 1-4 4-8 0-8 32-43 14-21 8-11 13-18 21-25 3-5 + 13-15 3 2 1-3 8-9 7-19 12-25 1-3 4-8 2-9 34-35 15-18 9-10 14-16 26-30 4 + 15 4 3 2 13 19 17 5 9 10 35 16 10 16 31-35 4-5 + 15-16 4 2 2-5 9-11 17-26 15-19 3-5 6-12 6-9 34-36 16-22 8-10 14-15 36-40 6+16 4-5 2-5 2-3 10-14 25-28 16-22 6-9 5-9 4-6 30-39 17-20 9-11 16-18 41-45 6-7 + 17-18 5 2 2 11-12 28-29 28 6 3-7 6 32-36 8-16 7-9 15-19 46-50 6-7 + 15-16 6 2 2 11-12 34-35 22 6-7 5 6 34-38 18-20 9-10 17-20 51-55 7 + 19 7 2 2 17 41 26 10 7 14 41 22 9 20 56-60 7 + 16 6-7 2 2 13 37 23 9 2 9 35 20 8 18 61-65 7+16 7 2 2 13 40 24 10 4 5 36 19 8 17 gill rakers on the lower arch of the 1st gill arch in tran- sitional specimens, but none on the upper arch. Epi- branchial gill rakers were present in all juveniles ex- amined (Table 4). The adult complement of 23-26 was present at about 50 mm. An uninterrupted row of medial gill rakers is present on the 3d and 4th arches in 17-mm juveniles, but those on the 1st and 2d arches appear at 30-35mm SL. There are 2 pseudobranchiae in the larva and transitional specimens. The number generally increased to 3 in juveniles of about 15mm SL, and 6-7 in juveniles >45mm SL. Dentition The single larval specimen is toothless, but all transition specimens have teeth on the den- tary, premaxillary, and basihyal, and, except for two individuals, on vomer and palatines as well (Table 3). Earliest appearances of maxillary, mid-dentary, and basibranchial teeth were in 16-17 mm juveniles. As in H. latifrons, there were more dentary than max- illary teeth in early stages; however, the number of maxillary teeth increased rapidly, becoming equal to that of the dentary by about 25 mm SL and more numerous after 35 mm (Table 4). Vomerine teeth numbered 2-5 and palatines 1-3 in the 16-50 mm SL range. In individuals 16 mm and larger, there were 3-12 teeth positioned around the perimeter of the basihyal, and 2-14 on the basibranchial. After a gradual increase, mid-dentary teeth numbered about 10 in individuals >50mm SL. Mesopterygoid teeth appeared after 65mm SL, and ectopterygoid teeth were found only in large adults. Photophores The posterior gular organ (G0 2 ), sub- opercular organ (SBO), intraventral organ (IVO), and the orbital organ (OO) are the only photophores in in- dividuals as large as 50mm; they are covered over or lost in adults. The IVO photophore is located behind the yolksac (Fig. 2A). It becomes enclosed in a sub- conical, black pouch with a silvery, inner lining during the transitional period. The wider end of the pouch faces anteriorly and is covered by a transparent mem- brane. The photophore is located at the narrower, posterior end of the pouch, and is directed anteriorly. The other photophores are similarly housed in anterior- ly facing, subconical pouches with a wider anterior opening and greater silvery surfaces. Adult photophores face ventrally. Most begin appear- ing in juveniles. At 50-60 mm, the following organs begin to form: the jugular organ (JO) located between the bases of the pectoral fins; the thoracic organ (THO) behind the JO; the midventral organ (MVO) behind the THO and just before the pelvic fin bases; a pair of supraventral organs (SVO) just dorsal to the base of each pelvic fin; and the supra-anal organs (SAO) on both sides of the anus. The JO photophore first appears as one or more longitudinal bars (as in H. latifrons), but at about 65 mm begins to transform into a short transverse bar. Two more transverse bars, the THO and MVO, form behind the JO on the widely flattened ventral margin. Also appearing by 65 mm are several branchiostegal organs (BRO) on the branchiostegal rays, a postorbital organ (POO) just behind the eye, a pair of postanal organs (PAO) at about the middle of the anal fin, and an infracaudal organ (ICO) ventrally on the caudal peduncle. The anterior gular organ Matsui: Description of young platytroctids from northeastern Pacific Ocean 217 YS 15 30 40 50 60 70 80 200 SL(mm) Figure 3 Length-frequency distribution of Holtbyrnia latifrons in SI0 collections, 1950-73. 30 n p i i w i i p V pi YS 15 30 40 60 80 100 120 140 160 180 >200 SL(mm) Figure 4 Length-frequency distribution of Sagamichthys abei in SIO collections, 1949-79. (GO]), located just behind the mandibular symphysis, and the pectoral organ (PO) on the pectoral fins, ap- pear by 75 mm. The anal organ (AO), located just before the anus, appears in large adults. Remarks Some young specimens of//, latifrons and S. abei with nearly exhausted yolksacs retain char- acteristics of the earliest stages, i.e., fragile mouth, short head, and toothless mouth. Only the fin rays are relatively well developed. They appear to be starved in spite of the presence of yolk material and may in- dicate that they need to feed while the yolk is present. The severely underdeveloped head and mouth indicate that development either lagged from the very early stages or had regressed from a more advanced stage. These larvae were assumed to be atypical, and were not included in the larval description. Distribution In SIO samples, H. latifrons larger than 30 mm are rare and individuals >50mm are nearly absent (Fig. 3); relatively small H. latifrons may avoid nets as large as a 3-m IKMT. Decline in size of S. abei in our samples is more gradual. The collection contains a number of individuals as large as 80 mm and a few even larger (Fig. 4). These results point to the inadequacy of nets 4- 4n JAN Bn | FEB U. 8-, MAR 3 - | < 4- 3 - -L— JUN JUL > 16 Q -, APR MAY 8 AUG SEP OCT NOV DEC T~1 YS 15 30 40 50 60 YS 15 30 40 50 60 SL(mm) Figure 5 Length-frequency distribution of Holtbyrnia latifrons in SIO collections, 1950-73, presented by month of capture. as large as 3-m IKMT in sampling larger juveniles of both species. The presence of individuals smaller than 20 mm dur- ing most of the year is interpreted as year-round spawning (Figs. 5, 6). These figures represent samples collected over many years pooled by month. Depth distribution for both species is estimated to be 300-1000 m (Matsui and Rosenblatt 1987). On MLRG-SIO cruises, 63 opening-closing Leavitt net tows made in this depth zone collected 12 specimens of H. latifrons and 3 of 5. abei. Fourteen net tows sampling shallower than 300 m and 30 tows sampling deeper than 1000 m on these cruises failed to catch 218 Fishery Bulletin 89(2). 1991 JAN Ll_j^^ FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC YS 15 30 40 100 120 SL(mm) Figure 6 Length-frequency distribution of Sagamichthys abei in SIO collections, 1949-79, presented by month of capture. o 100 200 300 m too | 500- g Jc 600- Q. 03 O 700 800 - 900 - 1000- 1280 ^ — 16 "I J-S H S Lw\A-Lw^ 20 22 24 26 28 133 217 YOLK SAC JUVENILE Body Length (mm) ADULT Figure 7 Sampling depths of opening-closing plankton net tows which captured Holtbyrnia latifrons (H) and Sagamichthys abei (S). Table 5 Time and depth of capture of Holtbyrnia latifrons in opening-closing net tows. Asterisks = yolksac stage. Standard length (SL) in mm. SL Date Time Depth (m) *14 2/09/72 0921-1021 450-740 *14 . • 14 - *14 4/09/72 1146-1355 580-860 14 - •15 4/08/72 1639-1901 580-720 16 - - 28 - - *15 2/02/71 1745-1845 669-706 18 • - 558-595 17 12/20/71 1748-1848 350-520 18 2/02/71 2320-0025 433-573 *15 4/13/71 0013-0152 286-563 either of these species. Samples of H. latifrons were from nets that had sampled depths of 290-860 m, and S. abei of 330-860 m (Fig. 7). All of these individuals were 20 mm and smaller, providing evidence that young stages occur over the entire depth range of the species. Large larval size and early appearance of presumed swimming and foraging capabilities, noted in this study, are apparent specializations for developing at these nutrient-poor depths. Table 6 Time and depth of 'apture of Sagamichthys abei in opening-closing nets . Standard length (SL) in mm. SL Date Time Depth (m) 20 2/09/72 0921-1021 333-450 14 4/08/72 1146-1355 580-860 133 11/04/69 1230-1300 615-680 15 4/08/72 1639-1901 580-720 213 3/28/70 2246-2350 250-500 Absence of day and night differences in depth of cap- ture (Tables 5, 6) add to the previous evidence (Matsui and Rosenblatt 1987) suggesting the absence of diel migration in platytroctids. Hart (1973) and Fitch and Lavenberg (1968) mention the migration of young 5. abei to within 200 m of the surface at night, but did not give the source of their information. However, records of platytroctids from depths 200 m and shal- lower are extremely rare and the single record from the Pacific Ocean may be an error (Matsui and Rosen- blatt 1987). Hydrographic data (Fig. 8) taken in an area where most Leavitt net tows were made show that samples represented in Figure 7 were taken below the thermo- cline at depths where the temperature range was ~4-8°C. CalCOFI (California Cooperative Oceanic Matsui: Description of young platytroctids from northeastern Pacific Ocean 219 S(% o )33.0 33.5 34.0 34.5 TCC) 5 10 15 35.0 35.5 36.0 20 25 30 1 % • • • 250 » • * • 500 t/i 5 750 f tA i Q. Ixl *A a 1000 1250 — •* Temp. * Salinity • 2 o 2 my L ? | 2 \ 3 4 5 6 1 1 1 1 Figure 8 Temperature, oxygen, and salinity distribution at Station 7403G (32°32.4'N, 117°34.5'W), taken 27 March 1974 in the San Diego Trough. Fisheries Investigation) station data from the area show seasonal changes in temperature below 300 m averaging less than 0.5°C (Lynn et al. 1982). Oxygen values of less than 1 mL/L occurred at these depths (Fig. 8). Gill filaments are highly developed (for platy- troctids) in both species and are well adapted for this environment (Matsui and Rosenblatt 1987). Acknowledgments This study was supported by the Marine Life Research Program, the Scripps Institution of Oceanography's component of the California Cooperative Oceanic Fisheries Investigation. I thank Dr. Richard Rosenblatt for loan of specimens from the SIO Fish Collection, and his staff for cura- torial assistance. I am grateful for advice received from Dr. Geoffrey Moser, and for his comments on an earlier draft; and to Barbara Sumida MacCall for useful com- ments on illustrations of the larvae and juveniles. Other figures were prepared by the Marine Life Research Il- lustration department. I wish to thank Ron McCon- naughey for his able assistance during the MLRG cruises, and other participants too numerous to be named here. Citations Beebe. W. 1933 Deep-sea fishes of the Bermuda oceanographic expedi- tions. Family Alepocephalidae. Zoologica (NY) 16(2):15-93. Fitch, J.E., and R.J. Lavenberg 1968 Deep-water fishes of California. Univ. Calif. Press, Berkeley, 155 p. Hart, J.L. 1973 Pacific fishes of Canada. Fish. Res. Board Can. Bull. 180, 740 p. Leavitt, B.B. 1938 The quantitative vertical distribution of macroplankton in the Atlantic Ocean basin. Biol. Bull. (Woods Hole) 74:376-394. Lynn, R.J.. K.A. Bliss, and L.E. Eber 1982 Vertical and horizontal distributions of seasonal mean temperature, salinity, sigma-t, stability, dynamic height, oxy- gen, and oxygen saturation in the California Current, 1950-1978. Calif. Coop. Oceanic Fish. Invest. Atlas 30, 513 p. Matsui, T., and R.H. Rosenblatt 1971 Ontogenetic changes in patterns of light organs in sear- sids and the taxonomy of Sagamichthys and Persparsia. Copeia 1971:440-448. 1987 Review of the deep-sea fish family Platytroctidae (Pisces: Salmoniformes). Bull. Scripps Inst. Oceanogr. Univ. Calif. 26:1-159. Mayer, R.F., and T.T. Nalbant 1972 Additional species of fishes in the fauna of Peru Trench. Results of the 11th cruise of R/V "Anton Bruun", 1965. Rev. Roum. Biol. Ser. Zool. 17(3): 159-165. McGowan, J. A., and D.M. Brown 1966 A new opening-closing paired zooplankton net. Scripps Inst. Oceanogr., SIO Ref. 66-23, La Jolla, CA, 56 p. Parr. A.E. 1960 The fishes of the family Searsidae. Dana-Rep. Carlsberg Found. 51, 108 p. Abstract.- Length-at-sexual- maturity and spawning periodicity of the tuna baitfish Encrasicholina devisi, E. heterolobus, SprateUoides delicatulus, S. gracilis, S. lewisi, and Archamia zosterophora were studied at two exploited fishing grounds and one unexploited site in the Solomon Islands. All species became sexually mature and capable of spawning at 70% of the largest size, except the apogonid A zosterophora which ma- tured at a larger size (80%). There was little site-related variability in length-at-flrst-spawning, although S. lewisi from Tulagi grew to a larger size and was larger than S. lewisi from other sites when it spawned for the first time. There was no evidence that length-at-first-spawning was af- fected by commercial baitfishing. The timing and intensity of spawn- ing of each species were extreme- ly variable. All species spawned throughout the year, with one or two periods of more intense activity. The spawning peaks of the same species at different sites did not coincide, and no proximate stimuli correlated with spawning by any species at all sites. The timing of major spawning events was not random, nor did fish spawn as soon as they reached ma- turity. Spawning events at the three sites correlate with particular en- vironmental conditions, especially moon phase and, less importantly, rainfall and temperature. These re- sults are not consistent with the hypothesis that spawning is timed to maximize either local dispersal or the potential for larvae to find suitable food. Lack of clear proximate stimuli for spawning among the six species examined makes it difficult to predict the timing of major spawning events by these species. Maturation, Spawning Seasonality, and Proximate Spawning Stimuli of Six Species of Tuna Baitfish in the Solomon Islands David A. Milton Stephen J.M. Blaber CSIRO Division of Fisheries, Marine Laboratories PO Box 120, Cleveland. Queensland 4163. Australia Manuscript accepted 17 December 1990. Fishery Bulletin, U.S. 89:221-237 (1991). The pole-and-line fisheries for skip- jack tuna Katsuwonus pelamis in the Pacific are dependent on adequate supplies of suitable bait. Engraulids (genus Encrasicholina) and dussumi- erids (genus SprateUoides) are the basis of the Solomon Islands tuna baitfishery, the largest in the region with catches of over 2000 1 annually (Anon. 1988). Knowledge of the reproductive biol- ogy of the main bait species may be important in developing management regimes to minimize the impact of the baitfishery on these species. Some aspects have been studied in the Sol- omon Islands (Evans and Nichols 1984) and elsewhere in the Pacific (Tester 1955; Tiews et al. 1971; Leary et al. 1975; Dalzell and Wankowski 1980; Conand 1985; Dalzell 1985, 1986, 1987ab; Lewis et al. 1983; McCarthy 1985; Clarke 1987) and in southeast Asia (Dharmamba 1960; Tham 1965; Luther 1979; Chen 1984, 1986). These studies suggest both genera spawn year-round, with peri- ods of increased spawning during spring and summer (Leary et al. 1975, Tiews et al. 1971, Luther 1979) or with the change of monsoon (Dal- zell and Wankowski 1980, Dalzell 1987b) or periods of high zooplankton production (Sitthichockpan 1972). How- ever, timing of peak spawning is vari- able, both temporally (e.g., Dalzell 1987b) and between regions. There have been several reviews (e.g., Scott 1979, Lam 1983, Bye 1984) of the importance of various cues which stimulate gonadal devel- opment and cause fish to spawn (proximate factors). Among temper- ate species, temperature and light are the most common cues (Scott 1979, Bye 1984). Other cues, such as food supply, moon phase, and rain- fall, have also been suggested as important for spawning by tropical marine fishes (Johannes 1978, Lam 1983, Walsh 1987). However, the proximate stimuli that arouse in- creased spawning activity among baitfish remain obscure. Tester (1955) found that variations in egg production by Encrasicholina pur- pureas in Hawaii could not be ade- quately explained by temperature, salinity, or moon phase. Similarly, Muller (1976) showed that fluctua- tions in salinity and zooplankton biomass accounted for only 30% of the variation in egg production of E. heterolobus at Palau. Such poor cor- relations suggest that spawning may be random, or fish may begin to spawn as soon as they are physio- logically capable of doing so. The length at which Encrasicho- lina and SprateUoides become sex- ually mature appears to be variable both between and within countries of the southwestern Pacific (Dalzell and Wankowski 1980, Conand 1985, Dal- zell 1985, McCarthy 1985, Dalzell 1987ab, Wright 1989), and these fish may adjust their life-history param- eters to changes in their demography 221 222 Fishery Bulletin 89(2), 199) or environment (Stearns and Crandall 1984). Such variability could have impor- tant implications for the baitfishing in- dustry. Juveniles could be caught before they have had a chance to spawn, because the liftnets used in the baitfishery are not size-selective and catch all sizes of En- crasicholina and Spratelloides (see Evans and Nichols 1984). The aims of this study were, first, to examine the length at maturity and the spawning seasonality of the six most abundant baitfish species (Encrasicho- lina devisi, E. heterolobus, Spratelloides delicatulus, S. lewisi, S. gracilis, and the apogonid Archamia zosterophora); sec- ondly, to determine whether spawning is random or correlated with environmental cycles; and, thirdly, to assess the effects, if any, of the commercial fishery on these reproductive parameters. Methods and materials Sampling Samples of six species of baitfish were collected from three sites each month for two years. Fish from com- mercial bait catches came from Munda and Tulagi, and fish from an unexploited control site came from Vona Vona, Solomon Islands. Sites and sampling methods are described in Milton et al (1990). Samples were usually collected between 2100 and 2200 hours. Two sites, Munda and Vona Vona, were in enclosed coral reef lagoons with little water movement and were ap- proximately 20 km apart. The other site, Tulagi, con- sisted of a series of protected bays which opened into a narrow channel between two islands and was located over 300km southeast of the other sites. Each month a random sample of over 100 fish of each species from each site was preserved in 4% formaldehyde and taken to the laboratory for analysis. On each sampling occa- sion, the water temperature, cloud cover, moon phase, and wind speed and direction were recorded. At least two 5-minute horizontal plankton tows (~100m 3 ) were made with a 250-jjmm mesh net (0.5 m diameter) 1 hour prior to fish collection. The daily rainfall at a village adjacent to each site was recorded throughout the study. Laboratory analyses Fish were measured (standard length (SL) + 0.5 mm) and weighed (± 0.001 g) and the gonads were removed and weighed ( ± 0.001 g). The gonads of up to 10 females of each species and of any other females with enlarged Criteria used foi Table 1 staging female gonads of baitfish. Stage Histology (1) Immature Oogonia present (2) Developing/resting Previtellogenic oocytes (3) Maturing Yolk precursor stage; some non-staining yolk (4) Ripe Non-staining yolk; developed chorion (5) Running ripe Homogeneous red-staining yolk; oocytes hydrated; development complete (6) Spent Atresion of ripe oocytes plus previtellogenic oocytes; presence of post-vitellogenic follicles oocytes were randomly subsampled from each site sample each month. Gonads were embedded in paraf- fin wax, sectioned at 9fim, and stained with Ehrlich's haemotoxylin and eosin (McManus and Mowry 1964). Gonad maturation stages were defined following Cyrus and Blaber (1984) and Hunter and Goldberg (1980). Each gonad was staged, based on the relative numbers of cells at each developmental stage (Young et al. 1987; Table 1), and the presence of any postovulatory follicles was noted. Gonosomatic indices (GSI) were calculated as the ratio of wet gonad weight to somatic weight (total weight minus gonad weight) expressed as a percentage. Size-at-sexual-maturity was determined from the length at which a fish developed ripe eggs (Table 1: stage 4). Among fish that were only examined macro- scopically, the criterion for sexual maturity was a gonosomatic index greater than the minimum GSI of fish that were shown histologically to have ripe eggs. Fish examined histologically were considered to be in spawning condition if more than 85% of the eggs were fully hydrated, running-ripe (Table 1: late stage 5). The mean GSI ± 2 standard errors of these fish was calculated for each species at each site. The GSI value of the lower 95% confidence limit of the mean GSI of spawning fish was used as the indicator of spawning among fish examined macroscopically. The proportion in spawning condition in each sample was then calcu- lated from the fraction of the sample (examined both histologically and macroscopically) with a GSI greater than this value. This criterion was used, as it was a con- servative estimate of the real proportion spawning. Plankton samples were split in half with a Folsom plankton splitter and one fraction was dried to a con- stant mass to provide an estimate of relative zooplank- ton biomass. The other fraction was sorted and the number of Encrasicholina eggs and larvae (Delsman 1931), the number of other eggs, larvae, and potential Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 223 prey (from those found by Milton et al. 1990) were counted. Spratelloides eggs are demersal (Leis and Trnski 1989) so were not sampled by this method. Data analyses The proportion of each 1 mm length-class that were sex- ually mature was compared between sites. The normal approximation to the binomial distribution was used to estimate the 95% confidence limits of the proportion mature in any length-class. Confidence limits of the estimated proportion of the mature population spawn- ing each month were calculated in a similar way. To assess whether the timing of spawning was ran- dom, we calculated the proportion of the population spawning from the fraction of the entire sample of each species at each site during the study. This proportion was then compared with the proportion spawning each month. Monthly proportions greater than the 95% con- fidence limits to the normal approximation to a bi- nomial distribution were scored as a plus sign. A non- parametric runs test (Sokal and Rohlf 1981) was used to test whether the distribution of plus signs was random. The relationship between the proportion of fish in each monthly sample (of each species at each site) that were longer than the mean adult length at that site, and the proportion in that sample that were spawning, was examined using linear regression. A significant positive relationship between these proportions was used to assess whether fish spawned as soon as they were physiologically capable of doing so. The relationships between possible proximate stimuli and the proportion spawning were compared by step- wise regression analysis. Variables that improved the fit were added to the model until the best three- variable model was obtained or the most significant fit was found. The proportion of each sample of each species in spawning condition was transformed by its arcsine square-root to reduce possible bias due to an excess of low values (Sokal and Rohlf 1981). The following 11 variables were compared for each species at each site: (1) sea-surface temperature, (2,3) prey biomass and density, (4) moon phase, (5) tide range, (6) cloud cover, (7) wind speed, and (8-11) rainfall between samples. Previous studies of spawning by these species (Dalzell 1985, 1987b) and other tropical inshore fishes (e.g., Johannes 1978, Walsh 1987) suggested that these vari- ables may be important cues for these species. Initial- ly, salinity and current speed were also measured, but as they varied little, they were not included in the analysis. The variable moon phase was calculated by fitting a curve of the form y = sin(2nx) + cos(2nx) where x = number of days since the last full moon prior to the start of sampling divided by 29.5 (days in a lunar month). This variable has higher values about the new and full moon. Rainfall data were regressed in several ways to assess the influence they may have on baitfish spawning: total rainfall since previous sample (usually 1 month)(Total); number of days since rain (Days); number of days since rain >25mm (Days 25 mm); and number of days of rain since previous sample (Rain). Rainfall, cloud cover, and wind variables were trans- formed by their square root to normalize skewed values. Except for moon phase, positive relationships between the proportion of each species spawning and proximate variables are indicative of greater spawn- ing activity at higher values. To assess whether all variables were independent, proximate variables were correlated with one another. Principal component analysis (Sokal and Rohlf 1981) was also used to help identify variables that covaried within and between sites. A subset of proximate vari- ables that behaved independently was identified and analysed separately. Where a potential stimulus had been measured in several ways (e.g., rainfall), the most biologically appropriate was chosen. Results Physical environment The mean sea-surface temperature and monthly rain- fall at each site varied seasonally during the sampling period (Figs. 1-3). The temperature ranged from 27.5° to 32.5°C at all sites, with lower temperatures during the dry season (May-September). Temperatures were lower in 1987 than 1988 at all sites. Rainfall occurred in all months at all sites. The amount of rainfall and its monthly distribution pattern were similar at the two closest sites, Munda and Vona Vona (r 2 0.48, P< 0.05), but Tulagi had a higher rainfall and different pat- tern of distribution. Maturation A total of 1 159 fish of six baitfish species from the three sites were examined histologically, including over 200 of each of the four most abundant species: Encrasi- cholina devisi, E. heterolobus, Spratelloides delicatulus, and S. lewisi. Encrasichollna Histological examination of ripe ova- ries of E. devisi and E. heterolobus showed that almost all oocytes were in the most advanced stage of develop- ment, with a few at the yolk precursor stage (Table 1) or less developed. Although many females (~15%) of those examined had hydrated eggs, no post-vitellogenic follicles or spent fish (Stage 6) were observed. 224 Fishery Bulletin 89(2). 1991 ll MAMJJASONDJFMAMJJASONDJFMAM 1987 1988 1989 Time (months) Figure 1 Mean monthly sea temperature and rainfall at Munda, Solomon Is., March 1987-May 1989. (J MAMJJASONDJFMAMJJASONDJFMAM 1987 1988 1989 Time (months) Figure 3 Mean monthly sea temperature and rainfall at Tulagi, Solomon Is.. March 1987-May 1989. 600 -p 500 w 400 MAMJ JASONDJFMAMJ JASONDJ FMA 1987 1988 1989 Time (months) Figure 2 Mean monthly sea temperature and rainfall at Vona Vona, Solomon Is.. March 1987-May 1989. In both Encrasicholina species, fish with a GSI greater than 2% (Table 2) were sexually mature. The size at maturation of E. devisi was similar at all sites with 50% of the fish being sexually mature at 44-45 mm (Fig. 4a). At Munda and Vona Vona, fish beyond this length were capable of spawning, but at Tulagi the smallest E. devisi in spawning condition was 52 mm (Table 2). Sexual maturity in E. heterolobus was reached at approximately 43 mm at Munda and Vona Vona (Fig. 4b) and 45 mmat Tulagi. Hydrated eggs were not observed in E. heterolobus less than 50 mm, except at Munda where the smallest potential spawner was 45 mm (Table 2). Spratelloldes The three Spratelloides species at the three sites were mature at similar lengths. Spratel- loides gracilis and S. lewisi were sexually mature and had hydrated eggs at 35 mm (Table 2, Fig. 4c), except at Tulagi where the smallest S. lewisi with running ripe eggs was 40 mm. Spratelloides delicatulus reached sex- ual maturity at 37 mm (Fig. 4d), and running- ripe eggs were found in fish beyond this length. However, their length-at-maturity was not significantly different from the other Spratelloides species (P>0.1). For all species, a gonosomatic index of over 2% correlated with fish having ripe eggs (stage 4) in the ovary. Most oocytes in the ovaries of females of these species were at a Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 225 Table 2 Minimum gonosomatic index (GSI) values used as criteria to estimate proportion of sexually mature fish (stage 3) and proportion of fish spawning (late stage 5) at each site, and minimum size (SL, mm) of fish in spawning condition (based on values obtained from fish examined histologically). N = sample size. Site Sexual maturity GSI (%) N Spawning Species GSI (%) Length N Archamia zosterophora All 2.5 57 3.8 37 23 Encrasicholina devisi Munda Vona Vona Tulagi 2.0 1.8 1.8 86 80 58 9.0 7.0 5.7 45 45 52 39 41 11 Encrasicholina heterolobics Munda Vona Vona Tulagi 2.0 1.9 1.6 134 135 84 8.5 8.8 10.0 45 50 51 26 24 31 Spratelloides delicatulus Munda Vona Vona Tulagi 2.0 1.0 1.2 60 35 43 5.7 5.9 5.5 37 37 38 34 23 25 Spratelloides lewisi Munda Vona Vona Tulagi 1.5 2.8 1.0 235 107 24 13.3 11.2 12.5 35 35 40 210 90 10 Spratelloides gracilis All 2.0 12 11.9 35 6 60 30 40 50 60 Length (mm) 70 100- (c) 80- s 60" / S. gracilis o c5 a. 40" 20- n - S. lewisi 70 100 (b)£ heterolobus 100 J* 60 60 S. delicatulus A. zosterophora 40 50 60 Length (mm) 70 70 Figure 4 Logistic curves of best fit which describe the change in proportion of ripe eggs (stage 4) in the ovaries of six species of baitfish with increasing fish length. 226 Fishery Bulletin 89(2). 1991 (a) Munda £ devisi fcUv^ MAMJJASONDJFMAMJJASONDJFMAM (b) Vona Vona 1 f f r ^N t MAMJJASONDJFMAMJJASONDJFMAM (c) Tulagi MAMJJASONDJFMAMJJASONDJFMAM 1987 1988 1989 Time (months) 5 a 0) a. E heterolobus MAMJJASONDJFMAMJJASONDJFMAM T+T MAMJJASONDJFMAMJJASONDJFMAM /K-K . . MAMJJASONDJFMAMJJASONDJFMAM 1987 1988 1989 Time (months) Figure 5 Monthly variation in proportion (±95% confidence limits) of female Encrasicholina devisi and E. heterolobus spawning at three sites in the Solomon Islands, March 1987-May 1989. similar stage of development. Fish with spent ovaries (Stage 6) were observed but were rare (<1%). Archamia zosterophora This species matured at 37 mm and was capable of spawning at this length (Table 2, Fig. 4d). A gonosomatic index value greater than 2.5% corresponded with sexual maturity (Table 2). Spawning seasons All non-parametric tests showed a significant deviation from random spawning for all species at all sites (P< 0.05). The distribution of deviations (plus signs) was either clumped or regular. For no species at any site were there significant positive correlations between the proportion of larger fish and the proportion spawning (P>0.05). Encrasicholina Spawning activity by the two Encra- sicholina species showed both seasonal and interannual variation (Fig. 5). Both species had one or two major peaks in spawning each year. However, the pattern was different each year. Both species had a major peak in spawning activity early in the year (March-May) dur- ing 1987 and 1988, except for E. devisi at Munda in 1987 when most fish spawned later (Sept.-Oct.). The peak spawning was usually followed by several months when a small proportion of the population was spawn- ing. Patterns of spawning at Munda and Vona Vona were more similar than at Tulagi. Encrasicholina Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 227 Figure 6 Monthly variation in proportion (±95% confidence limits) of female Spratelloides delicatulus and S. lewisi spawning at three sites in the Solomon Islands, March 1987-May 1989. (a) Munda S. delicatulus MAMJ J ASOND J FMAMJ JASONDJ FMAM (b) Vona Vona MAMJ JASONDJ FMAMJ JASONDJ FMAM (c) Tulagi X MAMJ JASONDJ FMAMJ JASONDJ FMAM 1987 1988 1989 Time (months) S. lewisi MAMJ JASONDJ FMAMJ JASONDJ FMAM C C 5 a. 0) o a MAMJJASONDJFMAMJJASONDJFMAM tt«tT MAMJ JASONDJ FMAMJ J ASONDJ FMAM 1987 1988 1989 Time (months) heterolobus also had fewer, smaller peaks in spawning activity during most of 1987 at Tulagi and at all sites during the first five months of 1989. Encrasicholina devisi in Munda and Vona Vona did not have a March- May spawning peak in 1989. Spratelloides Two Spratelloides species also showed inter- and intraspecific differences in spawning activ- ity between sites and between years (Fig. 6). Spratel- loides delicatulus spawned continuously during 1987 at both Munda and Vona Vona, but the proportion spawning declined in the next 17 months. At Tulagi, the pattern was reversed, with less spawning during 1987 than in 1988-89 when there was a single major protracted spawning season from December to March (Fig. 6c). Spratelloides lewisi showed seasonal spawn- ing activity at all sites. Fish from Vona Vona and Tulagi had a similar pattern during 1987-88, with increased spawning activity from October to May. However, the proportion spawning at Tulagi (50%) was much lower. The spawning activity of S. lewisi at Munda showed no seasonal increase during 1987, although during 1988 a higher proportion was spawning during the middle of the year (Fig. 6). Too few S. gracilis and A. zosterophora could be ob- tained to determine their spawning seasons. However, a proportion of the fish sampled of both species was in spawning condition for several months of the year (Fig. 7). Archamia zosterophora data from Munda showed that some spawning activity occurred through- 228 Fishery Bulletin 89(2), 1991 100- (a) Munda 80- A zosterophora 60" 40" 20" 1 „ n- ,,,,/ V MAM J J ASONDJ F MAM J J ASONDJFMAM 100- (b) Vona Vona O) c 80- C g. 60 " /4. zosterophora C 40" 0) o n 3 20 " / \ ,,,/ y i i~N i i » i i i i i \ MAM J J ASONDJ F MAM J J ASONDJFMAM 100 80 60" 40 20 (c) Vona Vona S gracilis t MAMJJ ASONDJ FMAMJJ ASONDJ FMAM 1987 1988 1989 Time (months) Figure 7 Monthly variation in proportion ( ± 95% confi- dence limits) of female Archamia zosterophora at Munda and Vona Vona (a,b), and female Spratelloides gracilis at Vona Vona (c), March 1987-May 1989. out the year. At Vona Vona, however, there was a peak in spawning activity in samples taken in October 1988. The spawning data for S. gracilis are consistent with data for other Spratelloides at Vona Vona, with a decline in spawning activity during late 1988 and 1989 (Fig. 7). Eggs and larvae Encrasicholina eggs were present in the zooplankton at each site most months of the sampling period (Fig. 8). The abundance of eggs did not vary greatly at each site during the period, al- though the samples from Vona Vona had fewer eggs than the other sites. Overall, the mean Encrasicholina egg density for the entire sampling period varied from 0.08 ± 0.03/m 3 at Vona Vona to 0.43 ± 0.10 at Munda. The overall mean teleost egg density was also lower at Vona Vona (1.34 ± 0.21/m 3 ) and highest at Munda (4.98 ± /m 3 ). Encrasicholina egg density did not cor- relate with total egg density at any site (P>0.3) or with the proportion of spawning E. devisi or E. heterolobus (P>0.5). No Spratelloides larvae were found, although apogonid larvae were present but could not be iden- tified to species. The density of Encrasicholina larvae followed a similar pattern at all sites (Fig. 8): larvae were pres- ent in most months, increasing every three or four months. In contrast to egg densities, Encrasicholina larvae reached higher densities at Vona Vona (0.05 ± 0.01/m 3 ) than at Munda (0.03 ± 0.006/m 3 ). Other fish larvae were also in higher densities at Vona Vona (1.15 ±0.27) than at the other sites (Munda 0.93 ± 0.26 and Tulagi 0.60 ± 0.22). The density of Encrasi- cholina larvae was significantly correlated with the density of other fish larvae at Munda (r 2 0.62, P<0.05), but not at other sites. Proximate spawning stimuli To assess the independence of the 11 proximate stimuli, all variables were correlated with one another (Table 3). At all sites, most measures of rainfall were signif- icantly correlated (P<0.05). Total rainfall was nega- tively correlated with moon phase at Vona Vona and Tulagi, and with zooplankton biomass at Tulagi. Cloud cover was correlated with total rain and days of rain at Tulagi (P<0.05). Zooplankton biomass and density were correlated at Munda and Tulagi. However, there were no consistent correlations at all sites between non- rainfall variables (Table 3). Principal component analysis was also performed separately on the proximate variables from each site. Rainfall variables at each site had loadings on the first three factors that were similar in magnitude and direc- tion. Zooplankton density and biomass also covaried at each site. Temperature, moon phase, tide, and wind had loadings that varied independently for the first three factors at each site. These variables, zooplankton den- sity, and the number of days of rain (which covaried least with other rainfall variables) were used in a separate stepwise regression analysis. Encrasicholina Analysis of the relationships between all 11 environmental parameters and the proportion of each sample spawning showed no consistent pattern for either species of Encrasicholina (Table 4). At Munda, spawning of E. devisi positively correlated Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 229 < (a) Munda eggs I t , , , T , fOvT-f-r-r . 0.3 0.2 0.1 Larvae AMJ JASONDJFMAM (b) Vona Vona AMJJASONDJFMAM 7 Pt 'l 'l' l ff^ i ^ T -r-T^ AMJJASONDJFMAM AMJJASONDJFMAM 0.0-'-' — r AMJJASONDJFMAM Time AMJJASONDJFMAM Time Figure 8 Variation in monthly abundance of Encrasicholina eggs and larvae ( ± range) at three sites in the Solomon Islands, April 1988-May 1989. with wind and time since rain. At Vona Vona there was a negative correlation with wind strength and positive correlation with zooplankton biomass and tide range. Spawning at Tulagi correlated most strongly with time since heavy rainfall (>25mm), temperature, and zoo- plankton biomass (Table 4). When all data were in- cluded, time since heavy rain was the only significant correlate. Stepwise regression analysis of the six independent proximate stimuli (Table 5) showed a much poorer fit. The only site that showed a significant relationship was Vona Vona, where full moon, greater tidal range, and low wind accounted for 40% of the variation in spawn- ing oiE. devisi. Although not significant, days of rain had a similar coefficient in the equations of best fit at both Munda and Tulagi (Table 5), which suggests that rainfall had a similar effect on spawning at these sites. The significant proximate stimuli for E. heterolobus differed between sites (Table 4). The best fit was ob- tained at Vona Vona (r 2 0.44, P<0.01) where greater spawning occurred when cloud cover was low and moon phase approached full. Spawning was nega- tively related to total rainfall and days since rain at Munda (r 2 0.3, P<0.05), and there was a negative 230 Fishery Bulletin 89(2). 1991 Table 3 Correlations between independent environmental variables *P<0.05 . See " Data analyses section for explanation of variables. Moon Tide Cloud Wind Rain* Days Variable Temp. Biomass Density phase range cover speed Total Days 25 mm Rain Munda Temperature x -0.02 -0.12 -0.32 -' -0.20 0.41 0.18 0.46* 0.04 -0.19 0.37 Biomass x 0.73* -0.21 -0.08 0.21 -0.38 0.08 -0.36 0.00 0.30 Density x 0.05 0.23 0.29 -0.23 -0.03 -0.22 0.05 -0.01 Moon phase X 0.06 0.01 0.50* -0.29 0.3 0.51* -0.29 Tide range X -0.07 0.07 -0.09 0.47* 0.12 -0.05 Cloud cover X 0.25 0.30 0.01 0.11 0.17 Wind speed X 0.16 0.33 0.23 -0.11 Total* X -0.21 -0.62* 0.69* Days X 0.52* -0.31 Days 25 mm X -0.54* Rain X Vona Vona Temperature x -0.05 -0.46* -0.25 0.08 0.38 0.06 0.25 -0.19 0.10 0.24 Biomass x 0.31 -0.03 -0.32 0.21 0.39 -0.09 -0.15 -0.41 0.08 Density X 0.19 -0.27 -0.15 0.26 -0.19 -0.03 -0.25 0.04 Moon phase X 0.08 -0.29 -0.09 -0.55* 0.26 0.42* -0.36 Tide range X -0.05 -0.34 -0.30 0.28 0.46* -0.34 Cloud cover X 0.49* 0.10 -0.21 0.03 0.25 Wind speed X 0.16 -0.06 -0.28 0.12 Total* X -0.49* -0.56* 0.62* Days X 0.33 -0.83* Days 25 mm X -0.44* Rain X Tulagi Temperature x -0.22 -0.20 -0.22 -0.33 0.10 -0.11 0.33 -0.40 -0.21 0.11 Biomass x 0.70* 0.41* 0.24 -0.33 -0.15 -0.45* 0.14 0.09 -0.44* Density x 0.19 0.22 -0.45* -0.45* -0.40* 0.49 -0.05 -0.52* Moon phase X 0.22 -0.38 0.04 -0.51* -0.00 0.10 -0.35 Tide range X -0.18 -0.19 -0.18 0.28 -0.13 -0.26 Cloud cover X 0.39 0.52* -0.33 -0.25 0.72* Wind speed X 0.06 -0.24 0.10 0.21 Total* X -0.36 -0.42* 0.67* Days X 0.15 -0.44* Days 25 mm X -0.29 Rain Total rainfall since previous sample X * Total Days = No. days since rain Days 25 mm = No. days since rain >25mm Rain No. days of rain since previous sample. relationship with tide range at Tulagi. Overall, the most significant variable was total rainfall. The results of the regression analysis of the independent stimuli (Table 5) were similar, except at Vona Vona where a four- variable model including zooplankton density, wind, and days of rain gave the best fit. Spratelloides For S. delicatulus, moon phase was highly correlated with spawning and accounted for at least 26% of the variation in spawning periodicity at all sites (Table 4). Increased cloud cover and reduced tidal range were also correlated with spawning at Vona Vona and Tulagi. However, when data from all sites were combined, the most significant correlates were Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 231 Table 4 Stepwise multiple regresssion of best fit of 11 proximate variables related to the proportion of each species spawning at each site. Maximum number of variables allowed was three. Negative sign preceding a variable indicates negative corre lation; r. = partial cor- relation coefficient, r 2 = overall correlation coefficient, P = significance level, N = sample size. Species Site Environmental parameter r 2 p r 2 P N Encrasicholina devisi Munda - Days since 25 mm rain Days since rain Wind speed 0.16 0.10 0.09 0.35 <0.05 22 Vona Vona Wind speed Zooplankton biomass Tide range 0.33 0.11 0.06 0.50 <0.01 21 Tulagi Days since 25 mm rain Temperature Zooplankton biomass 0.31 0.25 0.03 0.59 <0.005 19 Overall — Days since 25 mm rain Days since rain 0.12 0.02 0.14 <0.01 62 Encrasicholina heterolobus Munda - Total rainfall Days since rain 0.19 0.11 0.30 <0.05 22 Vona Vona — Cloud cover Moon phase 0.32 0.12 0.44 <0.01 21 Tulagi — Tide range 0.24 0.24 <0.05 16 Overall " Total rainfall Temperature 0.11 0.03 0.14 <0.05 59 Spratelloides delicatulus Munda Moon phase 0.28 0.28 <0.01 22 Vona Vona - Moon phase Tide range Cloud cover 0.26 0.13 0.11 0.50 <0.01 22 Tulagi Moon phase Cloud cover 0.51 0.09 0.60 < 0.005 17 Overall - Tide range Zooplankton biomass Cloud cover 0.05 0.04 0.04 0.13 <0.05 61 Spratelloides lewisi Munda Zooplankton biomass Days since rain Cloud cover 0.22 0.10 0.09 0.41 <0.05 18 Vona Vona - Cloud cover Wind speed Moon phase 0.29 0.19 0.14 0.62 <0.01 15 Tulagi - Temperature Zooplankton biomass Days since rain 0.39 0.16 0.11 0.66 <0.01 14 Overall Temperature Zooplankton biomass 0.13 0.07 0.20 <0.01 47 Spratelloides gracilis Vona Vona Moon phase Temperature Zooplankton biomass 0.62 0.10 0.10 0.82 <0.05 8 Archamia zosterophora Munda - Days since 25 mm rain Days since rain 0.49 0.16 0.65 <0.01 11 Vona Vona Zooplankton biomass 0.42 0.42 <0.05 10 Overall Days since 25 mm rain Zooplankton biomass 0.30 0.11 0.41 <0.05 21 negative tide range, cloud cover, and negative zoo- plankton biomass. When the number of variables were reduced (Table 5), wind and days of rain at Vona Vona and temperature at Tulagi were also significant (P< 0.05). Among samples from months where spawning had been detected, moon phase and days of rain showed the best fit. There was significant positive correlation between spawning of S. lewisi and zooplankton biomass and days since rain at Munda (Table 4). Lower temperature, 232 Fishery Bulletin 89(2). 1991 Table 5 Stepwise linear regression mc idels of best fit for six independent proximate stimuli, y = percent spawning; t = temperature; d = zooplankton density; m = moon phase; ti = tida range; w = wind speed; r = days of rain between samples; r 2 = multiple regres- sion coefficient; P = significance level; N = sample size. Overall includes all samples; spawning includes only months when spawning was detected. Species Site Model r 2 P N Encrasicholina devisi Munda y = 0.58 - 0.02r 0.14 <0.08 22 Vona Vona y = 3480m + 0.29ti - 0.15w - 3479 0.40 <0.05 21 Tulagi y = 0.14t - 0.0003d - 0.18r - 2.89 0.33 <0.07 19 Spawning y = 0.52 - 0.006r 0.02 <0.44 40 Overall y = 0.50 - O.OOOld - O.Olr 0.07 <0.10 62 Encrasicholina heterolobus Munda y = 0.92 - 0.15r 0.08 <0.22 22 Vona Vona y = 3741m + O.OOOld - 0.26w - 0.08r - 3742 0.50 <0.05 21 Tulagi y = 0.79 - 0.92ti 0.24 <0.05 16 Spawning y = O.lOt - 0.57ti - 0.02r - 1.73 0.27 <0.05 33 Overall y = 2459m - 0.42ti - 0.02w - O.Olr - 2458 0.16 <0.05 59 Spratelloides Micatulus Munda y = 7084m - O.OOOld - 0.37ti - 7083 0.38 <0.05 22 Vona Vona y = 10759m + 0.27w + 0.13r - 10759 0.49 <0.01 22 Tulagi y = 8825 + 0.1 It - 8828m 0.51 < 0.005 17 Spawning y = 3510m + 0.02r - 3510 0.16 <0.05 38 Overall y = 0.57 - 0.36ti 0.05 <0.11 61 Spratelloides lewisi Munda y = 0.0003d + 0.17r - 0.39 0.24 <0.13 18 Vona Vona y = 5217 - 5216m 0.15 <0.15 15 Tulagi y = 1.40 - 0.04t - 0.25ti 0.26 <0.17 14 Spawning y = 0.02r - O.OOOld + 0.34 0.30 <0.01 29 Overall y = 0.08t - 0.03w + O.Olr - 2.34 0.23 <0.01 47 Spratelloides gracilis Vona Vona y = 12588m + 0.56ti - 12587 0.86 <0.05 8 Archamia zosterophora Munda y = 4262 - 4262m - 0.49ti 0.50 <0.06 11 Vona Vona y = 0.08t - 1.98ti - 0.80 0.94 <0.001 10 Spawning y = 3957 - 3956m 0.30 <0.16 8 Overall y = 0.72 - 0.79ti 0.46 <0.001 21 low zooplankton biomass, and recent rain explained 66% of the variation in spawning periodicity at Tulagi (P<0.01). Cloudy conditions, light winds, and waning moon phases were the variables most correlated with spawning in S. lewisi at Vona Vona (Table 4). However, the combined data showed that temperature and zoo- plankton biomass were the most significant stimuli. These conflicting results were reflected in the second analysis, where the stimuli chosen could not explain a significant amount of the variation in spawning at any site (Table 5). Correlations between environmental stimuli and the spawning periodicity of S. gracilis could only be ana- lysed for fish from Vona Vona. Most variation could be explained by moon phase, temperature, tide, and zooplankton biomass (Tables 4, 5). Most fish spawned at full moon and when temperature, tidal range, and prey biomass were high. Archamia zosterophora This species was most fre- quently caught at Munda and Vona Vona where it spawned most often during periods of high rainfall (at Munda), and when zooplankton biomass was high (at Vona Vona) (Table 4). These variables were also the most significant when all data were combined. When these variables were excluded from the analysis (Table 5), lower tidal range and higher temperatures (Vona Vona) were the only significant stimuli. Discussion Size-at-sexual-maturity of some Encrasicholina and Spratelloides species from the Solomon Islands varied both locally and compared with other studies elsewhere in the South Pacific. Encrasicholina heterolobus and S. gracilis reached sexual maturity at smaller lengths than previously reported (Tham 1965; Tiews et al. 1971; Dalzell and Wankowski 1980; Conand 1985; Dalzell 1985, 1987b). Such differences may be partly an artifact of the dif- ferent sexual maturity criteria used. The present study, Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 233 unlike others, used histological examination of gonads to verify macroscopic stages. In the Solomon Islands, Encrasicholina devisi reached sexual maturity at 45 mm, and some fish at this size were in spawning condition. This is consistent with results from Papua New Guinea (Dalzell and Wankow- ski 1980), New Caledonia (Conand 1985), and southern India (Luther 1979). Spratelloides delicatulus also showed little difference throughout the region in length-at-sexual-maturity (Lewis et al. 1983, Conand 1985, McCarthy 1985). Spratelloides lewisi showed local variation in length- at-sexual-maturity. Fish from Tulagi were not sexual- ly mature at less than 40 mm (compared with 35 mm at other sites) and they also grew to a greater size than fish from the other sites (unpubl. data). Length-at-first-spawning, however, was similar for both Encrasicholina species and S. delicatulus throughout the region (Dalzell and Wankowski 1980, Lewis et al. 1983, Conand 1985), except for E. hetero- lobus at Munda, where fish were in spawning condi- tion at a smaller length than at the other sites in the Solomon Islands. Both Spratelloides lewisi and S. gracilis showed variation in length-at-first-spawning between sites. At Tulagi, S. lewisi did not spawn until a much greater length was attained. These results are consistent with those of Dalzell (1987), who found a difference of 9 mm in the length-at-first-spawning in two populations of S. lewisi in Papua New Guinea waters. Dalzell (1985) also found that Papua New Guinea S. gracilis did not develop ripe eggs until 44 mm, which was much larger than the length-at-first-spawning of fish from the Solomon Islands (35 mm). Unfavorable conditions for reproduction may delay the onset of gonadal develop- ment in these species at some sites to help offset reproductive uncertainty (Mann and Mills 1979). No comparative data on length-at-sexual-maturity of A. zosterophora are available, and there are few data for other similar-sized apogonids. However, this species matures at almost 80% of maximum size, which is larger than in the clupeoids (70%). The subtropical Australian species Apogon fasciatus also matures at about 70% of maximum size (90 mm) near Brisbane (K. Warburton, Zool. Dep., Univ. Queensland, Brisbane, Australia, unpubl. data). The relationship between length-at-first-spawning and maximum length of the Solomon Islands baitfish closely fits that found for other clupeoids (Beverton 1963). Beverton (1963) showed that length-at-first- spawning among clupeoids is closely proportional to maximum length, with smaller species spawning at a smaller size (relative to their maximum) than larger ones (Blaxter and Hunter 1982). Longhurst and Pauly (1987) hypothesize that length-at-sexual-maturity and maximum length are determined by the interactions of oxygen supply and demand. Any species of fish living in cold water should grow to a greater size and mature at a larger size than the same species in warm water, given similar food supplies. Water temperature at Tulagi was consistently 1-2°C colder than at the other sites, and this may account for differences in these parameters in S. lewisi at this site. The regional dif- ferences seen in E. heterolobus and S. gracilis also support this idea, with fish from higher latitudes maturing at greater lengths. However, similar patterns were not found in the other species. Availability of food also must play an important role by affecting growth rates. Many clupeoids, including most engraulids, are multi- ple spawners (Blaxter and Hunter 1982), and studies of other engraulids suggest they spawn batches of eggs every 2-10 days (Hunter and Goldberg 1980, Alheit et al. 1984, Clarke 1987). Smaller species (e.g., Encrasi- cholina purpureus) spawn more frequently during the peak spawning period (as often as every 2 days; Clarke 1987). The spawning frequency of multiple spawning fish can be determined by the presence or absence of postvitellogenic follicles in the ovaries; their presence indicates that a fish has spawned within the previous 24-48 hours (Hunter and Goldberg 1980). Clarke (1987) reported that postvitellogenic follicles were distinguish- able in Encrasicholina purpureus up to 16 hours after spawning. In our study, we did not find either post- vitellogenic follicles or a continuous egg-size distribu- tion indicative of multiple spawning (Blaxter and Hunter 1982). However, the similarity in reproductive behavior of Encrasicholina species and the presence of eggs in the plankton throughout most of the year suggest that E. devisi and E. heterolobus are also multi- ple spawners. Our data on Encrasicholina devisi and E. heterolobus were consistent with that of Leary et al. (1975) on Encrasicholina purpureus. We found, as had Leary et al., only two egg sizes: one advanced and one with all eggs at the yolk-precursor stage of develop- ment. Leary et al. (1975) interpreted these results to indicate that E. purpureus spawned once in its lifetime. However, Clarke (1987) found that E. purpureus eggs could mature very rapidly, and hence timing of sam- pling was critical. Running- ripe eggs were only present shortly before spawning, which occurred one or two hours after sunset. If E. devisi and E. heterolobus are batch spawners similar to E. purpureus, then we would expect to collect only fish about to spawn in our samples, as the postvitellogenic follicles of fish that spawned the previous night would have degraded (Clarke 1987). However, spawning is probably less fre- quent than in the smaller E. purpureus because greater energy is required by larger fish to maintain continuous spawning (Hunter and Leong 1981). 234 Fishery Bulletin 89(2). 1991 Ovaries of the three species of Spratelloides and A. zosterophora all contained only a single size-group of developing oocytes, which suggests that they spawn all the eggs in the ovaries at once. Whether a female develops another batch of eggs after spawning was not determined. However, given the high proportion of female Spratelloides spawning at any time, it seems probable that each female produces more than,one batch of eggs. Most apogonids are mouth-brooders (Thresher 1984), and other species have been found with eggs in their buccal cavity that are at more than one stage of development, which suggests that they are multiple spawners (Thresher 1982). Flexibility in reproduction is well documented among a range of animals and usually involves a direct physi- ological response to nutrient level or some associated environmental cue (Giesel 1976). Extended breeding seasons are a common phenomenon among tropical fishes, particularly coral reef fish (Munro et al. 1973, Russell et al. 1977, Johannes 1978, Lowe-McConnell 1979, Walsh 1987). The present work, and other studies of Encrasicholina, Spratelloides, and apogonid repro- duction (Leary et al. 1975; Russell et al. 1977; Dalzell and Wankowski 1980; Dalzell 1985, 1987ab; Conand 1985; McCarthy 1985; Clarke 1987), found that the spawning season of these fish is protracted, with some individuals in the population spawning at any time dur- ing the year. There are also periods when most of the population is spawning. If reproduction were controlled only by endogenous cycles and females spawned as soon as they were physiologically capable, then as the proportion of larger fish increased there should be a greater proportion of spawning fish in the population. Examination of our data failed to find any significant relationship between the proportion of larger fish and the proportion spawning for any species at any site. Non-parametric tests also showed that the frequency of major spawnings was not random. Hence it is unlike- ly that spawning periodicities shown by baitfish are a result of intrinsic mechanisms. The timing and inten- sity of periods of increased spawning activity appear to be highly variable, linked to exogenous stimuli and with no endogenous rhythms. Of the several hypotheses put forward to explain the timing of fish reproduction, one of the more widely ac- cepted is the "match-mismatch" hypothesis of Cushing (1967). This proposes that if the timing of reproduc- tion coincides with peaks in the plankton cycle, larval survival is enhanced. Timing of peak plankton produc- tion does not necessarily occur at exactly the same time each year, so if fish are serial spawners they can ex- ploit the plankton cycle well (e.g., California sardine) by spawning during any of the months of spring or sum- mer when food becomes superabundant (Cushing 1975). Much of the data to support this hypothesis comes from temperate waters, but data on the reproduction of tropical clupeoids from open oceans also support this hypothesis (Longhurst 1971, Roy et al. 1989). Johannes (1978) proposed that coastal tropical fish tend to spawn at times and locations that will reduce predation on larvae by transporting them out of the adult habitat, while enabling them to return to suitable areas for postlarval settlement. At many sites, periods of major spawning coincide with low winds and full moon. The baitfish species studied here spawn in lagoons fringed by coral reefs, where water currents are negligible. Both adults and larvae are pelagic and live in the deeper waters of the lagoon. When the lar- vae are abundant, adults prey on them (Milton et al. 1990). If deeper waters of the lagoon are the favored habitat, then according to Johannes' hypothesis (1978) adults should spawn when local flushing is greatest (spring tides) but regional flushing is least (low winds) to ensure larvae develop in the same area. Although baitfish spawn throughout the year, with no consistent seasonal pattern, there appears to be some regularity in their spawning. At each site, cer- tain proximate factors were significant for more than one species. This suggests that if these fish are re- sponding to exogenous spawning stimuli, their influ- ence must be interposed by local hydrography and topography. At Vona Vona, spawning by five of the six species examined was correlated with moon phase. Both Encrasicholina and two Spratelloides species spawned around the full moon, while S. lewisi spawned around the new moon. Both Encrasicholina species also spawned when wind strength was reduced. The spawning around phases of the moon has often been assumed to be related to increased tidal exchange dur- ing spring tides (e.g., Johannes 1978, Walsh 1987). However, in this study greater tidal range at any site was not directly correlated with moon phase (Table 3). Vona Vona is an enclosed lagoon with many islands and patch reefs. In this habitat, spawning around the full or new moon does not appear to be related to the poten- tial to flush eggs and larvae to more favorable habitats by tidal water movement (Johannes 1978). Possibly, spawning around the full moon when the wind is re- duced may increase spawning success by increasing visibility at night (during spawning) and reducing the dispersal of eggs, and hence increasing the chances of fertilization. This hypothesis would also explain the strong negative relationship between spawning by A. zosterophora and tidal range at this site. Moon phase was the single variable most often cor- related with spawning in this study, both for five spe- cies at one site (Vona Vona) and for one species at dif- ferent sites (S. delicatulus). However, our data do not indicate a strong relationship with moon phase for either Encrasicholina species at Munda or Tulagi. This Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 235 suggests that either local conditions exert a strong in- fluence on the timing of reproduction in these species or that at these sites reproduction is not strongly linked to particular environmental events. Wright (1989) also found no relationship between spawning by E. hetero- lobus in Indonesia and temperature, rainfall, or tidal phase, and suggested that these factors were not in- fluencing spawning. However, among the other species, temperature and reduced tidal range appear to be more important at Tulagi. Higher temperatures should allow increased growth under favorable conditions, and reduced tidal exchange would reduce egg movement away from favorable habitats. While rainfall has been suggested as a proximate fac- tor influencing spawning by Encrasicholina species in Papua New Guinea (Dalzell and Wankowski 1980; Dalzell 1984, 1987b), in our study rainfall was only weakly correlated with spawning of Encrasicholina at Munda (Tables 4, 5). Spawning appeared to be greater during periods of lower rainfall. Positive relationships between spawning and time since rain or less rain may be indirectly linked to phytoplankton production, which is determined by light or the depth of mixing (Wyatt 1980). Variations in wind strength and cloudiness will cause major variations in the onset and end of the phytoplankton production cycle (Blaxter and Hunter 1982). If rainfall at Munda is linked to the phytoplank- ton cycle, spawning during periods of low rainfall would be consistent with Cushing's (1967) hypothesis. Some previous studies in southeast Asia and Papua New Guinea (Tiews et al. 1971, Sitthichockpan 1972, Dalzell 1987b) have suggested that Encrasicholina spawned more intensively during the months of peak zooplank- ton production. The variation in spawning frequency observed among the species in this study suggests that each responds differently to local conditions, reacting to those vari- ables most appropriate to maximize reproductive suc- cess in the immediate environment (Bye 1984). No ob- vious differences were detected in baitfish spawning patterns between the exploited fishing grounds and the unexploited site. If baitfish numbers are higher at the unexploited site, this suggests, in turn, that differences in the observed spawning patterns were not density- dependent. Lack of clear proximate stimuli for spawn- ing among the six species examined makes it difficult to predict the timing of major spawning events by these species. However, protracted spawning and the baitfish population's adaptation to local conditions suggests that these species should be resilient to increased fishing mortality. Factors affecting larval survival and growth may be more important in determining recruit- ment to the fishery. Acknowledgments We thank N. Rawlinson, G. Tiroba, and J. Leqata of the Fisheries Division, Solomon Islands, for assistance in the field and laboratory, and J. Kerr for statistical advice. Drs. R. Thresher, R. Johannes, and R. Harden Jones provided useful comments on the manuscript. This work is part of a CSIRO/Solomon Islands govern- ment collaborative research project funded by the Australian Centre of Agricultural Research (ACIAR project 8543). Citations Alheit, J., B. Algre, V.H. Alarcon, and B. Macewicz 1984 Spawning frequency and sex-ratio in the Peruvian an- chovy, Engraulis ringens. Calif. Coop. Oceanic Fish. Invest. Rep. 25:43-52. Anonymous 1988 Solomon Islands Fisheries Division annual report. Minist. Nat. Resour., Honiara. Beverton, R.J.H. 1963 Maturation, growth and mortality of clupeid and en- graulid stocks in relation to fishing. Rapp. P-V. Reun. Cons. Int. Explor. Mer 154:44-67. Blaxter, J.S.H., and J.R. Hunter 1982 The biology of the clupeiod fishes. Adv. Mar. Biol. 20: 1-224. Bye, V.J. 1984 The role of environmental factors in the timing of repro- ductive cycles. In Potts, G.W., and R.J. Wooton (eds.), Fish reproduction: Strategies and tactics, p. 188-205. Academic Press, London. Chen, T.S. 1984 The suitable fishing season of larval anchovy in Fang- Liao and Lin-Yan by studying on maturity and spawning of anchovy Stolephorus zollingeri (Bleeker). Bull. Taiwan Fish. Res. Inst. 37:53-66. 1986 Study on maturation and spawning of Stolephorus hetero- lobus in south of Taiwan. Bull. Taiwan Fish. Res. Inst. 40: 59-66. Clarke, T.A. 1987 Fecundity and spawning frequency of the Hawaiian an- chovy or Nehu, Encrasicholina purpurea. Fish. Bull., U.S. 85:127-138. Conand, F. 1985 Biology of the small pelagic fishes of the lagoon of New Caledonia used as bait fish for tuna fishing. In Proc. Fifth Int. Coral Reef Congr., p. 463-467. Fifth Int. Coral Reef. Symp. Exec. Comm., Tahiti. Cushing, D.H. 1967 The grouping of herring populations. J. Mar. Biol. Assoc. U.K. 47:193-208. 1975 Marine ecology and fisheries. Cambridge Univ. Press, Cambridge, 278 p. Cyrus, D.P., and S.J.M. Blaber 1984 The reproductive biology of Gerres in Natal estuaries. J. Fish. Biol 24:491-504. Dalzell, P. 1984 The population biology and management of bait-fish in Papua New Guinea waters. Dep. Primary Ind. Fish. Res. Rep. (Port Moresby) 84-05, 59 p. 236 Fishery Bulletin 89(2). 1991 1985 Some aspects of the reproductive biology of Spratelloides gracilis (Schlegel) in the Ysabel Passage, Papua New Guinea. J. Fish. Biol. 27:229-237. 1986 The 1985 bait fishing season at Ysabel Passage. Dep. Primary Ind. Fish. Div. Tech. Rep. (Port Moresby) 86/9, 27 p. 1987a Notes on the biology of Spratelloides lewisi (Wongra- tana), a recent described species of sprat from Papua New Guinea waters. J. Fish. Biol. 30:691-700. 1987b Some aspects of the reproductive biology of stolephprid anchovies from northern Papua New Guinea. Asian Fish. Sci. 1:91-106. Dalzell, P., and J.W.J. Wankowski 1980 The biology, population dynamics, and fisheries dynamics of exploited stocks of three baitfish species, Stolephorus heterolobus, S. devisi and Spratelloides gracilis, in Ysabel Passage New Ireland Province Papua New Guinea. Dep. Primary Ind. Fish. Res. Bull. (Port Moresby) 22:1-124. Delsman, H.C. 1931 Fish eggs and larva from the Java Sea. 17. The genus Stolephorus. Treubia 13:217-243. Dharmamba, M. 1960 Studies on the maturation and spawning of some com- mon clupeoids of Lawson's Bay, Waltair. Indian J. Fish. 6: 374-388. Evans, D.W., and P.V. Nichols 1984 The baitfishery of Solomon Islands. Fish. Dep. Minist. Nat. Resour., Honiara, 177 p. Giesel, J.T. 1976 Reproductive strategies as adaptations to life in temporal- ly heterogeneous environments. Annu. Rev. Ecol. Syst. 7: 57-79. Hunter, J.R., and S.R. Goldberg 1980 Spawning incidence and batch fecundity in northern an- chovy Engraulis mordax. Fish. Bull., U.S. 77:641-652. Hunter, J.R., and R. Leong 1981 The spawning energetics of female northern anchovy Engraulis mordax. Fish. Bull., U.S. 79:215-230. Johannes, R.E. 1978 Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fish. 3:65-84. Lam, T.J. 1983 Environmental influences on gonadal activity in fish. In Hoar, W.S., D.J. Randall, and E.M. Donaldson (eds.), Fish physiology, Vol. 9B: Behaviour and fertility control, p. 65-116. Academic Press, NY. Leary, D.F., G.I. Murphy, and M. Miller 1975 Fecundity and length at first spawning of the Hawaiian anchovy, or Nehu (Stolephorus purpureus) in Kaneohe Bay, Oahu. Pac. Sci. 29:171-180. Leis, J.M., and T. Trnski 1989 The larvae of Indo-Pacific shorefishes. New South Wales Univ. Press, Sydney, 371 p. Lewis, A.D., S.Sharma, J. Prakash, and B. Tikomainiusiladi 1983 The Fiji baitfishery 1981-82, with notes on the biology of the gold spot herring Herklotsichthys quadrimaculatus (Clupeidae), and the blue sprat Spratelloides delicatulus (Dussumieriidae). Fish. Div. Minist. Agric. Fish. Tech. Rep. Fiji 6, 35 p. Longhurst, A.R. 1971 The clupeoid resources of tropical seas. Oceanogr. Mar. Biol. Annu. Rev. 9:349-385. Longhurst, A.R., and D. Pauly 1987 Ecology of tropical oceans. Academic Press, San Diego, 407 p. Lowe-McConnell. R.H. 1979 Ecological aspects of seasonality in fishes of tropical waters. In Miller, P.J. (ed.), Fish phenology: Anabolic adap- tiveness in teleosts, p. 219-242. Zool. Soc. Lond. Symp. 44. Academic Press, London. Luther, G. 1979 Anchovy fisheries of southwest coast of India with notes on the characteristics of the resources. Indian J. Fish. 26:25-39. Mann. R.H.K., and C.A. Mills 1979 Demographic aspects of fish fecundity. In Miller, P.J. (ed.), Fish phenology: Anabolic adaptiveness in teleosts, p. 161- 177. Zool. Soc. Lond. Symp. 44. Academic Press, London. McCarthy, D. 1985 Fishery dynamics and biology of the major wild baitfish species particularly Spratelloides delicatulus, from Tarawa, Kiribati. Atoll Res. Dev. Unit, Univ. South Pacific, Suva, Fuji, 53 p. McManus, J.F.A., and R.W. Mowry 1964 Staining methods: Histological and histochemical. Harper & Row, NY, 423 p. Milton, D.A., S.J.M. Blaber, and N.R. Rawlinson 1990 Diet and prey selection of six species of tuna baitfish in three coral reef lagoons in Solomon Islands. J. Fish. Biol. 37:205-224. Muller, R.C. 1976 Population biology of Stolephorus heterolobus (Pisces: Engraulidae) in Palau, western Caroline Islands. Ph.D thesis, Univ. Hawaii, Honolulu, 178 p. Munro, J.L., V.C. Gaut, R. Thompson, and P.H. Reeson 1973 The spawning seasons of Carribean fishes. J. Fish. Biol. 5:69-84. Roy, C, P. Cury, A. Fontana, and H. Belveze 1989 Spatio-temporal reproductive strategies of the clupeoids in West African upwelling areas. Aquat. Living Resour. 2: 21-29. Russell, B.C., G.R.V. Anderson, and F.H. Talbot 1977 Seasonality and recruitment of coral reef fishes. Aust. J. Mar. Freshwater Res. 28:521-528. Scott, D.B.C. 1979 Environmental timing and the control of reproduction in teleost fish. Symp. Zool. Soc. Lond. 44:105-132. Sitthichockpan, S. 1972 A study on the spawning periods and spawning ground of Stolephorus spp. off the west coast of the Gulf of Thailand, 1968-1969. Third symp. mar. fish., Mar. Fish. Lab., Bangkok, 6 p. Sokal, R.R., and F.J. Rohlf 1981 Biometry, 2d ed. Freeman and Co., NY, 859 p. Stearns, S.C., and R.E. Crandall 1984 Plasticity for age and size at sexual maturity: A life- history response to unavoidable stress. In Potts, G.W., and R.J. Wootton (eds.), Fish reproduction: Strategies and tactics, p. 13-33. Academic Press, London. Tester, A.L. 1955 Variation in egg and larva production of the anchovy, Stolephorus purpureus Fowler, in Kaneohe Bay, Oahu, dur- ing 1950-1952. Pac. Sci. 9:31-41. Tham, A.K. 1965 Notes on biology of the anchovy, Stolephorus pseudo- heterolobus Hardenberg. Bull. Natl. Mus. Singapore 33:23-26. Thresher, R.E. 1982 Reef fish. Palmetto Publ. Co., St. Petersburg, FL. 171 p. 1984 Reproduction in reef fishes. T.F.H. Publ., Neptune City, NJ, 399 p. Milton and Blaber: Sexual maturity and spawning of tuna baitfish in the Solomon Islands 237 Tiews, K., I. A. Ronquillo, and L.M. Santos 1971 On the biology of anchovies (Stolephorus) in Philippine waters. Philipp. J. Fish. 9:92-123. Walsh, W.J. 1987 Patterns of recruitment and spawning in Hawaiian reef fishes. Environ. Biol. Fish. 18:257-276. Wright, P. 1989 The growth and reproductive biology of Stolephorus heterolobus (Engraulididae). M.Phil, thesis, Univ. Newcastle on Tyne, 112 p. Wyatt, T. 1980 The growth seasons in the sea. J. Plankton Res. 2:81-97. Young, J.W., S.J.M. Blaber, and R. Rose 1987 Reproductive biology of three species of midwater fishes associated with the continental slope of eastern Tasmania, Australia. Mar. Biol. (Berl.) 95:323-332. Abstract. - A biochemical exam- ination of otolith growth-somatic growth relationship was conducted in rainbow trout Oncorhynchus my- kiss. The rate of otolith growth was defined by calcium deposition on oto- liths in an in vitro isolated prepar- ation of otolith-containing sacculi. Somatic growth was estimated by RNA-DNA ratios in white trunk muscle. Rainbow trout weighing ap- proximately 120 g were starved for 5 days and then fed commercial trout pellets once a day. They were sam- pled on days 1, 2, 3, and 5 after star- vation, and on days 1, 2, 3, and 4 after feeding. In a separate experi- ment, fish were sampled at 6-hour in- tervals of 1000, 1600, 2200, and 0400 hours over a 24-hour period. Otolith and somatic growth showed a posi- tive relationship, both decreasing from 2 days after starvation and re- covering on day 4 after feeding. In otoliths, however, the starvation-in- duced decrease in calcium deposition was transiently restored on day 1 after feeding, followed by a decrease on day 2. The diel relationship be- tween otolith and somatic growth was coupled, showing minimum levels at 2200 hours. These results suggest that otolith growth ordinar- ily reflects somatic growth rates on the daily basis. Biochemical Relationship Between Otolith and Somatic Growth \n the Rainbow Trout Oncorhynchus my kiss: Consequence of Starvation, Resumed Feeding, and Diel Variations Yasuo Mugiya Hirotaka Oka Faculty of Fisheries, Hokkaido University Mmato-3. Hakodate 041, Japan Manuscript accepted 17 December 1990. Fishery Bulletin, U.S. 89:239-245 (1991). Teleost otoliths are calcium carbon- ate concretions which have incre- ments consisting of a bipartite struc- ture of light and dark rings. These structures are known to be formed on a daily basis (Pannella 1971). Since otolith and fish size are highly cor- related for a variety of marine and freshwater species (Campana and Neilson 1985), it is possible to esti- mate growth-rate histories of individ- ual fish by measuring the width of otolith increments. Volk et al. (1984) averaged otolith increment widths over each week and found a linear regression of mean increment width with somatic growth rate in chum salmon Oncorhynchus keta under controlled feeding. Nishimura and Yamada (1988) successfully back- calculated the growth rate of walleye pollock Theragra chalcogramma by using mean otolith width measured every 10 increments. However, it re- mains to be shown whether the rate of otolith growth reflects somatic growth rates in terms of daily or sub- daily trends. Recently several workers (Secor and Dean 1989, Reznick et al. 1989, Wright et al. 1990) reported an un- coupling of the relationship between otolith and somatic growth in fish populations experiencing slow and fast somatic growth. Similar uncou- pling was exaggeratedly induced in hypophysectomized goldfish Caras- sius auratus, in which somatic growth in length was completely inhibited but otolith continued to grow at a re- duced rate (Mugiya 1990). Since the otolith (sagitta) occurs within the sacculus which is anatom- ically closed in rainbow trout Onco- rhynchus mykiss, it is feasible to take out the otolith-containing sacculus without any leakage of the endolymph (Mugiya 1984). An in vitro prepara- tion of the isolated sacculi was used for indicating the current growth rate of otoliths, which reflected the in vivo physiological state at the time when the fish were sampled (Mugiya 1984, 1987). Somatic growth is a balance between catabolic and ana- bolic components in protein meta- bolism (Miglavs and Jobling 1989). Since accretive growth should be directly associated with protein syn- thetic capacity, RNA-DNA ratios in muscle are widely used as an indi- cator of the short-term or current somatic growth rate (Bulow 1987). The aim of the present study is to clarify the biochemical relationship between otolith and somatic growth on a daily basis, using starved and then fed rainbow trout. The otolith growth-fish growth relationship was also examined at 6-hour intervals 239 240 Fishery Bulletin 89(2). 1991 over a 24-hour period. Otolith and somatic growth were defined by the rate of in vitro calcium deposition on otoliths and RNA-DNA ratios in white trunk muscle, respectively. Serum calcium concentrations were ex- amined for diel variations based on previous work on the relationship between serum calcium and otolith growth (Mugiya 1984). Materials and methods We performed two experiments with food and diel effects on somatic and otolith growth. Immature rain- bow trout Oncorhynchus mykiss weighing 100-130g were obtained from a trout farm and acclimated to ex- perimental conditions for at least 4 weeks before use. Throughout the acclimation and experimental periods, they were maintained in a pair of outside concrete ponds (3.8m 2 x 0.7m in depth) which were supplied with running water at 14 ± 0.5°C, and fed commercial trout pellets once a day at around 1500 hours, unless otherwise stated. The ration fed was about 1.5% of body weight. Starvation and refeeding This experiment was carried out in October 1988. Dusk occurred at 1700 hours and dawn at 0545 hours. Sixty fish were randomly divided into two groups (30 fish per group), and separately assigned to one compartment of the paired ponds. One group was starved for 5 days and then fed. During these periods, they were sampled on days 1 (43 hours from last feeding), 2, 3, and 5 after starvation, and on days 1 (19 hours from the first refeeding), 2, 3, and 4 after refeeding. The other group was fed throughout the period and sampled as the con- trol on the same time schedule, except on the third day of starvation and on the second day of refeeding when only the experimental group was sampled. Sampling was conducted alternately between experimental and control groups at 0945-1045 hours every day. At each sampling, 4-6 fish were gently netted one at a time, and immediately bled by cutting the tail of the fish. After bleeding, a pair of sacculi containing otoliths were isolated, placed in an incubation medium, and then the next fish was netted. The remaining part of the body was stored at -40°C and analyzed within 5 days for RNA-DNA ratios. Data from the control group were pooled for statistical analyses, as little difference was found among the sampling days. Diel variation This experiment was carried out in December 1988. Dusk occurred in 1600 hours and dawn at 0700 hours. To examine the diel relationship between otolith and somatic growth, 25 fish were stocked in each compart- ment of the paired ponds, and 7 fish each were ran- domly sampled at 6-hour intervals of 1000, 1600, 2200, and 0400 hours over a 24-hour period. Sampling was conducted alternately from one compartment at 1000 and 2200 hours, and from the other compartment at 1600 and 0400 hours. This sampling regime is recom- mended for minimizing handling effects. After fish were netted one at a time, blood was immediately col- lected from the caudal vessels by cutting the tail of the fish and draining it into test tubes. After centrifuga- tion, the separated sera were stored at -40°C for 6-24 hours and analyzed for total calcium concentrations by flame photometry using an atomic absorption spectro- photometer (Hitachi #518). After the blood collection, sacculi were isolated for incubation, and the remain- ing part of the body was stored for RNA and DNA analyses. In this experiment, the fish were starved throughout the sampling day. Otolith incubation Otolith-containing sacculi were isolated according to a previously described technique (Mugiya 1987). Isolated sacculi were incubated in 50 mL of a Ringer solution (Mugiya 1986) containing 45 CA (New England Nu- clear) at a concentration of 1 x 10 4 Bq/mL. Incubation was carried out with oxygenation at 14°C for 2 hours. After incubation, sacculi were rinsed several times in 45 CA-free Ringer solution and the otoliths were sep- arated under a binocular microscope. The separated otoliths were lightly rinsed in water, placed in each counting vial, dried overnight at 80°C, and then weighed. They were solubilized in a mixture of 0.2 mL perchloric acid and 0.2 mL hydrogen peroxide (Mugiya 1987), and added to Scintisol EX-H (Wako Chem.) for radioactive counting (liquid scintillation spectrometer, Aloka LSC-673). The rate of otolith growth was evalu- ated in terms of microgrammes of calcium deposited per unit otolith weight. RNA and DIMA determinations RNA-DNA ratios were estimated in the white muscle collected from an area of the dorsoanterior trunk, using a modification of the Schmidt and Thannhauser method (Buckley and Bulow 1987). Briefly, white muscle (l.OOg) was homogenized with cold distilled water and adjusted to 10.0 mL. A 1.4-mL aliquot of the homog- enate was used for the estimation of nucleic acids. RNA and DNA were purified with 0.6 N cold HC10 4 , and then extracted with 0.3N KOH and 0.6N hot HC10 4 , respectively. Both acids were quantified from the ab- sorbance of the extracts at 260 nm. RNA and DNA Mugiya and Oka: Biochemical relationship between otolith and somatic growth in Oncorhynchus mykiss 241 120 100 J 80 c o I 60 a T3 40 20 t ; K * 12 3 5 ^ ' 2 3 4 Days aftec starvation or refeeding Figure 1 Percentage effects of starvation and refeeding on in vitro calcium deposition on otoliths in Oncorhynchus mykiss. Con- trol values are pooled and the mean level (horizontal bar) is shown with SE of 48 otoliths. Vertical bars represent mean 1 SE of 8-10 otoliths. Arrow indicates time when fish were refed. *P<0.05. standard solutions were prepared using RNA from yeast and DNA from salmon sperm (Wako Chem.). Statistics Student's (-test for unpaired observations was applied to assess statistical significance of differences between mean values. Significance was accepted at P<0.05. Results Starvation and refeeding The rate of in vitro calcium deposition on otoliths ranged from 0.06 to 0.08^g/mg-hour in the control group. These data are pooled and presented as the mean level with standard errors (Fig. 1). Effects of starvation and refeeding on the rate are expressed as a percentage against the control level. Starvation in- duced an inhibitory effect (P<0.05) on the rate of cal- cium deposition on otoliths by day 2, decreasing to approximately half of the control. This level remained almost unchanged until day 5, the last day of starva- tion. On refeeding, the starvation-induced decrease recovered to the control level as early as day 1 . How- ever, this recovery was transient, followed by a sig- nificantly (P<0.05) reduced calcium deposition on day 2. This reduced calcium deposition recovered to the control level on day 4 after refeeding (Fig. 1). Table 1 Effects of starvation and refeeding on muscle RNA and DNA concentrations (mg/g) in rainbow trout Oncorhyn- chus mykiss. RNA DNA Control 1.51 + 0.03 0.61 1 0.02 Starvation (days) 1 1.50 ±0.06 0.56 1 0.02 2 1.2110.08* 0.5810.01 3 0.9010.04** 0.65 1 0.02 5 0.8910.04** 0.68 1 0.04 Refeeding (days) 1 0.8810.07** 0.65 1 0.06 2 0.9710.06** 0.61 1 0.02 3 1.0710.02** 0.6610.01 4 1.5710.04 t SE of 26 and 4- 0.63 1 0.02 -6 fish for control Values are mean : and experimental groups, respectively. *P<0.05; **P<0.01. pJq 12 3 5 1 2 3 4 Days after starvation or refeeding Figure 2 Percentage effects of starvation and refeeding on RNA-DNA ratios in white trunk muscle in Oncorhynchus mykiss. Con- trol values are pooled and the mean level (horizontal bar) is shown with SE of 26 fish. Vertical bars represent mean 1 SE of 4 or 6 fish. Arrow indicates time when fish were refed. *P<0.05; **P<0.01. Muscle RNA and DNA concentrations and their ratios are presented in Table 1 and Figure 2, respec- tively. Starvation and refeeding affected RNA concen- trations, but DNA concentrations remained constant during the experiment (Table 1). Therefore, changes in RNA-DNA ratios are primarily attributable to changes in RNA concentrations. 242 Fishery Bulletin 89(2). 1991 1000 '600 2200 0400 Time of day (hours) Figure 3 Diel variations in serum calcium concentra- tions (upper graph) and in vitro calcium deposition on otoliths (lower graph) in On- corhynchus mykiss. Each plotted value represents mean ± SE of 7 fish for serum calcium and of 12-14 otoliths. *P<0.05 for 1600 hours; **P< 0.001 for 1600 hours. Table 2 Diel variations in muscle RNA and DNA concentrations (mg/g) in rainbow trout Oncorhynchus mykiss. Time of day (h) RNA DNA 1000 1600 2200 0400 1.26 ±0.08 1.30 ± 0.08 0.66 ± 0.04* 0.80 ± 0.04 0.66 ± 0.04 0.72 ± 0.02 0.76 ± 0.06 0.68 ± 0.04 Values are mean ± SE of 7 fish. *P<0.01 for 1600 hours. RNA-DNA ratios ranged from 1.91 to 2.73, with a mean value of 2.44 in the control group. One day's star- vation had no effect on the ratio (Fig. 2). The first significant (P<0.05) effect occurred on day 2 after star- vation, followed by further decrease to approximately half of the control on day 3. This reduced level appar- ently remained essentially unchanged through day 5. In contrast to the response observed for otoliths, re- feeding had no effect on the recovery in RNA-DNA ratios the next day (Fig. 2). Ratios increased gradual- ly with feeding and recovered to the control level on day 4 after refeeding. 2 2 2 1 8 o S 1 6 < I 14 < rf .2 1 08 1000 1600 2200 0400 Time of day (hours) Figure 4 Diel variations in RNA-DNA ratios in white trunk muscle in Oncorhynchus mykiss. Each plotted value represents mean ± SE of 7 fish. *P<0.001 for 1600 hours. Diel variation Serum calcium concentrations varied dielly by approx- imately 4%, and this variation was statistically signif- icant at P = 0.05 (Fig. 3). Calcium concentrations were high during the daytime, but a nadir occurred at 2200 hours. The profile of diel variations in calcium deposition on otoliths was essentially the same as that of serum calcium concentrations (Fig. 3). The high rate of cal- cium deposition at 1600 hours rapidly decreased by 70% (P< 0.001) to a nadir at 2200 hours, followed by an increase toward 0400 hours. Diel variations in muscle RNA and DNA concentra- tions and their ratios are presented in Table 2 and Figure 4, respectively. Diel variations were significant in RNA concentrations and RNA-DNA ratios. The pro- file of variations in the ratios was similar to that of calcium deposition on otoliths: high ratios during the daytime were followed by a steep decrease between 1600 and 2200 hours. The lowest ratio occurred at 2200 hours, and this ratio was highly significant (P<0.001) compared with the ratio at 1600 hours. Differences between 1000 and 1600 hours were not statistically significant in either otolith or muscle, showing that the criterion of similar growth profiles was fulfilled. A comparison of variations in calcium deposition on otoliths and RNA-DNA ratios in muscle showed posi- tive relationships through several days after starvation and during the diel experiment. However, uncoupling was found in the recovery processes with refeeding: calcium deposition on otoliths transiently recovered on day 1 after refeeding, while RNA-DNA ratios did not (compare Figures 1 and 2). Mugiya and Oka: Biochemical relationship between otolith and somatic growth in Oncorhynchus mykiss 243 Discussion RNA-DNA ratios in muscle have been widely accepted as an index of current somatic growth rates in various species of marine and freshwater fish (Bulow 1987). These ratios are affected by various factors such as season (Bulow et al. 1981), ration size (Bulow 1970, Buckley 1979, Wilder and Stanley 1983, Jiirss et al. 1986), temperature, salinity (Jiirss et al. 1987), and lunar cycles (Farbridge and Leatherland 1987). Various toxicants also reduce the ratios (Barron and Adelman 1984). The present study presents an additional find- ing with regard to variations of RNA-DNA ratios: the ratios had distinct diel variations, showing higher values during daytime than nighttime. It is desirable to confirm such a profile of variations by further ex- aminations at shorter and longer time-intervals. Endocrinologically, RNA-DNA ratios are under the control of growth-regulating hormones. Hypophysec- tomy reduced the ratios, and replacement therapy with beef growth hormone restored the ratios to a normal level in bullheads Ictalurus melas (Kayes 1979). Orca- dian periodicities in the surge of growth-regulating hormones are well documented in higher vertebrates (Kato et al. 1982). Although few comparable references are available in fish, cyclic variations in growth hor- mone have been reported in plasma and pituitary levels in salmonids (Leatherland et al. 1974, Leatherland and Nuti 1982, Bates et al. 1989). Therefore, the cyclic surge of this hormone in combination with other hor- mones is probably a cause for diel variations in RNA- DNA ratios in the present fish. Effects of starvation or restricted food on RNA-DNA ratios have been repeatedly reported. A 4-week star- vation induced an approximately 40% decrease in the ratios in rainbow trout Oncorhynchus mykiss (Jiirss et al. 1986). Brook trout Salvelinus fontinalis also had the ratio reduced by 44% after 22 days of restricted feed- ing (Wilder and Stanley 1983). Bulow (1970) showed that RNA-DNA ratios directly reflected different soma- tic growth rates induced by nutritional manipulation. However, it is not clear how fast food deprivation affects this ratio. Although the time course of the ef- fect will depend on various factors such as tempera- ture, fish sizes, sexual status, etc., the present study revealed that the first effect of starvation on RNA- DNA ratios in muscle appeared on day 2 (67 hours from last feeding) in adult rainbow trout Oncorhynchus mykiss. This inhibitory effect was exaggerated with time, but the ratios never decreased below half the con- trol level even on day 15 after starvation (data are not presented here). This level (55% of the control) is similar to trout values in the literature mentioned above, and therefore it appears to be a basal level in RNA-DNA ratios in white muscle of rainbow trout. Starvation-induced decrease in RNA-DNA ratios recovered with the resumption of daily feeding at a level of 1.5% of body weight. The recovery processes were rather steady, taking 4 days in the present study. Bulow (1970) showed a more rapid recovery in golden shiners Notemigonus crysoleucas: 2 days feeding at 6% body weight seemed to be enough for the recovery of a starvation-induced decrease in ratios to the normal level. This disparity may be explained by the amount of food used and/or species difference. Lied et al. (1983) reported that a starvation-induced decrease in RNA- DNA ratios recovered to the normal level within 8 hours after refeeding ad libitum in cod Gadus morhua. However, it is uncertain whether this recovery was due to the result of food assimilation or to diel variations in the ratios, because control values were not available at each determination time in the study. Recently Miglavs and Jobling (1989) reported that a growth spurt (compensatory growth) transiently oc- curred following realimentation after a period of food restriction in juvenile Arctic char Salvelinus alpinus. Nevertheless, no corresponding increase occurred in tissue RNA-DNA ratios. This led them to present a limit to using the ratios as a growth index. However, since their somatic growth rate was calculated from a cumulated increase in body weight for 28 days, it is uncertain whether or not the rate remained high on the last day, when the RNA-DNA ratios were determined. When somatic and otolith growth rates were bio- chemically compared at 6-hour intervals over a 24-hour period, or after food deprivation, they were coupled at a qualitative level. In spite of these positive relation- ships, evidence of an allometric relationship between otolith and somatic growth has been accumulated, especially under suboptimal conditions. For example, Mosegaard et al. (1988) examined the effect of tem- perature, fish size, and somatic growth rate on otolith growth rate in Arctic char Salvelinus alpinus and found an uncoupling between somatic and otolith growth rates at hyperoptimal temperatures. Based on these results, they suggested that metabolic activity, not necessarily somatic growth rate, governs otolith growth rate. Somatic growth rate results mainly from the balance between protein synthesis and degradation, and hyper- optimal temperatures would accelerate both compo- nents, especially degradation, resulting in no somatic growth (Houlihan et al. 1988). However, since somatic growth is comparable with components from metabolic rates within the range of optimal temperatures (Webb 1978), the muscle RNA-DNA ratio, an index for pro- tein synthesis, will be a reflection of metabolic com- ponents at an appropriate temperature. Therefore, it appears reasonable to use this ratio for examining the relationship between somatic and otolith growth rates, even if otolith growth is a function of metabolic rate. 244 Fishery Bulletin 89(2). 1991 In the present study, the transient recovery of otolith growth occurred on the first day following refeeding. Although the definite reason for this remains unex- plained, an accelerated recovery process after growth suppression is well known as "repletion" in mammalian skeletal tissue formation (Linkhart et al. 1988). Mech- anisms for repletion would differ depending on tissues. Since the otolith is an acellular product secreted by otolith-forming cells (Saitoh and Yamada 1989), a possi- ble explanation for otolith repletion is as follows: Star- vation may interrupt the releasing activity partly due to the reduced processing of the secretory product. This would inversely result in some accumulation of otolith precursor materials in the cells (Anderson and Capen 1976). Refeeding stimulates release of the accumulated precursors through some factors (e.g., calcium-calmo- dulin interaction; Mugiya 1986) regulating the secre- tory activity. A steady-state recovery in synthesis and processing of the secretory product is probably a time- consuming process, running parallel to the recovery of RNA-DNA ratios in muscle. These possibilities await further examination. In the present study, we have demonstrated corre- sponding variations in otolith calcification and RNA- DNA ratios in muscle, but a causal relationship be- tween specific growth rate and otolith calcification remains to be studied. Acknowledgments The present study was financially supported in part by Grant No. 01560197 from the Ministry of Education of Japan. Citations Anderson, M.P., and C.C. Capen 1976 Ultrastructural evaluation of parathyroid and ultimobran- chial glands in iguanas with experimental nutritional osteo- dystrophy. Gen. Comp. Endocrinol. 30:209-222. Barron, M.G., and I.R. Adelman 1984 Nucleic acid, protein content, and growth of larval fish sublethally exposed to various toxicants. Can. J. Fish. Aquat. Sci. 41:141-150. Bates, D.J., B.A. Barrett, and B.A. McKeown 1989 Daily variation in plasma growth hormone of juvenile coho salmon, Oncorhynchus kisutch. Can. J. Zool. 67:1246-1248. Buckley, L.J. 1979 Relationships between RNA-DNA ratio, prey density, and growth rate in Atlantic cod (Gadus morhua) larvae. J. Fish. Res. Board Can. 36:1497-1502. Buckley, L.J., and F.J. Bulow 1987 Techniques for the estimation of RNA, DNA, and pro- tein in fish. In Summerfelt, R.C., and G.E. Hall (eds.), Age and growth of fish, p. 345-354. Iowa State Univ. Press, Ames. Bulow, F.J. 1970 RNA-DNA ratios as indicators of recent growth rates of a fish. J. Fish. Res. Board Can. 27:2343-2349. 1987 RNA-DNA ratios as indicators of growth in fish: A review. In Summerfelt, R.C., and G.E. Hall (eds.), Age and growth of fish, p. 45-64. Iowa State Univ. Press, Ames. Bulow, F.J., M.E. Zeman, J.R. Winningham, and W.F. Hudson 1981 Seasonal variations in RNA-DNA ratios and in indicators of feeding, reproduction, energy storage, and condition in a population of bluegill, Lepomis macrochirus Rafinesque. J. Fish Biol. 18:237-244. Campana, S.E., and J.D. Neilson 1985 Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1032. Farbridge, K.J., and J.F. Leatherland 1987 Lunar cycles of coho salmon, Oncorhynchus kisutch . II. Scale amino acid uptake, nucleic acids, metabolic reserves and plasma thyroid hormones. J. Exp. Biol. 129:179-189. Houlihan, D.F., S.J. Hall, C. Gray, and B.S. Noble 1988 Growth rates and protein turnover in Atlantic cod, Gadus morhua. Can. J. Fish. Aquat. Sci. 45:951-964. Jurss, K., Th. Bittorf, and Th. Vokler 1986 Influence of salinity and food deprivation on growth, RNA/DNA ratio and certain enzyme activities in rainbow trout (Salmo gairdneri Richardson). Comp. Biochem. Physiol. 83B: 425-433. Jurss, K., Th. Bittorf, Th. Vokler. and R. Wacke 1987 Effects of temperature, food deprivation and certain en- zyme activities in rainbow trout (Salmo gairdneri Richardson). Comp. Biochem. Physiol. 87B:241-253. Kato, Y., H. Katakami, S. Hiroto, A. Shimatsu, N. Matsushita, and H. Imura 1982 The circadian and ultradian related periodicity of pituitary hormone secretion in man and rats. PNE (Protein Nucleic Acid Enzyme) 27:233-245. Kayes, T. 1979 Effects of hypophysectomy and beef growth hormone replacement therapy, over time and at various dosages, on body weight and total RNA-DNA levels in the black bullhead (Icta- lurus melas). Gen. Comp. Endocrinol. 37:321-332. Leatherland, J.F., and R.A. Nuti 1982 Diurnal variation in somatotrop activity and correlated changes in plasma free fatty acids and tissue lipid levels in rain- bow trout Salmo gairdneri. J. Interdiscip. Cycle Res. 13: 219-228. Leatherland, J.F., B.A. McKeown, and T.M. John 1974 Circadian rhythm of plasma prolactin, growth hormone, glucose, and free fatty acid in juvenile kokanee salmon, Onco- rhynchus nerka. Comp. Biochem. Physiol. 47A:821-828. Lied, E., G. Rosenlung, B. Lund, and A. von der Decken 1983 Effect of starvation and refeeding on in vitro protein syn- thesis in white trunk muscle of Atlantic cod (Gadus morhua). Comp. Biochem. Physiol. 76B:777-781. Linkhart, T.A., S. Mohan, and D.J. Baylink 1988 Bone repletion in vitro: Evidence for a locally regulated bone repair response to PTH treatment. Bone 9:371-379. Miglavs, I., and M. Jobling 1989 Effects of feeding regime on food consumption, growth rates and tissue nucleic acids in juvenile Arctic charr, Salvelinus alpinus, with particular respect to compensatory growth. J. Fish Biol. 34:9478957. Mosegaard, H., H. Svedang, and K. Taberman 1988 Uncoupling of somatic and otolith growth rates in Arc- tic char (Salvelinus alpinus) as an effect of differences in tem- perature response. Can. J. Fish. Aquat. Sci. 45:1514-1524. Mugiya and Oka: Biochemical relationship between otolith and somatic growth in Oncorhynchus mykiss 245 Mugiya, Y. 1984 Diurnal rhythm in otolith formation in the rainbow trout, Salmo gairdneri: Seasonal reversal of the rhythm in relation to plasma calcium concentrations. Comp. Biochem. Physiol. 78A:289-293. 1986 Effects of calmodulin inhibitors and other metabolic modulators on in vitro otolith formation in the rainbow trout, Salmo gairdnerii. Comp. Biochem. Physiol. 84A:57-60. 1987 Phase difference between calcification and organic matrix formation in the diurnal growth of otoliths in the rainbow trout, Salmo gairdneri. Fish. Bull., U.S. 85:395-401. 1990 Long-term effects of hypophysectomy on the growth and calcification of otoliths and scales in the goldfish, Carassius auratus. Zool. Sci. (Tokyo) 7:273-279. Nishimura, A., and J. Yamada 1988 Geographical differences in early growth of walleye pollock Theragra ehaleogramma, estimated by back-calculation of oto- liths daily growth increments. Mar. Biol. (Berl.) 97:459-465. Pannella, G. 1971 Fish otoliths: Daily growth layers and periodical patterns. Science (NY) 173:1124-1127. Reznick, D., E. Lindbeck, and H. Bryga 1989 Slower growth results in larger otoliths: An experimen- tal test with guppies (Poecilia reticulata). Can. J. Fish. Aquat. Sci. 46:108-112. Saitoh. S., and J. Yamada 1989 infrastructure of the saccular epithelium and the otolithic membrane in relation to otolith growth in tilapia, Oreochromis niloticus (Teleostei: Cichlidae). Trans. Am. Microsc. Soc. 108:223-238. Secor, D.H., and J.M. Dean 1989 Somatic growth effects on the otolith-fish size relation- ship in young pond-reared striped bass, Morone saxatilis. Can. J. Fish. Aquat. Sci. 46:113-121. Volk, E.C., R.C. Wissmar, C.A. Simenstad, and D.M. Eggers 1984 Relationship between otolith microstructure and the growth of juvenile chum salmon (Oncorhynchus keta) under dif- ferent prey rations. Can. J. Fish. Aquat. Sci. 41:126-133. Webb, P.W. 1978 Partitioning of energy into metabolism and growth. In Gerking, S.H. (ed.), Ecology of freshwater fish production, p. 184-214. Blackwell Sci. Publ., Oxford. Wilder, LB., and J.G. Stanley 1983 RNA-DNA ratio as an index to growth in salmonid fishes in the laboratory and in streams contaminated by carbaryl. J. Fish Biol. 22:165-172. 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. Abstract. - Western Atlantic tonguefishes of the Sympkurus plagu- sia (Schneider, in Bloch and Schnei- der 1801) complex are distinguished from other Atlantic Sympkurus spe- cies by the possession of 12 caudal fin rays, a 1-4-3 pattern of interdigi- tation of dorsal-fin pterygiophores and neural spines, absence of a pupil- lary operculum, reduced or absent dentition on ocular-side jaws, and an unpigmented peritoneum. Consider- able taxonomic uncertainty has been associated with nominal species of this complex, but the most common practice has been to recognize one widespread species (S. plagusia) with two subspecies ranging from the Ca- ribbean southward to Uruguay, and a second species, S. civitatium Gins- burg 1951, occurring in inshore areas along the southeastern and Gulf of Mexico coasts of the United States and northern Mexico. The validity of S. civitatium is confirmed in this study. Examination of tonguefishes from the Caribbean and southward indicates that specimens previously identified as S. plagusia do not com- prise one species with two allopatric subspecies, but rather four largely sympatric, albeit not necessarily syn- topic, species. Sympkurus plagusia, the first described species in this complex, occurs in inshore habitats ranging from the Caribbean to Rio de Janeiro. Sympkurus tessellatus (Quoy and Gaimard 1824), removed from the synonymy of S. plagusia, occurs in nearshore, estuarine, and neritic waters throughout the Carib- bean southwards to Uruguay. Two new species, S. oeulellus occurring in neritic waters off northern South America (Guyana to northern Brazil), and S. caribbeanus (nearshore habi- tats throughout the Caribbean), are described and figured. A key to the western Atlantic species of this com- plex is provided. Western Atlantic Tonguefishes of the Symphurus plagusia Complex (Cynoglossidae: Pleuronectiformes), with Descriptions of Two New Species* Thomas A. Munroe Systematics Laboratory, National Marine Fisheries Service, NOAA National Museum of Natural History, Washington. DC 20560 Shallow-water symphurine tongue- fishes possessing 12 caudal fin rays, a 1-4-3 pattern of interdigitation of dorsal pterygiophores and neural spines, with reduced or absent den- tition on ocular-side jaws, an unpig- mented peritoneum, and lacking a pupillary operculum comprise the Sym- phurus plagusia (Schneider, in Bloch and Schneider 1801) complex.** Five western Atlantic and several eastern Pacific species of tonguefishes are recognized in this complex. Through- out the western Atlantic, from North Carolina, U.S.A., to Uruguay (Gins- burg 1951, Menezes and Benvegnii 1976, Munroe 1987), these common- ly collected tonguefishes are abun- dant locally in estuarine and near- shore habitats as well as on sandy or muddy substrates on the inner con- tinental shelf (Meek and Hildebrand 1928, Ginsburg 1951, Lowe-McCon- nell 1962, Caldwell 1966, Cervigon 1966, Carvalho et al. 1968, Palacio 1974, Menezes and Benvegnii 1976, Lema and Oliveria 1977, Lema et al. 1980, Munroe 1987). Manuscript accepted 31 December 1990. Fishery Bulletin, U.S. 89:247-287(1991). * Contribution no. 1659 of the Virginia Insti- tute of Marine Science, College of William and Mary. * * There is another western Atlantic tongue- fish, S. playiusa (Linnaeus), completely allo- patric from S. plagusia, which unfortunately has a nearly identical spelling for its specific epithet. It is emphasized that these are com- pletely different and distinctive species that should not be confused because of similar- ities in their names. Nomenclatural uncertainty and questions regarding taxonomic valid- ity have been associated with these western Atlantic tonguefishes since the first description of a species from Jamaica by Browne (1756). Much of the confusion centers on species col- lected in shallow waters of the Carib- bean and coastal seas of Central America and much of South America. At least ten combinations of names have been used for these tropical western Atlantic, shallow-water tonguefishes. Historically (Kaup 1858, Jordan and Evermann 1898, Chabanaud 1949), Atlantic members of this spe- cies complex were long regarded as comprising populations of a single widespread, polytypic species, Sym- phurus plagusia. This nomenclatural arrangement began with Kaup (1858) and has continued to the present (Jor- dan and Goss 1889, Jordan and Ever- mann 1898, Ginsburg 1951, Menezes and Benvegnii 1976, Rosa 1980, Lu- cena and Lucena 1982). Ginsburg considered the tropical western At- lantic members of this complex to represent two allopatric subspecies, and his newly-described S. civitatium with its disjunct northern distribu- tion, perhaps representing a third subspecies of one wide-ranging poly- typic species. However, the most re- cent review of Symphurus of south- ern South America (Menezes and Benvegnii 1976) questioned recog- 247 248 Fishery Bulletin 89(2), 1991 nition of only one species in the tropical western Atlan- tic region. My examination of approximately 1000 specimens of Symphurus possessing 12 caudal fin rays and a 1-4-3 interdigitation pattern, collected in inshore waters from North Carolina, throughout the Gulf of Mexico and Caribbean, to Uruguay, reveals that previous studies failed to recognize the presence of multiple sym- patric species among their material. Neither the hypothesis of multiple populations within a single polytypic species, envisioned especially by Jordan and co-workers, nor Ginsburg's hypothesis of one wide- spread polytypic species comprised of allopatric sub- species, adequately explain the divergent morpho- logical variation observed in the specimens and the sympatric (sometimes syntopic) occurrences of speci- mens with different morphological attributes. Instead, the present study recognizes not one, but rather five, western Atlantic members of the S. plagusia complex, which are somewhat phenetically similar species that differ in morphology and pigmentation. Four of these species have largely sympatric, but not necessarily syn- topic, distributions. The fifth species, S. civitatium Ginsburg 1951, is completely allopatric to the others, occurring along the southeastern and Gulf coasts of the United States and northern Mexico. Three species in this assemblage, S. plagusia, S. tessellatus (Quoy and Gaimard 1824), and S. civitatium, were described previously. Two additional species are described herein. Available taxonomic and ecological information is summarized, differential diagnoses are provided for each species, and a key to identification of the five species is included. Materials and methods Methods for counts and measurements and general terminology follow Munroe and Mahadeva (1989) and Munroe (1990). Meristic data, exclusive of scale counts, were taken from radiographs. ID pattern refers to the pattern of interdigitation of dorsal pterygiophores and neural spines. In species accounts, total ranges for meristic features are presented first, followed by modal counts when data were sufficient. Measurements less than 150 mm were taken to the nearest 0.1 mm with dial calipers or ocular micrometer. Measurements over 150 mm were taken to the nearest mm with a steel ruler. Measurements are expressed either as thousandths of standard length (SL) or thou- sandths of head length (HL). Morphometric abbreviations ABL anal fin length BD body depth CD chin depth CFL caudal fin length DBL dorsal fin length ED eye diameter HL head length HW head width LHL lower head lobe width OPUL width of upper opercular lobe OPLL width of lower opercular lobe PA pelvic to anal fin length PAL preanal length PDL predorsal length PL pelvic fin length POL postorbital length SNL snout length UHL upper head lobe width UJL upper jaw length All descriptions of pigmentation are based on fishes fixed in formalin and stored in ethyl or isopropyl alcohol. Maturity was estimated by macroscopic examination of stages of developing ova and extent of posterior elongation of the ovaries (ovaries of mature females are sometimes conspicuous through the body wall in transmitted light; in immature females and large females, ovaries are best observed by dissection). Since no obvious differences in male testicular size were ap- parent, estimates of maturity were based entirely on females. Immature females were those with non- elongate or only partially elongate ovaries. Mature females had fully elongate ovaries. Gravid females were those individuals with enlarged ovaries filled with large, macroscopically visible ova. When available, depth-of-capture information (con- verted to the nearest meter) was recorded and sum- marized for specimens listed in the "Material exam- ined" sections in each species account. If depth of capture included a range of depths over which the nets were towed, a mean depth for that particular trawl was calculated. Synonomies are selective for S. plagusia and S. tessellatus because of the numerous locality citations; synonomies are presumed to be complete for the other species. Because of their common occurrence, of Symphurus plagusia. Abbrevia- tions defined in Methods section.) Character Neotype Range Mean SD HL/HW 1.2 1.2-1.3 1.2 0.38 POL 663 630-714 665.8 25.18 SNL 228 205-250 229.1 15.62 UJL 213 200-250 227.6 14.99 ED 79 64-95 81.9 9.57 CD 213 222-374 275.1 40.32 OPLL 272 250-346 296.9 29.10 OPUL 223 169-272 211.9 27.24 UHL 703 695-926 843.1 63.80 LHL 530 427-606 510.3 53.23 lower eye. Ocular-side lower jaw with distinct, fleshy ridge near posterior margin (Fig. 4a). Chin depth 222-374 HL, x 275. Lower eye small (64-95 HL, x 82), spherical; upper eye usually anterior to lower eye; eyes not covered with scales; usually 1-2 small ctenoid scales in narrow interorbital space. Pupillary oper- culum absent. Length of dorsal fin base 950-977 SL, x 967. Dorsal fin origin placed far forward (Fig. 4a), usually at vertical through anterior margin of upper eye or with first and sometimes second fin rays insert- ing anterior to vertical through anterior margin of upper eye; predorsal length 23-50 SL, x 33. Length of anal fin base 758-802 SL, x 786. Scales absent on blind sides of dorsal and anal fin rays. Pelvic fin length 51-73 SL, x 64; longest pelvic fin ray reaching base of first or occasionally second anal fin ray; pelvic to anal fin distance 38-60 SL, x 50. Posterior pelvic fin ray connected to body by delicate membrane terminating immediately anterior to anus or occasionally extending posteriorly almost to origin of anal fin base (membrane torn in many specimens). Caudal fin length moderate (88-111 SL, x 100). Teeth well developed on blind-side jaws. Ocular-side dentary without teeth or with short row of small teeth developed only on anterior one-half to one-third; pre- maxilla on ocular side usually with small, single, mostly incomplete row of slender teeth anterior to vertical equal with anterior nostril. Scales large, ctenoid on both sides of body. Pigmentation Ocular surface usually uniformly light- brown or yellowish, occasionally with 8-14 narrow, faint crossbands. Crossbands not continued onto dorsal and anal fins; mostly complete in anterior trunk region; on remainder of body obvious only as vertical mark- ings at body margin along dorsal and anal fin bases. 258 Fishery Bulletin 89(2). 1991 Table 1 1 Bathymetric distribution (meters) for five species of the western Atlantic Symphurus plagusia complex cent occurrence of individuals. Numbers represent the per- Species N Depth 1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 110 civitatium 216 tessellatus 349 oculellus 57 plagusia 25 caribbeanus 94 2 10 4 80 51 26 26 19 32 34 2 12 17 26 > 24 9 16 5 5 19 5 9 2 0.5 8 40 4 0.5 6 0.3 4 2 Blind side creamy-white. Peritoneum unpigmented. Pigmentation of outer surface of ocular-side opercle usually same as that of body; occasionally with dusky blotch on upper opercular lobe due to pigment on in- ner lining of ocular-side opercle showing through to outer surface. Inner lining of opercle and isthmus on ocular side usually dusky; some specimens with dark- brown pigmentation on inner opercular lining; inner opercle and isthmus on blind side usually unpigmented or occasionally with small patch of pepper-dot pigmen- tation on ventral margin. Usually with slight pigment band on ocular-side upper lip and diffuse pattern of melanophores on lower lip. Dorsal and anal fins dusky; fin rays streaked with pigment darker brown than that of connecting mem- brane; sometimes with alternating series of darker- pigmented rays (usually 2-3 in succession) separated by about 4-5 successive, lighter-pigmented rays. Basal half (scale-covered) of caudal fin dark-brown; fin rays in distal half streaked with pigment. Size and sexual maturation (Fig. 2) Symphurus pla- gusia is a medium-sized tonguefish. The largest of five males examined was 130mm SL; the largest of 23 females was only slightly smaller (127mm SL). Sexual maturity occurs at a relatively large body size in this species. All females larger than 80 mm SL were mature. All but one female (79.3 mmSL) smaller than 80 mm SL were immature, with gonads undergoing elongation without ripening ova or with ovaries bare- ly elongating. Geographic distribution (Fig. 3) Widely distributed in shallow waters of the tropical western Atlantic. In the northern portion of its distribution, this species occurs in Puerto Rico, Haiti, and Hispaniola, but is unknown from the Bahamas (Bohlke and Chaplin 1968). Along the continental margin of Central America, S. plagusia has been collected at Belize, Nicaragua, Costa Rica, and Panama, while further south it ranges along the Atlantic coast of Colombia and coastal regions in Surinam, Tobago, and Brazil at least as far south as Rio de Janeiro. Bathymetric distribution Symphurus plagusia is a shallow-water species (1-51 m) most commonly in- habiting depths between the shoreline and 10 m (Table 11), where (20/25, 80%) of specimens examined were taken. All life-history stages occur in these shallow areas and only occasionally were individuals taken at deeper locations (one specimen at 51m, three speci- mens at 40 m, and one specimen at 37 m). Ecology Little is known concerning the biology of S. plagusia. Its general rarity in collections indicates that it occurs in rarely-sampled habitats. Remarks The earliest description of a western Atlan- tic, shallow-water, 12-caudal-rayed tonguefish is of a specimen collected in Jamaica that Browne first described (1756) as Plagusia and later (1789) as Pleu- ronectes plagusia. He described this specimen as a small sinistral flatfish with dorsal, anal, and caudal fins united (tail ending in sharp point), lacking pectoral fins and lateral lines. His description was clearly that of a tonguefish, but he provided no figure or diagnostic characters to unequivocally identify his specimen. Browne's names were later suppressed under the plenary powers for nomenclatorial purposes in Opinion 89 of the Commission for Zoological Nomenclature (see Hemming and Noakes 1958). In 1801, Schneider first made Browne's tonguefish Pleuronectes plagusia available as a binomial. Schnei- der's Pleuronectes plagusia was based entirely on the description of the tonguefish from Jamaica in Browne's works (1756, 1789). The description by Schneider (in Bloch and Schneider 1801) is identical to that provided by Browne and, in addition, all indications are that Schneider did not directly examine any specimens of this species. Dr. H.-J. Paepke (Mus. fur Naturkunde der Humboldt-Universitat zu Berlin, Zoologisches Mus., Invalidenstrasse 43, Berlin DDR 1040, pers. Munroe: Western Atlantic tonguefishes of the Symphurus plagusia complex 259 commun. 8 Nov. 1986) informs me that no remarks were made in Bloch's ledger to indicate that specimens were available for examination when Schneider wrote the description of Pleuronectes plagusia. Additional- ly, Paepke also stated that there are no specimens of this species in the Bloch and Schneider collection. Therefore, it appears that the description of Pleuro- nectes plagusia Schneider, in Bloch and Schneider 1801, was copied directly from Browne's work and that no type exists for this species. Although quite vague, the original description of Pleuronectes plagusia by Schneider does refer to a species of Symphurus and is the oldest available name for a tropical, western Atlantic species in the genus. This name represents the oldest binomial generally con- sidered to represent a member of this species group and has been the one name most consistently applied to any shallow-water tonguefish possessing 12 caudal fin rays. In order to stabilize the nomenclature for this species, it is necessary to designate a neotype. Since the original description is based on a specimen from Jamaica, a topotypic specimen would be the most ap- propriate neotype. Unfortunately, no specimens of S. plagusia from Jamaica were available to Munroe (1987) and several more recent attempts to procure a speci- men during the present study have also been unsuc- cessful. All tonguefishes collected from Jamaican waters that I have examined are specimens of S. tessellatus trawled at depths generally exceeding those usually occupied by S. plagusia. Therefore, designation of a neotype for S. plagusia, based on a topotype specimen from Jamaica, is not possible. Instead, ANSP 132030, a mature female measuring 103.2mm SL, col- lected by beach seine at Puerto Yabucoa, Puerto Rico, 24-27 July 1974, is selected as the neotype for this species. Meristic features for this specimen are: ID pattern 1-4-3; caudal fin rays 12; dorsal fin rays 93; anal fin rays 78; total vertebrae 50; longitudinal scales 79; transverse scale count 39; and 18 scale rows on head posterior to eyes. Many authors have included Achirus ornatus La- cepede 1802 in the synonymy of Symphurus plagusia. The description of this species is very brief and does not include figures or locality data, and it is unknown if any type(s) exists. The information provided is that the fish was donated to France by Holland, and has the following characteristics: dorsal and anal fins joined, 95 dorsal fin rays, 82 anal fin rays, 8 or 9 dark trans- verse bands, and a lateral line on each side. Notably absent in Lacepede 's account is the caudal-fin-ray count for this specimen. The lateral line referred to in the description may refer to the mid-lateral junction of the myomeres that is apparent on some tonguefish speci- mens (especially those partially dehydrated during preservation). Based on counts listed by Lacepede, it is possible his specimen is a S. plagusia (sensu strictu). However, the dark, transverse bands and meristic features listed in the description of Achirus ornatus could also apply to several other western Atlantic tonguefishes. Among shallow-water species possessing 12 caudal fin rays, the data fit at least three species: S. caribbeanus (described below), S. plagusia (Schnei- der, in Bloch and Schneider 1801), and S. tessellatus (Quoy and Gaimard 1824). Of these, the description is more typical of S. tessellatus, especially the reference to darkly-pigmented crossbands. Nonetheless, the exact identity of Achirus ornatus Lacepede cannot be determined from the vague original description, par- ticularly given the unknown site of capture for the specimen on which this name is based. Achirus ornatus Lacepede 1802 is therefore regarded as a nomen dubium. In 1824, Quoy and Gaimard described Plagusia tes- sellata from Rio de Janeiro Bay ( = Guanabara Bay), Brazil. Although no figure of this specimen was pro- vided, the descriptive account of meristic features, color pattern, and other characters leave little doubt as to the identity of the species. Quoy and Gaimard described the dorsal fin as originating above the eyes and consisting of 99 rays; the anal fin has 78 rays. The color is described as brown with small transverse bands of the same color. Although no type exists for this species (M.L. Bauchot, Ichtyologie Generale et Appliquee, 43 Cuvier, Mus. Natl. Hist. Nat., Paris Cedex 05 75231, pers. commun. 23 June 1982), the original description is sufficient to allow identifica- tion of this species. Unfortunately, most authors begin- ning with Kaup (1858) and continuing to Ginsburg (1951) regarded this species as a junior synonym of S. plagusia (Schneider, in Bloch and Schneider 1801). It is unlikely that the specimen described by Quoy and Gaimard belongs to S. plagusia (sensu strictu), because the S. tessellatus specimen has slightly higher meristic features, darker banding, and the dorsal fin origin is described as being above and not anterior to the eyes, which is the typical condition found in S. plagusia. A second nominal species of tonguefish from Brazil- ian waters, Plagusia brasiliensis, described by Agassiz (in Spix and Agassiz 1829-1831), has also been placed in the synonymy of S. plagusia. The possible holotype or syntype (MHNN 691; see Kottelat 1984, 1988) was illustrated and an adequate description provided. The specimen has 99 dorsal fin rays, 83 anal fin rays, 12 caudal fin rays, 53 total vertebrae, several small ctenoid scales on the blind sides of the dorsal and anal fin rays, the dorsal fin origin at the vertical through the front margin of the pupil of the upper eye, and a relatively large eye (10.6% HL). It agrees in all these features with S. tessellatus and is removed from the synonymy 260 Fishery Bulletin 89(2), 1991 of S. plagusia and placed in the synonymy of S. tessel- latus (see below). Beginning with Kaup (1858), all previously described species of western Atlantic, shallow-water tonguefishes possessing 12 caudal fin rays were regarded as a single species. Kaup placed Achirus ornatus Lacepede, Pla- gusia brasiliensis Agassiz, in Spix and Agassiz (Kaup cited authorship of this species as Cuvier, in Spix), and Plagusia tessellatus Quoy and Gaimard (Kaup listed Valenciennes as the author of this name) in synonymy and proposed the new combination Aphoristia ornata to accommodate a single, widespread species ranging from the Caribbean to southern South America. Gun- ther (1862) regarded Aphoristia ornata as including the nominal species Pleuronectes plagusia Browne, Plagu- sia brasiliensis Agassiz, and Plagusia tessellatus Quoy and Gaimard. Subsequent authors, including Jordan and Goss (1889) and Jordan and Evermann (1898), until Ginsburg (1951), continued to include three species in the syn- onymy of Symphurus ( = Aphoristia) plagusia (Schnei- der, in Bloch and Schneider): P. tessellatus, P. brasili- ensis, and A. ornatus. Jordan and his co-workers, and other researchers, still recognized only one widespread, polytypic species of shallow-water, 12-caudal-rayed Symphurus occurring in the western Atlantic. Gins- burg (1951), although continuing to regard all Carib- bean and South American specimens as representing a single widespread, polytypic species, S. plagusia (Schneider, in Bloch and Schneider 1801), allocated his study specimens to two allopatric subspecies. He con- sidered S. plagusia plagusia as a northern subspecies ranging from the West Indies to Central America that was characterized by somewhat lower meristic fea- tures. The second subspecies, S. p. tessellatus, with a more southern distribution along the coasts of Brazil and Uruguay had higher meristic features. In this revi- sion, Ginsburg also described a second species (S. civi- tatum = civitatium, this study) of shallow-water, 12- caudal-rayed tonguefish from continental seas off the southeastern United States and Gulf of Mexico. He equivocated in his description of this new species, stating that his S. civitatium could also be recognized as a third northern subspecies of a widespread, poly- typic S. plagusia. Subsequent workers have utilized subspecies desig- nations proposed by Ginsburg for Caribbean and South American shallow-water, 12-caudal-rayed tonguefishes and have used the name S. civitatium for specimens collected in the Gulf of Mexico and along the south- eastern coast of the United States. More recently, however, several studies have noted that both nominal subspecies of S. plagusia occur sympatrically in north- ern South America. For example, Carvalho et al. (1968) found both subspecies in northern Brazil, and Palacio (1974) reported both subspecies from the Colombian Caribbean. In their revision of western South Atlantic tongue- fishes, Menezes and Benvegnu (1976) reported that all their specimens (collected mostly in offshore habitats by trawling) were quite similar, lacking variation re- ported for specimens collected in more northern regions. Using the name S. plagusia, they considered their specimens to represent a single taxon but also pointed out that the sympatric occurrence of both subspecies in other South American localities indicated that the subspecific status designated by Ginsburg should be reexamined. In examining material of S. plagusia, I successfully located 19 of the 25 specimens listed by Ginsburg (1951) in his account of 5. plagusia plagusia. These include representatives of four species: twelve S. tessellatus, one S. caribbeanus, one S. parvus Ginsburg 1951, with only five actually S. plagusia (sensu strictu). The twelve specimens of S. tessellatus incorrectly identified asS. p. plagusia by Ginsburg are small juveniles collected from shallow-water habitats. These, as well as many of the remaining 25 specimens that Ginsburg included in his account of S. p. plagusia, had been collected in the latter part of the last century and during the early 1920s. Most of these older specimens were completely devoid of any obvious pigmentation pattern. As a con- sequence, the specimens provided little clue that more than a single species was represented in these shallow- water collections. Additionally, since most of Gins- burg's Caribbean and Central American specimens came from shallow-water collections, very few adult S. tessellatus were available to his study. Therefore, he was unable to unravel size-related differences among the three sympatric species in this complex that occur in this region (the two smaller species, S. plagusia and S. caribbeanus, and the much larger S. tessellatus). Ginsburg did not list catalog numbers for 34 speci- mens identified as S. p. tessellatus in his study, so that it is difficult to ascertain if more than one species was included in his account of this subspecies. Of the eight lots designated as 5. p. tessellatus by Ginsburg that I have examined, all are one species, S. tessellatus (Quoy and Gaimard 1824). It is highly probable, therefore, that Ginsburg's S. p. tessellatus are equivalent to S. tessellatus (Quoy and Gaimard 1824) in the present study. Comparisons Symphurus plagusia most closely resembles S. civitatium but differs from that species in its modally higher meristic features (total vertebrae modally 49-50 vs. 47-49 in S. civitatium; dorsal fin rays 89-97 vs. 86-93; anal fin rays 73-81 vs. 70-78); and degree of development of sexually dimorphic color- ation. In S. plagusia, both sexes are more or less Munroe: Western Atlantic tonguefishes of the Symphurus plagusia complex 261 uniformly pigmented with only slight evidence of band- ing on the body, and with vertical fins of both sexes uniformly colored with no darkening in the posterior portion of the body. In contrast, in S. civitatium there is considerably more pronounced sexual dimorphism in pigmentation. Females tend to have well-developed crossbands on the body whereas in males the cross- bands are less conspicuous. In male S. civitatium, posterior portions of the dorsal and anal fins are noticeably darkened with black pigment (black pigment absent in females). Symphurus plagusia of all sizes are usually collected with juveniles and small adults of S. tessellatus. Despite overall similarities in meristic features, the two species are quite distinctive. Symphurus plagusia is uniform- ly colored with only faint, narrow crossbands in some individuals, has a well-developed fleshy ridge on the ocular-side lower jaw (Fig. 4a), and this species lacks a striking black pigment spot on the outer opercle (some individuals have a diffuse blotch on the inner opercle where the pigmentation on the inner surface of the ocular-side opercle shows through). In S. tessel- latus, in contrast, all individuals have a bold pattern of wide crossbands, a prominent black spot on the outer surface of the opercle, and lack a fleshy ridge on the ocular-side lower jaw (Fig. 4c). Symphurus plagusia also has a smaller eye (6.4-9.5, * 8.2% HL vs. 7.9- 11.4, x 9.5% HL in S. tessellatus) and lacks the small ctenoid scales on the posterior fin rays on the blind side of the dorsal and anal fins that are present in S. tessel- latus larger than about 70 mm SL. Symphurus plagusia also has modally lower meristic values (total vertebrae 49-50 vs. 50-53 in 5. tessellatus; dorsal fin rays 89-97 vs. 91-102 (usually 93-101); anal fin rays 73-81 vs. 77-86 in S. tessellatus). Symphurus plagusia differs further from S. tessella- tus in the almost squarish snout of S. plagusia, which contrasts with the more pointed snout of 5. tessellatus (compare Figs. 4a and 4c). Also, in S. plagusia the dorsal fin origin is usually anterior to the vertical through the eye, while in 60 mm SL) S. plagiusa, there are 4-8 ctenoid scales on the blind sides of the dorsal and anal fin rays (scales usually absent altogether, or occasionally 1-2 scales along bases of fin rays in S. plagusia). Symphurus plagusia is not easily confused with other Atlantic species (S. kyaropterygium Menezes and Ben- vegnu, S. trewavasae Chabanaud, 5. normani Chaba- naud, S. piger (Goode and Bean), S. nigrescens Rafi- nesque, S. pusillus (Goode and Bean), S. lubbocki Munroe, and S. reticulatus Munroe) with which it over- laps in meristic features. Symphurus plagusia differs from all of these in its 1-4-3 ID pattern (versus 1-4-2 in S. kyaropterygium; 1-3-3 in S. trewavasae and S. nor- mani; 1-3-2 in S. piger, S. nigrescens, S. pusillus, S. reticulatus, and S. lubbocki). In addition, S. plagusia differs from the South Atlantic S. kyaropterygium in caudal-fin-ray number (12 vs. 10) and in lacking the pupillary operculum and dark pigment blotch on the caudal extremity that are present in S. kyaroptery- gium. It differs from the South Atlantic S. trewavasae in its caudal-fin-ray count (12 vs. 10) and smaller eye (64-95 HL vs. 114-162 HL in S. trewavasae). Symphu- rus plagusia lacks scales on the blind sides of the dor- sal and anal fin rays and the spotted peritoneum that are present in S. normani. Symphurus plagusia dif- fers from the 1-3-2 species (except S. lubbocki and S. reticulatus) in lacking a black peritoneum. Symphurus plagusia differs from S. lubbocki and S. reticulatus in having no dentition or greatly reduced dentition on ocular-side jaws (versus ocular-side jaws with complete or nearly complete row of teeth in S. lubbocki and S. reticulatus), its much larger size (130 vs. <50mm SL), and pigmentation (dark- or light-brown, usually without crossbands, and with uniformly-pigmented fins, versus light-brown or yellowish body with incomplete cross- bands in S. lubbocki, and dark, chocolate-brown body with X- and Y-shaped markings and vertical fins with alternating series of blotches and unpigmented areas in S. reticulatus). Meristic values of 5. plagusia, overlap those of six eastern Pacific species possessing either a 1-4-3 or 1-5-3 ID pattern, including S. leei Jordan and Bollman, S. atricaudus (Jordan and Gilbert), S. melanurus Clark, 5. williamsi Jordan and Culver, S. fasciolaris Gilbert, and S. melasmatotheca Munroe and Nizinski. Of these species, S. plagusia is most similar to S. melanurus in that both species possess a fleshy ridge on the ocular- side lower jaw, and in both the first dorsal fin ray reaches a vertical equal with or anterior to the anterior margin of the upper eye. The two species are distin- guished in that S. plagusia lacks a pupillary operculum (versus a weakly-developed pupillary operculum usually present in S. melanurus), has fewer scales in a longi- tudinal series (79-89 vs. 89-108 in S. melanurus), has a lightly-pigmented inner lining on the blind-side oper- cle (versus darkly-pigmented inner lining on the blind- side opercle in S. melanurus), and in S. plagusia the posterior dorsal and anal fins and the caudal fin are not darker than the anterior regions (versus progres- sive darkening in posterior dorsal and anal fins and darkly-pigmented caudal fin in S. melanurus). Symphu- rus plagusia differs from the remaining five eastern Pacific species with comparable meristic values in lack- ing a pupillary operculum (present in the others) and in having a fleshy ridge on the ocular-side lower jaw (absent in the others). Symphurus plagusia is further distinguished from S. fasciolaris and S. melasmato- theca in possessing 12 caudal fin rays (versus 10 and 1 1 in S. fasciolaris and S. melasmatotheca, respectively) and in lacking an ocellated spot on the caudal fin (present in S. fasciolaris) or pigmented peritoneum (present in S. melasmatotheca). Symphurus plagusia differs further from S. atricaudus and S. williamsi in lacking small ctenoid scales on the blind sides of the dorsal and anal fin rays (present in these other species). Munroe: Western Atlantic tonguefishes of the Symphurus plagusia complex 263 From S. leei, S. plagusia is further distinguished in hav- ing the head length considerably smaller than the head width (versus head length usually equal to or greater than head width in S. leei), in having a smaller eye (12-18 SL vs. 22-27 SL in S. leei), and in having an unpigmented peritoneum (versus black or heavily spotted in S. leei). Material examined 45 specimens (19.5-130mm SL). Counted and measured (15 specimens, 57.4-130. 3mm SL) Puerto Rico: ANSP 132030; Neotype; (102.9); Puerto Yabucoa, one- half mile east of Playa de Guayanes, Municipio de Yabucoa; collected by J.J. Loos, 24-27 Jul 1973. FMNH 3286; (83.1); Mayaguez; 20 Jan 1899. FMNH 61572; (117.0); Allasco Bay; 10 Jan 1954. UF 12059; (127.1); beach at Mani, just N of Mayaguez; 16 Apr 1964. Costa Rica: UF 10762; (79.6); Tortuguero Lagoon, Limon Province; Aug 1963. Panama: GCRL 15694; (57.4); Canal Zone; 8 Feb 1977. Trinidad: UPRM 1828; (89.4); Icacos Bay; 4 May 1964. Brazil: FMNH 88853; 2(120.0-130.3); 2°09'S, 42°44'W; 40m; 10 Mar 1963. UFPB 884; (101.4); Praiade Jacare; 13 Nov 1981. UFPB 896; 3(79.3-87.5); Rio Paraiba do Norte; 30 Jul 1981. ANSP 121326; 2(112.9-118.3); Atafona (23°02'S, 44°01'W); Jul-Aug 1963. Counted (27 specimens, 17 lots) Puerto Rico: ANSP 118542; (47.4); Puerto Yabucoa; 25 Jan 1971. ANSP 129952; (112.7); Puerto Yabucoa; 21 Jul 1973. ANSP 129985; (98.2); Puerto Yabucoa; 25 Jul 1973. USNM 50178; 4(61.9-83.1); Ponce; 31 Jan 1899. Haiti: UF 33896; 5(82.8-90.8); 2 km NW of Port Salut; sandy beach near eelgrass bed; 7 Apr 1979. Belize: FMNH 97492; (49.3); Belizean Beach, 4.5 mi on Western Highway; 16 Apr 1973. FMNH 97493; 2(52.1-68.2); Belizean Beach, 4.5 mi on Western Highway; 16 Apr 1973. FMNH 97494; (25.1); Belize City, St. John's College Beach; 16 Apr 1973. FMNH 97495; (36.4); Belize City, St. John's College Beach, mangroves and beach; 3 Aug 1973. Honduras: FMNH 94818; (54.3); Brus Lagoon; 1 m; 10 May 1975. FMNH 94822; (34.8); Roatan; 1 m; 1 May 1975. FMNH 97490; 3(19.5-61.3); Stann Creek District, along Pelican Beach, 17-33 m N of Pelican Beach Motel; 30 Mar 1973. FMNH 97491; (40.0); Stann Creek District, along Pelican Beach, 17-33 m N of Pelican Beach Motel; 15 Apr 1973. Panama: UMML 34347; (121.5); 7°42'N, 57°32'W; 27m; 15 Jul 1968. USNM 81654; (54.9); Colon; 5 Jan 1911. French Guiana: USNM 236252; (110.5); 6°34'N, 54°28'W; 37m; 28 Jun 1972. USNM 291331; (79.4); 5°30'N, 52°10'W; 51m; 12 Sep 1958. Other material examined (3 specimens, 1 lot) Tobago: USNM 313648; 3(19.5-20.3); Bloody Bay, 11°18'14"N, 60°37'46"W; 3m; 13 Sep 1990. Symphurus civitatium Ginsburg 1951 Figures 1 b-c, 2, 4b Synonymy Symphurus piger (not of Goode and Bean). Baughman 1950:137 (inner harbor, Freeport, Texas). Symphurus civitatum Ginsburg 1951:198 (counts, figure, included in key; Gulf of Mexico and south- eastern coasts of the United States; see "Remarks" about emendation of specific name). Springer and Bullis 1956:65 (Gulf of Mexico; list of Oregon sta- tions at which this species was collected). Resen- dez 1979:646 (occurrence in lagunas El Carmen, La Machona, and Lagunade Terminos, northern Mexico). Symphurus civitatus. Briggs 1958:297 (summary of distribution records; North Carolina to Florida and widespread in Gulf of Mexico; see "Remarks" about emendation of specific name). Roithmayr 1965:22 (included in industrial bottomfish catch in north- central Gulf of Mexico). Struhsaker 1969:298 (rare- ly occurring [< 10% of the tows] in demersal fish com- munity of continental shelf from North Carolina to central Florida). Swingle 1971:65 (listed, offshore waters of Alabama). Topp and Hoff 1972:78 (gen- eral absence from west Florida shelf). Miller and Jorgenson 1973:305 (meristic features reported for four specimens). Chittenden and McEachran 1976: 93; 99 (abundance on continental shelf of northwest- ern Gulf of Mexico). Walls 1976:390 (counts, fig- ure, in key; suggested synonymy with S. plagiusa). Schwartz et al. 1981:32 (Cape Fear River, North Carolina). McCaffrey 1981:204 (in part) (abundance and distribution in northeastern Gulf of Mexico). Darcy and Gutherz 1984:104 (collected on west Florida continental shelf). Diagnosis Symphurus civitatium is identified by the combination of: a predominant 1-4-3 ID pattern; 12 caudal fin rays; an unpigmented peritoneum; absence of a pupillary operculum; relatively small eye (70-110 HL, x 88); a fleshy ridge on the ocular-side lower jaw; 86-93 dorsal fin rays; 70-78 anal fin rays; 46-50, usu- ally 47-49 total vertebrae; 66-83, usually 74-82, scales in longitudinal series; relatively short jaws usually extending posteriorly to a vertical line through the middle of the pupil of the lower eye, or sometimes ex- tending to the vertical line through the posterior margin of the pupil of the lower eye; dorsal fin origin usually situated at the vertical anterior to the front margin of the upper eye, or occasionally only reaching the vertical line through the front margin of the pupil of the upper eye; scales usually absent on the blind sides of the dorsal and anal fin rays (occasionally with 1-3 small scales at proximal bases of fin rays but without scales on distal portions of fin rays); ocular surface 264 Fishery Bulletin 89(2), 1991 usually light- to dark-brown, occasionally with 6-14 nar- row, dark-brown crossbands; outer surface of ocular- side opercle without black blotch, pigmentation usual- ly same as that of body (some specimens with dusky blotch on upper opercular lobe as a consequence of pig- ment on inner lining of ocular-side opercle showing through to outer surface); inner lining of ocular-side opercle and isthmus usually heavily pigmented, that of blind side usually unpigmented. Dorsal and anal fins considerably darker posteriorly, without an alternating series of pigmented blotches and unpigmented areas. Description A medium-sized tonguefish reaching maximum lengths of approximately 152 mm SL. The usual ID pattern (Table 1) is 1-4-3 (127/177 specimens), less frequently 1-4-2 (19/177) and 1-5-2 (7/177). Caudal fin rays usually 12 (163/171), infrequently 11 (Table 2). Dorsal fin rays 86-93 (Table 3). Anal fin rays 70-78 (Table 4). Total vertebrae 46-50, usually 47-49 (170/ 174) (Table 5); abdominal vertebrae 3 + 6. Hypurals 4. Longitudinal scale rows 66-83, usually 74-82 (Table 6). Scale rows on head posterior to lower orbit 16-20, usually 17-19 (Table 7). Transverse scales 26-39, usual- ly 31-38 (Table 8). Proportional measurements appear in Tables 12 and 13. Body relatively deep (247-328 SL, x 307) with greatest depth in anterior one-third of body. Preanal length 147-238 SL, x 202. Head moderately short, 170-219 SL, x 191; considerably shorter than body depth. Head wide (212-271 SL, * 238); usually great- er than head length (HW/HL 1.0-1.5, x 1.2); lower head lobe (87-118 SL, x 104) narrower than upper head lobe (139-184 SL, x 159). Lower opercular lobe of ocular side (253-388 HL, x 321) considerably wider than upper opercular lobe (178-329 HL, x 230). Post- orbital length 117-187 SL, x 134. Snout relatively short (169-231 HL, x 206), somewhat square (Fig. 4b); covered with small ctenoid scales. Anterior nostril, when depressed posteriorly, not reaching anterior margin of lower eye. Dermal papillae well developed on blind side of snout and chin regions. Jaws relative- ly short; upper jaw length 181-289 HL, x 228; pos- terior extension of maxilla usually reaching to vertical line through middle of pupil of lower eye or sometimes to vertical through posterior margin of pupil of lower eye. Chin depth 225-331 HL, x 268. Ocular-side lower jaw with distinct, fleshy ridge near posterior margin (Fig. 4b). Lower eye relatively small (70-110 HL, x 88); upper eye slightly anterior to lower eye; eyes not covered with scales; usually only 1-3 small ctenoid scales in narrow interorbital space. Pupillary oper- culum absent. Length of dorsal fin base 925-982 SL, x 963. Dorsal fin origin (Fig. 4b) usually situated at vertical line anterior to front margin of upper eye; oc- casionally dorsal fin origin only reaching vertical line Table 12 Summary of morphometries expressed in thousandths of stan- dard length, except SL (in mm), for the holotype (USNM 155227) and 29 additional specimens oiSymph irus civitatium. (Abbreviations defined in Methods section.) Character Holotype N Range Mean SD SL 110.3 30 48.8-149.3 108.6 24.96 BD 304 30 247-328 306.9 16.42 PDL 43 29 22-46 34.5 6.78 PAL 210 30 147-238 202.4 19.75 DBL 957 30 925-982 963.0 13.22 ABL 787 30 745-891 797.9 31.36 PL 68 30 49-85 63.0 8.83 PA 39 26 33-74 44.7 10.23 CFL 124 29 87-124 108.6 9.32 HL 200 30 170-219 191.3 11.55 HW 240 30 212-271 238.3 14.42 POL 134 30 117-187 134.3 13.64 SNL 41 30 31-47 39.5 3.55 UJL 54 30 37-54 43.5 4.88 ED 16 30 13-21 16.6 2.01 CD 62 29 39-68 50.6 7.32 UHL 159 30 139-184 158.9 11.60 LHL 103 30 87-118 103.9 6.86 through front margin of pupil of upper eye; predorsal length 22-46 SL, i 34. Scales usually absent on blind sides of dorsal and anal fin rays; occasionally with 1-3 small scales at proximal bases of fin rays but without scales on distal portions of fin rays. Pelvic fin length 49-85 SL, x 63; longest pelvic fin ray usually reaching base of first or occasionally second anal fin ray; pelvic to anal fin distance 33-74 SL, x 45. Posteriormost pelvic fin ray connected to body by delicate membrane terminating immediately anterior to anus or occasional- ly extending posteriorly almost to origin of anal fin base (membrane torn in most specimens). Caudal fin rela- tively short, 87-124 SL, x 109. Teeth well developed on blind-side jaws. Dentary on ocular side with single, mostly incomplete row of slender teeth on anterior one-third of jaw. Premaxilla on ocular side with only short row of teeth on anterior one-third; or occasionally lacking teeth altogether. Scales large, ctenoid on both sides of body. Pigmentation Body coloration generally similar for both sexes (dichromatic differences in pigmentation are discussed below). Ocular surface light- to dark-brown; sometimes with dark-brown crossbands continuous across the body. Crossbands, when developed, narrow, 6-14 in number, sometimes sharply contrasting (espe- cially in mature females), otherwise faint and barely perceptible against dark body coloration. Crossbands not continued onto dorsal and anal fins. First band Munroe: Western Atlantic tonguefishes of the Symphurus plagusia complex 265 Table 13 Summary of morphometries expressed in thousandths of head length (except HW/HL) for the holotype (USNM 155227) and 29 additional specimens of Symphurus civitatium. (Abbrevia- tions definec in Methods section.) Character Holotype N Range Mean SD HW/HL 1.2 30 1.0-1.5 1.2 0.10 POL 670 30 645-740 692.0 27.45 SNL 204 30 169-231 206.4 13.84 UJL 272 30 181-289 227.7 24.14 ED 81 30 70-110 87.8 10.26 CD 308 29 225-331 267.9 30.55 OPLL 371 29 253-388 321.3 34.80 OPUL 217 29 178-329 230.0 34.86 UHL 792 30 636-996 832.6 82.14 LHL 516 30 462-629 543.0 46.00 crossing head short distance anterior to opercular open- ing. Crossbands along trunk 3-6 scale rows wide. Two posteriormost bands immediately anterior to caudal fin base often conjoined. Blind side off-white. Peritoneum unpigmented. Dorsal margin of outer surface of ocular- side opercle often with dusky blotch due to dark pig- mentation of inner lining of opercle showing through to outer surface. Inner lining of opercle and isthmus on ocular side usually heavily pigmented; lining of blind- side opercle and blind-side isthmus usually unpig- mented. Band of pigmentation usually developed on ocular-side upper lip; lower lip on ocular side frequently spotted but usually without definite band. Pigmentation of dorsal and anal fins generally similar in both sexes, but usually more intense in males. All dorsal and anal fin rays on anterior two-thirds of body streaked with brown pigment similar in shade and in- tensity to body color. Fin rays completely pigmented other than for extreme distal tips, which are unpig- mented. Membranes of anterior three-fourths of fins unpigmented. Caudal fin and dorsal and anal fins on posterior one-third of body more heavily pigmented and considerably darker than anterior two-thirds of fin. Fin membranes on posterior quarter of body heavily pig- mented. Basal one-third of caudal fin more lightly pigmented than posterior two-thirds of fin. Distal tips of middle caudal fin rays unpigmented, or with tips of middle caudal fin rays streaked with pigment but mem- brane unpigmented. Size and sexual maturity (Fig. 2) The largest fish ex- amined, a female (152 mm SL), was only slightly larger than the largest male (149 mm SL). Most specimens ex- amined ranged in size from 80 to 140 mm SL. Females mature at sizes usually larger than 90 mm SL. Of 86 females examined, only three (83-95 mm SL) were immature. The two smallest gravid females were 80-90 mm SL, whereas the majority of gravid females were usually larger (91-140mm SL). Geographic distribution (Fig. 5) Western North Atlantic from Cape Hatteras, North Carolina, to the Yucatan Peninsula of Mexico. There is a single record for this species from Bermuda (ANSP 137573). In the Gulf of Mexico, S. civitatium occurs most commonly west of Apalachicola Bay in northern Florida (Springer and Bullis 1956, Chittenden and McEachran 1976, McCaffrey 1981). It is one of the most commonly col- lected tonguefishes on inner continental shelf regions in the central and western portions of the Gulf of Mex- ico from Alabama to Texas. Along northern Mexico, this species occurs coastally on sandy substrates at least as far south as coastal lagoons (lagunas El Carmen y La Machona, Laguna de Terminos) in Tabasco and Campeche, Mexico (Resendez 1979) and on the con- tinental shelf of the southern Gulf of Mexico (Cabo Rojo, Veracruz to Sabuncuy, Yucatan). Collection data for 347 specimens from this study reveal a general absence of this species from the western Florida shelf. Only two collections record this species from the Tortugas region off southern Florida. Topp and Hoff (1972) also noted the general absence of this species along the west Florida shelf and found just a single record for S. civitatium in the eastern Gulf of Mexico (St. Joseph Bay; from Ginsburg 1951). Neither their efforts during the Hourglass cruises on the continental shelf off west Florida nor other studies (Moe and Martin 1965, Ogren and Brusher 1977, Naughton and Saloman 1978) have collected this species in coastal and continental shelf habitats off west Florida. Furthermore, Darcy and Gutherz (1984) reported taking only a single specimen during 338 10-minute bottom trawls in 9-193 m on the west Florida shelf. Symphurus civitatium occurs on sand or silt sub- strates throughout its range. The geographic and bathymetric distribution of this species apparently coin- cides with the distribution of terrigenous, quartzite sandy, and silty substrates on the inner continental shelf. The scarcity of this species on the west Florida shelf and Yucatan Peninsula may reflect the striking- ly different substrate compositions there. Along the west Florida shelf, primarily in depths of 55-92 m, Topp and Hoff (1972) reported that substrates consist of lithified sediments of cemented lime, including (1) near- shore deposits of cemented shell beachrock, (2) lime- stone, ranging from soft marl to conglomeritic and foraminiferal limestone, (3) small patches of living and dead coral, and (4) calcareous algae. They noted that substrates off the Yucatan Peninsula are similar in composition to those of the west Florida shelf. In 266 Fishery Bulletin 89(2). 1991 contrast, in the central and western Gulf of Mexico from the Mississippi Delta to Cabo Rojo, Veracruz, where S. civitatium is very abundant, substrates on the inner shelf consist largely of terrigenous quartzite sands, silts, and clays delivered primarily by Mississippi and Rio Grande rivers (van Andel 1960). Substrate preference may also affect the distribution of S. civitatium in coastal seas off the southeastern United States. The depth of occurrence (ll-40m, see below) for S. civitatium apparently coincides with sand- silt substrates on the inner portions of the shelf, and this species is absent from live-bottom habitats occur- ring at similar depths (Struhsaker 1969). The specimen of S. civitatum reported by Lazzaro (1977:69) from the continental shelf off Uruguay is neither this species nor any other of the Symphurus plagusia complex (those possessing 12 caudal fin rays, a 1-4-3 ID pattern, and an unpigmented peritoneum). The body shape evident in the photograph, meristic features, and great depth of occurrence (183 m) indicate the specimen is probably S. ginsburgi Menezes and Benvegnu. Bathymetric distribution Although S. civitatium has been collected over a wide depth range of 1-73 m (Table 11), its center of abundance, based on overall frequen- cy of capture and general abundance, occurs between 11 and 45m. Approximately 91% (199/216) of the specimens examined in the present study were cap- tured at these depths. The deepest captures were at 73 and 62 m, where a single fish was taken each time. It is unusual for adult S. civitatium to occur in shallow, inshore regions. Of four fish collected shallower than 10 m, three were small juveniles (< 35 mm SL). Recent- ly, several small juveniles (22-24 mm SL) have been col- lected in the Cape Fear River estuary, North Carolina (S.W. Ross, Dep. Nat. Resour., Morehead City, NC, pers. commun. 24 July 1985). It is not known how regularly this species occurs in these inshore areas or whether the recent captures represent isolated occur- rences; however, Schwartz et al. (1981) listed this species as rare in the Cape Fear River estuary. Sea- sonal occurrence and abundance of juveniles in near- shore waters need further investigation. McCaffrey (1981) reported capture depths of 30-187 m for 23 specimens purported to be S. civitatium taken on the continental shelf in the northeastern Gulf of Mexico between 84°30' and 89°00'W longitudes. Nine- teen of these specimens were collected between 80 and 187 m, depths considerably deeper than records for specimens I examined. Not all specimens identified as S. civitatium by McCaffrey were preserved and curated in collections (at least nine were indicated as having been discarded). One retained specimen (UF 70946) collected at 45 m, is S. civitatium, but two other specimens (UF 70885) from Tursiops Station 7019-07 with an estimated depth of 187 m are, rather, S. pusillus (Goode and Bean) and an undescribed spe- cies (species C of Munroe 1987), which occur common- ly at depths similar to that reported for this station (187 m). Given the complexity of the series identified as .S. civitatium in McCaffrey's study, the very deep captures reported (80-187 m) for S. civitatium are probably erroneous. Remarks In the original description of S. civitatum, Ginsburg (1951:198) stated that this species and S. plagusia differed enough to consider them distinct species but that it was possible that they represented subspecies of a more widespread polytypic species. It was shown earlier (see "Remarks" section in account of S. plagusia) that Ginsburg had more than one species in his account of S. plagusia, and therefore his sub- specific designations for this species were unfounded. Symphurus civitatium is consequently recognized as a species within this complex of morphologically similar species of Symphurus. The etymology of the name civitatum, applied by Ginsburg to this species, is unclear from the original description. The name may have been derived from the genitive plural of civitas (meaning "of the citizenry"). Following this assumption, the proper genitive plural is civitatium, not civitatum as indicated by Ginsburg (G.C. Steyskal, Dep. Entomology, Natl. Mus. Nat. Hist., Wash. DC, pers. commun. 1989). Thus, the spell- ing of the specific name for this species remains un- changed regardless of generic assignment of the species. Therefore, spelling changes such asS. civitatus (Bailey et al. 1960, and other checklists of common and scientific names) are incorrect. Comparisons Symphurus civitatium is most similar to, but has a completely allopatric distribution from, the Caribbean and South Atlantic species S. plagusia. Differences between these two species were discussed in the "Comparisons" section under the account for S. plagusia. Meristic features of 70mm SL); lacking a fleshy ridge on the ocular-side lower jaw; 91-102 dorsal fin rays; 77-86, usually 78-84, anal fin rays; 49-54, usually 50-53, total vertebrae; 81-96, usually 83-93, scales in a longitudinal series; moderately long jaws usually ex- tending to the vertical line through the middle or posterior margin of pupil of lower eye; moderately- sized eye (79-114 HL, x 95) without pupillary opercu- lum; dorsal fin origin reaching the vertical line through the anterior margin of the upper eye, or occasionally only reaching the vertical through the middle of the upper eye; ocular surface dark- to light-brown, with 5-9 well-developed, sharply contrasting, relatively wide, dark-brown crossbands on head and trunk; inner lining of opercle and isthmus heavily pigmented on both sides of body; dorsal and anal fins without an alternating series of pigmented blotches and unpigmented areas; anterior dorsal and anal fin rays usually streaked with brown pigment; fin rays and membranes of dorsal and anal fins on posterior two-thirds of body becoming pro- gressively darker posteriorly; males with posteriormost regions of fins almost uniformly black, while in females, posterior portions of fins, although darker than an- terior regions, usually dark-brown and not as intensive- ly pigmented as in mature males. Description A large tonguefish, attaining adult sizes to 220mm SL. ID pattern (Table 1) usually 1-4-3 (170/231 specimens), less frequently 1-5-3 (14), 1-4-2 270 Fishery- Bulletin 89(2). 1991 Table 14 Summary of morphometries expressed as thousandths of stan- dard length (except SL in mm) for Symphurus tessellatus {N 22) and the possible holotype (MHNN 691) of Plagusia brasiliensis (a junior subjective synonym). (Abbreviations defined in Methods section.) Character Range Mean SD P. brasiliensis SL 97.9-203 145.0 27.66 140.3 BD 247-312 280.2 18.82 262 PDL 32-48 41.7 4.48 32 PAL 181-227 204.7 10.58 217 DBL 952-968 958.3 4.48 968 ABL 771-876 798.0 22.90 793 PL 44-73 59.0 6.47 43 PA 27-56 41.5 6.01 — CFL 72-118 90.9 10.36 88 HL 170-199 186.6 7.37 175 HW 193-247 218.6 15.58 209 POL 117-135 125.9 5.38 117 SNL 35-46 40.3 2.55 32 UJL 41-52 46.3 3.12 41 ED 15-21 17.6 2.04 18 CD 33-63 46.4 6.51 36 UHL 113-163 143.3 12.03 120 LHL 80-114 97.8 10.56 94 (10), or 1-3-3 (8). Caudal fin rays usually 12 (207/224), less frequently 10,11, or 13 (Table 2). Dorsal fin rays 91-102 (Table 3). Anal fin rays 77-86, usually 78-84 (Table 4). Total vertebrae 49-54, usually 50-53 (228/ 233) (Table 5). Hypurals 4. Longitudinal scale rows 81-96, usually 83-93 (Table 6). Scale rows on head posterior to lower orbit 18-23, usually 20-22 (Table 7). Transverse scales 38-45 (Table 8). Proportional measurements appear in Tables 14 and 15. Body relatively elongate, only moderately deep (247-312 SL, x 280); greatest depth usually occurring in anterior third of body. Preanal length 181-227 SL, x 205. Head relatively short (170-199 SL, x 187); con- siderably shorter than body depth. Head relatively wide (193-247_SL, x 219), wider than head length (HW/HL 1.1-1.4, x 1.2); lower head lobe (80-114 SL, x 98) nar- rower in width than upper head lobe (113-163 SL, x 143). Lower opercular lobe on ocular side (243-359 HL, x 307) greater in width than upper opercular lobe (161-252, x 206). Postorbital length 117-135 SL, x 126. Snout (Fig. 4c) moderately long and somewhat pointed (196-231 HL, x 216); covered with small cte- noid scales. Anterior nostril, when depressed posterior- ly, not reaching anterior margin of lower eye. Dermal papillae well developed on blind side of snout and chin regions, but not particularly dense, occasionally extend- ing onto ocular-side snout. Jaws moderately long; upper-jaw length 222-278 HL, x 248; -posterior exten- Table 15 Summary of morphometries expressed as thousandths of head length (except HL and HW) for Symphurus tessellatus (N 22) and the possible holotype (MHNN 691) of Plagusia brasilien- sis (a junior subjective synonym). (Abbreviations defined in Methods section.) Character Range Mean SD P. brasiliensis HL/HW 1.1-1.4 1.2 0.08 1.2 POL 593-723 674.9 25.07 669 SNL 196-231 215.7 9.25 184 UJL 222-278 248.1 15.58 237 ED 79-114 95.2 10.06 106 CD 173-322 245.0 31.85 204 OPLL 243-359 306.8 31.68 — OPUL 161-252 205.7 24.03 — UHL 682-891 774.2 56.18 690 LHL 422-593 523.1 46.26 539 sion of maxilla usually reaching to the vertical through the middle or posterior margin of pupil of lower eye. Ocular-side lower jaw lacking a fleshy ridge near pos- terior margin (Fig. 4c). Chin depth 173-322 HL, x 245. Lower eye moderate in size (79-114 HL, x 95); upper eye usually slightly anterior to lower eye; eyes not covered with scales; usually 1-3 small ctenoid scales in narrow interorbital space. Pupillary operculum ab- sent. Length of dorsal fin base 952-968 SL, x 958. Dorsal fin origin (Fig. 4c) usually reaching to vertical line through anterior margin of upper eye, or occa- sionally only reaching vertical line through middle of upper eye; predorsal length 32-48 SL, x 42. Length of anal fin base 771-876 SL, x 798. Four to eight scales present on blind sides of dorsal and anal fin rays (best developed on fin rays in posterior third of body of specimens > 70 mm SL). Pelvic fin length 44-73 SL, x 59; longest pelvic fin ray usually reaching base of first anal fin ray or occasionally falling short of that point; pelvic to anal fin distance 27-56 SL, x 42. Pos- teriormost pelvic fin ray connected to body by delicate membrane terminating immediately anterior to anus or occasionally extending posteriorly almost to origin of anal fin base (membrane torn in most specimens). Caudal fin relatively short, 72-118 SL, x 91. Teeth well developed on blind-side jaws. Dentary on ocular side usually with single, mostly incomplete row of slender teeth; premaxilla on ocular side either with very short row of teeth anterior to vertical line equal with anterior nostril or lacking teeth altogether. Scales large, strongly ctenoid on both sides of body. Pigmentation General pattern of body pigmentation similar in both sexes at all sizes but usually more in- tense in sexually mature males. Males, especially those Munroe: Western Atlantic tonguefishes of the Symphurus plagusia complex 271 in breeding condition (collected with gravid females), usually with more intense banding, dark-black fins, dark-black spot on ocular-side opercle, and, additional- ly, some specimens with irregularly-shaped, black pig- ment patches on posterior one-half of blind side of body. In contrast, mature females also with crossbands but less conspicuous than in males and with posterior por- tions of fins dark-brown but usually not black. Females lack black pigment patches on blind side observed in males. Ocular-surface background pigmentation ranging from dark- to light-brown. Body usually with 5-9 (usually 5-7) well-developed, sharply contrasting, rela- tively wide, dark-brown crossbands on head and trunk. First two bands relatively consistent in position; first crossing head immediately posterior to eyes; second crossing body immediately behind opercular opening. Crossbands on trunk variable in number and degree of completeness, especially those between opercular opening and point about equal to two-thirds of trunk length. Males usually with 3-4 well-developed and lesser number of incomplete bands along trunk. Two posteriormost bands, just anterior to caudal fin base, slightly arched and usually darker than others on body. Blind side usually uniformly creamy-white; some mature males with irregular patches of black pig- ment on caudal one-third of blind side. Peritoneum unpigmented. Outer surface of ocular-side opercle usually with distinct, dark-brown or b